14022 lines
458 KiB
JSON
14022 lines
458 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.3086864060199574,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 5.000000000000001e-07,
|
|
"logits/chosen": -2.586496591567993,
|
|
"logits/rejected": -2.58866286277771,
|
|
"logps/chosen": -140.71868896484375,
|
|
"logps/rejected": -141.9235382080078,
|
|
"loss": 0.6941,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.005378031637519598,
|
|
"rewards/margins": -0.0017773156287148595,
|
|
"rewards/rejected": -0.0036007165908813477,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.0000000000000002e-06,
|
|
"logits/chosen": -2.7250607013702393,
|
|
"logits/rejected": -2.737359046936035,
|
|
"logps/chosen": -239.8214111328125,
|
|
"logps/rejected": -245.04498291015625,
|
|
"loss": 0.69,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.0041117193177342415,
|
|
"rewards/margins": 0.006983352825045586,
|
|
"rewards/rejected": -0.002871633041650057,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.5e-06,
|
|
"logits/chosen": -2.407137393951416,
|
|
"logits/rejected": -2.3473005294799805,
|
|
"logps/chosen": -184.03599548339844,
|
|
"logps/rejected": -155.21461486816406,
|
|
"loss": 0.6886,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": 0.003687595948576927,
|
|
"rewards/margins": 0.009552741423249245,
|
|
"rewards/rejected": -0.0058651454746723175,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0000000000000003e-06,
|
|
"logits/chosen": -2.7127842903137207,
|
|
"logits/rejected": -2.777282953262329,
|
|
"logps/chosen": -171.6352081298828,
|
|
"logps/rejected": -191.40785217285156,
|
|
"loss": 0.7021,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.004523229319602251,
|
|
"rewards/margins": -0.01729598268866539,
|
|
"rewards/rejected": 0.012772750109434128,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5e-06,
|
|
"logits/chosen": -2.4465034008026123,
|
|
"logits/rejected": -2.466231107711792,
|
|
"logps/chosen": -254.792724609375,
|
|
"logps/rejected": -245.39108276367188,
|
|
"loss": 0.6988,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.004072976298630238,
|
|
"rewards/margins": -0.010413647629320621,
|
|
"rewards/rejected": 0.006340669468045235,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3e-06,
|
|
"logits/chosen": -2.2196829319000244,
|
|
"logits/rejected": -2.2641539573669434,
|
|
"logps/chosen": -179.39495849609375,
|
|
"logps/rejected": -220.39300537109375,
|
|
"loss": 0.6891,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.01630725897848606,
|
|
"rewards/margins": 0.008536052890121937,
|
|
"rewards/rejected": 0.00777120515704155,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5000000000000004e-06,
|
|
"logits/chosen": -2.3294830322265625,
|
|
"logits/rejected": -2.380467653274536,
|
|
"logps/chosen": -194.7043914794922,
|
|
"logps/rejected": -199.31422424316406,
|
|
"loss": 0.6919,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.005030441097915173,
|
|
"rewards/margins": 0.0030234563164412975,
|
|
"rewards/rejected": 0.0020069843158125877,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.000000000000001e-06,
|
|
"logits/chosen": -2.590843677520752,
|
|
"logits/rejected": -2.5107312202453613,
|
|
"logps/chosen": -215.23556518554688,
|
|
"logps/rejected": -189.38424682617188,
|
|
"loss": 0.6995,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.0025288108736276627,
|
|
"rewards/margins": -0.012505888007581234,
|
|
"rewards/rejected": 0.009977078065276146,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.5e-06,
|
|
"logits/chosen": -2.659663200378418,
|
|
"logits/rejected": -2.6844048500061035,
|
|
"logps/chosen": -229.0231170654297,
|
|
"logps/rejected": -205.54054260253906,
|
|
"loss": 0.7032,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.014649391174316406,
|
|
"rewards/margins": -0.019562961533665657,
|
|
"rewards/rejected": 0.0049135684967041016,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 5e-06,
|
|
"logits/chosen": -2.6927154064178467,
|
|
"logits/rejected": -2.6597423553466797,
|
|
"logps/chosen": -207.30467224121094,
|
|
"logps/rejected": -197.54656982421875,
|
|
"loss": 0.692,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0019803522154688835,
|
|
"rewards/margins": 0.002554560313001275,
|
|
"rewards/rejected": -0.004534911829978228,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 5.500000000000001e-06,
|
|
"logits/chosen": -2.6474952697753906,
|
|
"logits/rejected": -2.6725170612335205,
|
|
"logps/chosen": -231.35108947753906,
|
|
"logps/rejected": -274.9834289550781,
|
|
"loss": 0.6924,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.0055945878848433495,
|
|
"rewards/margins": 0.0016048436518758535,
|
|
"rewards/rejected": 0.003989744000136852,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 6e-06,
|
|
"logits/chosen": -2.2198939323425293,
|
|
"logits/rejected": -2.2628731727600098,
|
|
"logps/chosen": -198.29022216796875,
|
|
"logps/rejected": -170.82257080078125,
|
|
"loss": 0.688,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0031775482930243015,
|
|
"rewards/margins": 0.010572671890258789,
|
|
"rewards/rejected": -0.013750219717621803,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 6.5000000000000004e-06,
|
|
"logits/chosen": -2.7613415718078613,
|
|
"logits/rejected": -2.8206562995910645,
|
|
"logps/chosen": -239.75253295898438,
|
|
"logps/rejected": -238.97320556640625,
|
|
"loss": 0.6999,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.0029205563478171825,
|
|
"rewards/margins": -0.013042164966464043,
|
|
"rewards/rejected": 0.010121609084308147,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 7.000000000000001e-06,
|
|
"logits/chosen": -2.0942578315734863,
|
|
"logits/rejected": -2.017871856689453,
|
|
"logps/chosen": -184.2721710205078,
|
|
"logps/rejected": -175.73641967773438,
|
|
"loss": 0.6872,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.004038786515593529,
|
|
"rewards/margins": 0.01193075068295002,
|
|
"rewards/rejected": -0.015969539061188698,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 7.5e-06,
|
|
"logits/chosen": -2.526791572570801,
|
|
"logits/rejected": -2.5704824924468994,
|
|
"logps/chosen": -200.25115966796875,
|
|
"logps/rejected": -223.4153289794922,
|
|
"loss": 0.6917,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": 0.00700113782659173,
|
|
"rewards/margins": 0.003126000752672553,
|
|
"rewards/rejected": 0.003875136375427246,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.000000000000001e-06,
|
|
"logits/chosen": -2.5145041942596436,
|
|
"logits/rejected": -2.478262424468994,
|
|
"logps/chosen": -169.4720001220703,
|
|
"logps/rejected": -179.3622283935547,
|
|
"loss": 0.687,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.00018756365170702338,
|
|
"rewards/margins": 0.012972594238817692,
|
|
"rewards/rejected": -0.013160157017409801,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.500000000000002e-06,
|
|
"logits/chosen": -2.1660895347595215,
|
|
"logits/rejected": -2.1799678802490234,
|
|
"logps/chosen": -158.1959686279297,
|
|
"logps/rejected": -225.02365112304688,
|
|
"loss": 0.687,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": 0.01399233564734459,
|
|
"rewards/margins": 0.01266777515411377,
|
|
"rewards/rejected": 0.001324558281339705,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9e-06,
|
|
"logits/chosen": -2.417539596557617,
|
|
"logits/rejected": -2.4626646041870117,
|
|
"logps/chosen": -170.4958038330078,
|
|
"logps/rejected": -178.34300231933594,
|
|
"loss": 0.6877,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": 0.0049651628360152245,
|
|
"rewards/margins": 0.011217641644179821,
|
|
"rewards/rejected": -0.006252479739487171,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.5e-06,
|
|
"logits/chosen": -2.6470260620117188,
|
|
"logits/rejected": -2.7213802337646484,
|
|
"logps/chosen": -179.53497314453125,
|
|
"logps/rejected": -200.94956970214844,
|
|
"loss": 0.69,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.0027004480361938477,
|
|
"rewards/margins": 0.006518864538520575,
|
|
"rewards/rejected": -0.00921931304037571,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1e-05,
|
|
"logits/chosen": -2.4962029457092285,
|
|
"logits/rejected": -2.5313973426818848,
|
|
"logps/chosen": -208.752685546875,
|
|
"logps/rejected": -244.1136474609375,
|
|
"loss": 0.683,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": 0.009996438398957253,
|
|
"rewards/margins": 0.020868420600891113,
|
|
"rewards/rejected": -0.01087198406457901,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.05e-05,
|
|
"logits/chosen": -2.531207799911499,
|
|
"logits/rejected": -2.5359435081481934,
|
|
"logps/chosen": -194.47694396972656,
|
|
"logps/rejected": -185.80142211914062,
|
|
"loss": 0.6895,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.002445125486701727,
|
|
"rewards/margins": 0.0077151767909526825,
|
|
"rewards/rejected": -0.005270051304250956,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.1000000000000001e-05,
|
|
"logits/chosen": -2.430760622024536,
|
|
"logits/rejected": -2.395613431930542,
|
|
"logps/chosen": -144.2230987548828,
|
|
"logps/rejected": -148.8902587890625,
|
|
"loss": 0.6942,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.00968785211443901,
|
|
"rewards/margins": -0.001990032149478793,
|
|
"rewards/rejected": -0.007697821129113436,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.1500000000000002e-05,
|
|
"logits/chosen": -2.4417223930358887,
|
|
"logits/rejected": -2.441425323486328,
|
|
"logps/chosen": -165.82875061035156,
|
|
"logps/rejected": -178.82815551757812,
|
|
"loss": 0.6886,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": 0.0060547590255737305,
|
|
"rewards/margins": 0.009464193135499954,
|
|
"rewards/rejected": -0.0034094336442649364,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.2e-05,
|
|
"logits/chosen": -2.3494763374328613,
|
|
"logits/rejected": -2.40555477142334,
|
|
"logps/chosen": -133.24130249023438,
|
|
"logps/rejected": -151.85446166992188,
|
|
"loss": 0.6986,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.011995697394013405,
|
|
"rewards/margins": -0.010638857260346413,
|
|
"rewards/rejected": -0.0013568403664976358,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.25e-05,
|
|
"logits/chosen": -2.3450560569763184,
|
|
"logits/rejected": -2.5124807357788086,
|
|
"logps/chosen": -242.19821166992188,
|
|
"logps/rejected": -259.1798095703125,
|
|
"loss": 0.6912,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.013580609112977982,
|
|
"rewards/margins": 0.004313135519623756,
|
|
"rewards/rejected": -0.017893744632601738,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1.3000000000000001e-05,
|
|
"logits/chosen": -2.5465288162231445,
|
|
"logits/rejected": -2.5980730056762695,
|
|
"logps/chosen": -202.898193359375,
|
|
"logps/rejected": -248.03968811035156,
|
|
"loss": 0.6854,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": 0.0018544672057032585,
|
|
"rewards/margins": 0.016238166019320488,
|
|
"rewards/rejected": -0.01438369695097208,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.3500000000000001e-05,
|
|
"logits/chosen": -2.733663320541382,
|
|
"logits/rejected": -2.712355375289917,
|
|
"logps/chosen": -205.80429077148438,
|
|
"logps/rejected": -174.95797729492188,
|
|
"loss": 0.6933,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -8.790497668087482e-05,
|
|
"rewards/margins": 2.1266518160700798e-05,
|
|
"rewards/rejected": -0.0001091718440875411,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.4000000000000001e-05,
|
|
"logits/chosen": -2.550997257232666,
|
|
"logits/rejected": -2.4390013217926025,
|
|
"logps/chosen": -171.59786987304688,
|
|
"logps/rejected": -176.4530029296875,
|
|
"loss": 0.6931,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.02987699769437313,
|
|
"rewards/margins": 0.0004779808223247528,
|
|
"rewards/rejected": -0.030354974791407585,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.45e-05,
|
|
"logits/chosen": -2.4879837036132812,
|
|
"logits/rejected": -2.480923652648926,
|
|
"logps/chosen": -149.10079956054688,
|
|
"logps/rejected": -162.999267578125,
|
|
"loss": 0.6834,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.005112767685204744,
|
|
"rewards/margins": 0.01993861421942711,
|
|
"rewards/rejected": -0.025051379576325417,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.5e-05,
|
|
"logits/chosen": -2.3427562713623047,
|
|
"logits/rejected": -2.5123889446258545,
|
|
"logps/chosen": -160.62950134277344,
|
|
"logps/rejected": -201.14747619628906,
|
|
"loss": 0.6952,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.015586448833346367,
|
|
"rewards/margins": -0.003298282390460372,
|
|
"rewards/rejected": -0.012288165278732777,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.55e-05,
|
|
"logits/chosen": -2.2548415660858154,
|
|
"logits/rejected": -2.3827996253967285,
|
|
"logps/chosen": -126.6205825805664,
|
|
"logps/rejected": -145.7274627685547,
|
|
"loss": 0.6883,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.02384941652417183,
|
|
"rewards/margins": 0.010512137785553932,
|
|
"rewards/rejected": -0.03436155617237091,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.6000000000000003e-05,
|
|
"logits/chosen": -2.5090818405151367,
|
|
"logits/rejected": -2.560624122619629,
|
|
"logps/chosen": -182.53944396972656,
|
|
"logps/rejected": -208.08763122558594,
|
|
"loss": 0.6877,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.0014843230601400137,
|
|
"rewards/margins": 0.011496281251311302,
|
|
"rewards/rejected": -0.01298060454428196,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.65e-05,
|
|
"logits/chosen": -2.800107479095459,
|
|
"logits/rejected": -2.7355594635009766,
|
|
"logps/chosen": -207.99392700195312,
|
|
"logps/rejected": -208.31610107421875,
|
|
"loss": 0.6927,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.014949416741728783,
|
|
"rewards/margins": 0.0012244456447660923,
|
|
"rewards/rejected": -0.016173863783478737,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"logits/chosen": -2.6325690746307373,
|
|
"logits/rejected": -2.630256175994873,
|
|
"logps/chosen": -196.8790283203125,
|
|
"logps/rejected": -202.53794860839844,
|
|
"loss": 0.7116,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.041500113904476166,
|
|
"rewards/margins": -0.03608906269073486,
|
|
"rewards/rejected": -0.005411052145063877,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1.75e-05,
|
|
"logits/chosen": -2.476334571838379,
|
|
"logits/rejected": -2.473987579345703,
|
|
"logps/chosen": -256.1812438964844,
|
|
"logps/rejected": -250.76165771484375,
|
|
"loss": 0.6882,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.01104288175702095,
|
|
"rewards/margins": 0.010350894182920456,
|
|
"rewards/rejected": -0.021393775939941406,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1.8e-05,
|
|
"logits/chosen": -2.342329502105713,
|
|
"logits/rejected": -2.3805429935455322,
|
|
"logps/chosen": -164.03692626953125,
|
|
"logps/rejected": -217.39849853515625,
|
|
"loss": 0.688,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.009070659056305885,
|
|
"rewards/margins": 0.010929775424301624,
|
|
"rewards/rejected": -0.020000435411930084,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1.85e-05,
|
|
"logits/chosen": -2.4253318309783936,
|
|
"logits/rejected": -2.4550042152404785,
|
|
"logps/chosen": -146.70420837402344,
|
|
"logps/rejected": -129.80674743652344,
|
|
"loss": 0.6845,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.031095195561647415,
|
|
"rewards/margins": 0.017679547891020775,
|
|
"rewards/rejected": -0.04877474159002304,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1.9e-05,
|
|
"logits/chosen": -2.4936635494232178,
|
|
"logits/rejected": -2.5476386547088623,
|
|
"logps/chosen": -164.6866455078125,
|
|
"logps/rejected": -198.52139282226562,
|
|
"loss": 0.6895,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.025130080059170723,
|
|
"rewards/margins": 0.007643342949450016,
|
|
"rewards/rejected": -0.03277342766523361,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1.9500000000000003e-05,
|
|
"logits/chosen": -2.4614553451538086,
|
|
"logits/rejected": -2.4693663120269775,
|
|
"logps/chosen": -172.24966430664062,
|
|
"logps/rejected": -166.52056884765625,
|
|
"loss": 0.6975,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.027715325355529785,
|
|
"rewards/margins": -0.00841212272644043,
|
|
"rewards/rejected": -0.019303202629089355,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 2e-05,
|
|
"logits/chosen": -2.3024024963378906,
|
|
"logits/rejected": -2.2906301021575928,
|
|
"logps/chosen": -243.98683166503906,
|
|
"logps/rejected": -199.12408447265625,
|
|
"loss": 0.6979,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.028235863894224167,
|
|
"rewards/margins": -0.00847182422876358,
|
|
"rewards/rejected": -0.019764041528105736,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 2.05e-05,
|
|
"logits/chosen": -2.660562753677368,
|
|
"logits/rejected": -2.6754074096679688,
|
|
"logps/chosen": -156.67514038085938,
|
|
"logps/rejected": -147.5068359375,
|
|
"loss": 0.6887,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.03220677375793457,
|
|
"rewards/margins": 0.009226083755493164,
|
|
"rewards/rejected": -0.041432857513427734,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 2.1e-05,
|
|
"logits/chosen": -2.482787847518921,
|
|
"logits/rejected": -2.4663233757019043,
|
|
"logps/chosen": -207.6859588623047,
|
|
"logps/rejected": -182.35931396484375,
|
|
"loss": 0.6875,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.014715791679918766,
|
|
"rewards/margins": 0.012153576128184795,
|
|
"rewards/rejected": -0.02686937153339386,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.15e-05,
|
|
"logits/chosen": -2.6544148921966553,
|
|
"logits/rejected": -2.717411756515503,
|
|
"logps/chosen": -197.96400451660156,
|
|
"logps/rejected": -223.0219268798828,
|
|
"loss": 0.691,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.024796580895781517,
|
|
"rewards/margins": 0.005184031091630459,
|
|
"rewards/rejected": -0.0299806110560894,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.2000000000000003e-05,
|
|
"logits/chosen": -2.561187744140625,
|
|
"logits/rejected": -2.6224913597106934,
|
|
"logps/chosen": -194.60935974121094,
|
|
"logps/rejected": -194.01577758789062,
|
|
"loss": 0.6971,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.020246006548404694,
|
|
"rewards/margins": -0.0076716188341379166,
|
|
"rewards/rejected": -0.012574386782944202,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.25e-05,
|
|
"logits/chosen": -2.5130224227905273,
|
|
"logits/rejected": -2.672990083694458,
|
|
"logps/chosen": -154.0092010498047,
|
|
"logps/rejected": -190.70748901367188,
|
|
"loss": 0.6818,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.018519926816225052,
|
|
"rewards/margins": 0.023535681888461113,
|
|
"rewards/rejected": -0.042055610567331314,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.3000000000000003e-05,
|
|
"logits/chosen": -2.738673686981201,
|
|
"logits/rejected": -2.82279896736145,
|
|
"logps/chosen": -174.6641387939453,
|
|
"logps/rejected": -163.0157928466797,
|
|
"loss": 0.6967,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.014652942307293415,
|
|
"rewards/margins": -0.006221937946975231,
|
|
"rewards/rejected": -0.008431006222963333,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.35e-05,
|
|
"logits/chosen": -2.581393241882324,
|
|
"logits/rejected": -2.5734691619873047,
|
|
"logps/chosen": -137.8702392578125,
|
|
"logps/rejected": -154.74896240234375,
|
|
"loss": 0.6887,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.007426738273352385,
|
|
"rewards/margins": 0.009324884042143822,
|
|
"rewards/rejected": -0.016751624643802643,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.4e-05,
|
|
"logits/chosen": -2.544344425201416,
|
|
"logits/rejected": -2.690079689025879,
|
|
"logps/chosen": -185.8907012939453,
|
|
"logps/rejected": -195.9602508544922,
|
|
"loss": 0.6898,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.02690129354596138,
|
|
"rewards/margins": 0.007246016524732113,
|
|
"rewards/rejected": -0.03414731100201607,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 2.45e-05,
|
|
"logits/chosen": -2.583885669708252,
|
|
"logits/rejected": -2.609614849090576,
|
|
"logps/chosen": -215.93264770507812,
|
|
"logps/rejected": -234.49636840820312,
|
|
"loss": 0.6858,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.022774625569581985,
|
|
"rewards/margins": 0.015723586082458496,
|
|
"rewards/rejected": -0.03849821165204048,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.5e-05,
|
|
"logits/chosen": -2.6080195903778076,
|
|
"logits/rejected": -2.655003070831299,
|
|
"logps/chosen": -229.5977020263672,
|
|
"logps/rejected": -211.47779846191406,
|
|
"loss": 0.6925,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.026349734514951706,
|
|
"rewards/margins": 0.0026414887979626656,
|
|
"rewards/rejected": -0.028991222381591797,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.5500000000000003e-05,
|
|
"logits/chosen": -2.180418014526367,
|
|
"logits/rejected": -2.210296869277954,
|
|
"logps/chosen": -221.1846466064453,
|
|
"logps/rejected": -185.85655212402344,
|
|
"loss": 0.6891,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.013596129603683949,
|
|
"rewards/margins": 0.009130668826401234,
|
|
"rewards/rejected": -0.022726796567440033,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.6000000000000002e-05,
|
|
"logits/chosen": -2.647798538208008,
|
|
"logits/rejected": -2.715623617172241,
|
|
"logps/chosen": -186.01138305664062,
|
|
"logps/rejected": -180.89259338378906,
|
|
"loss": 0.6785,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": 0.019299650564789772,
|
|
"rewards/margins": 0.030588869005441666,
|
|
"rewards/rejected": -0.011289214715361595,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.6500000000000004e-05,
|
|
"logits/chosen": -2.570209503173828,
|
|
"logits/rejected": -2.5728917121887207,
|
|
"logps/chosen": -209.2962188720703,
|
|
"logps/rejected": -216.60145568847656,
|
|
"loss": 0.6824,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.005204535089433193,
|
|
"rewards/margins": 0.021862652152776718,
|
|
"rewards/rejected": -0.027067184448242188,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.7000000000000002e-05,
|
|
"logits/chosen": -2.672147512435913,
|
|
"logits/rejected": -2.678018808364868,
|
|
"logps/chosen": -167.4713134765625,
|
|
"logps/rejected": -170.08120727539062,
|
|
"loss": 0.691,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.013661455363035202,
|
|
"rewards/margins": 0.004836535546928644,
|
|
"rewards/rejected": -0.018497992306947708,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.7500000000000004e-05,
|
|
"logits/chosen": -2.4589505195617676,
|
|
"logits/rejected": -2.462191343307495,
|
|
"logps/chosen": -200.73976135253906,
|
|
"logps/rejected": -195.30276489257812,
|
|
"loss": 0.704,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.03748317062854767,
|
|
"rewards/margins": -0.020142268389463425,
|
|
"rewards/rejected": -0.017340898513793945,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.8000000000000003e-05,
|
|
"logits/chosen": -2.230989933013916,
|
|
"logits/rejected": -2.2140636444091797,
|
|
"logps/chosen": -143.33631896972656,
|
|
"logps/rejected": -136.0796661376953,
|
|
"loss": 0.6989,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.020144915208220482,
|
|
"rewards/margins": -0.01022801361978054,
|
|
"rewards/rejected": -0.009916901588439941,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 2.8499999999999998e-05,
|
|
"logits/chosen": -2.5017194747924805,
|
|
"logits/rejected": -2.58662486076355,
|
|
"logps/chosen": -167.50990295410156,
|
|
"logps/rejected": -208.83607482910156,
|
|
"loss": 0.6808,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0027861359994858503,
|
|
"rewards/margins": 0.02604994922876358,
|
|
"rewards/rejected": -0.02883608266711235,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 2.9e-05,
|
|
"logits/chosen": -2.53066349029541,
|
|
"logits/rejected": -2.608471393585205,
|
|
"logps/chosen": -241.85113525390625,
|
|
"logps/rejected": -260.0408630371094,
|
|
"loss": 0.6939,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.024016117677092552,
|
|
"rewards/margins": -0.0005100721027702093,
|
|
"rewards/rejected": -0.023506049066781998,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 2.95e-05,
|
|
"logits/chosen": -2.2705140113830566,
|
|
"logits/rejected": -2.4387779235839844,
|
|
"logps/chosen": -179.22540283203125,
|
|
"logps/rejected": -182.5544891357422,
|
|
"loss": 0.6775,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.00418963422998786,
|
|
"rewards/margins": 0.033549048006534576,
|
|
"rewards/rejected": -0.02935941144824028,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3e-05,
|
|
"logits/chosen": -2.46372652053833,
|
|
"logits/rejected": -2.513279438018799,
|
|
"logps/chosen": -230.43133544921875,
|
|
"logps/rejected": -227.0118865966797,
|
|
"loss": 0.6761,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.010919665917754173,
|
|
"rewards/margins": 0.03570995107293129,
|
|
"rewards/rejected": -0.04662961885333061,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.05e-05,
|
|
"logits/chosen": -2.530576467514038,
|
|
"logits/rejected": -2.5415897369384766,
|
|
"logps/chosen": -200.46568298339844,
|
|
"logps/rejected": -220.82989501953125,
|
|
"loss": 0.6794,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.0384901762008667,
|
|
"rewards/margins": 0.03156058490276337,
|
|
"rewards/rejected": -0.07005076855421066,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.1e-05,
|
|
"logits/chosen": -2.5888161659240723,
|
|
"logits/rejected": -2.639291286468506,
|
|
"logps/chosen": -132.55116271972656,
|
|
"logps/rejected": -167.082275390625,
|
|
"loss": 0.7042,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.03156990930438042,
|
|
"rewards/margins": -0.018804360181093216,
|
|
"rewards/rejected": -0.0127655528485775,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.15e-05,
|
|
"logits/chosen": -2.46744704246521,
|
|
"logits/rejected": -2.494837522506714,
|
|
"logps/chosen": -210.9451141357422,
|
|
"logps/rejected": -211.72433471679688,
|
|
"loss": 0.6799,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.0017502065747976303,
|
|
"rewards/margins": 0.027423406019806862,
|
|
"rewards/rejected": -0.02567320130765438,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.2000000000000005e-05,
|
|
"logits/chosen": -2.4783239364624023,
|
|
"logits/rejected": -2.490576982498169,
|
|
"logps/chosen": -197.71629333496094,
|
|
"logps/rejected": -207.23048400878906,
|
|
"loss": 0.6864,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.01412363164126873,
|
|
"rewards/margins": 0.01561451144516468,
|
|
"rewards/rejected": -0.029738139361143112,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.2500000000000004e-05,
|
|
"logits/chosen": -2.3918569087982178,
|
|
"logits/rejected": -2.4426636695861816,
|
|
"logps/chosen": -164.30029296875,
|
|
"logps/rejected": -201.85992431640625,
|
|
"loss": 0.6908,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.025302361696958542,
|
|
"rewards/margins": 0.009253643453121185,
|
|
"rewards/rejected": -0.034556008875370026,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.3e-05,
|
|
"logits/chosen": -2.4934239387512207,
|
|
"logits/rejected": -2.5188210010528564,
|
|
"logps/chosen": -178.12030029296875,
|
|
"logps/rejected": -211.23947143554688,
|
|
"loss": 0.6908,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.05651288107037544,
|
|
"rewards/margins": 0.0065690516494214535,
|
|
"rewards/rejected": -0.06308193504810333,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.35e-05,
|
|
"logits/chosen": -2.415985345840454,
|
|
"logits/rejected": -2.3687984943389893,
|
|
"logps/chosen": -211.91453552246094,
|
|
"logps/rejected": -216.65591430664062,
|
|
"loss": 0.6808,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.045519113540649414,
|
|
"rewards/margins": 0.028419354930520058,
|
|
"rewards/rejected": -0.07393846660852432,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.4000000000000007e-05,
|
|
"logits/chosen": -2.37481427192688,
|
|
"logits/rejected": -2.4693892002105713,
|
|
"logps/chosen": -172.92459106445312,
|
|
"logps/rejected": -209.3857421875,
|
|
"loss": 0.6801,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.03707296401262283,
|
|
"rewards/margins": 0.027949143201112747,
|
|
"rewards/rejected": -0.06502211093902588,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.45e-05,
|
|
"logits/chosen": -2.3615283966064453,
|
|
"logits/rejected": -2.3259239196777344,
|
|
"logps/chosen": -154.25360107421875,
|
|
"logps/rejected": -161.130126953125,
|
|
"loss": 0.7022,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.04533124342560768,
|
|
"rewards/margins": -0.016323519870638847,
|
|
"rewards/rejected": -0.029007721692323685,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.5e-05,
|
|
"logits/chosen": -2.64551043510437,
|
|
"logits/rejected": -2.7289645671844482,
|
|
"logps/chosen": -200.1268768310547,
|
|
"logps/rejected": -225.6541748046875,
|
|
"loss": 0.6993,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.0578455924987793,
|
|
"rewards/margins": -0.008899472653865814,
|
|
"rewards/rejected": -0.04894612357020378,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.55e-05,
|
|
"logits/chosen": -2.703721046447754,
|
|
"logits/rejected": -2.669755220413208,
|
|
"logps/chosen": -208.99668884277344,
|
|
"logps/rejected": -164.8292694091797,
|
|
"loss": 0.6756,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.022734597325325012,
|
|
"rewards/margins": 0.03882308304309845,
|
|
"rewards/rejected": -0.06155767664313316,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6e-05,
|
|
"logits/chosen": -2.6480765342712402,
|
|
"logits/rejected": -2.696770191192627,
|
|
"logps/chosen": -172.99888610839844,
|
|
"logps/rejected": -201.09979248046875,
|
|
"loss": 0.7063,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.053855180740356445,
|
|
"rewards/margins": -0.024987507611513138,
|
|
"rewards/rejected": -0.028867674991488457,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.65e-05,
|
|
"logits/chosen": -2.6311445236206055,
|
|
"logits/rejected": -2.6575632095336914,
|
|
"logps/chosen": -251.72789001464844,
|
|
"logps/rejected": -271.0470886230469,
|
|
"loss": 0.6867,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.03289387375116348,
|
|
"rewards/margins": 0.014702942222356796,
|
|
"rewards/rejected": -0.04759681224822998,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7e-05,
|
|
"logits/chosen": -2.3990683555603027,
|
|
"logits/rejected": -2.502671480178833,
|
|
"logps/chosen": -197.92665100097656,
|
|
"logps/rejected": -211.00967407226562,
|
|
"loss": 0.6841,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.014272330328822136,
|
|
"rewards/margins": 0.020740672945976257,
|
|
"rewards/rejected": -0.03501300886273384,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7500000000000003e-05,
|
|
"logits/chosen": -2.575502872467041,
|
|
"logits/rejected": -2.527902364730835,
|
|
"logps/chosen": -214.5215301513672,
|
|
"logps/rejected": -201.05194091796875,
|
|
"loss": 0.6651,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.006246686447411776,
|
|
"rewards/margins": 0.059338975697755814,
|
|
"rewards/rejected": -0.065585657954216,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.8e-05,
|
|
"logits/chosen": -2.217294692993164,
|
|
"logits/rejected": -2.223574638366699,
|
|
"logps/chosen": -174.7819366455078,
|
|
"logps/rejected": -178.09852600097656,
|
|
"loss": 0.7013,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.09115451574325562,
|
|
"rewards/margins": -0.012327454052865505,
|
|
"rewards/rejected": -0.07882705330848694,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.85e-05,
|
|
"logits/chosen": -2.518965721130371,
|
|
"logits/rejected": -2.443446159362793,
|
|
"logps/chosen": -201.1251220703125,
|
|
"logps/rejected": -170.15533447265625,
|
|
"loss": 0.7029,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.04162323474884033,
|
|
"rewards/margins": -0.015459035523235798,
|
|
"rewards/rejected": -0.02616419643163681,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.9000000000000006e-05,
|
|
"logits/chosen": -2.521008253097534,
|
|
"logits/rejected": -2.6866934299468994,
|
|
"logps/chosen": -269.85284423828125,
|
|
"logps/rejected": -240.82626342773438,
|
|
"loss": 0.6928,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.05473289266228676,
|
|
"rewards/margins": 0.00423135980963707,
|
|
"rewards/rejected": -0.05896425247192383,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.9500000000000005e-05,
|
|
"logits/chosen": -2.3015501499176025,
|
|
"logits/rejected": -2.351773500442505,
|
|
"logps/chosen": -225.17691040039062,
|
|
"logps/rejected": -249.8846435546875,
|
|
"loss": 0.6873,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.06839534640312195,
|
|
"rewards/margins": 0.014748764224350452,
|
|
"rewards/rejected": -0.08314411342144012,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 4e-05,
|
|
"logits/chosen": -2.6201720237731934,
|
|
"logits/rejected": -2.680964708328247,
|
|
"logps/chosen": -225.08645629882812,
|
|
"logps/rejected": -232.00096130371094,
|
|
"loss": 0.6739,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.06486918777227402,
|
|
"rewards/margins": 0.04111311584711075,
|
|
"rewards/rejected": -0.10598230361938477,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.05e-05,
|
|
"logits/chosen": -2.5782110691070557,
|
|
"logits/rejected": -2.5646634101867676,
|
|
"logps/chosen": -249.90745544433594,
|
|
"logps/rejected": -257.51275634765625,
|
|
"loss": 0.6976,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.08720846474170685,
|
|
"rewards/margins": -0.0017095585353672504,
|
|
"rewards/rejected": -0.08549890667200089,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.1e-05,
|
|
"logits/chosen": -2.5777688026428223,
|
|
"logits/rejected": -2.665797472000122,
|
|
"logps/chosen": -171.82347106933594,
|
|
"logps/rejected": -211.63311767578125,
|
|
"loss": 0.6883,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.08461908996105194,
|
|
"rewards/margins": 0.013853237964212894,
|
|
"rewards/rejected": -0.09847234189510345,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.15e-05,
|
|
"logits/chosen": -2.3364946842193604,
|
|
"logits/rejected": -2.3084681034088135,
|
|
"logps/chosen": -193.4939727783203,
|
|
"logps/rejected": -184.15200805664062,
|
|
"loss": 0.717,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.0981694683432579,
|
|
"rewards/margins": -0.04480106756091118,
|
|
"rewards/rejected": -0.053368404507637024,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.2e-05,
|
|
"logits/chosen": -2.445204257965088,
|
|
"logits/rejected": -2.5341625213623047,
|
|
"logps/chosen": -174.59225463867188,
|
|
"logps/rejected": -210.94137573242188,
|
|
"loss": 0.699,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.09675168246030807,
|
|
"rewards/margins": -0.00985276885330677,
|
|
"rewards/rejected": -0.08689891546964645,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.25e-05,
|
|
"logits/chosen": -2.4798362255096436,
|
|
"logits/rejected": -2.4760632514953613,
|
|
"logps/chosen": -148.0768585205078,
|
|
"logps/rejected": -176.31700134277344,
|
|
"loss": 0.6945,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.07728591561317444,
|
|
"rewards/margins": 0.0020556673407554626,
|
|
"rewards/rejected": -0.0793415829539299,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.3e-05,
|
|
"logits/chosen": -2.5911381244659424,
|
|
"logits/rejected": -2.630659341812134,
|
|
"logps/chosen": -203.1630859375,
|
|
"logps/rejected": -183.4739227294922,
|
|
"loss": 0.653,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.04599933326244354,
|
|
"rewards/margins": 0.08558819442987442,
|
|
"rewards/rejected": -0.13158753514289856,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 4.35e-05,
|
|
"logits/chosen": -2.600797414779663,
|
|
"logits/rejected": -2.6318492889404297,
|
|
"logps/chosen": -156.8225860595703,
|
|
"logps/rejected": -161.6145477294922,
|
|
"loss": 0.6705,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.07066044956445694,
|
|
"rewards/margins": 0.05500438064336777,
|
|
"rewards/rejected": -0.1256648302078247,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.4000000000000006e-05,
|
|
"logits/chosen": -2.2124781608581543,
|
|
"logits/rejected": -2.259019374847412,
|
|
"logps/chosen": -142.0679473876953,
|
|
"logps/rejected": -177.9369659423828,
|
|
"loss": 0.6949,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.13633377850055695,
|
|
"rewards/margins": 0.000554969534277916,
|
|
"rewards/rejected": -0.1368887573480606,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.4500000000000004e-05,
|
|
"logits/chosen": -2.396183729171753,
|
|
"logits/rejected": -2.533904552459717,
|
|
"logps/chosen": -171.38790893554688,
|
|
"logps/rejected": -202.30979919433594,
|
|
"loss": 0.6554,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.09015192836523056,
|
|
"rewards/margins": 0.08280421048402786,
|
|
"rewards/rejected": -0.17295613884925842,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.5e-05,
|
|
"logits/chosen": -2.640425443649292,
|
|
"logits/rejected": -2.674539804458618,
|
|
"logps/chosen": -175.3508758544922,
|
|
"logps/rejected": -180.97915649414062,
|
|
"loss": 0.6768,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.1077599823474884,
|
|
"rewards/margins": 0.03759467601776123,
|
|
"rewards/rejected": -0.14535464346408844,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.55e-05,
|
|
"logits/chosen": -2.5857696533203125,
|
|
"logits/rejected": -2.5822975635528564,
|
|
"logps/chosen": -176.55982971191406,
|
|
"logps/rejected": -196.5354461669922,
|
|
"loss": 0.7,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.09103889763355255,
|
|
"rewards/margins": -0.009319041855633259,
|
|
"rewards/rejected": -0.08171986043453217,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.600000000000001e-05,
|
|
"logits/chosen": -2.660163402557373,
|
|
"logits/rejected": -2.5935940742492676,
|
|
"logps/chosen": -233.04049682617188,
|
|
"logps/rejected": -212.1655731201172,
|
|
"loss": 0.6863,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.118477463722229,
|
|
"rewards/margins": 0.017601586878299713,
|
|
"rewards/rejected": -0.13607905805110931,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.6500000000000005e-05,
|
|
"logits/chosen": -2.5211241245269775,
|
|
"logits/rejected": -2.6064364910125732,
|
|
"logps/chosen": -176.89476013183594,
|
|
"logps/rejected": -187.39271545410156,
|
|
"loss": 0.6736,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.07700353115797043,
|
|
"rewards/margins": 0.04530329257249832,
|
|
"rewards/rejected": -0.12230683863162994,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.7e-05,
|
|
"logits/chosen": -2.6894078254699707,
|
|
"logits/rejected": -2.6806745529174805,
|
|
"logps/chosen": -244.7545623779297,
|
|
"logps/rejected": -212.83453369140625,
|
|
"loss": 0.6446,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.10891007632017136,
|
|
"rewards/margins": 0.10942523181438446,
|
|
"rewards/rejected": -0.2183353304862976,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 4.75e-05,
|
|
"logits/chosen": -2.4758429527282715,
|
|
"logits/rejected": -2.5564448833465576,
|
|
"logps/chosen": -161.29554748535156,
|
|
"logps/rejected": -185.48199462890625,
|
|
"loss": 0.6991,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.152861088514328,
|
|
"rewards/margins": -0.0035788798704743385,
|
|
"rewards/rejected": -0.14928221702575684,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.8e-05,
|
|
"logits/chosen": -2.47318959236145,
|
|
"logits/rejected": -2.439746856689453,
|
|
"logps/chosen": -241.09283447265625,
|
|
"logps/rejected": -218.68580627441406,
|
|
"loss": 0.7017,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.1952984780073166,
|
|
"rewards/margins": -0.01051153801381588,
|
|
"rewards/rejected": -0.18478693068027496,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.85e-05,
|
|
"logits/chosen": -2.541466474533081,
|
|
"logits/rejected": -2.56956148147583,
|
|
"logps/chosen": -180.84725952148438,
|
|
"logps/rejected": -211.02352905273438,
|
|
"loss": 0.6627,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.15894030034542084,
|
|
"rewards/margins": 0.06507530808448792,
|
|
"rewards/rejected": -0.22401559352874756,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.9e-05,
|
|
"logits/chosen": -2.226611852645874,
|
|
"logits/rejected": -2.359558343887329,
|
|
"logps/chosen": -145.781494140625,
|
|
"logps/rejected": -181.21531677246094,
|
|
"loss": 0.6689,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.177232027053833,
|
|
"rewards/margins": 0.058251187205314636,
|
|
"rewards/rejected": -0.23548321425914764,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"logits/chosen": -2.5605039596557617,
|
|
"logits/rejected": -2.641690492630005,
|
|
"logps/chosen": -156.43446350097656,
|
|
"logps/rejected": -171.6468505859375,
|
|
"loss": 0.6801,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.23164159059524536,
|
|
"rewards/margins": 0.03432049974799156,
|
|
"rewards/rejected": -0.2659620940685272,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 5e-05,
|
|
"logits/chosen": -2.7840309143066406,
|
|
"logits/rejected": -2.8028616905212402,
|
|
"logps/chosen": -221.6836395263672,
|
|
"logps/rejected": -247.433349609375,
|
|
"loss": 0.647,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.14767608046531677,
|
|
"rewards/margins": 0.1068597286939621,
|
|
"rewards/rejected": -0.25453582406044006,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.99999978299634e-05,
|
|
"logits/chosen": -2.1325531005859375,
|
|
"logits/rejected": -2.1438934803009033,
|
|
"logps/chosen": -171.369140625,
|
|
"logps/rejected": -184.96827697753906,
|
|
"loss": 0.6417,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.18850459158420563,
|
|
"rewards/margins": 0.12018898129463196,
|
|
"rewards/rejected": -0.3086935877799988,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.999999131985394e-05,
|
|
"logits/chosen": -2.6781508922576904,
|
|
"logits/rejected": -2.5359549522399902,
|
|
"logps/chosen": -232.26742553710938,
|
|
"logps/rejected": -214.5188446044922,
|
|
"loss": 0.7372,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.32363808155059814,
|
|
"rewards/margins": -0.07766950130462646,
|
|
"rewards/rejected": -0.2459685504436493,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 4.999998046967279e-05,
|
|
"logits/chosen": -2.54486346244812,
|
|
"logits/rejected": -2.6447086334228516,
|
|
"logps/chosen": -179.6219482421875,
|
|
"logps/rejected": -222.37010192871094,
|
|
"loss": 0.6585,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.23128211498260498,
|
|
"rewards/margins": 0.08162947744131088,
|
|
"rewards/rejected": -0.31291159987449646,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.9999965279421804e-05,
|
|
"logits/chosen": -2.4281625747680664,
|
|
"logits/rejected": -2.5068893432617188,
|
|
"logps/chosen": -163.3339385986328,
|
|
"logps/rejected": -180.06222534179688,
|
|
"loss": 0.6739,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.22076301276683807,
|
|
"rewards/margins": 0.05238480493426323,
|
|
"rewards/rejected": -0.2731478214263916,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.999994574910364e-05,
|
|
"logits/chosen": -2.5497522354125977,
|
|
"logits/rejected": -2.558523178100586,
|
|
"logps/chosen": -214.64537048339844,
|
|
"logps/rejected": -210.29678344726562,
|
|
"loss": 0.6691,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.19515390694141388,
|
|
"rewards/margins": 0.06168302148580551,
|
|
"rewards/rejected": -0.2568369209766388,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.999992187872167e-05,
|
|
"logits/chosen": -2.6358773708343506,
|
|
"logits/rejected": -2.672025680541992,
|
|
"logps/chosen": -235.18109130859375,
|
|
"logps/rejected": -233.78045654296875,
|
|
"loss": 0.7203,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.25453630089759827,
|
|
"rewards/margins": -0.03941688686609268,
|
|
"rewards/rejected": -0.215119406580925,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.9999893668280043e-05,
|
|
"logits/chosen": -2.305128812789917,
|
|
"logits/rejected": -2.384552478790283,
|
|
"logps/chosen": -168.1253662109375,
|
|
"logps/rejected": -210.5357666015625,
|
|
"loss": 0.6558,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.30615508556365967,
|
|
"rewards/margins": 0.10809167474508286,
|
|
"rewards/rejected": -0.4142467677593231,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.999986111778367e-05,
|
|
"logits/chosen": -2.4207632541656494,
|
|
"logits/rejected": -2.4426565170288086,
|
|
"logps/chosen": -187.8740234375,
|
|
"logps/rejected": -188.6339569091797,
|
|
"loss": 0.6383,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.2796909809112549,
|
|
"rewards/margins": 0.13547499477863312,
|
|
"rewards/rejected": -0.4151660203933716,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.999982422723818e-05,
|
|
"logits/chosen": -2.505187749862671,
|
|
"logits/rejected": -2.4738471508026123,
|
|
"logps/chosen": -187.51710510253906,
|
|
"logps/rejected": -227.13916015625,
|
|
"loss": 0.7216,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5267550945281982,
|
|
"rewards/margins": -0.03421187028288841,
|
|
"rewards/rejected": -0.4925432801246643,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 4.9999782996649994e-05,
|
|
"logits/chosen": -2.4367451667785645,
|
|
"logits/rejected": -2.4629135131835938,
|
|
"logps/chosen": -221.5149688720703,
|
|
"logps/rejected": -255.66635131835938,
|
|
"loss": 0.6923,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4186938405036926,
|
|
"rewards/margins": 0.03650692105293274,
|
|
"rewards/rejected": -0.45520079135894775,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999973742602626e-05,
|
|
"logits/chosen": -2.6166908740997314,
|
|
"logits/rejected": -2.638493061065674,
|
|
"logps/chosen": -212.32894897460938,
|
|
"logps/rejected": -216.49720764160156,
|
|
"loss": 0.6789,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.47340863943099976,
|
|
"rewards/margins": 0.059217117726802826,
|
|
"rewards/rejected": -0.5326257944107056,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999968751537489e-05,
|
|
"logits/chosen": -2.549008846282959,
|
|
"logits/rejected": -2.6095943450927734,
|
|
"logps/chosen": -205.5695343017578,
|
|
"logps/rejected": -196.99383544921875,
|
|
"loss": 0.7338,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.47608208656311035,
|
|
"rewards/margins": -0.05748309940099716,
|
|
"rewards/rejected": -0.4185989201068878,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.9999633264704564e-05,
|
|
"logits/chosen": -2.2947440147399902,
|
|
"logits/rejected": -2.240459680557251,
|
|
"logps/chosen": -203.31182861328125,
|
|
"logps/rejected": -175.49354553222656,
|
|
"loss": 0.7139,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.39874717593193054,
|
|
"rewards/margins": -0.014322709292173386,
|
|
"rewards/rejected": -0.38442447781562805,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999957467402468e-05,
|
|
"logits/chosen": -2.118520736694336,
|
|
"logits/rejected": -2.0125582218170166,
|
|
"logps/chosen": -159.09979248046875,
|
|
"logps/rejected": -151.80679321289062,
|
|
"loss": 0.6981,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.43373599648475647,
|
|
"rewards/margins": 0.0064653148874640465,
|
|
"rewards/rejected": -0.4402012825012207,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.9999511743345426e-05,
|
|
"logits/chosen": -2.251662254333496,
|
|
"logits/rejected": -2.343283176422119,
|
|
"logps/chosen": -166.46377563476562,
|
|
"logps/rejected": -213.02813720703125,
|
|
"loss": 0.6281,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.44466906785964966,
|
|
"rewards/margins": 0.18120086193084717,
|
|
"rewards/rejected": -0.625869870185852,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999944447267771e-05,
|
|
"logits/chosen": -2.347130298614502,
|
|
"logits/rejected": -2.342409610748291,
|
|
"logps/chosen": -216.30650329589844,
|
|
"logps/rejected": -210.64273071289062,
|
|
"loss": 0.7966,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6098219752311707,
|
|
"rewards/margins": -0.12615808844566345,
|
|
"rewards/rejected": -0.4836638867855072,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999937286203322e-05,
|
|
"logits/chosen": -2.3265388011932373,
|
|
"logits/rejected": -2.3534939289093018,
|
|
"logps/chosen": -204.47706604003906,
|
|
"logps/rejected": -186.42286682128906,
|
|
"loss": 0.6674,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.291274756193161,
|
|
"rewards/margins": 0.0668107196688652,
|
|
"rewards/rejected": -0.3580854535102844,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.999929691142439e-05,
|
|
"logits/chosen": -2.062469720840454,
|
|
"logits/rejected": -2.2004528045654297,
|
|
"logps/chosen": -144.79664611816406,
|
|
"logps/rejected": -149.7763671875,
|
|
"loss": 0.6225,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.46431243419647217,
|
|
"rewards/margins": 0.1997339278459549,
|
|
"rewards/rejected": -0.6640464067459106,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.99992166208644e-05,
|
|
"logits/chosen": -2.2899487018585205,
|
|
"logits/rejected": -2.2257750034332275,
|
|
"logps/chosen": -208.61517333984375,
|
|
"logps/rejected": -200.343505859375,
|
|
"loss": 0.7233,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5118069052696228,
|
|
"rewards/margins": -0.018164699897170067,
|
|
"rewards/rejected": -0.4936422109603882,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.999913199036719e-05,
|
|
"logits/chosen": -2.520472288131714,
|
|
"logits/rejected": -2.612544298171997,
|
|
"logps/chosen": -160.17779541015625,
|
|
"logps/rejected": -191.498291015625,
|
|
"loss": 0.6421,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4993933439254761,
|
|
"rewards/margins": 0.15375012159347534,
|
|
"rewards/rejected": -0.6531434059143066,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.9999043019947454e-05,
|
|
"logits/chosen": -2.5590789318084717,
|
|
"logits/rejected": -2.594666004180908,
|
|
"logps/chosen": -219.09494018554688,
|
|
"logps/rejected": -271.0511474609375,
|
|
"loss": 0.6746,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.37969574332237244,
|
|
"rewards/margins": 0.06747917830944061,
|
|
"rewards/rejected": -0.44717487692832947,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.9998949709620636e-05,
|
|
"logits/chosen": -2.292649269104004,
|
|
"logits/rejected": -2.4197192192077637,
|
|
"logps/chosen": -156.77658081054688,
|
|
"logps/rejected": -181.56752014160156,
|
|
"loss": 0.6224,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.24466285109519958,
|
|
"rewards/margins": 0.17808005213737488,
|
|
"rewards/rejected": -0.4227428734302521,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.999885205940293e-05,
|
|
"logits/chosen": -2.526571750640869,
|
|
"logits/rejected": -2.6078643798828125,
|
|
"logps/chosen": -188.48431396484375,
|
|
"logps/rejected": -209.88192749023438,
|
|
"loss": 0.6138,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.38001808524131775,
|
|
"rewards/margins": 0.21444422006607056,
|
|
"rewards/rejected": -0.5944622755050659,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.9998750069311306e-05,
|
|
"logits/chosen": -2.477735757827759,
|
|
"logits/rejected": -2.524329900741577,
|
|
"logps/chosen": -170.30780029296875,
|
|
"logps/rejected": -167.05538940429688,
|
|
"loss": 0.7105,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.45474332571029663,
|
|
"rewards/margins": -0.0021874159574508667,
|
|
"rewards/rejected": -0.45255595445632935,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.999864373936345e-05,
|
|
"logits/chosen": -2.486284017562866,
|
|
"logits/rejected": -2.511571168899536,
|
|
"logps/chosen": -213.28150939941406,
|
|
"logps/rejected": -220.744140625,
|
|
"loss": 0.7154,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5180980563163757,
|
|
"rewards/margins": -0.013586281798779964,
|
|
"rewards/rejected": -0.5045117139816284,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.999853306957783e-05,
|
|
"logits/chosen": -2.149216651916504,
|
|
"logits/rejected": -2.331148862838745,
|
|
"logps/chosen": -212.50167846679688,
|
|
"logps/rejected": -229.76211547851562,
|
|
"loss": 0.6783,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.45535096526145935,
|
|
"rewards/margins": 0.13961246609687805,
|
|
"rewards/rejected": -0.5949634909629822,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.9998418059973654e-05,
|
|
"logits/chosen": -2.3633456230163574,
|
|
"logits/rejected": -2.3958113193511963,
|
|
"logps/chosen": -260.77191162109375,
|
|
"logps/rejected": -269.7747497558594,
|
|
"loss": 0.624,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3385522663593292,
|
|
"rewards/margins": 0.154096320271492,
|
|
"rewards/rejected": -0.4926486015319824,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.99982987105709e-05,
|
|
"logits/chosen": -2.4027421474456787,
|
|
"logits/rejected": -2.47273588180542,
|
|
"logps/chosen": -190.19549560546875,
|
|
"logps/rejected": -199.79869079589844,
|
|
"loss": 0.647,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.45771288871765137,
|
|
"rewards/margins": 0.16400352120399475,
|
|
"rewards/rejected": -0.6217163801193237,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.999817502139027e-05,
|
|
"logits/chosen": -2.459967851638794,
|
|
"logits/rejected": -2.5030205249786377,
|
|
"logps/chosen": -152.8588104248047,
|
|
"logps/rejected": -151.28167724609375,
|
|
"loss": 0.682,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4902128279209137,
|
|
"rewards/margins": 0.07620880752801895,
|
|
"rewards/rejected": -0.566421627998352,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.999804699245325e-05,
|
|
"logits/chosen": -2.613358497619629,
|
|
"logits/rejected": -2.6325273513793945,
|
|
"logps/chosen": -240.51898193359375,
|
|
"logps/rejected": -209.2674560546875,
|
|
"loss": 0.6862,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5984245538711548,
|
|
"rewards/margins": 0.08296503126621246,
|
|
"rewards/rejected": -0.6813895106315613,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.999791462378206e-05,
|
|
"logits/chosen": -2.523002862930298,
|
|
"logits/rejected": -2.544940948486328,
|
|
"logps/chosen": -209.53164672851562,
|
|
"logps/rejected": -211.6482391357422,
|
|
"loss": 0.6617,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4731237292289734,
|
|
"rewards/margins": 0.139340341091156,
|
|
"rewards/rejected": -0.6124641299247742,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.999777791539968e-05,
|
|
"logits/chosen": -2.3467066287994385,
|
|
"logits/rejected": -2.4002304077148438,
|
|
"logps/chosen": -229.88079833984375,
|
|
"logps/rejected": -236.3416290283203,
|
|
"loss": 0.6545,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5287673473358154,
|
|
"rewards/margins": 0.11849495768547058,
|
|
"rewards/rejected": -0.6472623348236084,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.9997636867329844e-05,
|
|
"logits/chosen": -2.3951451778411865,
|
|
"logits/rejected": -2.3899013996124268,
|
|
"logps/chosen": -230.42471313476562,
|
|
"logps/rejected": -204.70712280273438,
|
|
"loss": 0.6886,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7095022797584534,
|
|
"rewards/margins": 0.06169421225786209,
|
|
"rewards/rejected": -0.7711963653564453,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.999749147959703e-05,
|
|
"logits/chosen": -2.4759864807128906,
|
|
"logits/rejected": -2.461402177810669,
|
|
"logps/chosen": -208.62255859375,
|
|
"logps/rejected": -215.98049926757812,
|
|
"loss": 0.6288,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.47989076375961304,
|
|
"rewards/margins": 0.18254442512989044,
|
|
"rewards/rejected": -0.6624351143836975,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.99973417522265e-05,
|
|
"logits/chosen": -2.6637802124023438,
|
|
"logits/rejected": -2.6086812019348145,
|
|
"logps/chosen": -201.55992126464844,
|
|
"logps/rejected": -185.3202667236328,
|
|
"loss": 0.7158,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5636375546455383,
|
|
"rewards/margins": 0.09092222899198532,
|
|
"rewards/rejected": -0.6545597910881042,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.9997187685244234e-05,
|
|
"logits/chosen": -2.386868953704834,
|
|
"logits/rejected": -2.438833475112915,
|
|
"logps/chosen": -149.93809509277344,
|
|
"logps/rejected": -164.56199645996094,
|
|
"loss": 0.7271,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.43254008889198303,
|
|
"rewards/margins": -0.0416962131857872,
|
|
"rewards/rejected": -0.39084386825561523,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.999702927867698e-05,
|
|
"logits/chosen": -2.382352113723755,
|
|
"logits/rejected": -2.447472095489502,
|
|
"logps/chosen": -195.46499633789062,
|
|
"logps/rejected": -191.3668212890625,
|
|
"loss": 0.7562,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5873669385910034,
|
|
"rewards/margins": -0.07318583130836487,
|
|
"rewards/rejected": -0.5141811370849609,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.999686653255222e-05,
|
|
"logits/chosen": -2.6838254928588867,
|
|
"logits/rejected": -2.735358476638794,
|
|
"logps/chosen": -232.84722900390625,
|
|
"logps/rejected": -253.7335205078125,
|
|
"loss": 0.5769,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5806477069854736,
|
|
"rewards/margins": 0.33124226331710815,
|
|
"rewards/rejected": -0.9118900299072266,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.9996699446898235e-05,
|
|
"logits/chosen": -2.3523752689361572,
|
|
"logits/rejected": -2.455883026123047,
|
|
"logps/chosen": -158.80831909179688,
|
|
"logps/rejected": -215.85203552246094,
|
|
"loss": 0.7447,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5600742697715759,
|
|
"rewards/margins": 0.03419441357254982,
|
|
"rewards/rejected": -0.5942687392234802,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.999652802174402e-05,
|
|
"logits/chosen": -2.5024056434631348,
|
|
"logits/rejected": -2.4992480278015137,
|
|
"logps/chosen": -233.17120361328125,
|
|
"logps/rejected": -249.13568115234375,
|
|
"loss": 0.6822,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7229613065719604,
|
|
"rewards/margins": 0.06977183371782303,
|
|
"rewards/rejected": -0.7927330732345581,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.999635225711933e-05,
|
|
"logits/chosen": -2.3752949237823486,
|
|
"logits/rejected": -2.499027729034424,
|
|
"logps/chosen": -165.4285888671875,
|
|
"logps/rejected": -213.1377716064453,
|
|
"loss": 0.6571,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4794481098651886,
|
|
"rewards/margins": 0.1226089596748352,
|
|
"rewards/rejected": -0.602057158946991,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999617215305468e-05,
|
|
"logits/chosen": -2.578481674194336,
|
|
"logits/rejected": -2.5905745029449463,
|
|
"logps/chosen": -160.28704833984375,
|
|
"logps/rejected": -186.60206604003906,
|
|
"loss": 0.5771,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.379621684551239,
|
|
"rewards/margins": 0.28185874223709106,
|
|
"rewards/rejected": -0.6614804863929749,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999598770958134e-05,
|
|
"logits/chosen": -2.1814939975738525,
|
|
"logits/rejected": -2.273460865020752,
|
|
"logps/chosen": -178.5402069091797,
|
|
"logps/rejected": -246.2064208984375,
|
|
"loss": 0.6741,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.47223299741744995,
|
|
"rewards/margins": 0.11296035349369049,
|
|
"rewards/rejected": -0.5851933360099792,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999579892673133e-05,
|
|
"logits/chosen": -2.6444742679595947,
|
|
"logits/rejected": -2.651982545852661,
|
|
"logps/chosen": -191.74351501464844,
|
|
"logps/rejected": -191.7657928466797,
|
|
"loss": 0.6952,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6643162965774536,
|
|
"rewards/margins": 0.027530232444405556,
|
|
"rewards/rejected": -0.691846489906311,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.9995605804537426e-05,
|
|
"logits/chosen": -2.4908974170684814,
|
|
"logits/rejected": -2.550518274307251,
|
|
"logps/chosen": -188.7380828857422,
|
|
"logps/rejected": -211.574951171875,
|
|
"loss": 0.5798,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3170633912086487,
|
|
"rewards/margins": 0.3337726294994354,
|
|
"rewards/rejected": -0.6508359313011169,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999540834303315e-05,
|
|
"logits/chosen": -2.403130054473877,
|
|
"logits/rejected": -2.5044126510620117,
|
|
"logps/chosen": -216.05221557617188,
|
|
"logps/rejected": -224.658203125,
|
|
"loss": 0.5719,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5969533920288086,
|
|
"rewards/margins": 0.4197196662425995,
|
|
"rewards/rejected": -1.0166730880737305,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999520654225278e-05,
|
|
"logits/chosen": -2.5251312255859375,
|
|
"logits/rejected": -2.6076760292053223,
|
|
"logps/chosen": -171.32049560546875,
|
|
"logps/rejected": -189.77108764648438,
|
|
"loss": 0.5929,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5219372510910034,
|
|
"rewards/margins": 0.3483983874320984,
|
|
"rewards/rejected": -0.8703356385231018,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.9995000402231354e-05,
|
|
"logits/chosen": -2.564474105834961,
|
|
"logits/rejected": -2.510350227355957,
|
|
"logps/chosen": -219.81732177734375,
|
|
"logps/rejected": -202.83493041992188,
|
|
"loss": 0.7793,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.7294387817382812,
|
|
"rewards/margins": -0.11600115895271301,
|
|
"rewards/rejected": -0.6134375333786011,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999478992300466e-05,
|
|
"logits/chosen": -2.5275182723999023,
|
|
"logits/rejected": -2.6380527019500732,
|
|
"logps/chosen": -165.94358825683594,
|
|
"logps/rejected": -212.28086853027344,
|
|
"loss": 0.565,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4685615599155426,
|
|
"rewards/margins": 0.44902655482292175,
|
|
"rewards/rejected": -0.9175881147384644,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999457510460923e-05,
|
|
"logits/chosen": -2.407663106918335,
|
|
"logits/rejected": -2.5699005126953125,
|
|
"logps/chosen": -167.9168701171875,
|
|
"logps/rejected": -214.4752197265625,
|
|
"loss": 0.5801,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.48059117794036865,
|
|
"rewards/margins": 0.3546959161758423,
|
|
"rewards/rejected": -0.8352870345115662,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999435594708236e-05,
|
|
"logits/chosen": -2.3764097690582275,
|
|
"logits/rejected": -2.3675248622894287,
|
|
"logps/chosen": -201.25814819335938,
|
|
"logps/rejected": -229.6868133544922,
|
|
"loss": 0.6714,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6616516709327698,
|
|
"rewards/margins": 0.10846880823373795,
|
|
"rewards/rejected": -0.7701205611228943,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999413245046211e-05,
|
|
"logits/chosen": -2.5667271614074707,
|
|
"logits/rejected": -2.5309338569641113,
|
|
"logps/chosen": -210.30267333984375,
|
|
"logps/rejected": -206.71014404296875,
|
|
"loss": 0.5848,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.572422444820404,
|
|
"rewards/margins": 0.30542662739753723,
|
|
"rewards/rejected": -0.8778490424156189,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.9993904614787254e-05,
|
|
"logits/chosen": -2.463524341583252,
|
|
"logits/rejected": -2.530627965927124,
|
|
"logps/chosen": -179.8057861328125,
|
|
"logps/rejected": -200.56394958496094,
|
|
"loss": 0.6231,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6925724148750305,
|
|
"rewards/margins": 0.21750670671463013,
|
|
"rewards/rejected": -0.9100791215896606,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999367244009736e-05,
|
|
"logits/chosen": -2.4720635414123535,
|
|
"logits/rejected": -2.5370302200317383,
|
|
"logps/chosen": -194.29440307617188,
|
|
"logps/rejected": -196.02944946289062,
|
|
"loss": 0.8265,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -1.0657517910003662,
|
|
"rewards/margins": -0.16590073704719543,
|
|
"rewards/rejected": -0.8998512029647827,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999343592643274e-05,
|
|
"logits/chosen": -2.3978195190429688,
|
|
"logits/rejected": -2.397984266281128,
|
|
"logps/chosen": -229.4518585205078,
|
|
"logps/rejected": -217.56869506835938,
|
|
"loss": 0.6598,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6397801637649536,
|
|
"rewards/margins": 0.10495337843894958,
|
|
"rewards/rejected": -0.7447335720062256,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.999319507383444e-05,
|
|
"logits/chosen": -2.3354411125183105,
|
|
"logits/rejected": -2.4454195499420166,
|
|
"logps/chosen": -149.3455810546875,
|
|
"logps/rejected": -187.3743438720703,
|
|
"loss": 0.5981,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.7036041617393494,
|
|
"rewards/margins": 0.3835078775882721,
|
|
"rewards/rejected": -1.0871120691299438,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.999294988234428e-05,
|
|
"logits/chosen": -2.491671323776245,
|
|
"logits/rejected": -2.550612449645996,
|
|
"logps/chosen": -180.36758422851562,
|
|
"logps/rejected": -251.5354461669922,
|
|
"loss": 0.6038,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.7723766565322876,
|
|
"rewards/margins": 0.2752906382083893,
|
|
"rewards/rejected": -1.047667384147644,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.999270035200483e-05,
|
|
"logits/chosen": -2.492790937423706,
|
|
"logits/rejected": -2.365570306777954,
|
|
"logps/chosen": -195.1597900390625,
|
|
"logps/rejected": -185.96673583984375,
|
|
"loss": 0.7169,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -1.2230509519577026,
|
|
"rewards/margins": 0.14210942387580872,
|
|
"rewards/rejected": -1.3651604652404785,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.99924464828594e-05,
|
|
"logits/chosen": -2.412907838821411,
|
|
"logits/rejected": -2.508164882659912,
|
|
"logps/chosen": -196.51846313476562,
|
|
"logps/rejected": -218.0734405517578,
|
|
"loss": 0.5942,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.813502848148346,
|
|
"rewards/margins": 0.3059760332107544,
|
|
"rewards/rejected": -1.1194789409637451,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.9992188274952064e-05,
|
|
"logits/chosen": -2.4737277030944824,
|
|
"logits/rejected": -2.5720887184143066,
|
|
"logps/chosen": -155.16171264648438,
|
|
"logps/rejected": -195.86997985839844,
|
|
"loss": 0.7379,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.8073078989982605,
|
|
"rewards/margins": 0.11351939290761948,
|
|
"rewards/rejected": -0.920827329158783,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.999192572832765e-05,
|
|
"logits/chosen": -2.5303239822387695,
|
|
"logits/rejected": -2.5628015995025635,
|
|
"logps/chosen": -281.7694396972656,
|
|
"logps/rejected": -303.6089782714844,
|
|
"loss": 0.6735,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6562893986701965,
|
|
"rewards/margins": 0.12363787740468979,
|
|
"rewards/rejected": -0.7799273133277893,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.999165884303174e-05,
|
|
"logits/chosen": -2.481058359146118,
|
|
"logits/rejected": -2.4526753425598145,
|
|
"logps/chosen": -182.2716064453125,
|
|
"logps/rejected": -242.58828735351562,
|
|
"loss": 0.7317,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7687569260597229,
|
|
"rewards/margins": -0.007903970777988434,
|
|
"rewards/rejected": -0.7608529329299927,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.999138761911066e-05,
|
|
"logits/chosen": -2.459987163543701,
|
|
"logits/rejected": -2.5180883407592773,
|
|
"logps/chosen": -209.83102416992188,
|
|
"logps/rejected": -216.1147918701172,
|
|
"loss": 0.7265,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -1.0292317867279053,
|
|
"rewards/margins": 0.04909829795360565,
|
|
"rewards/rejected": -1.0783300399780273,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.99911120566115e-05,
|
|
"logits/chosen": -2.5340957641601562,
|
|
"logits/rejected": -2.654463052749634,
|
|
"logps/chosen": -210.88929748535156,
|
|
"logps/rejected": -198.64614868164062,
|
|
"loss": 0.7475,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -1.2450902462005615,
|
|
"rewards/margins": 0.012418277561664581,
|
|
"rewards/rejected": -1.2575085163116455,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.99908321555821e-05,
|
|
"logits/chosen": -2.4791364669799805,
|
|
"logits/rejected": -2.5890302658081055,
|
|
"logps/chosen": -192.5817108154297,
|
|
"logps/rejected": -224.9603271484375,
|
|
"loss": 0.6089,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.9010199308395386,
|
|
"rewards/margins": 0.4015497863292694,
|
|
"rewards/rejected": -1.3025696277618408,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.999054791607105e-05,
|
|
"logits/chosen": -2.428842067718506,
|
|
"logits/rejected": -2.372598171234131,
|
|
"logps/chosen": -221.89833068847656,
|
|
"logps/rejected": -212.29595947265625,
|
|
"loss": 0.7305,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -1.0515369176864624,
|
|
"rewards/margins": 0.017772257328033447,
|
|
"rewards/rejected": -1.0693092346191406,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.999025933812769e-05,
|
|
"logits/chosen": -2.50471830368042,
|
|
"logits/rejected": -2.5308945178985596,
|
|
"logps/chosen": -198.7537384033203,
|
|
"logps/rejected": -236.5478515625,
|
|
"loss": 0.6487,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.8205583095550537,
|
|
"rewards/margins": 0.18831993639469147,
|
|
"rewards/rejected": -1.008878231048584,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.9989966421802114e-05,
|
|
"logits/chosen": -2.410905122756958,
|
|
"logits/rejected": -2.5401611328125,
|
|
"logps/chosen": -202.42991638183594,
|
|
"logps/rejected": -219.6848602294922,
|
|
"loss": 0.6257,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.9180124402046204,
|
|
"rewards/margins": 0.24881067872047424,
|
|
"rewards/rejected": -1.166823148727417,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.998966916714519e-05,
|
|
"logits/chosen": -2.4587323665618896,
|
|
"logits/rejected": -2.5724194049835205,
|
|
"logps/chosen": -154.88470458984375,
|
|
"logps/rejected": -172.47268676757812,
|
|
"loss": 0.618,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7627564072608948,
|
|
"rewards/margins": 0.294120192527771,
|
|
"rewards/rejected": -1.0568766593933105,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.998936757420851e-05,
|
|
"logits/chosen": -2.284227132797241,
|
|
"logits/rejected": -2.380892515182495,
|
|
"logps/chosen": -138.282958984375,
|
|
"logps/rejected": -180.36441040039062,
|
|
"loss": 0.6089,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7448742389678955,
|
|
"rewards/margins": 0.4253446161746979,
|
|
"rewards/rejected": -1.1702189445495605,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.9989061643044434e-05,
|
|
"logits/chosen": -2.5783276557922363,
|
|
"logits/rejected": -2.5471560955047607,
|
|
"logps/chosen": -186.851318359375,
|
|
"logps/rejected": -193.73837280273438,
|
|
"loss": 0.7108,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -1.0091472864151,
|
|
"rewards/margins": 0.033400412648916245,
|
|
"rewards/rejected": -1.0425477027893066,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.9988751373706075e-05,
|
|
"logits/chosen": -2.398940086364746,
|
|
"logits/rejected": -2.3498027324676514,
|
|
"logps/chosen": -171.69107055664062,
|
|
"logps/rejected": -209.19424438476562,
|
|
"loss": 0.7045,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.9662235975265503,
|
|
"rewards/margins": 0.137020543217659,
|
|
"rewards/rejected": -1.1032441854476929,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.9988436766247284e-05,
|
|
"logits/chosen": -2.415847063064575,
|
|
"logits/rejected": -2.3855319023132324,
|
|
"logps/chosen": -228.1761932373047,
|
|
"logps/rejected": -195.01991271972656,
|
|
"loss": 0.6696,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.9869641661643982,
|
|
"rewards/margins": 0.11401443928480148,
|
|
"rewards/rejected": -1.1009787321090698,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.9988117820722704e-05,
|
|
"logits/chosen": -2.638780117034912,
|
|
"logits/rejected": -2.607104539871216,
|
|
"logps/chosen": -234.91183471679688,
|
|
"logps/rejected": -203.67996215820312,
|
|
"loss": 0.9279,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -1.2375922203063965,
|
|
"rewards/margins": -0.281325101852417,
|
|
"rewards/rejected": -0.956267237663269,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.998779453718768e-05,
|
|
"logits/chosen": -2.254124879837036,
|
|
"logits/rejected": -2.429471969604492,
|
|
"logps/chosen": -235.76698303222656,
|
|
"logps/rejected": -262.7959899902344,
|
|
"loss": 0.6238,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -1.0454847812652588,
|
|
"rewards/margins": 0.3030804395675659,
|
|
"rewards/rejected": -1.3485652208328247,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.9987466915698346e-05,
|
|
"logits/chosen": -2.276986598968506,
|
|
"logits/rejected": -2.337780475616455,
|
|
"logps/chosen": -138.3777313232422,
|
|
"logps/rejected": -166.30181884765625,
|
|
"loss": 0.5682,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.5320202708244324,
|
|
"rewards/margins": 0.315082311630249,
|
|
"rewards/rejected": -0.8471025228500366,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.998713495631156e-05,
|
|
"logits/chosen": -2.420923948287964,
|
|
"logits/rejected": -2.565662384033203,
|
|
"logps/chosen": -180.71446228027344,
|
|
"logps/rejected": -192.2035369873047,
|
|
"loss": 0.647,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.8985862731933594,
|
|
"rewards/margins": 0.2061445415019989,
|
|
"rewards/rejected": -1.1047309637069702,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.998679865908499e-05,
|
|
"logits/chosen": -2.5881500244140625,
|
|
"logits/rejected": -2.6503522396087646,
|
|
"logps/chosen": -183.07540893554688,
|
|
"logps/rejected": -233.943603515625,
|
|
"loss": 0.6889,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -1.1374621391296387,
|
|
"rewards/margins": 0.19141921401023865,
|
|
"rewards/rejected": -1.3288813829421997,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.9986458024076984e-05,
|
|
"logits/chosen": -2.4952712059020996,
|
|
"logits/rejected": -2.50299072265625,
|
|
"logps/chosen": -192.31314086914062,
|
|
"logps/rejected": -197.241455078125,
|
|
"loss": 0.7729,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -1.13700532913208,
|
|
"rewards/margins": -0.10435198992490768,
|
|
"rewards/rejected": -1.0326533317565918,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.998611305134669e-05,
|
|
"logits/chosen": -2.5952248573303223,
|
|
"logits/rejected": -2.600719928741455,
|
|
"logps/chosen": -191.4530487060547,
|
|
"logps/rejected": -212.82955932617188,
|
|
"loss": 0.674,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.9059873819351196,
|
|
"rewards/margins": 0.1599172055721283,
|
|
"rewards/rejected": -1.0659046173095703,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.9985763740954e-05,
|
|
"logits/chosen": -2.4821391105651855,
|
|
"logits/rejected": -2.4574878215789795,
|
|
"logps/chosen": -264.071533203125,
|
|
"logps/rejected": -256.51611328125,
|
|
"loss": 0.7333,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -1.0796548128128052,
|
|
"rewards/margins": 0.06612614542245865,
|
|
"rewards/rejected": -1.1457810401916504,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.9985410092959553e-05,
|
|
"logits/chosen": -2.4601786136627197,
|
|
"logits/rejected": -2.519011974334717,
|
|
"logps/chosen": -158.57687377929688,
|
|
"logps/rejected": -173.29666137695312,
|
|
"loss": 0.7693,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.7817510366439819,
|
|
"rewards/margins": -0.05254516005516052,
|
|
"rewards/rejected": -0.729205846786499,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.998505210742472e-05,
|
|
"logits/chosen": -2.6099700927734375,
|
|
"logits/rejected": -2.58197021484375,
|
|
"logps/chosen": -195.2617950439453,
|
|
"logps/rejected": -188.54165649414062,
|
|
"loss": 0.9197,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -1.1366631984710693,
|
|
"rewards/margins": -0.32979804277420044,
|
|
"rewards/rejected": -0.8068650960922241,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.9984689784411686e-05,
|
|
"logits/chosen": -2.679287910461426,
|
|
"logits/rejected": -2.5679209232330322,
|
|
"logps/chosen": -211.75323486328125,
|
|
"logps/rejected": -245.02603149414062,
|
|
"loss": 0.6702,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.9691303372383118,
|
|
"rewards/margins": 0.17891569435596466,
|
|
"rewards/rejected": -1.1480460166931152,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.9984323123983334e-05,
|
|
"logits/chosen": -2.537733316421509,
|
|
"logits/rejected": -2.6055424213409424,
|
|
"logps/chosen": -151.74288940429688,
|
|
"logps/rejected": -250.8179168701172,
|
|
"loss": 0.6366,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.928146243095398,
|
|
"rewards/margins": 0.24633805453777313,
|
|
"rewards/rejected": -1.174484372138977,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.998395212620332e-05,
|
|
"logits/chosen": -2.4368271827697754,
|
|
"logits/rejected": -2.529536247253418,
|
|
"logps/chosen": -207.87677001953125,
|
|
"logps/rejected": -229.0767822265625,
|
|
"loss": 0.6199,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7506311535835266,
|
|
"rewards/margins": 0.24917976558208466,
|
|
"rewards/rejected": -0.9998109340667725,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.998357679113603e-05,
|
|
"logits/chosen": -2.5217807292938232,
|
|
"logits/rejected": -2.6954245567321777,
|
|
"logps/chosen": -192.20407104492188,
|
|
"logps/rejected": -200.1387176513672,
|
|
"loss": 0.6013,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6829215288162231,
|
|
"rewards/margins": 0.2588457763195038,
|
|
"rewards/rejected": -0.9417673349380493,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.9983197118846655e-05,
|
|
"logits/chosen": -2.4711546897888184,
|
|
"logits/rejected": -2.4839439392089844,
|
|
"logps/chosen": -160.9691925048828,
|
|
"logps/rejected": -167.21469116210938,
|
|
"loss": 0.7301,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7170782685279846,
|
|
"rewards/margins": 0.04331670328974724,
|
|
"rewards/rejected": -0.7603949904441833,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.9982813109401096e-05,
|
|
"logits/chosen": -2.3429412841796875,
|
|
"logits/rejected": -2.3843770027160645,
|
|
"logps/chosen": -221.22274780273438,
|
|
"logps/rejected": -246.99005126953125,
|
|
"loss": 0.7711,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -1.114179015159607,
|
|
"rewards/margins": -0.07230883836746216,
|
|
"rewards/rejected": -1.0418701171875,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.998242476286601e-05,
|
|
"logits/chosen": -2.7864720821380615,
|
|
"logits/rejected": -2.7745487689971924,
|
|
"logps/chosen": -194.08201599121094,
|
|
"logps/rejected": -226.28591918945312,
|
|
"loss": 0.71,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.8893502354621887,
|
|
"rewards/margins": 0.06053268164396286,
|
|
"rewards/rejected": -0.949882984161377,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.998203207930882e-05,
|
|
"logits/chosen": -2.6679065227508545,
|
|
"logits/rejected": -2.711076498031616,
|
|
"logps/chosen": -190.53013610839844,
|
|
"logps/rejected": -200.187744140625,
|
|
"loss": 0.8201,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -1.0141409635543823,
|
|
"rewards/margins": -0.17862781882286072,
|
|
"rewards/rejected": -0.8355131149291992,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.998163505879769e-05,
|
|
"logits/chosen": -2.7404122352600098,
|
|
"logits/rejected": -2.732025146484375,
|
|
"logps/chosen": -178.48135375976562,
|
|
"logps/rejected": -183.08074951171875,
|
|
"loss": 0.7078,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.8886659145355225,
|
|
"rewards/margins": 0.043551601469516754,
|
|
"rewards/rejected": -0.9322174787521362,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.998123370140156e-05,
|
|
"logits/chosen": -2.6989643573760986,
|
|
"logits/rejected": -2.7124342918395996,
|
|
"logps/chosen": -223.03314208984375,
|
|
"logps/rejected": -221.12120056152344,
|
|
"loss": 0.6559,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.764815628528595,
|
|
"rewards/margins": 0.1835068166255951,
|
|
"rewards/rejected": -0.9483224153518677,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.99808280071901e-05,
|
|
"logits/chosen": -2.618727684020996,
|
|
"logits/rejected": -2.6078970432281494,
|
|
"logps/chosen": -176.1873779296875,
|
|
"logps/rejected": -199.52859497070312,
|
|
"loss": 0.7406,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.8613646626472473,
|
|
"rewards/margins": -0.05548467859625816,
|
|
"rewards/rejected": -0.805880069732666,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.9980417976233735e-05,
|
|
"logits/chosen": -2.6192498207092285,
|
|
"logits/rejected": -2.732285261154175,
|
|
"logps/chosen": -211.3184814453125,
|
|
"logps/rejected": -195.73056030273438,
|
|
"loss": 0.7921,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.9030998945236206,
|
|
"rewards/margins": -0.1349114179611206,
|
|
"rewards/rejected": -0.7681884765625,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.9980003608603656e-05,
|
|
"logits/chosen": -2.5657694339752197,
|
|
"logits/rejected": -2.5230712890625,
|
|
"logps/chosen": -203.0819091796875,
|
|
"logps/rejected": -226.5617218017578,
|
|
"loss": 0.6618,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.609666109085083,
|
|
"rewards/margins": 0.08627453446388245,
|
|
"rewards/rejected": -0.6959406137466431,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.997958490437178e-05,
|
|
"logits/chosen": -2.641559362411499,
|
|
"logits/rejected": -2.76076602935791,
|
|
"logps/chosen": -169.1241455078125,
|
|
"logps/rejected": -183.4690399169922,
|
|
"loss": 0.7612,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7050959467887878,
|
|
"rewards/margins": -0.03217136114835739,
|
|
"rewards/rejected": -0.6729245781898499,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.9979161863610816e-05,
|
|
"logits/chosen": -2.441647529602051,
|
|
"logits/rejected": -2.467510223388672,
|
|
"logps/chosen": -143.13748168945312,
|
|
"logps/rejected": -172.0884246826172,
|
|
"loss": 0.6647,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5210601091384888,
|
|
"rewards/margins": 0.14004816114902496,
|
|
"rewards/rejected": -0.6611082553863525,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.99787344863942e-05,
|
|
"logits/chosen": -2.627023458480835,
|
|
"logits/rejected": -2.5984585285186768,
|
|
"logps/chosen": -203.31980895996094,
|
|
"logps/rejected": -218.84060668945312,
|
|
"loss": 0.6172,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5382375717163086,
|
|
"rewards/margins": 0.20881405472755432,
|
|
"rewards/rejected": -0.7470515966415405,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.997830277279612e-05,
|
|
"logits/chosen": -2.595003604888916,
|
|
"logits/rejected": -2.722808361053467,
|
|
"logps/chosen": -223.01495361328125,
|
|
"logps/rejected": -240.90130615234375,
|
|
"loss": 0.7333,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7480457425117493,
|
|
"rewards/margins": -0.029575012624263763,
|
|
"rewards/rejected": -0.7184706926345825,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.997786672289152e-05,
|
|
"logits/chosen": -2.6220602989196777,
|
|
"logits/rejected": -2.6606202125549316,
|
|
"logps/chosen": -204.98797607421875,
|
|
"logps/rejected": -206.25839233398438,
|
|
"loss": 0.7368,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.6979295611381531,
|
|
"rewards/margins": -0.07002773880958557,
|
|
"rewards/rejected": -0.6279018521308899,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.997742633675612e-05,
|
|
"logits/chosen": -2.547811508178711,
|
|
"logits/rejected": -2.551816463470459,
|
|
"logps/chosen": -196.38748168945312,
|
|
"logps/rejected": -222.4247589111328,
|
|
"loss": 0.6402,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5409201383590698,
|
|
"rewards/margins": 0.19667872786521912,
|
|
"rewards/rejected": -0.7375988364219666,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.9976981614466344e-05,
|
|
"logits/chosen": -2.2461767196655273,
|
|
"logits/rejected": -2.2968175411224365,
|
|
"logps/chosen": -185.43954467773438,
|
|
"logps/rejected": -196.11392211914062,
|
|
"loss": 0.7743,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5852402448654175,
|
|
"rewards/margins": -0.08380623161792755,
|
|
"rewards/rejected": -0.5014340281486511,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.997653255609942e-05,
|
|
"logits/chosen": -2.4160966873168945,
|
|
"logits/rejected": -2.389375925064087,
|
|
"logps/chosen": -189.16510009765625,
|
|
"logps/rejected": -202.54649353027344,
|
|
"loss": 0.7306,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.729293167591095,
|
|
"rewards/margins": -0.006243497133255005,
|
|
"rewards/rejected": -0.7230496406555176,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.997607916173329e-05,
|
|
"logits/chosen": -2.6787378787994385,
|
|
"logits/rejected": -2.6836445331573486,
|
|
"logps/chosen": -177.60980224609375,
|
|
"logps/rejected": -177.10487365722656,
|
|
"loss": 0.765,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5183064341545105,
|
|
"rewards/margins": -0.11038734763860703,
|
|
"rewards/rejected": -0.4079190790653229,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.997562143144668e-05,
|
|
"logits/chosen": -2.559563398361206,
|
|
"logits/rejected": -2.5205612182617188,
|
|
"logps/chosen": -208.58892822265625,
|
|
"logps/rejected": -248.25648498535156,
|
|
"loss": 0.6524,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5725008845329285,
|
|
"rewards/margins": 0.11152593046426773,
|
|
"rewards/rejected": -0.6840267777442932,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.997515936531903e-05,
|
|
"logits/chosen": -2.6034438610076904,
|
|
"logits/rejected": -2.528724193572998,
|
|
"logps/chosen": -159.13522338867188,
|
|
"logps/rejected": -135.77056884765625,
|
|
"loss": 0.7413,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5185827016830444,
|
|
"rewards/margins": -0.04235066473484039,
|
|
"rewards/rejected": -0.47623202204704285,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.9974692963430595e-05,
|
|
"logits/chosen": -2.3184618949890137,
|
|
"logits/rejected": -2.3163304328918457,
|
|
"logps/chosen": -130.70643615722656,
|
|
"logps/rejected": -142.0390625,
|
|
"loss": 0.7362,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6057229042053223,
|
|
"rewards/margins": -0.014669769443571568,
|
|
"rewards/rejected": -0.5910531282424927,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.99742222258623e-05,
|
|
"logits/chosen": -2.660839796066284,
|
|
"logits/rejected": -2.6470634937286377,
|
|
"logps/chosen": -221.08424377441406,
|
|
"logps/rejected": -213.25003051757812,
|
|
"loss": 0.6922,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.5536531209945679,
|
|
"rewards/margins": 0.022671688348054886,
|
|
"rewards/rejected": -0.5763247609138489,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.997374715269589e-05,
|
|
"logits/chosen": -2.2223334312438965,
|
|
"logits/rejected": -2.2852203845977783,
|
|
"logps/chosen": -172.71421813964844,
|
|
"logps/rejected": -206.3236541748047,
|
|
"loss": 0.6722,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5476804375648499,
|
|
"rewards/margins": 0.10758630931377411,
|
|
"rewards/rejected": -0.6552667617797852,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.997326774401383e-05,
|
|
"logits/chosen": -2.262892961502075,
|
|
"logits/rejected": -2.343332052230835,
|
|
"logps/chosen": -152.0110626220703,
|
|
"logps/rejected": -178.68345642089844,
|
|
"loss": 0.694,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.48357081413269043,
|
|
"rewards/margins": 0.07582694292068481,
|
|
"rewards/rejected": -0.55939781665802,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.9972783999899366e-05,
|
|
"logits/chosen": -2.2378625869750977,
|
|
"logits/rejected": -2.2833354473114014,
|
|
"logps/chosen": -154.71405029296875,
|
|
"logps/rejected": -177.74502563476562,
|
|
"loss": 0.709,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.5034060478210449,
|
|
"rewards/margins": -0.0023919548839330673,
|
|
"rewards/rejected": -0.5010141134262085,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.997229592043645e-05,
|
|
"logits/chosen": -2.4182140827178955,
|
|
"logits/rejected": -2.477410078048706,
|
|
"logps/chosen": -217.6938018798828,
|
|
"logps/rejected": -236.60140991210938,
|
|
"loss": 0.7474,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.42065930366516113,
|
|
"rewards/margins": -0.08166539669036865,
|
|
"rewards/rejected": -0.33899393677711487,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.997180350570984e-05,
|
|
"logits/chosen": -2.6971006393432617,
|
|
"logits/rejected": -2.709580659866333,
|
|
"logps/chosen": -231.80714416503906,
|
|
"logps/rejected": -234.41290283203125,
|
|
"loss": 0.6593,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4388115406036377,
|
|
"rewards/margins": 0.1023733913898468,
|
|
"rewards/rejected": -0.5411849617958069,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.9971306755804995e-05,
|
|
"logits/chosen": -2.600799083709717,
|
|
"logits/rejected": -2.5639290809631348,
|
|
"logps/chosen": -211.75051879882812,
|
|
"logps/rejected": -185.4350128173828,
|
|
"loss": 0.6654,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3958936929702759,
|
|
"rewards/margins": 0.08459535241127014,
|
|
"rewards/rejected": -0.4804890751838684,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.997080567080817e-05,
|
|
"logits/chosen": -2.2965052127838135,
|
|
"logits/rejected": -2.3450510501861572,
|
|
"logps/chosen": -170.9476318359375,
|
|
"logps/rejected": -174.3068389892578,
|
|
"loss": 0.6604,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5262787938117981,
|
|
"rewards/margins": 0.09660908579826355,
|
|
"rewards/rejected": -0.6228878498077393,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.9970300250806346e-05,
|
|
"logits/chosen": -2.5405097007751465,
|
|
"logits/rejected": -2.5496726036071777,
|
|
"logps/chosen": -266.0992736816406,
|
|
"logps/rejected": -260.38275146484375,
|
|
"loss": 0.7048,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5492247939109802,
|
|
"rewards/margins": 0.042401984333992004,
|
|
"rewards/rejected": -0.591626763343811,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.996979049588727e-05,
|
|
"logits/chosen": -2.502474308013916,
|
|
"logits/rejected": -2.439131736755371,
|
|
"logps/chosen": -227.49476623535156,
|
|
"logps/rejected": -206.38299560546875,
|
|
"loss": 0.711,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5280985236167908,
|
|
"rewards/margins": 0.010388553142547607,
|
|
"rewards/rejected": -0.5384870767593384,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.996927640613944e-05,
|
|
"logits/chosen": -2.3517448902130127,
|
|
"logits/rejected": -2.5404696464538574,
|
|
"logps/chosen": -162.60267639160156,
|
|
"logps/rejected": -216.92562866210938,
|
|
"loss": 0.6095,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.2759331464767456,
|
|
"rewards/margins": 0.19823111593723297,
|
|
"rewards/rejected": -0.474164217710495,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.99687579816521e-05,
|
|
"logits/chosen": -2.411978006362915,
|
|
"logits/rejected": -2.5019097328186035,
|
|
"logps/chosen": -201.91445922851562,
|
|
"logps/rejected": -211.31448364257812,
|
|
"loss": 0.6797,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4115257263183594,
|
|
"rewards/margins": 0.07280793786048889,
|
|
"rewards/rejected": -0.48433366417884827,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.9968235222515246e-05,
|
|
"logits/chosen": -2.4315428733825684,
|
|
"logits/rejected": -2.4550862312316895,
|
|
"logps/chosen": -162.66555786132812,
|
|
"logps/rejected": -156.19947814941406,
|
|
"loss": 0.6482,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.36304140090942383,
|
|
"rewards/margins": 0.12375527620315552,
|
|
"rewards/rejected": -0.48679661750793457,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.996770812881964e-05,
|
|
"logits/chosen": -2.5614113807678223,
|
|
"logits/rejected": -2.5391292572021484,
|
|
"logps/chosen": -214.3634033203125,
|
|
"logps/rejected": -209.9477996826172,
|
|
"loss": 0.7018,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.319223552942276,
|
|
"rewards/margins": 0.018056731671094894,
|
|
"rewards/rejected": -0.3372803032398224,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.9967176700656776e-05,
|
|
"logits/chosen": -2.438467264175415,
|
|
"logits/rejected": -2.4638776779174805,
|
|
"logps/chosen": -227.1153564453125,
|
|
"logps/rejected": -219.0038299560547,
|
|
"loss": 0.6209,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.45765751600265503,
|
|
"rewards/margins": 0.19506625831127167,
|
|
"rewards/rejected": -0.6527237892150879,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.996664093811892e-05,
|
|
"logits/chosen": -2.5968563556671143,
|
|
"logits/rejected": -2.5226142406463623,
|
|
"logps/chosen": -184.9003143310547,
|
|
"logps/rejected": -202.8119354248047,
|
|
"loss": 0.5973,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.36458614468574524,
|
|
"rewards/margins": 0.22884023189544678,
|
|
"rewards/rejected": -0.5934264063835144,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.996610084129908e-05,
|
|
"logits/chosen": -2.356076955795288,
|
|
"logits/rejected": -2.464617967605591,
|
|
"logps/chosen": -233.56741333007812,
|
|
"logps/rejected": -252.82614135742188,
|
|
"loss": 0.7174,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.475138783454895,
|
|
"rewards/margins": 0.015569500625133514,
|
|
"rewards/rejected": -0.49070829153060913,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996555641029101e-05,
|
|
"logits/chosen": -2.4807751178741455,
|
|
"logits/rejected": -2.5052688121795654,
|
|
"logps/chosen": -218.75250244140625,
|
|
"logps/rejected": -202.75149536132812,
|
|
"loss": 0.7541,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5363876819610596,
|
|
"rewards/margins": -0.06761842221021652,
|
|
"rewards/rejected": -0.4687691926956177,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996500764518923e-05,
|
|
"logits/chosen": -2.5997631549835205,
|
|
"logits/rejected": -2.6119892597198486,
|
|
"logps/chosen": -212.090087890625,
|
|
"logps/rejected": -220.1112823486328,
|
|
"loss": 0.7185,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.5663864016532898,
|
|
"rewards/margins": -0.016858436167240143,
|
|
"rewards/rejected": -0.5495280027389526,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.9964454546089026e-05,
|
|
"logits/chosen": -2.25441837310791,
|
|
"logits/rejected": -2.251708507537842,
|
|
"logps/chosen": -216.14996337890625,
|
|
"logps/rejected": -218.1949005126953,
|
|
"loss": 0.7617,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.46758705377578735,
|
|
"rewards/margins": -0.09628176689147949,
|
|
"rewards/rejected": -0.37130531668663025,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996389711308639e-05,
|
|
"logits/chosen": -2.475829601287842,
|
|
"logits/rejected": -2.4804282188415527,
|
|
"logps/chosen": -197.95965576171875,
|
|
"logps/rejected": -227.29605102539062,
|
|
"loss": 0.6275,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2889898717403412,
|
|
"rewards/margins": 0.18446122109889984,
|
|
"rewards/rejected": -0.47345107793807983,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996333534627809e-05,
|
|
"logits/chosen": -2.4503121376037598,
|
|
"logits/rejected": -2.40040922164917,
|
|
"logps/chosen": -190.1997833251953,
|
|
"logps/rejected": -170.68679809570312,
|
|
"loss": 0.7168,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4417100250720978,
|
|
"rewards/margins": -0.0062090009450912476,
|
|
"rewards/rejected": -0.4355010688304901,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996276924576169e-05,
|
|
"logits/chosen": -2.5104312896728516,
|
|
"logits/rejected": -2.5510809421539307,
|
|
"logps/chosen": -134.1800079345703,
|
|
"logps/rejected": -153.13365173339844,
|
|
"loss": 0.7468,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.42858752608299255,
|
|
"rewards/margins": -0.08554935455322266,
|
|
"rewards/rejected": -0.3430381715297699,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996219881163543e-05,
|
|
"logits/chosen": -2.456216335296631,
|
|
"logits/rejected": -2.4285402297973633,
|
|
"logps/chosen": -209.37985229492188,
|
|
"logps/rejected": -195.06568908691406,
|
|
"loss": 0.7159,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.46927523612976074,
|
|
"rewards/margins": -0.0166518222540617,
|
|
"rewards/rejected": -0.4526233971118927,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.996162404399835e-05,
|
|
"logits/chosen": -2.450845956802368,
|
|
"logits/rejected": -2.4209864139556885,
|
|
"logps/chosen": -198.9377899169922,
|
|
"logps/rejected": -203.7089385986328,
|
|
"loss": 0.6708,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4800047278404236,
|
|
"rewards/margins": 0.09784451127052307,
|
|
"rewards/rejected": -0.5778492093086243,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.996104494295024e-05,
|
|
"logits/chosen": -2.5744130611419678,
|
|
"logits/rejected": -2.641494035720825,
|
|
"logps/chosen": -182.59878540039062,
|
|
"logps/rejected": -207.31797790527344,
|
|
"loss": 0.6749,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4756154417991638,
|
|
"rewards/margins": 0.08592241257429123,
|
|
"rewards/rejected": -0.5615378618240356,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.996046150859161e-05,
|
|
"logits/chosen": -2.0956687927246094,
|
|
"logits/rejected": -2.1018872261047363,
|
|
"logps/chosen": -200.86215209960938,
|
|
"logps/rejected": -221.92138671875,
|
|
"loss": 0.6445,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.488430380821228,
|
|
"rewards/margins": 0.1436658799648285,
|
|
"rewards/rejected": -0.6320962905883789,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.9959873741023774e-05,
|
|
"logits/chosen": -2.3680927753448486,
|
|
"logits/rejected": -2.415205240249634,
|
|
"logps/chosen": -163.829833984375,
|
|
"logps/rejected": -182.74481201171875,
|
|
"loss": 0.6839,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.35495901107788086,
|
|
"rewards/margins": 0.03828255832195282,
|
|
"rewards/rejected": -0.3932415544986725,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.995928164034876e-05,
|
|
"logits/chosen": -2.5983476638793945,
|
|
"logits/rejected": -2.7171671390533447,
|
|
"logps/chosen": -189.3108673095703,
|
|
"logps/rejected": -195.94888305664062,
|
|
"loss": 0.6611,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4241011142730713,
|
|
"rewards/margins": 0.0877021998167038,
|
|
"rewards/rejected": -0.5118032693862915,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.995868520666936e-05,
|
|
"logits/chosen": -2.510660171508789,
|
|
"logits/rejected": -2.5224432945251465,
|
|
"logps/chosen": -227.51687622070312,
|
|
"logps/rejected": -231.05897521972656,
|
|
"loss": 0.7269,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.40695255994796753,
|
|
"rewards/margins": -0.046068765223026276,
|
|
"rewards/rejected": -0.36088380217552185,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.9958084440089095e-05,
|
|
"logits/chosen": -2.256051540374756,
|
|
"logits/rejected": -2.2608485221862793,
|
|
"logps/chosen": -209.19252014160156,
|
|
"logps/rejected": -206.2628631591797,
|
|
"loss": 0.6338,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3002999424934387,
|
|
"rewards/margins": 0.18500341475009918,
|
|
"rewards/rejected": -0.4853033423423767,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.995747934071229e-05,
|
|
"logits/chosen": -2.8373260498046875,
|
|
"logits/rejected": -2.8713366985321045,
|
|
"logps/chosen": -186.55723571777344,
|
|
"logps/rejected": -221.16917419433594,
|
|
"loss": 0.6489,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3408931493759155,
|
|
"rewards/margins": 0.18887601792812347,
|
|
"rewards/rejected": -0.5297691822052002,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995686990864398e-05,
|
|
"logits/chosen": -2.328822135925293,
|
|
"logits/rejected": -2.4501209259033203,
|
|
"logps/chosen": -164.64404296875,
|
|
"logps/rejected": -164.53396606445312,
|
|
"loss": 0.6872,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5059252977371216,
|
|
"rewards/margins": 0.046310462057590485,
|
|
"rewards/rejected": -0.5522357225418091,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995625614398996e-05,
|
|
"logits/chosen": -2.254148006439209,
|
|
"logits/rejected": -2.45564603805542,
|
|
"logps/chosen": -215.61080932617188,
|
|
"logps/rejected": -236.83526611328125,
|
|
"loss": 0.618,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3230094313621521,
|
|
"rewards/margins": 0.2154739797115326,
|
|
"rewards/rejected": -0.5384833812713623,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995563804685678e-05,
|
|
"logits/chosen": -2.4298973083496094,
|
|
"logits/rejected": -2.6152379512786865,
|
|
"logps/chosen": -171.45925903320312,
|
|
"logps/rejected": -215.27342224121094,
|
|
"loss": 0.6461,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5956869721412659,
|
|
"rewards/margins": 0.1696903109550476,
|
|
"rewards/rejected": -0.7653772830963135,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995501561735176e-05,
|
|
"logits/chosen": -2.5064899921417236,
|
|
"logits/rejected": -2.667478561401367,
|
|
"logps/chosen": -201.28167724609375,
|
|
"logps/rejected": -200.6697998046875,
|
|
"loss": 0.7252,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.50804603099823,
|
|
"rewards/margins": -0.021207518875598907,
|
|
"rewards/rejected": -0.48683851957321167,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995438885558294e-05,
|
|
"logits/chosen": -2.575934886932373,
|
|
"logits/rejected": -2.5915310382843018,
|
|
"logps/chosen": -241.2596893310547,
|
|
"logps/rejected": -236.16864013671875,
|
|
"loss": 0.6113,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.41618824005126953,
|
|
"rewards/margins": 0.21841469407081604,
|
|
"rewards/rejected": -0.6346028447151184,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995375776165913e-05,
|
|
"logits/chosen": -2.6145882606506348,
|
|
"logits/rejected": -2.6391665935516357,
|
|
"logps/chosen": -230.53524780273438,
|
|
"logps/rejected": -228.5028076171875,
|
|
"loss": 0.7431,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5892909169197083,
|
|
"rewards/margins": -0.044729601591825485,
|
|
"rewards/rejected": -0.5445613265037537,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.995312233568989e-05,
|
|
"logits/chosen": -2.279015302658081,
|
|
"logits/rejected": -2.4627904891967773,
|
|
"logps/chosen": -205.08226013183594,
|
|
"logps/rejected": -245.180908203125,
|
|
"loss": 0.6492,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.37771591544151306,
|
|
"rewards/margins": 0.17377406358718872,
|
|
"rewards/rejected": -0.5514900088310242,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.9952482577785545e-05,
|
|
"logits/chosen": -2.4535770416259766,
|
|
"logits/rejected": -2.5108795166015625,
|
|
"logps/chosen": -186.94288635253906,
|
|
"logps/rejected": -191.04786682128906,
|
|
"loss": 0.7335,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6674529314041138,
|
|
"rewards/margins": -0.03079444169998169,
|
|
"rewards/rejected": -0.6366585493087769,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.9951838488057134e-05,
|
|
"logits/chosen": -2.2177467346191406,
|
|
"logits/rejected": -2.3985226154327393,
|
|
"logps/chosen": -151.24977111816406,
|
|
"logps/rejected": -185.8676300048828,
|
|
"loss": 0.6516,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3900182843208313,
|
|
"rewards/margins": 0.13504652678966522,
|
|
"rewards/rejected": -0.5250648260116577,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.9951190066616495e-05,
|
|
"logits/chosen": -2.42427396774292,
|
|
"logits/rejected": -2.4583323001861572,
|
|
"logps/chosen": -255.53150939941406,
|
|
"logps/rejected": -261.932373046875,
|
|
"loss": 0.6442,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3976048231124878,
|
|
"rewards/margins": 0.12972235679626465,
|
|
"rewards/rejected": -0.5273271799087524,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.995053731357618e-05,
|
|
"logits/chosen": -2.462376832962036,
|
|
"logits/rejected": -2.568171262741089,
|
|
"logps/chosen": -197.1337432861328,
|
|
"logps/rejected": -202.1964569091797,
|
|
"loss": 0.6419,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6016325354576111,
|
|
"rewards/margins": 0.15742458403110504,
|
|
"rewards/rejected": -0.7590572237968445,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.9949880229049526e-05,
|
|
"logits/chosen": -2.4696521759033203,
|
|
"logits/rejected": -2.4089298248291016,
|
|
"logps/chosen": -187.97976684570312,
|
|
"logps/rejected": -178.5655059814453,
|
|
"loss": 0.7341,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.43453776836395264,
|
|
"rewards/margins": -0.025840837508440018,
|
|
"rewards/rejected": -0.4086969494819641,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.994921881315059e-05,
|
|
"logits/chosen": -2.292558431625366,
|
|
"logits/rejected": -2.199901580810547,
|
|
"logps/chosen": -192.6279754638672,
|
|
"logps/rejected": -203.34487915039062,
|
|
"loss": 0.6863,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.48495134711265564,
|
|
"rewards/margins": 0.07142490148544312,
|
|
"rewards/rejected": -0.5563762784004211,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.9948553065994197e-05,
|
|
"logits/chosen": -2.231633424758911,
|
|
"logits/rejected": -2.2734503746032715,
|
|
"logps/chosen": -184.4136962890625,
|
|
"logps/rejected": -194.4268341064453,
|
|
"loss": 0.6465,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.44051364064216614,
|
|
"rewards/margins": 0.19739174842834473,
|
|
"rewards/rejected": -0.6379053592681885,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.994788298769593e-05,
|
|
"logits/chosen": -2.5150344371795654,
|
|
"logits/rejected": -2.4598844051361084,
|
|
"logps/chosen": -238.438232421875,
|
|
"logps/rejected": -217.13897705078125,
|
|
"loss": 0.6945,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.43908870220184326,
|
|
"rewards/margins": 0.03722059726715088,
|
|
"rewards/rejected": -0.47630926966667175,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.994720857837211e-05,
|
|
"logits/chosen": -2.2310400009155273,
|
|
"logits/rejected": -2.1882810592651367,
|
|
"logps/chosen": -184.02981567382812,
|
|
"logps/rejected": -208.0157470703125,
|
|
"loss": 0.666,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5550325512886047,
|
|
"rewards/margins": 0.08377734571695328,
|
|
"rewards/rejected": -0.638809859752655,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.994652983813982e-05,
|
|
"logits/chosen": -2.4377846717834473,
|
|
"logits/rejected": -2.5074000358581543,
|
|
"logps/chosen": -187.0503387451172,
|
|
"logps/rejected": -195.54049682617188,
|
|
"loss": 0.5873,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5530596971511841,
|
|
"rewards/margins": 0.2584923803806305,
|
|
"rewards/rejected": -0.811552107334137,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.994584676711689e-05,
|
|
"logits/chosen": -2.464853286743164,
|
|
"logits/rejected": -2.504178047180176,
|
|
"logps/chosen": -186.5401611328125,
|
|
"logps/rejected": -227.3590087890625,
|
|
"loss": 0.7102,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5265690088272095,
|
|
"rewards/margins": 0.009033482521772385,
|
|
"rewards/rejected": -0.5356025695800781,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.994515936542191e-05,
|
|
"logits/chosen": -2.2838244438171387,
|
|
"logits/rejected": -2.3256828784942627,
|
|
"logps/chosen": -173.54270935058594,
|
|
"logps/rejected": -166.37547302246094,
|
|
"loss": 0.7262,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4617432951927185,
|
|
"rewards/margins": -0.03625304624438286,
|
|
"rewards/rejected": -0.42549026012420654,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.99444676331742e-05,
|
|
"logits/chosen": -2.407790422439575,
|
|
"logits/rejected": -2.3871612548828125,
|
|
"logps/chosen": -248.3179931640625,
|
|
"logps/rejected": -235.79884338378906,
|
|
"loss": 0.6761,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.458143413066864,
|
|
"rewards/margins": 0.10666979104280472,
|
|
"rewards/rejected": -0.5648132562637329,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.9943771570493856e-05,
|
|
"logits/chosen": -2.560073137283325,
|
|
"logits/rejected": -2.51522159576416,
|
|
"logps/chosen": -191.05734252929688,
|
|
"logps/rejected": -195.287841796875,
|
|
"loss": 0.7288,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6246404051780701,
|
|
"rewards/margins": -0.01798657327890396,
|
|
"rewards/rejected": -0.6066538095474243,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.9943071177501724e-05,
|
|
"logits/chosen": -2.364337682723999,
|
|
"logits/rejected": -2.39058780670166,
|
|
"logps/chosen": -200.97915649414062,
|
|
"logps/rejected": -183.07151794433594,
|
|
"loss": 0.6839,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.5552209615707397,
|
|
"rewards/margins": 0.09874990582466125,
|
|
"rewards/rejected": -0.6539708375930786,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.994236645431938e-05,
|
|
"logits/chosen": -2.6555843353271484,
|
|
"logits/rejected": -2.7670748233795166,
|
|
"logps/chosen": -244.5951690673828,
|
|
"logps/rejected": -246.04124450683594,
|
|
"loss": 0.688,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.45607852935791016,
|
|
"rewards/margins": 0.08103760331869125,
|
|
"rewards/rejected": -0.5371161103248596,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.994165740106918e-05,
|
|
"logits/chosen": -2.527970314025879,
|
|
"logits/rejected": -2.6416518688201904,
|
|
"logps/chosen": -231.80966186523438,
|
|
"logps/rejected": -246.3621826171875,
|
|
"loss": 0.6086,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.5444769263267517,
|
|
"rewards/margins": 0.21878674626350403,
|
|
"rewards/rejected": -0.7632635831832886,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.99409440178742e-05,
|
|
"logits/chosen": -2.584487199783325,
|
|
"logits/rejected": -2.5451977252960205,
|
|
"logps/chosen": -184.02330017089844,
|
|
"logps/rejected": -187.47161865234375,
|
|
"loss": 0.7151,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6932749152183533,
|
|
"rewards/margins": -0.005857221782207489,
|
|
"rewards/rejected": -0.6874176263809204,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.9940226304858296e-05,
|
|
"logits/chosen": -2.556795597076416,
|
|
"logits/rejected": -2.5516929626464844,
|
|
"logps/chosen": -187.25588989257812,
|
|
"logps/rejected": -206.25067138671875,
|
|
"loss": 0.7035,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.563089907169342,
|
|
"rewards/margins": 0.009414192289113998,
|
|
"rewards/rejected": -0.5725040435791016,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.993950426214606e-05,
|
|
"logits/chosen": -2.3932271003723145,
|
|
"logits/rejected": -2.393653154373169,
|
|
"logps/chosen": -218.82334899902344,
|
|
"logps/rejected": -226.3861541748047,
|
|
"loss": 0.6841,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6887062788009644,
|
|
"rewards/margins": 0.031964417546987534,
|
|
"rewards/rejected": -0.720670759677887,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.993877788986285e-05,
|
|
"logits/chosen": -2.3766613006591797,
|
|
"logits/rejected": -2.507246971130371,
|
|
"logps/chosen": -165.4411163330078,
|
|
"logps/rejected": -199.56741333007812,
|
|
"loss": 0.575,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.49406641721725464,
|
|
"rewards/margins": 0.3596181869506836,
|
|
"rewards/rejected": -0.853684663772583,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.9938047188134776e-05,
|
|
"logits/chosen": -2.4775924682617188,
|
|
"logits/rejected": -2.4857354164123535,
|
|
"logps/chosen": -195.7847442626953,
|
|
"logps/rejected": -216.16220092773438,
|
|
"loss": 0.6795,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5620791912078857,
|
|
"rewards/margins": 0.07040975242853165,
|
|
"rewards/rejected": -0.632489025592804,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.993731215708866e-05,
|
|
"logits/chosen": -2.637573719024658,
|
|
"logits/rejected": -2.700226306915283,
|
|
"logps/chosen": -200.697509765625,
|
|
"logps/rejected": -214.4364471435547,
|
|
"loss": 0.6027,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5127323269844055,
|
|
"rewards/margins": 0.24725483357906342,
|
|
"rewards/rejected": -0.7599871754646301,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.993657279685212e-05,
|
|
"logits/chosen": -2.4737141132354736,
|
|
"logits/rejected": -2.576711654663086,
|
|
"logps/chosen": -198.31393432617188,
|
|
"logps/rejected": -219.9713592529297,
|
|
"loss": 0.7162,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7182321548461914,
|
|
"rewards/margins": 0.04057084769010544,
|
|
"rewards/rejected": -0.7588030099868774,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.9935829107553516e-05,
|
|
"logits/chosen": -2.3163719177246094,
|
|
"logits/rejected": -2.3124592304229736,
|
|
"logps/chosen": -198.31076049804688,
|
|
"logps/rejected": -210.8530731201172,
|
|
"loss": 0.6581,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5568541288375854,
|
|
"rewards/margins": 0.1273648738861084,
|
|
"rewards/rejected": -0.6842190623283386,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.993508108932195e-05,
|
|
"logits/chosen": -2.4512782096862793,
|
|
"logits/rejected": -2.508169651031494,
|
|
"logps/chosen": -202.17613220214844,
|
|
"logps/rejected": -203.27789306640625,
|
|
"loss": 0.6612,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7900542616844177,
|
|
"rewards/margins": 0.12749117612838745,
|
|
"rewards/rejected": -0.91754549741745,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.9934328742287285e-05,
|
|
"logits/chosen": -2.391414165496826,
|
|
"logits/rejected": -2.3899242877960205,
|
|
"logps/chosen": -231.99913024902344,
|
|
"logps/rejected": -232.77243041992188,
|
|
"loss": 0.7863,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.6857910752296448,
|
|
"rewards/margins": -0.05944066494703293,
|
|
"rewards/rejected": -0.626350462436676,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.993357206658011e-05,
|
|
"logits/chosen": -2.5768861770629883,
|
|
"logits/rejected": -2.6738839149475098,
|
|
"logps/chosen": -175.33010864257812,
|
|
"logps/rejected": -185.26443481445312,
|
|
"loss": 0.7157,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6001064777374268,
|
|
"rewards/margins": 0.052334412932395935,
|
|
"rewards/rejected": -0.6524409651756287,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.993281106233182e-05,
|
|
"logits/chosen": -2.3842387199401855,
|
|
"logits/rejected": -2.425611972808838,
|
|
"logps/chosen": -165.09873962402344,
|
|
"logps/rejected": -185.34361267089844,
|
|
"loss": 0.5746,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.4550285339355469,
|
|
"rewards/margins": 0.2833191156387329,
|
|
"rewards/rejected": -0.738347589969635,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.9932045729674505e-05,
|
|
"logits/chosen": -2.370333671569824,
|
|
"logits/rejected": -2.343536376953125,
|
|
"logps/chosen": -211.58517456054688,
|
|
"logps/rejected": -191.9432373046875,
|
|
"loss": 0.7727,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.7500544786453247,
|
|
"rewards/margins": -0.05222831293940544,
|
|
"rewards/rejected": -0.697826087474823,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.993127606874104e-05,
|
|
"logits/chosen": -2.3698928356170654,
|
|
"logits/rejected": -2.459376573562622,
|
|
"logps/chosen": -241.92343139648438,
|
|
"logps/rejected": -265.6181945800781,
|
|
"loss": 0.6437,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.6877275705337524,
|
|
"rewards/margins": 0.16285517811775208,
|
|
"rewards/rejected": -0.8505828380584717,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9930502079665025e-05,
|
|
"logits/chosen": -2.3997507095336914,
|
|
"logits/rejected": -2.4078052043914795,
|
|
"logps/chosen": -228.6896514892578,
|
|
"logps/rejected": -221.53338623046875,
|
|
"loss": 0.7984,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6329948306083679,
|
|
"rewards/margins": -0.1108212098479271,
|
|
"rewards/rejected": -0.5221735835075378,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9929723762580835e-05,
|
|
"logits/chosen": -2.498244285583496,
|
|
"logits/rejected": -2.571398973464966,
|
|
"logps/chosen": -192.15635681152344,
|
|
"logps/rejected": -176.19012451171875,
|
|
"loss": 0.7497,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5479970574378967,
|
|
"rewards/margins": -0.0442587286233902,
|
|
"rewards/rejected": -0.5037383437156677,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9928941117623604e-05,
|
|
"logits/chosen": -2.6329376697540283,
|
|
"logits/rejected": -2.635847806930542,
|
|
"logps/chosen": -207.86293029785156,
|
|
"logps/rejected": -224.18490600585938,
|
|
"loss": 0.6649,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5313466787338257,
|
|
"rewards/margins": 0.07021217793226242,
|
|
"rewards/rejected": -0.6015588045120239,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9928154144929175e-05,
|
|
"logits/chosen": -2.5966103076934814,
|
|
"logits/rejected": -2.571523666381836,
|
|
"logps/chosen": -192.47384643554688,
|
|
"logps/rejected": -180.02134704589844,
|
|
"loss": 0.8799,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.8716793656349182,
|
|
"rewards/margins": -0.2290811538696289,
|
|
"rewards/rejected": -0.6425981521606445,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9927362844634186e-05,
|
|
"logits/chosen": -2.6513888835906982,
|
|
"logits/rejected": -2.606253147125244,
|
|
"logps/chosen": -208.56338500976562,
|
|
"logps/rejected": -200.686279296875,
|
|
"loss": 0.7609,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5643545389175415,
|
|
"rewards/margins": -0.08666092902421951,
|
|
"rewards/rejected": -0.4776935577392578,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9926567216876e-05,
|
|
"logits/chosen": -2.5372819900512695,
|
|
"logits/rejected": -2.5824756622314453,
|
|
"logps/chosen": -187.18386840820312,
|
|
"logps/rejected": -211.16427612304688,
|
|
"loss": 0.6838,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5520645380020142,
|
|
"rewards/margins": 0.10136236250400543,
|
|
"rewards/rejected": -0.6534268856048584,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.992576726179274e-05,
|
|
"logits/chosen": -2.5787792205810547,
|
|
"logits/rejected": -2.5688636302948,
|
|
"logps/chosen": -199.59490966796875,
|
|
"logps/rejected": -221.02023315429688,
|
|
"loss": 0.7033,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5855346918106079,
|
|
"rewards/margins": 0.010282933712005615,
|
|
"rewards/rejected": -0.5958175659179688,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.9924962979523296e-05,
|
|
"logits/chosen": -2.588792085647583,
|
|
"logits/rejected": -2.5663416385650635,
|
|
"logps/chosen": -197.04592895507812,
|
|
"logps/rejected": -184.49290466308594,
|
|
"loss": 0.6537,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.46529656648635864,
|
|
"rewards/margins": 0.1208263412117958,
|
|
"rewards/rejected": -0.5861228704452515,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.992415437020727e-05,
|
|
"logits/chosen": -2.6553092002868652,
|
|
"logits/rejected": -2.763751268386841,
|
|
"logps/chosen": -217.5886993408203,
|
|
"logps/rejected": -233.33489990234375,
|
|
"loss": 0.6027,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4947764575481415,
|
|
"rewards/margins": 0.2827147841453552,
|
|
"rewards/rejected": -0.7774912118911743,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.992334143398506e-05,
|
|
"logits/chosen": -2.5335421562194824,
|
|
"logits/rejected": -2.600069999694824,
|
|
"logps/chosen": -204.98912048339844,
|
|
"logps/rejected": -212.0712890625,
|
|
"loss": 0.6079,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5413306355476379,
|
|
"rewards/margins": 0.22219431400299072,
|
|
"rewards/rejected": -0.7635249495506287,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.992252417099778e-05,
|
|
"logits/chosen": -2.638511896133423,
|
|
"logits/rejected": -2.600456476211548,
|
|
"logps/chosen": -172.59390258789062,
|
|
"logps/rejected": -192.1689910888672,
|
|
"loss": 0.7194,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5202851891517639,
|
|
"rewards/margins": 0.017805464565753937,
|
|
"rewards/rejected": -0.5380906462669373,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.992170258138732e-05,
|
|
"logits/chosen": -2.5361521244049072,
|
|
"logits/rejected": -2.4191856384277344,
|
|
"logps/chosen": -177.23558044433594,
|
|
"logps/rejected": -169.40321350097656,
|
|
"loss": 0.6089,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.20310896635055542,
|
|
"rewards/margins": 0.2397383451461792,
|
|
"rewards/rejected": -0.4428473711013794,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.99208766652963e-05,
|
|
"logits/chosen": -2.727092742919922,
|
|
"logits/rejected": -2.613562822341919,
|
|
"logps/chosen": -228.9248504638672,
|
|
"logps/rejected": -222.22886657714844,
|
|
"loss": 0.6345,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5637404322624207,
|
|
"rewards/margins": 0.21417354047298431,
|
|
"rewards/rejected": -0.7779139280319214,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.99200464228681e-05,
|
|
"logits/chosen": -2.530290126800537,
|
|
"logits/rejected": -2.49137282371521,
|
|
"logps/chosen": -218.17156982421875,
|
|
"logps/rejected": -225.41952514648438,
|
|
"loss": 0.7465,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.567755937576294,
|
|
"rewards/margins": -0.07816261053085327,
|
|
"rewards/rejected": -0.4895933270454407,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.9919211854246874e-05,
|
|
"logits/chosen": -2.683086633682251,
|
|
"logits/rejected": -2.588696002960205,
|
|
"logps/chosen": -268.5683898925781,
|
|
"logps/rejected": -255.58445739746094,
|
|
"loss": 0.7069,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5820844769477844,
|
|
"rewards/margins": 0.0005566291511058807,
|
|
"rewards/rejected": -0.582641065120697,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.9918372959577486e-05,
|
|
"logits/chosen": -2.7265396118164062,
|
|
"logits/rejected": -2.7635843753814697,
|
|
"logps/chosen": -206.09573364257812,
|
|
"logps/rejected": -214.75784301757812,
|
|
"loss": 0.6975,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6429769396781921,
|
|
"rewards/margins": 0.1274566352367401,
|
|
"rewards/rejected": -0.7704335451126099,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.9917529739005574e-05,
|
|
"logits/chosen": -2.507643938064575,
|
|
"logits/rejected": -2.4392926692962646,
|
|
"logps/chosen": -219.877197265625,
|
|
"logps/rejected": -217.41537475585938,
|
|
"loss": 0.729,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5904738306999207,
|
|
"rewards/margins": -0.04813346266746521,
|
|
"rewards/rejected": -0.5423403382301331,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.991668219267752e-05,
|
|
"logits/chosen": -2.613016128540039,
|
|
"logits/rejected": -2.7239303588867188,
|
|
"logps/chosen": -235.21534729003906,
|
|
"logps/rejected": -266.6222839355469,
|
|
"loss": 0.7173,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6593362092971802,
|
|
"rewards/margins": 0.04644089192152023,
|
|
"rewards/rejected": -0.7057771682739258,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.991583032074047e-05,
|
|
"logits/chosen": -2.774909257888794,
|
|
"logits/rejected": -2.7659873962402344,
|
|
"logps/chosen": -250.04261779785156,
|
|
"logps/rejected": -242.6505584716797,
|
|
"loss": 0.7062,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6024019718170166,
|
|
"rewards/margins": 0.010418819263577461,
|
|
"rewards/rejected": -0.6128207445144653,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.99149741233423e-05,
|
|
"logits/chosen": -2.447873592376709,
|
|
"logits/rejected": -2.468458414077759,
|
|
"logps/chosen": -214.55784606933594,
|
|
"logps/rejected": -189.22238159179688,
|
|
"loss": 0.6822,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6992676854133606,
|
|
"rewards/margins": 0.09424494951963425,
|
|
"rewards/rejected": -0.7935126423835754,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.9914113600631665e-05,
|
|
"logits/chosen": -2.5641584396362305,
|
|
"logits/rejected": -2.602008104324341,
|
|
"logps/chosen": -310.73077392578125,
|
|
"logps/rejected": -290.17352294921875,
|
|
"loss": 0.6765,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5222407579421997,
|
|
"rewards/margins": 0.06505537033081055,
|
|
"rewards/rejected": -0.587296187877655,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.991324875275794e-05,
|
|
"logits/chosen": -2.5343639850616455,
|
|
"logits/rejected": -2.610719919204712,
|
|
"logps/chosen": -199.84573364257812,
|
|
"logps/rejected": -203.96478271484375,
|
|
"loss": 0.718,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5667111873626709,
|
|
"rewards/margins": -0.001658361405134201,
|
|
"rewards/rejected": -0.5650528073310852,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.991237957987127e-05,
|
|
"logits/chosen": -2.379605293273926,
|
|
"logits/rejected": -2.3678669929504395,
|
|
"logps/chosen": -211.86761474609375,
|
|
"logps/rejected": -213.41464233398438,
|
|
"loss": 0.6325,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.46346691250801086,
|
|
"rewards/margins": 0.20269399881362915,
|
|
"rewards/rejected": -0.6661609411239624,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.991150608212254e-05,
|
|
"logits/chosen": -2.6828835010528564,
|
|
"logits/rejected": -2.7126364707946777,
|
|
"logps/chosen": -196.61216735839844,
|
|
"logps/rejected": -223.1612548828125,
|
|
"loss": 0.7503,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5514590740203857,
|
|
"rewards/margins": -0.08642277121543884,
|
|
"rewards/rejected": -0.4650362730026245,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.9910628259663404e-05,
|
|
"logits/chosen": -2.4960732460021973,
|
|
"logits/rejected": -2.608139991760254,
|
|
"logps/chosen": -197.00277709960938,
|
|
"logps/rejected": -196.33804321289062,
|
|
"loss": 0.564,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.35482868552207947,
|
|
"rewards/margins": 0.38187745213508606,
|
|
"rewards/rejected": -0.7367061376571655,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.990974611264625e-05,
|
|
"logits/chosen": -2.3627309799194336,
|
|
"logits/rejected": -2.3418447971343994,
|
|
"logps/chosen": -232.52740478515625,
|
|
"logps/rejected": -227.15753173828125,
|
|
"loss": 0.6344,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.6158957481384277,
|
|
"rewards/margins": 0.18052013218402863,
|
|
"rewards/rejected": -0.7964158654212952,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.990885964122421e-05,
|
|
"logits/chosen": -2.594547748565674,
|
|
"logits/rejected": -2.6074790954589844,
|
|
"logps/chosen": -170.09300231933594,
|
|
"logps/rejected": -190.0100860595703,
|
|
"loss": 0.7467,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5137202739715576,
|
|
"rewards/margins": -0.07406175136566162,
|
|
"rewards/rejected": -0.439658522605896,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.990796884555119e-05,
|
|
"logits/chosen": -2.632927417755127,
|
|
"logits/rejected": -2.6694023609161377,
|
|
"logps/chosen": -215.45489501953125,
|
|
"logps/rejected": -203.8738555908203,
|
|
"loss": 0.6641,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4411565661430359,
|
|
"rewards/margins": 0.1269129067659378,
|
|
"rewards/rejected": -0.5680694580078125,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.9907073725781836e-05,
|
|
"logits/chosen": -2.4353699684143066,
|
|
"logits/rejected": -2.537163734436035,
|
|
"logps/chosen": -279.5519714355469,
|
|
"logps/rejected": -290.08135986328125,
|
|
"loss": 0.6756,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.32637274265289307,
|
|
"rewards/margins": 0.053382109850645065,
|
|
"rewards/rejected": -0.379754900932312,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.9906174282071535e-05,
|
|
"logits/chosen": -2.586729049682617,
|
|
"logits/rejected": -2.583314895629883,
|
|
"logps/chosen": -197.04298400878906,
|
|
"logps/rejected": -195.334228515625,
|
|
"loss": 0.6718,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.515608549118042,
|
|
"rewards/margins": 0.06315431743860245,
|
|
"rewards/rejected": -0.5787628293037415,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.990527051457644e-05,
|
|
"logits/chosen": -2.45725154876709,
|
|
"logits/rejected": -2.445600748062134,
|
|
"logps/chosen": -204.52867126464844,
|
|
"logps/rejected": -290.3397216796875,
|
|
"loss": 0.6908,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6991719007492065,
|
|
"rewards/margins": 0.08316923677921295,
|
|
"rewards/rejected": -0.7823411226272583,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.9904362423453446e-05,
|
|
"logits/chosen": -2.604722738265991,
|
|
"logits/rejected": -2.6120543479919434,
|
|
"logps/chosen": -193.1224822998047,
|
|
"logps/rejected": -189.89694213867188,
|
|
"loss": 0.6076,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3250601589679718,
|
|
"rewards/margins": 0.19382469356060028,
|
|
"rewards/rejected": -0.5188848376274109,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.990345000886019e-05,
|
|
"logits/chosen": -2.6838831901550293,
|
|
"logits/rejected": -2.6718058586120605,
|
|
"logps/chosen": -177.02090454101562,
|
|
"logps/rejected": -173.667724609375,
|
|
"loss": 0.6993,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5827906131744385,
|
|
"rewards/margins": 0.027324199676513672,
|
|
"rewards/rejected": -0.6101148128509521,
|
|
"step": 311
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.990253327095509e-05,
|
|
"logits/chosen": -2.60197377204895,
|
|
"logits/rejected": -2.522291660308838,
|
|
"logps/chosen": -173.03811645507812,
|
|
"logps/rejected": -149.1508026123047,
|
|
"loss": 0.7939,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.4563007950782776,
|
|
"rewards/margins": -0.13545696437358856,
|
|
"rewards/rejected": -0.32084381580352783,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.9901612209897275e-05,
|
|
"logits/chosen": -2.643686532974243,
|
|
"logits/rejected": -2.6443119049072266,
|
|
"logps/chosen": -187.93931579589844,
|
|
"logps/rejected": -208.4308624267578,
|
|
"loss": 0.6686,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6419711709022522,
|
|
"rewards/margins": 0.10658866912126541,
|
|
"rewards/rejected": -0.748559832572937,
|
|
"step": 313
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.990068682584666e-05,
|
|
"logits/chosen": -2.469681739807129,
|
|
"logits/rejected": -2.4439382553100586,
|
|
"logps/chosen": -186.7505340576172,
|
|
"logps/rejected": -204.6378936767578,
|
|
"loss": 0.7857,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5965012311935425,
|
|
"rewards/margins": -0.12374071031808853,
|
|
"rewards/rejected": -0.47276046872138977,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.989975711896388e-05,
|
|
"logits/chosen": -2.3346047401428223,
|
|
"logits/rejected": -2.323143720626831,
|
|
"logps/chosen": -223.70867919921875,
|
|
"logps/rejected": -238.58311462402344,
|
|
"loss": 0.7061,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5982979536056519,
|
|
"rewards/margins": -0.005420597270131111,
|
|
"rewards/rejected": -0.5928773283958435,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.989882308941034e-05,
|
|
"logits/chosen": -2.759549379348755,
|
|
"logits/rejected": -2.7803878784179688,
|
|
"logps/chosen": -239.48947143554688,
|
|
"logps/rejected": -181.70718383789062,
|
|
"loss": 0.6671,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.46423962712287903,
|
|
"rewards/margins": 0.11677178740501404,
|
|
"rewards/rejected": -0.5810113549232483,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.9897884737348196e-05,
|
|
"logits/chosen": -2.485264778137207,
|
|
"logits/rejected": -2.5983753204345703,
|
|
"logps/chosen": -175.43971252441406,
|
|
"logps/rejected": -219.08517456054688,
|
|
"loss": 0.7124,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5487624406814575,
|
|
"rewards/margins": 0.0002348199486732483,
|
|
"rewards/rejected": -0.5489972829818726,
|
|
"step": 317
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989694206294035e-05,
|
|
"logits/chosen": -2.6732518672943115,
|
|
"logits/rejected": -2.6291115283966064,
|
|
"logps/chosen": -193.59197998046875,
|
|
"logps/rejected": -207.9779052734375,
|
|
"loss": 0.6528,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4739716649055481,
|
|
"rewards/margins": 0.12173515558242798,
|
|
"rewards/rejected": -0.5957068204879761,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989599506635044e-05,
|
|
"logits/chosen": -2.4193854331970215,
|
|
"logits/rejected": -2.565293073654175,
|
|
"logps/chosen": -263.6156311035156,
|
|
"logps/rejected": -251.44192504882812,
|
|
"loss": 0.6307,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.47510460019111633,
|
|
"rewards/margins": 0.25979098677635193,
|
|
"rewards/rejected": -0.734895646572113,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989504374774288e-05,
|
|
"logits/chosen": -2.3681142330169678,
|
|
"logits/rejected": -2.4691953659057617,
|
|
"logps/chosen": -163.5777587890625,
|
|
"logps/rejected": -194.8612060546875,
|
|
"loss": 0.6185,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4559204876422882,
|
|
"rewards/margins": 0.19214320182800293,
|
|
"rewards/rejected": -0.648063600063324,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989408810728281e-05,
|
|
"logits/chosen": -2.477172613143921,
|
|
"logits/rejected": -2.5207676887512207,
|
|
"logps/chosen": -141.12867736816406,
|
|
"logps/rejected": -146.79791259765625,
|
|
"loss": 0.657,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.22462037205696106,
|
|
"rewards/margins": 0.09705987572669983,
|
|
"rewards/rejected": -0.3216802775859833,
|
|
"step": 321
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989312814513614e-05,
|
|
"logits/chosen": -2.534242630004883,
|
|
"logits/rejected": -2.47320294380188,
|
|
"logps/chosen": -223.2980499267578,
|
|
"logps/rejected": -224.9616241455078,
|
|
"loss": 0.7372,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5472955107688904,
|
|
"rewards/margins": -0.037168219685554504,
|
|
"rewards/rejected": -0.5101273059844971,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.989216386146953e-05,
|
|
"logits/chosen": -2.3438358306884766,
|
|
"logits/rejected": -2.623857021331787,
|
|
"logps/chosen": -190.59585571289062,
|
|
"logps/rejected": -234.1730194091797,
|
|
"loss": 0.6873,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3276049494743347,
|
|
"rewards/margins": 0.06552110612392426,
|
|
"rewards/rejected": -0.39312607049942017,
|
|
"step": 323
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.9891195256450366e-05,
|
|
"logits/chosen": -2.210477828979492,
|
|
"logits/rejected": -2.195955991744995,
|
|
"logps/chosen": -185.43902587890625,
|
|
"logps/rejected": -224.94488525390625,
|
|
"loss": 0.678,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.38528165221214294,
|
|
"rewards/margins": 0.09291733056306839,
|
|
"rewards/rejected": -0.47819897532463074,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.989022233024681e-05,
|
|
"logits/chosen": -2.7818360328674316,
|
|
"logits/rejected": -2.764118194580078,
|
|
"logps/chosen": -224.84347534179688,
|
|
"logps/rejected": -205.856201171875,
|
|
"loss": 0.6028,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3358863592147827,
|
|
"rewards/margins": 0.23018960654735565,
|
|
"rewards/rejected": -0.5660759806632996,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.9889245083027755e-05,
|
|
"logits/chosen": -2.3800809383392334,
|
|
"logits/rejected": -2.3793954849243164,
|
|
"logps/chosen": -168.0996551513672,
|
|
"logps/rejected": -197.4810791015625,
|
|
"loss": 0.7001,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.36885035037994385,
|
|
"rewards/margins": 0.015535619109869003,
|
|
"rewards/rejected": -0.3843860328197479,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.988826351496287e-05,
|
|
"logits/chosen": -2.5323662757873535,
|
|
"logits/rejected": -2.5999341011047363,
|
|
"logps/chosen": -215.6177978515625,
|
|
"logps/rejected": -255.65231323242188,
|
|
"loss": 0.6679,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6092446446418762,
|
|
"rewards/margins": 0.09709976613521576,
|
|
"rewards/rejected": -0.7063443660736084,
|
|
"step": 327
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.988727762622255e-05,
|
|
"logits/chosen": -2.7337067127227783,
|
|
"logits/rejected": -2.742793083190918,
|
|
"logps/chosen": -214.25506591796875,
|
|
"logps/rejected": -221.33468627929688,
|
|
"loss": 0.6614,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6302060484886169,
|
|
"rewards/margins": 0.12371520698070526,
|
|
"rewards/rejected": -0.7539212107658386,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.9886287416977936e-05,
|
|
"logits/chosen": -2.510671377182007,
|
|
"logits/rejected": -2.5760793685913086,
|
|
"logps/chosen": -169.99656677246094,
|
|
"logps/rejected": -181.1496124267578,
|
|
"loss": 0.6957,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4545098841190338,
|
|
"rewards/margins": 0.014530147425830364,
|
|
"rewards/rejected": -0.46904003620147705,
|
|
"step": 329
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.988529288740096e-05,
|
|
"logits/chosen": -2.6438777446746826,
|
|
"logits/rejected": -2.6174240112304688,
|
|
"logps/chosen": -187.69232177734375,
|
|
"logps/rejected": -198.16409301757812,
|
|
"loss": 0.6247,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.37409377098083496,
|
|
"rewards/margins": 0.1910334676504135,
|
|
"rewards/rejected": -0.5651272535324097,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.9884294037664245e-05,
|
|
"logits/chosen": -2.456332206726074,
|
|
"logits/rejected": -2.4327914714813232,
|
|
"logps/chosen": -221.3645782470703,
|
|
"logps/rejected": -220.4224853515625,
|
|
"loss": 0.726,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.9011685848236084,
|
|
"rewards/margins": 0.07650195062160492,
|
|
"rewards/rejected": -0.9776705503463745,
|
|
"step": 331
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.988329086794122e-05,
|
|
"logits/chosen": -2.463491439819336,
|
|
"logits/rejected": -2.49444842338562,
|
|
"logps/chosen": -197.96144104003906,
|
|
"logps/rejected": -214.404296875,
|
|
"loss": 0.6547,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.42488348484039307,
|
|
"rewards/margins": 0.11507010459899902,
|
|
"rewards/rejected": -0.5399536490440369,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.9882283378406015e-05,
|
|
"logits/chosen": -2.5106682777404785,
|
|
"logits/rejected": -2.5157594680786133,
|
|
"logps/chosen": -167.09947204589844,
|
|
"logps/rejected": -174.6180419921875,
|
|
"loss": 0.6179,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4473687410354614,
|
|
"rewards/margins": 0.22422267496585846,
|
|
"rewards/rejected": -0.6715914011001587,
|
|
"step": 333
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.988127156923355e-05,
|
|
"logits/chosen": -2.4497499465942383,
|
|
"logits/rejected": -2.491909980773926,
|
|
"logps/chosen": -167.1006622314453,
|
|
"logps/rejected": -165.177978515625,
|
|
"loss": 0.6855,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5522390604019165,
|
|
"rewards/margins": 0.06803999096155167,
|
|
"rewards/rejected": -0.6202789545059204,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.9880255440599476e-05,
|
|
"logits/chosen": -2.445422649383545,
|
|
"logits/rejected": -2.4025139808654785,
|
|
"logps/chosen": -199.83444213867188,
|
|
"logps/rejected": -203.16355895996094,
|
|
"loss": 0.7404,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.658203125,
|
|
"rewards/margins": -0.06064491346478462,
|
|
"rewards/rejected": -0.5975580811500549,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.987923499268018e-05,
|
|
"logits/chosen": -2.540513753890991,
|
|
"logits/rejected": -2.6426820755004883,
|
|
"logps/chosen": -181.93406677246094,
|
|
"logps/rejected": -203.6474609375,
|
|
"loss": 0.6465,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.45222824811935425,
|
|
"rewards/margins": 0.17588049173355103,
|
|
"rewards/rejected": -0.6281087398529053,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.987821022565284e-05,
|
|
"logits/chosen": -2.608978509902954,
|
|
"logits/rejected": -2.632282257080078,
|
|
"logps/chosen": -195.6852569580078,
|
|
"logps/rejected": -216.0919952392578,
|
|
"loss": 0.6137,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4888995289802551,
|
|
"rewards/margins": 0.3496631681919098,
|
|
"rewards/rejected": -0.8385627269744873,
|
|
"step": 337
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.987718113969534e-05,
|
|
"logits/chosen": -2.637908458709717,
|
|
"logits/rejected": -2.6849193572998047,
|
|
"logps/chosen": -307.9834289550781,
|
|
"logps/rejected": -321.95867919921875,
|
|
"loss": 0.5909,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.4099658131599426,
|
|
"rewards/margins": 0.3652358055114746,
|
|
"rewards/rejected": -0.7752016186714172,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.9876147734986335e-05,
|
|
"logits/chosen": -2.6805038452148438,
|
|
"logits/rejected": -2.659087657928467,
|
|
"logps/chosen": -235.20855712890625,
|
|
"logps/rejected": -222.38107299804688,
|
|
"loss": 0.7633,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7779167294502258,
|
|
"rewards/margins": -0.11177889257669449,
|
|
"rewards/rejected": -0.6661379337310791,
|
|
"step": 339
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.987511001170523e-05,
|
|
"logits/chosen": -2.4979588985443115,
|
|
"logits/rejected": -2.509103775024414,
|
|
"logps/chosen": -205.0665283203125,
|
|
"logps/rejected": -197.5661163330078,
|
|
"loss": 0.584,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.603629469871521,
|
|
"rewards/margins": 0.3201472759246826,
|
|
"rewards/rejected": -0.9237766861915588,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.987406797003218e-05,
|
|
"logits/chosen": -2.6034300327301025,
|
|
"logits/rejected": -2.606010913848877,
|
|
"logps/chosen": -212.59959411621094,
|
|
"logps/rejected": -266.6811218261719,
|
|
"loss": 0.7099,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.7883586287498474,
|
|
"rewards/margins": 0.047954261302948,
|
|
"rewards/rejected": -0.8363128900527954,
|
|
"step": 341
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.987302161014808e-05,
|
|
"logits/chosen": -2.26737380027771,
|
|
"logits/rejected": -2.2702267169952393,
|
|
"logps/chosen": -149.69259643554688,
|
|
"logps/rejected": -207.59146118164062,
|
|
"loss": 0.718,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6731799244880676,
|
|
"rewards/margins": 0.02036801353096962,
|
|
"rewards/rejected": -0.6935478448867798,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.9871970932234586e-05,
|
|
"logits/chosen": -2.3482635021209717,
|
|
"logits/rejected": -2.363096237182617,
|
|
"logps/chosen": -196.5614013671875,
|
|
"logps/rejected": -206.21414184570312,
|
|
"loss": 0.6163,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5754539966583252,
|
|
"rewards/margins": 0.2127307951450348,
|
|
"rewards/rejected": -0.7881847620010376,
|
|
"step": 343
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.9870915936474095e-05,
|
|
"logits/chosen": -2.7459514141082764,
|
|
"logits/rejected": -2.7396438121795654,
|
|
"logps/chosen": -206.6090087890625,
|
|
"logps/rejected": -211.6893768310547,
|
|
"loss": 0.6763,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7424849271774292,
|
|
"rewards/margins": 0.08226263523101807,
|
|
"rewards/rejected": -0.8247475624084473,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.986985662304976e-05,
|
|
"logits/chosen": -2.769526243209839,
|
|
"logits/rejected": -2.807554244995117,
|
|
"logps/chosen": -184.0075225830078,
|
|
"logps/rejected": -225.2920379638672,
|
|
"loss": 0.6217,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7790610194206238,
|
|
"rewards/margins": 0.24687500298023224,
|
|
"rewards/rejected": -1.0259360074996948,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.9868792992145484e-05,
|
|
"logits/chosen": -2.6607320308685303,
|
|
"logits/rejected": -2.6989316940307617,
|
|
"logps/chosen": -218.40530395507812,
|
|
"logps/rejected": -227.24240112304688,
|
|
"loss": 0.7374,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6406981945037842,
|
|
"rewards/margins": 0.009773064404726028,
|
|
"rewards/rejected": -0.6504712104797363,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.9867725043945904e-05,
|
|
"logits/chosen": -2.5469489097595215,
|
|
"logits/rejected": -2.539520740509033,
|
|
"logps/chosen": -222.92864990234375,
|
|
"logps/rejected": -236.92977905273438,
|
|
"loss": 0.6992,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.640974223613739,
|
|
"rewards/margins": 0.09771876782178879,
|
|
"rewards/rejected": -0.7386929988861084,
|
|
"step": 347
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.9866652778636436e-05,
|
|
"logits/chosen": -2.545572280883789,
|
|
"logits/rejected": -2.541029453277588,
|
|
"logps/chosen": -182.341552734375,
|
|
"logps/rejected": -197.535888671875,
|
|
"loss": 0.7149,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.7396100163459778,
|
|
"rewards/margins": 0.06020259112119675,
|
|
"rewards/rejected": -0.7998126149177551,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.986557619640322e-05,
|
|
"logits/chosen": -2.479189157485962,
|
|
"logits/rejected": -2.504956007003784,
|
|
"logps/chosen": -184.10472106933594,
|
|
"logps/rejected": -184.99070739746094,
|
|
"loss": 0.5709,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.6914107799530029,
|
|
"rewards/margins": 0.33003267645835876,
|
|
"rewards/rejected": -1.021443486213684,
|
|
"step": 349
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.986449529743314e-05,
|
|
"logits/chosen": -2.7267708778381348,
|
|
"logits/rejected": -2.711848735809326,
|
|
"logps/chosen": -225.98228454589844,
|
|
"logps/rejected": -204.31724548339844,
|
|
"loss": 0.7177,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.8359745740890503,
|
|
"rewards/margins": 0.011185385286808014,
|
|
"rewards/rejected": -0.8471599221229553,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.9863410081913875e-05,
|
|
"logits/chosen": -2.691619396209717,
|
|
"logits/rejected": -2.669311285018921,
|
|
"logps/chosen": -157.73721313476562,
|
|
"logps/rejected": -161.97061157226562,
|
|
"loss": 0.7285,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5839475393295288,
|
|
"rewards/margins": -0.027979355305433273,
|
|
"rewards/rejected": -0.555968165397644,
|
|
"step": 351
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.98623205500338e-05,
|
|
"logits/chosen": -2.6664159297943115,
|
|
"logits/rejected": -2.653743267059326,
|
|
"logps/chosen": -155.03836059570312,
|
|
"logps/rejected": -141.4946746826172,
|
|
"loss": 0.7569,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5298622250556946,
|
|
"rewards/margins": -0.08286149799823761,
|
|
"rewards/rejected": -0.44700077176094055,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.986122670198205e-05,
|
|
"logits/chosen": -2.588876485824585,
|
|
"logits/rejected": -2.706108808517456,
|
|
"logps/chosen": -170.23751831054688,
|
|
"logps/rejected": -172.2996826171875,
|
|
"loss": 0.8284,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.9628427624702454,
|
|
"rewards/margins": -0.18341386318206787,
|
|
"rewards/rejected": -0.779429018497467,
|
|
"step": 353
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.9860128537948555e-05,
|
|
"logits/chosen": -2.8683338165283203,
|
|
"logits/rejected": -2.814868211746216,
|
|
"logps/chosen": -194.464599609375,
|
|
"logps/rejected": -197.70077514648438,
|
|
"loss": 0.7405,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6296394467353821,
|
|
"rewards/margins": -0.0025831498205661774,
|
|
"rewards/rejected": -0.6270563006401062,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.9859026058123925e-05,
|
|
"logits/chosen": -2.978065013885498,
|
|
"logits/rejected": -2.9368367195129395,
|
|
"logps/chosen": -217.73974609375,
|
|
"logps/rejected": -220.19183349609375,
|
|
"loss": 0.7014,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6989356279373169,
|
|
"rewards/margins": 0.007979679852724075,
|
|
"rewards/rejected": -0.7069153189659119,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985791926269958e-05,
|
|
"logits/chosen": -2.666344165802002,
|
|
"logits/rejected": -2.8283472061157227,
|
|
"logps/chosen": -163.43722534179688,
|
|
"logps/rejected": -190.4180908203125,
|
|
"loss": 0.6778,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5998314023017883,
|
|
"rewards/margins": 0.11102181673049927,
|
|
"rewards/rejected": -0.7108532190322876,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985680815186764e-05,
|
|
"logits/chosen": -2.6322436332702637,
|
|
"logits/rejected": -2.6792352199554443,
|
|
"logps/chosen": -225.80972290039062,
|
|
"logps/rejected": -221.8551025390625,
|
|
"loss": 0.6464,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6963925957679749,
|
|
"rewards/margins": 0.1484571099281311,
|
|
"rewards/rejected": -0.844849705696106,
|
|
"step": 357
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985569272582101e-05,
|
|
"logits/chosen": -2.829981803894043,
|
|
"logits/rejected": -2.8077688217163086,
|
|
"logps/chosen": -245.94029235839844,
|
|
"logps/rejected": -279.1697082519531,
|
|
"loss": 0.6446,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6325417757034302,
|
|
"rewards/margins": 0.1735651046037674,
|
|
"rewards/rejected": -0.8061069250106812,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.9854572984753334e-05,
|
|
"logits/chosen": -2.8024842739105225,
|
|
"logits/rejected": -2.750588893890381,
|
|
"logps/chosen": -161.65240478515625,
|
|
"logps/rejected": -201.25054931640625,
|
|
"loss": 0.6502,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.46800869703292847,
|
|
"rewards/margins": 0.12990553677082062,
|
|
"rewards/rejected": -0.5979142189025879,
|
|
"step": 359
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985344892885899e-05,
|
|
"logits/chosen": -2.347231149673462,
|
|
"logits/rejected": -2.4064972400665283,
|
|
"logps/chosen": -158.72801208496094,
|
|
"logps/rejected": -167.15805053710938,
|
|
"loss": 0.7264,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.7969176173210144,
|
|
"rewards/margins": 0.016100093722343445,
|
|
"rewards/rejected": -0.8130176663398743,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985232055833313e-05,
|
|
"logits/chosen": -2.603712320327759,
|
|
"logits/rejected": -2.6721928119659424,
|
|
"logps/chosen": -211.3256378173828,
|
|
"logps/rejected": -189.5877685546875,
|
|
"loss": 0.7658,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6097235679626465,
|
|
"rewards/margins": -0.10059280693531036,
|
|
"rewards/rejected": -0.5091307163238525,
|
|
"step": 361
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.985118787337164e-05,
|
|
"logits/chosen": -2.6673433780670166,
|
|
"logits/rejected": -2.766993284225464,
|
|
"logps/chosen": -162.41415405273438,
|
|
"logps/rejected": -218.15077209472656,
|
|
"loss": 0.6236,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5241696238517761,
|
|
"rewards/margins": 0.28995370864868164,
|
|
"rewards/rejected": -0.8141233921051025,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.985005087417115e-05,
|
|
"logits/chosen": -2.4291069507598877,
|
|
"logits/rejected": -2.382821798324585,
|
|
"logps/chosen": -221.48736572265625,
|
|
"logps/rejected": -205.37940979003906,
|
|
"loss": 0.6503,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4502621591091156,
|
|
"rewards/margins": 0.2195020467042923,
|
|
"rewards/rejected": -0.6697642207145691,
|
|
"step": 363
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.984890956092905e-05,
|
|
"logits/chosen": -2.5559186935424805,
|
|
"logits/rejected": -2.496387481689453,
|
|
"logps/chosen": -253.63047790527344,
|
|
"logps/rejected": -244.56698608398438,
|
|
"loss": 0.9484,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.9435632228851318,
|
|
"rewards/margins": -0.268410325050354,
|
|
"rewards/rejected": -0.6751528978347778,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.984776393384348e-05,
|
|
"logits/chosen": -2.407367706298828,
|
|
"logits/rejected": -2.3134310245513916,
|
|
"logps/chosen": -173.61795043945312,
|
|
"logps/rejected": -157.41207885742188,
|
|
"loss": 0.7949,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5948637127876282,
|
|
"rewards/margins": -0.14962351322174072,
|
|
"rewards/rejected": -0.44524019956588745,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.984661399311332e-05,
|
|
"logits/chosen": -2.6404154300689697,
|
|
"logits/rejected": -2.7036595344543457,
|
|
"logps/chosen": -155.3403778076172,
|
|
"logps/rejected": -141.01382446289062,
|
|
"loss": 0.7475,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4795231521129608,
|
|
"rewards/margins": -0.06228278949856758,
|
|
"rewards/rejected": -0.4172403812408447,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.9845459738938204e-05,
|
|
"logits/chosen": -2.5403473377227783,
|
|
"logits/rejected": -2.5674545764923096,
|
|
"logps/chosen": -154.85910034179688,
|
|
"logps/rejected": -143.22906494140625,
|
|
"loss": 0.6505,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4790140390396118,
|
|
"rewards/margins": 0.1586410105228424,
|
|
"rewards/rejected": -0.6376550197601318,
|
|
"step": 367
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.9844301171518516e-05,
|
|
"logits/chosen": -2.6081302165985107,
|
|
"logits/rejected": -2.641979217529297,
|
|
"logps/chosen": -185.13006591796875,
|
|
"logps/rejected": -198.2803955078125,
|
|
"loss": 0.6726,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3776107430458069,
|
|
"rewards/margins": 0.08190791308879852,
|
|
"rewards/rejected": -0.4595187306404114,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.984313829105538e-05,
|
|
"logits/chosen": -2.6662757396698,
|
|
"logits/rejected": -2.7398993968963623,
|
|
"logps/chosen": -201.8738555908203,
|
|
"logps/rejected": -195.76390075683594,
|
|
"loss": 0.5956,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.38797736167907715,
|
|
"rewards/margins": 0.2769092321395874,
|
|
"rewards/rejected": -0.6648865342140198,
|
|
"step": 369
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.984197109775068e-05,
|
|
"logits/chosen": -2.4933226108551025,
|
|
"logits/rejected": -2.484591245651245,
|
|
"logps/chosen": -202.7592315673828,
|
|
"logps/rejected": -199.02105712890625,
|
|
"loss": 0.6536,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.35886555910110474,
|
|
"rewards/margins": 0.1531214416027069,
|
|
"rewards/rejected": -0.511987030506134,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.984079959180705e-05,
|
|
"logits/chosen": -2.3560287952423096,
|
|
"logits/rejected": -2.385936975479126,
|
|
"logps/chosen": -198.06044006347656,
|
|
"logps/rejected": -203.5518798828125,
|
|
"loss": 0.7203,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.35552191734313965,
|
|
"rewards/margins": 0.03736201673746109,
|
|
"rewards/rejected": -0.39288395643234253,
|
|
"step": 371
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.983962377342786e-05,
|
|
"logits/chosen": -2.456765651702881,
|
|
"logits/rejected": -2.522660970687866,
|
|
"logps/chosen": -236.66744995117188,
|
|
"logps/rejected": -241.58963012695312,
|
|
"loss": 0.6049,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.551971435546875,
|
|
"rewards/margins": 0.30021384358406067,
|
|
"rewards/rejected": -0.8521853089332581,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.983844364281723e-05,
|
|
"logits/chosen": -2.5745279788970947,
|
|
"logits/rejected": -2.5961902141571045,
|
|
"logps/chosen": -186.73411560058594,
|
|
"logps/rejected": -194.74974060058594,
|
|
"loss": 0.6354,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.312610000371933,
|
|
"rewards/margins": 0.14659735560417175,
|
|
"rewards/rejected": -0.45920735597610474,
|
|
"step": 373
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.983725920018004e-05,
|
|
"logits/chosen": -2.3747611045837402,
|
|
"logits/rejected": -2.4170022010803223,
|
|
"logps/chosen": -179.91162109375,
|
|
"logps/rejected": -170.89920043945312,
|
|
"loss": 0.6529,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.43839430809020996,
|
|
"rewards/margins": 0.16247621178627014,
|
|
"rewards/rejected": -0.6008704900741577,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.9836070445721924e-05,
|
|
"logits/chosen": -2.573714017868042,
|
|
"logits/rejected": -2.567589282989502,
|
|
"logps/chosen": -204.4001922607422,
|
|
"logps/rejected": -197.39869689941406,
|
|
"loss": 0.751,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.350676029920578,
|
|
"rewards/margins": -0.06606191396713257,
|
|
"rewards/rejected": -0.28461411595344543,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.983487737964924e-05,
|
|
"logits/chosen": -2.398193597793579,
|
|
"logits/rejected": -2.4593253135681152,
|
|
"logps/chosen": -199.0366973876953,
|
|
"logps/rejected": -199.51275634765625,
|
|
"loss": 0.6772,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.397927463054657,
|
|
"rewards/margins": 0.08031069487333298,
|
|
"rewards/rejected": -0.47823822498321533,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.9833680002169105e-05,
|
|
"logits/chosen": -2.5120034217834473,
|
|
"logits/rejected": -2.553926706314087,
|
|
"logps/chosen": -226.15040588378906,
|
|
"logps/rejected": -249.55255126953125,
|
|
"loss": 0.6996,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3695937395095825,
|
|
"rewards/margins": 0.0394027940928936,
|
|
"rewards/rejected": -0.4089965522289276,
|
|
"step": 377
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.983247831348939e-05,
|
|
"logits/chosen": -2.289033889770508,
|
|
"logits/rejected": -2.2501816749572754,
|
|
"logps/chosen": -201.75392150878906,
|
|
"logps/rejected": -217.8618621826172,
|
|
"loss": 0.6645,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.28367459774017334,
|
|
"rewards/margins": 0.1500779092311859,
|
|
"rewards/rejected": -0.43375247716903687,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.9831272313818716e-05,
|
|
"logits/chosen": -2.4449117183685303,
|
|
"logits/rejected": -2.522752046585083,
|
|
"logps/chosen": -201.63587951660156,
|
|
"logps/rejected": -215.33270263671875,
|
|
"loss": 0.6512,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.23286518454551697,
|
|
"rewards/margins": 0.127617746591568,
|
|
"rewards/rejected": -0.36048293113708496,
|
|
"step": 379
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.983006200336645e-05,
|
|
"logits/chosen": -2.4317235946655273,
|
|
"logits/rejected": -2.3696205615997314,
|
|
"logps/chosen": -243.03509521484375,
|
|
"logps/rejected": -186.61427307128906,
|
|
"loss": 0.7762,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5067302584648132,
|
|
"rewards/margins": -0.11989758908748627,
|
|
"rewards/rejected": -0.3868326246738434,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.98288473823427e-05,
|
|
"logits/chosen": -2.4131155014038086,
|
|
"logits/rejected": -2.498098611831665,
|
|
"logps/chosen": -190.73397827148438,
|
|
"logps/rejected": -213.9364471435547,
|
|
"loss": 0.6336,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.24977169930934906,
|
|
"rewards/margins": 0.1747249811887741,
|
|
"rewards/rejected": -0.42449668049812317,
|
|
"step": 381
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.982762845095833e-05,
|
|
"logits/chosen": -2.5675809383392334,
|
|
"logits/rejected": -2.6289899349212646,
|
|
"logps/chosen": -233.9481658935547,
|
|
"logps/rejected": -245.3311767578125,
|
|
"loss": 0.6879,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3384949564933777,
|
|
"rewards/margins": 0.07099074125289917,
|
|
"rewards/rejected": -0.40948566794395447,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.982640520942494e-05,
|
|
"logits/chosen": -2.3582189083099365,
|
|
"logits/rejected": -2.405932903289795,
|
|
"logps/chosen": -157.98828125,
|
|
"logps/rejected": -179.5477294921875,
|
|
"loss": 0.6256,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.30075955390930176,
|
|
"rewards/margins": 0.2314756065607071,
|
|
"rewards/rejected": -0.5322352051734924,
|
|
"step": 383
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.9825177657954914e-05,
|
|
"logits/chosen": -2.6346352100372314,
|
|
"logits/rejected": -2.7042407989501953,
|
|
"logps/chosen": -174.1151123046875,
|
|
"logps/rejected": -203.845458984375,
|
|
"loss": 0.6727,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3591979742050171,
|
|
"rewards/margins": 0.10004588216543198,
|
|
"rewards/rejected": -0.45924389362335205,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.982394579676133e-05,
|
|
"logits/chosen": -2.3420321941375732,
|
|
"logits/rejected": -2.3010876178741455,
|
|
"logps/chosen": -186.43467712402344,
|
|
"logps/rejected": -183.117431640625,
|
|
"loss": 0.6836,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3466501235961914,
|
|
"rewards/margins": 0.06071464717388153,
|
|
"rewards/rejected": -0.40736478567123413,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.982270962605806e-05,
|
|
"logits/chosen": -2.4120569229125977,
|
|
"logits/rejected": -2.449502944946289,
|
|
"logps/chosen": -168.2112579345703,
|
|
"logps/rejected": -227.885986328125,
|
|
"loss": 0.5666,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.22820252180099487,
|
|
"rewards/margins": 0.35979852080345154,
|
|
"rewards/rejected": -0.588001012802124,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.9821469146059704e-05,
|
|
"logits/chosen": -2.3874173164367676,
|
|
"logits/rejected": -2.4257519245147705,
|
|
"logps/chosen": -158.00424194335938,
|
|
"logps/rejected": -176.4921417236328,
|
|
"loss": 0.6581,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.28017669916152954,
|
|
"rewards/margins": 0.168703094124794,
|
|
"rewards/rejected": -0.44887974858283997,
|
|
"step": 387
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.982022435698161e-05,
|
|
"logits/chosen": -2.444685697555542,
|
|
"logits/rejected": -2.653350591659546,
|
|
"logps/chosen": -207.88034057617188,
|
|
"logps/rejected": -231.86582946777344,
|
|
"loss": 0.7048,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5877898931503296,
|
|
"rewards/margins": 0.10215066373348236,
|
|
"rewards/rejected": -0.6899405121803284,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.981897525903988e-05,
|
|
"logits/chosen": -2.5300025939941406,
|
|
"logits/rejected": -2.4865880012512207,
|
|
"logps/chosen": -155.8424835205078,
|
|
"logps/rejected": -152.6171875,
|
|
"loss": 0.8172,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.5343464612960815,
|
|
"rewards/margins": -0.17394664883613586,
|
|
"rewards/rejected": -0.3603998124599457,
|
|
"step": 389
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.981772185245135e-05,
|
|
"logits/chosen": -2.61796236038208,
|
|
"logits/rejected": -2.662163734436035,
|
|
"logps/chosen": -198.32403564453125,
|
|
"logps/rejected": -198.2169647216797,
|
|
"loss": 0.751,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4620332419872284,
|
|
"rewards/margins": -0.09868823736906052,
|
|
"rewards/rejected": -0.3633449673652649,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.981646413743363e-05,
|
|
"logits/chosen": -2.386425733566284,
|
|
"logits/rejected": -2.452526092529297,
|
|
"logps/chosen": -187.49838256835938,
|
|
"logps/rejected": -181.8330535888672,
|
|
"loss": 0.606,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.27474457025527954,
|
|
"rewards/margins": 0.2754724621772766,
|
|
"rewards/rejected": -0.5502170324325562,
|
|
"step": 391
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.981520211420506e-05,
|
|
"logits/chosen": -2.4169254302978516,
|
|
"logits/rejected": -2.4883086681365967,
|
|
"logps/chosen": -175.3878173828125,
|
|
"logps/rejected": -210.00119018554688,
|
|
"loss": 0.7642,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.532840371131897,
|
|
"rewards/margins": -0.09793217480182648,
|
|
"rewards/rejected": -0.4349081814289093,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.9813935782984724e-05,
|
|
"logits/chosen": -2.528134346008301,
|
|
"logits/rejected": -2.491497039794922,
|
|
"logps/chosen": -239.28878784179688,
|
|
"logps/rejected": -241.08883666992188,
|
|
"loss": 0.7798,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6152268648147583,
|
|
"rewards/margins": -0.10364706814289093,
|
|
"rewards/rejected": -0.5115798711776733,
|
|
"step": 393
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.9812665143992466e-05,
|
|
"logits/chosen": -2.5107626914978027,
|
|
"logits/rejected": -2.591826915740967,
|
|
"logps/chosen": -146.5967559814453,
|
|
"logps/rejected": -154.36585998535156,
|
|
"loss": 0.6751,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3082922697067261,
|
|
"rewards/margins": 0.11212008446455002,
|
|
"rewards/rejected": -0.4204123616218567,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.981139019744887e-05,
|
|
"logits/chosen": -2.5607235431671143,
|
|
"logits/rejected": -2.562152862548828,
|
|
"logps/chosen": -185.95343017578125,
|
|
"logps/rejected": -194.59483337402344,
|
|
"loss": 0.6469,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.2080746442079544,
|
|
"rewards/margins": 0.12636929750442505,
|
|
"rewards/rejected": -0.33444392681121826,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.981011094357527e-05,
|
|
"logits/chosen": -2.4996774196624756,
|
|
"logits/rejected": -2.5331335067749023,
|
|
"logps/chosen": -215.7071075439453,
|
|
"logps/rejected": -231.0633087158203,
|
|
"loss": 0.7228,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4276806116104126,
|
|
"rewards/margins": 0.03171852231025696,
|
|
"rewards/rejected": -0.45939913392066956,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.980882738259376e-05,
|
|
"logits/chosen": -2.3114399909973145,
|
|
"logits/rejected": -2.3782169818878174,
|
|
"logps/chosen": -169.04246520996094,
|
|
"logps/rejected": -188.5185089111328,
|
|
"loss": 0.7661,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.31136631965637207,
|
|
"rewards/margins": -0.09910126030445099,
|
|
"rewards/rejected": -0.21226505935192108,
|
|
"step": 397
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.980753951472715e-05,
|
|
"logits/chosen": -2.466881275177002,
|
|
"logits/rejected": -2.5022642612457275,
|
|
"logps/chosen": -168.63385009765625,
|
|
"logps/rejected": -192.87521362304688,
|
|
"loss": 0.6487,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3558157682418823,
|
|
"rewards/margins": 0.13948221504688263,
|
|
"rewards/rejected": -0.49529796838760376,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.980624734019903e-05,
|
|
"logits/chosen": -2.5671474933624268,
|
|
"logits/rejected": -2.6410720348358154,
|
|
"logps/chosen": -183.6251220703125,
|
|
"logps/rejected": -214.81272888183594,
|
|
"loss": 0.7078,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4311671555042267,
|
|
"rewards/margins": 0.0398234948515892,
|
|
"rewards/rejected": -0.4709906280040741,
|
|
"step": 399
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.980495085923372e-05,
|
|
"logits/chosen": -2.602818012237549,
|
|
"logits/rejected": -2.5820987224578857,
|
|
"logps/chosen": -167.52012634277344,
|
|
"logps/rejected": -180.94418334960938,
|
|
"loss": 0.7179,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.41106754541397095,
|
|
"rewards/margins": 0.013390736654400826,
|
|
"rewards/rejected": -0.42445826530456543,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.980365007205631e-05,
|
|
"logits/chosen": -2.4487011432647705,
|
|
"logits/rejected": -2.5343518257141113,
|
|
"logps/chosen": -201.9692840576172,
|
|
"logps/rejected": -244.95530700683594,
|
|
"loss": 0.5837,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3986441195011139,
|
|
"rewards/margins": 0.3018190264701843,
|
|
"rewards/rejected": -0.700463056564331,
|
|
"step": 401
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.980234497889259e-05,
|
|
"logits/chosen": -1.9015774726867676,
|
|
"logits/rejected": -1.9911203384399414,
|
|
"logps/chosen": -165.83837890625,
|
|
"logps/rejected": -178.608154296875,
|
|
"loss": 0.7531,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.47146371006965637,
|
|
"rewards/margins": -0.03502827137708664,
|
|
"rewards/rejected": -0.43643543124198914,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.980103557996915e-05,
|
|
"logits/chosen": -2.609720468521118,
|
|
"logits/rejected": -2.631443738937378,
|
|
"logps/chosen": -235.0199737548828,
|
|
"logps/rejected": -234.86245727539062,
|
|
"loss": 0.6376,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2306605726480484,
|
|
"rewards/margins": 0.16340221464633942,
|
|
"rewards/rejected": -0.3940627872943878,
|
|
"step": 403
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.9799721875513306e-05,
|
|
"logits/chosen": -2.283406972885132,
|
|
"logits/rejected": -2.386524200439453,
|
|
"logps/chosen": -226.02059936523438,
|
|
"logps/rejected": -220.51963806152344,
|
|
"loss": 0.7208,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5439196228981018,
|
|
"rewards/margins": 0.06111856549978256,
|
|
"rewards/rejected": -0.6050382256507874,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.979840386575311e-05,
|
|
"logits/chosen": -2.2238452434539795,
|
|
"logits/rejected": -2.3374826908111572,
|
|
"logps/chosen": -169.82667541503906,
|
|
"logps/rejected": -197.92726135253906,
|
|
"loss": 0.7292,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.46263885498046875,
|
|
"rewards/margins": 0.007239609956741333,
|
|
"rewards/rejected": -0.46987849473953247,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.979708155091737e-05,
|
|
"logits/chosen": -2.381389856338501,
|
|
"logits/rejected": -2.358883857727051,
|
|
"logps/chosen": -153.38771057128906,
|
|
"logps/rejected": -147.0846405029297,
|
|
"loss": 0.6555,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.2459157109260559,
|
|
"rewards/margins": 0.1274775117635727,
|
|
"rewards/rejected": -0.3733932375907898,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.979575493123566e-05,
|
|
"logits/chosen": -2.4942612648010254,
|
|
"logits/rejected": -2.4526309967041016,
|
|
"logps/chosen": -222.35780334472656,
|
|
"logps/rejected": -205.19630432128906,
|
|
"loss": 0.7278,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4296140670776367,
|
|
"rewards/margins": -0.018014922738075256,
|
|
"rewards/rejected": -0.41159915924072266,
|
|
"step": 407
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.979442400693827e-05,
|
|
"logits/chosen": -2.4821577072143555,
|
|
"logits/rejected": -2.4502267837524414,
|
|
"logps/chosen": -157.32427978515625,
|
|
"logps/rejected": -167.9297332763672,
|
|
"loss": 0.7119,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4199235439300537,
|
|
"rewards/margins": 0.024679476395249367,
|
|
"rewards/rejected": -0.4446030259132385,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.979308877825626e-05,
|
|
"logits/chosen": -2.1102793216705322,
|
|
"logits/rejected": -2.100888729095459,
|
|
"logps/chosen": -235.02105712890625,
|
|
"logps/rejected": -232.0230712890625,
|
|
"loss": 0.747,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4274646043777466,
|
|
"rewards/margins": 0.0070232609286904335,
|
|
"rewards/rejected": -0.43448784947395325,
|
|
"step": 409
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.9791749245421434e-05,
|
|
"logits/chosen": -2.4029147624969482,
|
|
"logits/rejected": -2.436694622039795,
|
|
"logps/chosen": -160.94296264648438,
|
|
"logps/rejected": -219.91183471679688,
|
|
"loss": 0.7659,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6857771873474121,
|
|
"rewards/margins": -0.07571852207183838,
|
|
"rewards/rejected": -0.6100587248802185,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.979040540866632e-05,
|
|
"logits/chosen": -2.423585891723633,
|
|
"logits/rejected": -2.583218574523926,
|
|
"logps/chosen": -152.8751220703125,
|
|
"logps/rejected": -195.98130798339844,
|
|
"loss": 0.6592,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.29784753918647766,
|
|
"rewards/margins": 0.13703800737857819,
|
|
"rewards/rejected": -0.43488556146621704,
|
|
"step": 411
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.9789057268224234e-05,
|
|
"logits/chosen": -2.0688211917877197,
|
|
"logits/rejected": -2.1250336170196533,
|
|
"logps/chosen": -192.48826599121094,
|
|
"logps/rejected": -219.7627716064453,
|
|
"loss": 0.7074,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.16141924262046814,
|
|
"rewards/margins": 0.035948604345321655,
|
|
"rewards/rejected": -0.197367861866951,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.978770482432921e-05,
|
|
"logits/chosen": -2.4140584468841553,
|
|
"logits/rejected": -2.394195079803467,
|
|
"logps/chosen": -177.2361297607422,
|
|
"logps/rejected": -201.13475036621094,
|
|
"loss": 0.7109,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.41423746943473816,
|
|
"rewards/margins": 0.01436915248632431,
|
|
"rewards/rejected": -0.42860662937164307,
|
|
"step": 413
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.9786348077216024e-05,
|
|
"logits/chosen": -2.399442672729492,
|
|
"logits/rejected": -2.4223692417144775,
|
|
"logps/chosen": -184.38839721679688,
|
|
"logps/rejected": -176.9226531982422,
|
|
"loss": 0.6572,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.2739775776863098,
|
|
"rewards/margins": 0.14736850559711456,
|
|
"rewards/rejected": -0.4213460683822632,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.9784987027120236e-05,
|
|
"logits/chosen": -2.331202745437622,
|
|
"logits/rejected": -2.3324198722839355,
|
|
"logps/chosen": -199.08740234375,
|
|
"logps/rejected": -171.50718688964844,
|
|
"loss": 0.7564,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4433014988899231,
|
|
"rewards/margins": -0.06983453035354614,
|
|
"rewards/rejected": -0.37346696853637695,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.9783621674278104e-05,
|
|
"logits/chosen": -2.402174949645996,
|
|
"logits/rejected": -2.5383033752441406,
|
|
"logps/chosen": -174.37954711914062,
|
|
"logps/rejected": -165.9345703125,
|
|
"loss": 0.6405,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5206565856933594,
|
|
"rewards/margins": 0.13473454117774963,
|
|
"rewards/rejected": -0.6553912162780762,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.978225201892667e-05,
|
|
"logits/chosen": -2.4111886024475098,
|
|
"logits/rejected": -2.3553833961486816,
|
|
"logps/chosen": -220.5755615234375,
|
|
"logps/rejected": -195.2405242919922,
|
|
"loss": 0.7531,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.4778462052345276,
|
|
"rewards/margins": -0.041694507002830505,
|
|
"rewards/rejected": -0.4361516833305359,
|
|
"step": 417
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.97808780613037e-05,
|
|
"logits/chosen": -2.307465076446533,
|
|
"logits/rejected": -2.3099443912506104,
|
|
"logps/chosen": -171.25372314453125,
|
|
"logps/rejected": -200.85305786132812,
|
|
"loss": 0.6326,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.445933997631073,
|
|
"rewards/margins": 0.2173961102962494,
|
|
"rewards/rejected": -0.663330078125,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.977949980164773e-05,
|
|
"logits/chosen": -2.3666648864746094,
|
|
"logits/rejected": -2.433722496032715,
|
|
"logps/chosen": -217.6551513671875,
|
|
"logps/rejected": -234.6415557861328,
|
|
"loss": 0.5846,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3358496427536011,
|
|
"rewards/margins": 0.309315949678421,
|
|
"rewards/rejected": -0.6451655626296997,
|
|
"step": 419
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.977811724019802e-05,
|
|
"logits/chosen": -2.336599111557007,
|
|
"logits/rejected": -2.409275770187378,
|
|
"logps/chosen": -191.2588653564453,
|
|
"logps/rejected": -208.126953125,
|
|
"loss": 0.6455,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4646199941635132,
|
|
"rewards/margins": 0.13836784660816193,
|
|
"rewards/rejected": -0.6029877662658691,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.9776730377194596e-05,
|
|
"logits/chosen": -2.1639509201049805,
|
|
"logits/rejected": -2.2069835662841797,
|
|
"logps/chosen": -203.13632202148438,
|
|
"logps/rejected": -169.3188018798828,
|
|
"loss": 0.7455,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.23145976662635803,
|
|
"rewards/margins": 0.013234168291091919,
|
|
"rewards/rejected": -0.24469394981861115,
|
|
"step": 421
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.9775339212878215e-05,
|
|
"logits/chosen": -2.1670339107513428,
|
|
"logits/rejected": -2.1488430500030518,
|
|
"logps/chosen": -166.50318908691406,
|
|
"logps/rejected": -169.4437713623047,
|
|
"loss": 0.698,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.2539255619049072,
|
|
"rewards/margins": 0.030521919950842857,
|
|
"rewards/rejected": -0.28444746136665344,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.977394374749039e-05,
|
|
"logits/chosen": -2.336040735244751,
|
|
"logits/rejected": -2.3590757846832275,
|
|
"logps/chosen": -150.48399353027344,
|
|
"logps/rejected": -154.54652404785156,
|
|
"loss": 0.667,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.30479297041893005,
|
|
"rewards/margins": 0.10228873789310455,
|
|
"rewards/rejected": -0.4070816934108734,
|
|
"step": 423
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.9772543981273374e-05,
|
|
"logits/chosen": -2.25268292427063,
|
|
"logits/rejected": -2.267082691192627,
|
|
"logps/chosen": -246.07342529296875,
|
|
"logps/rejected": -230.28424072265625,
|
|
"loss": 0.7204,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.42524397373199463,
|
|
"rewards/margins": -0.008991291746497154,
|
|
"rewards/rejected": -0.4162527322769165,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.977113991447017e-05,
|
|
"logits/chosen": -2.35402250289917,
|
|
"logits/rejected": -2.3599345684051514,
|
|
"logps/chosen": -163.17568969726562,
|
|
"logps/rejected": -175.94029235839844,
|
|
"loss": 0.7346,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.4037836790084839,
|
|
"rewards/margins": 0.02548382803797722,
|
|
"rewards/rejected": -0.4292675256729126,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.976973154732454e-05,
|
|
"logits/chosen": -2.3611769676208496,
|
|
"logits/rejected": -2.3299450874328613,
|
|
"logps/chosen": -205.6555633544922,
|
|
"logps/rejected": -213.8665771484375,
|
|
"loss": 0.7124,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.22228708863258362,
|
|
"rewards/margins": 0.031078480184078217,
|
|
"rewards/rejected": -0.25336551666259766,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.976831888008096e-05,
|
|
"logits/chosen": -2.316866636276245,
|
|
"logits/rejected": -2.299907684326172,
|
|
"logps/chosen": -168.08087158203125,
|
|
"logps/rejected": -160.0249786376953,
|
|
"loss": 0.6877,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.13129538297653198,
|
|
"rewards/margins": 0.04759233817458153,
|
|
"rewards/rejected": -0.17888770997524261,
|
|
"step": 427
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.976690191298469e-05,
|
|
"logits/chosen": -2.1294949054718018,
|
|
"logits/rejected": -2.1100010871887207,
|
|
"logps/chosen": -176.1876678466797,
|
|
"logps/rejected": -175.78619384765625,
|
|
"loss": 0.7597,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.16827335953712463,
|
|
"rewards/margins": -0.110628642141819,
|
|
"rewards/rejected": -0.05764475464820862,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.9765480646281716e-05,
|
|
"logits/chosen": -2.120499849319458,
|
|
"logits/rejected": -2.2186028957366943,
|
|
"logps/chosen": -178.4534912109375,
|
|
"logps/rejected": -200.6439971923828,
|
|
"loss": 0.6143,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.11943019181489944,
|
|
"rewards/margins": 0.28324708342552185,
|
|
"rewards/rejected": -0.4026772677898407,
|
|
"step": 429
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.976405508021877e-05,
|
|
"logits/chosen": -2.430680274963379,
|
|
"logits/rejected": -2.5126051902770996,
|
|
"logps/chosen": -201.0321502685547,
|
|
"logps/rejected": -231.42811584472656,
|
|
"loss": 0.7632,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.47548148036003113,
|
|
"rewards/margins": -0.046219632029533386,
|
|
"rewards/rejected": -0.42926183342933655,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.9762625215043334e-05,
|
|
"logits/chosen": -2.396533966064453,
|
|
"logits/rejected": -2.4125595092773438,
|
|
"logps/chosen": -236.99049377441406,
|
|
"logps/rejected": -253.93577575683594,
|
|
"loss": 0.7165,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4821574091911316,
|
|
"rewards/margins": -0.007029315456748009,
|
|
"rewards/rejected": -0.4751281142234802,
|
|
"step": 431
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9761191051003644e-05,
|
|
"logits/chosen": -2.1031394004821777,
|
|
"logits/rejected": -2.143960475921631,
|
|
"logps/chosen": -168.68551635742188,
|
|
"logps/rejected": -209.39537048339844,
|
|
"loss": 0.6219,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.1691710501909256,
|
|
"rewards/margins": 0.2633129060268402,
|
|
"rewards/rejected": -0.4324839413166046,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.975975258834867e-05,
|
|
"logits/chosen": -2.451735496520996,
|
|
"logits/rejected": -2.410836696624756,
|
|
"logps/chosen": -219.5807342529297,
|
|
"logps/rejected": -238.2914276123047,
|
|
"loss": 0.6939,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.26019036769866943,
|
|
"rewards/margins": 0.09737718105316162,
|
|
"rewards/rejected": -0.35756757855415344,
|
|
"step": 433
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9758309827328134e-05,
|
|
"logits/chosen": -2.3229422569274902,
|
|
"logits/rejected": -2.3514108657836914,
|
|
"logps/chosen": -181.05836486816406,
|
|
"logps/rejected": -211.49786376953125,
|
|
"loss": 0.7734,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.2722405195236206,
|
|
"rewards/margins": -0.07979361712932587,
|
|
"rewards/rejected": -0.19244688749313354,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9756862768192504e-05,
|
|
"logits/chosen": -2.3667490482330322,
|
|
"logits/rejected": -2.328806161880493,
|
|
"logps/chosen": -201.50511169433594,
|
|
"logps/rejected": -179.98643493652344,
|
|
"loss": 0.7353,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.3051126301288605,
|
|
"rewards/margins": -0.007734470069408417,
|
|
"rewards/rejected": -0.29737818241119385,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9755411411192996e-05,
|
|
"logits/chosen": -2.4068377017974854,
|
|
"logits/rejected": -2.532482862472534,
|
|
"logps/chosen": -186.65342712402344,
|
|
"logps/rejected": -205.83154296875,
|
|
"loss": 0.7144,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.17129601538181305,
|
|
"rewards/margins": -0.002290443517267704,
|
|
"rewards/rejected": -0.1690055876970291,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.975395575658156e-05,
|
|
"logits/chosen": -2.382935047149658,
|
|
"logits/rejected": -2.4527859687805176,
|
|
"logps/chosen": -200.10215759277344,
|
|
"logps/rejected": -207.49302673339844,
|
|
"loss": 0.7113,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.33162015676498413,
|
|
"rewards/margins": 0.029873518273234367,
|
|
"rewards/rejected": -0.36149367690086365,
|
|
"step": 437
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9752495804610916e-05,
|
|
"logits/chosen": -2.2459821701049805,
|
|
"logits/rejected": -2.2536497116088867,
|
|
"logps/chosen": -249.44638061523438,
|
|
"logps/rejected": -216.59483337402344,
|
|
"loss": 0.8425,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.3279973268508911,
|
|
"rewards/margins": -0.22765566408634186,
|
|
"rewards/rejected": -0.10034167766571045,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.9751031555534504e-05,
|
|
"logits/chosen": -2.416114330291748,
|
|
"logits/rejected": -2.5028560161590576,
|
|
"logps/chosen": -186.53961181640625,
|
|
"logps/rejected": -160.81617736816406,
|
|
"loss": 0.674,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.1654774248600006,
|
|
"rewards/margins": 0.07480952888727188,
|
|
"rewards/rejected": -0.2402869611978531,
|
|
"step": 439
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.9749563009606534e-05,
|
|
"logits/chosen": -2.4489054679870605,
|
|
"logits/rejected": -2.527320623397827,
|
|
"logps/chosen": -202.46507263183594,
|
|
"logps/rejected": -185.72853088378906,
|
|
"loss": 0.6135,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.22387953102588654,
|
|
"rewards/margins": 0.19476865231990814,
|
|
"rewards/rejected": -0.4186481833457947,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.9748090167081936e-05,
|
|
"logits/chosen": -2.147719144821167,
|
|
"logits/rejected": -2.249868154525757,
|
|
"logps/chosen": -152.69786071777344,
|
|
"logps/rejected": -190.48863220214844,
|
|
"loss": 0.6568,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.29554668068885803,
|
|
"rewards/margins": 0.11645788699388504,
|
|
"rewards/rejected": -0.4120045602321625,
|
|
"step": 441
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.974661302821641e-05,
|
|
"logits/chosen": -2.142338275909424,
|
|
"logits/rejected": -2.2330148220062256,
|
|
"logps/chosen": -157.3348388671875,
|
|
"logps/rejected": -194.91302490234375,
|
|
"loss": 0.7378,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.13629794120788574,
|
|
"rewards/margins": -0.030040191486477852,
|
|
"rewards/rejected": -0.10625775158405304,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.974513159326638e-05,
|
|
"logits/chosen": -2.551227569580078,
|
|
"logits/rejected": -2.5135083198547363,
|
|
"logps/chosen": -201.37612915039062,
|
|
"logps/rejected": -167.43287658691406,
|
|
"loss": 0.6417,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.25598540902137756,
|
|
"rewards/margins": 0.18629388511180878,
|
|
"rewards/rejected": -0.4422793388366699,
|
|
"step": 443
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.974364586248904e-05,
|
|
"logits/chosen": -2.174506425857544,
|
|
"logits/rejected": -2.1649577617645264,
|
|
"logps/chosen": -181.4037628173828,
|
|
"logps/rejected": -171.9210968017578,
|
|
"loss": 0.8572,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3619764745235443,
|
|
"rewards/margins": -0.19866357743740082,
|
|
"rewards/rejected": -0.1633128821849823,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.974215583614232e-05,
|
|
"logits/chosen": -2.3266444206237793,
|
|
"logits/rejected": -2.418710470199585,
|
|
"logps/chosen": -207.11448669433594,
|
|
"logps/rejected": -287.95574951171875,
|
|
"loss": 0.7347,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.41178515553474426,
|
|
"rewards/margins": -0.002230718731880188,
|
|
"rewards/rejected": -0.40955445170402527,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.974066151448488e-05,
|
|
"logits/chosen": -2.5598104000091553,
|
|
"logits/rejected": -2.559662342071533,
|
|
"logps/chosen": -231.59133911132812,
|
|
"logps/rejected": -239.7825469970703,
|
|
"loss": 0.7079,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.42168766260147095,
|
|
"rewards/margins": 0.02927880734205246,
|
|
"rewards/rejected": -0.4509664475917816,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.9739162897776146e-05,
|
|
"logits/chosen": -2.3569555282592773,
|
|
"logits/rejected": -2.3967788219451904,
|
|
"logps/chosen": -185.6054229736328,
|
|
"logps/rejected": -192.92169189453125,
|
|
"loss": 0.7437,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4934594929218292,
|
|
"rewards/margins": -0.05449381470680237,
|
|
"rewards/rejected": -0.43896567821502686,
|
|
"step": 447
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.973765998627628e-05,
|
|
"logits/chosen": -2.5715792179107666,
|
|
"logits/rejected": -2.5525949001312256,
|
|
"logps/chosen": -252.23037719726562,
|
|
"logps/rejected": -268.3929443359375,
|
|
"loss": 0.7203,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3784821331501007,
|
|
"rewards/margins": -0.01087496429681778,
|
|
"rewards/rejected": -0.3676071763038635,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.973615278024619e-05,
|
|
"logits/chosen": -2.441962718963623,
|
|
"logits/rejected": -2.481163501739502,
|
|
"logps/chosen": -161.66839599609375,
|
|
"logps/rejected": -192.90371704101562,
|
|
"loss": 0.6712,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3044736981391907,
|
|
"rewards/margins": 0.08197857439517975,
|
|
"rewards/rejected": -0.38645222783088684,
|
|
"step": 449
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.9734641279947535e-05,
|
|
"logits/chosen": -2.500675916671753,
|
|
"logits/rejected": -2.487545967102051,
|
|
"logps/chosen": -215.66336059570312,
|
|
"logps/rejected": -232.06967163085938,
|
|
"loss": 0.6796,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4935031235218048,
|
|
"rewards/margins": 0.12285958230495453,
|
|
"rewards/rejected": -0.6163626909255981,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.973312548564272e-05,
|
|
"logits/chosen": -2.4314112663269043,
|
|
"logits/rejected": -2.4294910430908203,
|
|
"logps/chosen": -170.31471252441406,
|
|
"logps/rejected": -159.71585083007812,
|
|
"loss": 0.7036,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.32274457812309265,
|
|
"rewards/margins": 0.005533523857593536,
|
|
"rewards/rejected": -0.328278124332428,
|
|
"step": 451
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.9731605397594884e-05,
|
|
"logits/chosen": -2.4075756072998047,
|
|
"logits/rejected": -2.32165789604187,
|
|
"logps/chosen": -186.2813720703125,
|
|
"logps/rejected": -199.41989135742188,
|
|
"loss": 0.655,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3598267138004303,
|
|
"rewards/margins": 0.11941798776388168,
|
|
"rewards/rejected": -0.4792447090148926,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.973008101606792e-05,
|
|
"logits/chosen": -2.422717571258545,
|
|
"logits/rejected": -2.5152597427368164,
|
|
"logps/chosen": -173.51361083984375,
|
|
"logps/rejected": -203.63848876953125,
|
|
"loss": 0.7204,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.25253090262413025,
|
|
"rewards/margins": 0.029171746224164963,
|
|
"rewards/rejected": -0.2817026376724243,
|
|
"step": 453
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.972855234132646e-05,
|
|
"logits/chosen": -2.517770767211914,
|
|
"logits/rejected": -2.501011848449707,
|
|
"logps/chosen": -202.05300903320312,
|
|
"logps/rejected": -191.65887451171875,
|
|
"loss": 0.7912,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.42360594868659973,
|
|
"rewards/margins": -0.1400626003742218,
|
|
"rewards/rejected": -0.28354334831237793,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.9727019373635895e-05,
|
|
"logits/chosen": -2.612367630004883,
|
|
"logits/rejected": -2.617859125137329,
|
|
"logps/chosen": -210.05746459960938,
|
|
"logps/rejected": -199.4877166748047,
|
|
"loss": 0.621,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.12031766027212143,
|
|
"rewards/margins": 0.193304181098938,
|
|
"rewards/rejected": -0.3136218190193176,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.972548211326235e-05,
|
|
"logits/chosen": -2.5540568828582764,
|
|
"logits/rejected": -2.587779998779297,
|
|
"logps/chosen": -183.94273376464844,
|
|
"logps/rejected": -219.0413818359375,
|
|
"loss": 0.6669,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.1985998898744583,
|
|
"rewards/margins": 0.12929418683052063,
|
|
"rewards/rejected": -0.32789406180381775,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.9723940560472705e-05,
|
|
"logits/chosen": -2.573051691055298,
|
|
"logits/rejected": -2.6791186332702637,
|
|
"logps/chosen": -193.5257568359375,
|
|
"logps/rejected": -189.68743896484375,
|
|
"loss": 0.7393,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.31839922070503235,
|
|
"rewards/margins": -0.04755366966128349,
|
|
"rewards/rejected": -0.27084553241729736,
|
|
"step": 457
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.972239471553457e-05,
|
|
"logits/chosen": -2.127181053161621,
|
|
"logits/rejected": -2.1366987228393555,
|
|
"logps/chosen": -180.3720703125,
|
|
"logps/rejected": -206.37628173828125,
|
|
"loss": 0.6858,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.14177769422531128,
|
|
"rewards/margins": 0.04955483227968216,
|
|
"rewards/rejected": -0.19133250415325165,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.97208445787163e-05,
|
|
"logits/chosen": -2.4841840267181396,
|
|
"logits/rejected": -2.576199531555176,
|
|
"logps/chosen": -175.8242645263672,
|
|
"logps/rejected": -205.9224090576172,
|
|
"loss": 0.7229,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.39802610874176025,
|
|
"rewards/margins": -0.00046849995851516724,
|
|
"rewards/rejected": -0.3975576162338257,
|
|
"step": 459
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.9719290150287026e-05,
|
|
"logits/chosen": -2.4974851608276367,
|
|
"logits/rejected": -2.552859306335449,
|
|
"logps/chosen": -204.64480590820312,
|
|
"logps/rejected": -229.373291015625,
|
|
"loss": 0.7028,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.09486077725887299,
|
|
"rewards/margins": 0.017538849264383316,
|
|
"rewards/rejected": -0.1123996153473854,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.9717731430516576e-05,
|
|
"logits/chosen": -2.5486257076263428,
|
|
"logits/rejected": -2.505126953125,
|
|
"logps/chosen": -216.0121307373047,
|
|
"logps/rejected": -204.5452880859375,
|
|
"loss": 0.6996,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.1612711399793625,
|
|
"rewards/margins": 0.06856641173362732,
|
|
"rewards/rejected": -0.229837566614151,
|
|
"step": 461
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.9716168419675555e-05,
|
|
"logits/chosen": -2.5866286754608154,
|
|
"logits/rejected": -2.667116403579712,
|
|
"logps/chosen": -198.39340209960938,
|
|
"logps/rejected": -250.51675415039062,
|
|
"loss": 0.6783,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3453162908554077,
|
|
"rewards/margins": 0.07628250867128372,
|
|
"rewards/rejected": -0.42159876227378845,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.9714601118035325e-05,
|
|
"logits/chosen": -2.339556932449341,
|
|
"logits/rejected": -2.2843379974365234,
|
|
"logps/chosen": -202.150634765625,
|
|
"logps/rejected": -192.74267578125,
|
|
"loss": 0.674,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.1799849569797516,
|
|
"rewards/margins": 0.09412900358438492,
|
|
"rewards/rejected": -0.2741139531135559,
|
|
"step": 463
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.971302952586796e-05,
|
|
"logits/chosen": -2.3514528274536133,
|
|
"logits/rejected": -2.313333749771118,
|
|
"logps/chosen": -216.48309326171875,
|
|
"logps/rejected": -205.1566925048828,
|
|
"loss": 0.7625,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.3306673467159271,
|
|
"rewards/margins": -0.07343533635139465,
|
|
"rewards/rejected": -0.25723204016685486,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.971145364344628e-05,
|
|
"logits/chosen": -2.475745916366577,
|
|
"logits/rejected": -2.5058016777038574,
|
|
"logps/chosen": -198.23568725585938,
|
|
"logps/rejected": -226.68544006347656,
|
|
"loss": 0.6412,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.38944318890571594,
|
|
"rewards/margins": 0.15158149600028992,
|
|
"rewards/rejected": -0.5410246849060059,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.970987347104389e-05,
|
|
"logits/chosen": -2.490628242492676,
|
|
"logits/rejected": -2.5108678340911865,
|
|
"logps/chosen": -180.23779296875,
|
|
"logps/rejected": -232.67515563964844,
|
|
"loss": 0.6863,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3212130665779114,
|
|
"rewards/margins": 0.09368535876274109,
|
|
"rewards/rejected": -0.41489848494529724,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.9708289008935096e-05,
|
|
"logits/chosen": -2.6224091053009033,
|
|
"logits/rejected": -2.6361618041992188,
|
|
"logps/chosen": -254.05960083007812,
|
|
"logps/rejected": -247.78634643554688,
|
|
"loss": 0.6909,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.40414828062057495,
|
|
"rewards/margins": 0.051116712391376495,
|
|
"rewards/rejected": -0.4552650451660156,
|
|
"step": 467
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.9706700257394966e-05,
|
|
"logits/chosen": -2.6673390865325928,
|
|
"logits/rejected": -2.6409029960632324,
|
|
"logps/chosen": -160.94146728515625,
|
|
"logps/rejected": -228.09808349609375,
|
|
"loss": 0.7103,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4117045998573303,
|
|
"rewards/margins": 0.047504693269729614,
|
|
"rewards/rejected": -0.45920926332473755,
|
|
"step": 468
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.970510721669932e-05,
|
|
"logits/chosen": -2.12448787689209,
|
|
"logits/rejected": -2.209928512573242,
|
|
"logps/chosen": -247.7496337890625,
|
|
"logps/rejected": -256.54254150390625,
|
|
"loss": 0.602,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.13797114789485931,
|
|
"rewards/margins": 0.25247055292129517,
|
|
"rewards/rejected": -0.3904416561126709,
|
|
"step": 469
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.97035098871247e-05,
|
|
"logits/chosen": -2.531524896621704,
|
|
"logits/rejected": -2.5506911277770996,
|
|
"logps/chosen": -239.89340209960938,
|
|
"logps/rejected": -232.53662109375,
|
|
"loss": 0.6477,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2409316897392273,
|
|
"rewards/margins": 0.11869845539331436,
|
|
"rewards/rejected": -0.35963016748428345,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.970190826894842e-05,
|
|
"logits/chosen": -2.4936788082122803,
|
|
"logits/rejected": -2.6132445335388184,
|
|
"logps/chosen": -193.3863067626953,
|
|
"logps/rejected": -196.8754119873047,
|
|
"loss": 0.6012,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.33218735456466675,
|
|
"rewards/margins": 0.244198739528656,
|
|
"rewards/rejected": -0.5763860940933228,
|
|
"step": 471
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.9700302362448517e-05,
|
|
"logits/chosen": -2.7358896732330322,
|
|
"logits/rejected": -2.860752582550049,
|
|
"logps/chosen": -193.43789672851562,
|
|
"logps/rejected": -240.47451782226562,
|
|
"loss": 0.6757,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.2744450271129608,
|
|
"rewards/margins": 0.07577859610319138,
|
|
"rewards/rejected": -0.3502236306667328,
|
|
"step": 472
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.9698692167903794e-05,
|
|
"logits/chosen": -2.581242799758911,
|
|
"logits/rejected": -2.7139995098114014,
|
|
"logps/chosen": -206.38645935058594,
|
|
"logps/rejected": -210.1461639404297,
|
|
"loss": 0.5954,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3665695786476135,
|
|
"rewards/margins": 0.26593247056007385,
|
|
"rewards/rejected": -0.632502019405365,
|
|
"step": 473
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.9697077685593766e-05,
|
|
"logits/chosen": -2.6083261966705322,
|
|
"logits/rejected": -2.6498053073883057,
|
|
"logps/chosen": -131.37237548828125,
|
|
"logps/rejected": -151.5311737060547,
|
|
"loss": 0.649,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.15027746558189392,
|
|
"rewards/margins": 0.126561239361763,
|
|
"rewards/rejected": -0.2768386900424957,
|
|
"step": 474
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.969545891579873e-05,
|
|
"logits/chosen": -2.359971046447754,
|
|
"logits/rejected": -2.3982698917388916,
|
|
"logps/chosen": -157.39181518554688,
|
|
"logps/rejected": -158.3177032470703,
|
|
"loss": 0.6222,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.10072011500597,
|
|
"rewards/margins": 0.24248063564300537,
|
|
"rewards/rejected": -0.34320077300071716,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.9693835858799696e-05,
|
|
"logits/chosen": -2.408668279647827,
|
|
"logits/rejected": -2.4540035724639893,
|
|
"logps/chosen": -190.36312866210938,
|
|
"logps/rejected": -204.20359802246094,
|
|
"loss": 0.6997,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.18467962741851807,
|
|
"rewards/margins": 0.015133664011955261,
|
|
"rewards/rejected": -0.19981329143047333,
|
|
"step": 476
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.9692208514878444e-05,
|
|
"logits/chosen": -2.270054340362549,
|
|
"logits/rejected": -2.177553653717041,
|
|
"logps/chosen": -197.311767578125,
|
|
"logps/rejected": -205.99948120117188,
|
|
"loss": 0.7255,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.2719980776309967,
|
|
"rewards/margins": -0.018820375204086304,
|
|
"rewards/rejected": -0.2531776428222656,
|
|
"step": 477
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.969057688431748e-05,
|
|
"logits/chosen": -2.4876270294189453,
|
|
"logits/rejected": -2.516055107116699,
|
|
"logps/chosen": -150.5498046875,
|
|
"logps/rejected": -173.20225524902344,
|
|
"loss": 0.6482,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2806074023246765,
|
|
"rewards/margins": 0.18431711196899414,
|
|
"rewards/rejected": -0.46492451429367065,
|
|
"step": 478
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.968894096740006e-05,
|
|
"logits/chosen": -2.699005365371704,
|
|
"logits/rejected": -2.706855297088623,
|
|
"logps/chosen": -208.8563232421875,
|
|
"logps/rejected": -229.45355224609375,
|
|
"loss": 0.6569,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3221612572669983,
|
|
"rewards/margins": 0.1469050496816635,
|
|
"rewards/rejected": -0.4690663516521454,
|
|
"step": 479
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.968730076441017e-05,
|
|
"logits/chosen": -2.752363681793213,
|
|
"logits/rejected": -2.8258612155914307,
|
|
"logps/chosen": -233.5675811767578,
|
|
"logps/rejected": -261.7062683105469,
|
|
"loss": 0.7128,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.604475200176239,
|
|
"rewards/margins": 0.022365085780620575,
|
|
"rewards/rejected": -0.6268402338027954,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.9685656275632575e-05,
|
|
"logits/chosen": -2.5177550315856934,
|
|
"logits/rejected": -2.510917901992798,
|
|
"logps/chosen": -217.02195739746094,
|
|
"logps/rejected": -189.55174255371094,
|
|
"loss": 0.6434,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3391067683696747,
|
|
"rewards/margins": 0.13497701287269592,
|
|
"rewards/rejected": -0.4740837812423706,
|
|
"step": 481
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.968400750135276e-05,
|
|
"logits/chosen": -2.540860652923584,
|
|
"logits/rejected": -2.582563877105713,
|
|
"logps/chosen": -194.17721557617188,
|
|
"logps/rejected": -198.79684448242188,
|
|
"loss": 0.6679,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.574460506439209,
|
|
"rewards/margins": 0.10229263454675674,
|
|
"rewards/rejected": -0.6767531633377075,
|
|
"step": 482
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.968235444185695e-05,
|
|
"logits/chosen": -2.7020184993743896,
|
|
"logits/rejected": -2.7175564765930176,
|
|
"logps/chosen": -211.08863830566406,
|
|
"logps/rejected": -235.41836547851562,
|
|
"loss": 0.7235,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5369959473609924,
|
|
"rewards/margins": 0.011504769325256348,
|
|
"rewards/rejected": -0.5485007762908936,
|
|
"step": 483
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.968069709743212e-05,
|
|
"logits/chosen": -2.4362404346466064,
|
|
"logits/rejected": -2.5247411727905273,
|
|
"logps/chosen": -149.7013397216797,
|
|
"logps/rejected": -185.23251342773438,
|
|
"loss": 0.5556,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.1017133966088295,
|
|
"rewards/margins": 0.3292151391506195,
|
|
"rewards/rejected": -0.4309285283088684,
|
|
"step": 484
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.9679035468365986e-05,
|
|
"logits/chosen": -2.3711907863616943,
|
|
"logits/rejected": -2.412506580352783,
|
|
"logps/chosen": -202.7202911376953,
|
|
"logps/rejected": -219.67977905273438,
|
|
"loss": 0.6658,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3317970931529999,
|
|
"rewards/margins": 0.12470696866512299,
|
|
"rewards/rejected": -0.45650404691696167,
|
|
"step": 485
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.967736955494703e-05,
|
|
"logits/chosen": -2.204392910003662,
|
|
"logits/rejected": -2.2046332359313965,
|
|
"logps/chosen": -195.41537475585938,
|
|
"logps/rejected": -196.8472442626953,
|
|
"loss": 0.7358,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.43655478954315186,
|
|
"rewards/margins": -0.000651337206363678,
|
|
"rewards/rejected": -0.4359034597873688,
|
|
"step": 486
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.9675699357464445e-05,
|
|
"logits/chosen": -2.198075771331787,
|
|
"logits/rejected": -2.1894662380218506,
|
|
"logps/chosen": -248.62149047851562,
|
|
"logps/rejected": -267.9733581542969,
|
|
"loss": 0.5947,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.14261871576309204,
|
|
"rewards/margins": 0.31586068868637085,
|
|
"rewards/rejected": -0.4584794044494629,
|
|
"step": 487
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.967402487620818e-05,
|
|
"logits/chosen": -2.5536584854125977,
|
|
"logits/rejected": -2.577322006225586,
|
|
"logps/chosen": -224.8876190185547,
|
|
"logps/rejected": -227.71383666992188,
|
|
"loss": 0.663,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.40421125292778015,
|
|
"rewards/margins": 0.11385571211576462,
|
|
"rewards/rejected": -0.5180670022964478,
|
|
"step": 488
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.9672346111468934e-05,
|
|
"logits/chosen": -2.508915424346924,
|
|
"logits/rejected": -2.624011993408203,
|
|
"logps/chosen": -181.42379760742188,
|
|
"logps/rejected": -231.30458068847656,
|
|
"loss": 0.6309,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.31741198897361755,
|
|
"rewards/margins": 0.24159038066864014,
|
|
"rewards/rejected": -0.5590023994445801,
|
|
"step": 489
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.967066306353816e-05,
|
|
"logits/chosen": -2.4504952430725098,
|
|
"logits/rejected": -2.4388234615325928,
|
|
"logps/chosen": -229.2549591064453,
|
|
"logps/rejected": -238.2850341796875,
|
|
"loss": 0.6922,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.47322726249694824,
|
|
"rewards/margins": 0.07009989768266678,
|
|
"rewards/rejected": -0.5433271527290344,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.966897573270801e-05,
|
|
"logits/chosen": -2.555962562561035,
|
|
"logits/rejected": -2.5629987716674805,
|
|
"logps/chosen": -172.14407348632812,
|
|
"logps/rejected": -168.1295166015625,
|
|
"loss": 0.7069,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2661295533180237,
|
|
"rewards/margins": 0.07032840698957443,
|
|
"rewards/rejected": -0.3364579379558563,
|
|
"step": 491
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.966728411927144e-05,
|
|
"logits/chosen": -2.4634742736816406,
|
|
"logits/rejected": -2.4717013835906982,
|
|
"logps/chosen": -206.756591796875,
|
|
"logps/rejected": -202.884033203125,
|
|
"loss": 0.7013,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5119624733924866,
|
|
"rewards/margins": 0.043263014405965805,
|
|
"rewards/rejected": -0.5552253723144531,
|
|
"step": 492
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.9665588223522096e-05,
|
|
"logits/chosen": -2.333207607269287,
|
|
"logits/rejected": -2.3922958374023438,
|
|
"logps/chosen": -210.04153442382812,
|
|
"logps/rejected": -239.76979064941406,
|
|
"loss": 0.7059,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.32687515020370483,
|
|
"rewards/margins": 0.09048546105623245,
|
|
"rewards/rejected": -0.41736066341400146,
|
|
"step": 493
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.96638880457544e-05,
|
|
"logits/chosen": -2.414531707763672,
|
|
"logits/rejected": -2.4198668003082275,
|
|
"logps/chosen": -225.68203735351562,
|
|
"logps/rejected": -252.04661560058594,
|
|
"loss": 0.6259,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.44954052567481995,
|
|
"rewards/margins": 0.19488830864429474,
|
|
"rewards/rejected": -0.6444287896156311,
|
|
"step": 494
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.9662183586263514e-05,
|
|
"logits/chosen": -2.523651361465454,
|
|
"logits/rejected": -2.4278194904327393,
|
|
"logps/chosen": -222.92274475097656,
|
|
"logps/rejected": -203.31668090820312,
|
|
"loss": 0.8255,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6954134106636047,
|
|
"rewards/margins": -0.12540754675865173,
|
|
"rewards/rejected": -0.5700058937072754,
|
|
"step": 495
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.966047484534533e-05,
|
|
"logits/chosen": -2.465467929840088,
|
|
"logits/rejected": -2.4773805141448975,
|
|
"logps/chosen": -215.38780212402344,
|
|
"logps/rejected": -222.23593139648438,
|
|
"loss": 0.6348,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.43675118684768677,
|
|
"rewards/margins": 0.19925275444984436,
|
|
"rewards/rejected": -0.6360039710998535,
|
|
"step": 496
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.965876182329648e-05,
|
|
"logits/chosen": -2.4241979122161865,
|
|
"logits/rejected": -2.4904754161834717,
|
|
"logps/chosen": -176.40281677246094,
|
|
"logps/rejected": -189.89398193359375,
|
|
"loss": 0.7316,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.35698071122169495,
|
|
"rewards/margins": 0.12830819189548492,
|
|
"rewards/rejected": -0.48528894782066345,
|
|
"step": 497
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.965704452041437e-05,
|
|
"logits/chosen": -2.4424967765808105,
|
|
"logits/rejected": -2.447554588317871,
|
|
"logps/chosen": -186.7346954345703,
|
|
"logps/rejected": -202.56729125976562,
|
|
"loss": 0.6988,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.379626601934433,
|
|
"rewards/margins": 0.05914265662431717,
|
|
"rewards/rejected": -0.43876922130584717,
|
|
"step": 498
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.9655322936997115e-05,
|
|
"logits/chosen": -2.384314775466919,
|
|
"logits/rejected": -2.6753270626068115,
|
|
"logps/chosen": -168.5001220703125,
|
|
"logps/rejected": -201.58462524414062,
|
|
"loss": 0.5711,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.30074769258499146,
|
|
"rewards/margins": 0.31708773970603943,
|
|
"rewards/rejected": -0.6178354024887085,
|
|
"step": 499
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.9653597073343594e-05,
|
|
"logits/chosen": -2.44107723236084,
|
|
"logits/rejected": -2.518571615219116,
|
|
"logps/chosen": -190.8749542236328,
|
|
"logps/rejected": -195.59471130371094,
|
|
"loss": 0.6465,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3541903495788574,
|
|
"rewards/margins": 0.2186165750026703,
|
|
"rewards/rejected": -0.5728069543838501,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.9651866929753424e-05,
|
|
"logits/chosen": -2.580932140350342,
|
|
"logits/rejected": -2.6375184059143066,
|
|
"logps/chosen": -201.61231994628906,
|
|
"logps/rejected": -195.1832275390625,
|
|
"loss": 0.6788,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4448980689048767,
|
|
"rewards/margins": 0.1541019082069397,
|
|
"rewards/rejected": -0.5990000367164612,
|
|
"step": 501
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.9650132506526944e-05,
|
|
"logits/chosen": -2.601297378540039,
|
|
"logits/rejected": -2.5584542751312256,
|
|
"logps/chosen": -188.46102905273438,
|
|
"logps/rejected": -176.1941680908203,
|
|
"loss": 0.6884,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.32628124952316284,
|
|
"rewards/margins": 0.053992465138435364,
|
|
"rewards/rejected": -0.3802737295627594,
|
|
"step": 502
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.9648393803965284e-05,
|
|
"logits/chosen": -2.594449281692505,
|
|
"logits/rejected": -2.6243650913238525,
|
|
"logps/chosen": -180.65838623046875,
|
|
"logps/rejected": -173.06399536132812,
|
|
"loss": 0.608,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3709149658679962,
|
|
"rewards/margins": 0.24527978897094727,
|
|
"rewards/rejected": -0.6161947846412659,
|
|
"step": 503
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.9646650822370265e-05,
|
|
"logits/chosen": -2.623140811920166,
|
|
"logits/rejected": -2.6087071895599365,
|
|
"logps/chosen": -175.10757446289062,
|
|
"logps/rejected": -198.5271759033203,
|
|
"loss": 0.6623,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.34849774837493896,
|
|
"rewards/margins": 0.2720443308353424,
|
|
"rewards/rejected": -0.6205421090126038,
|
|
"step": 504
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.964490356204449e-05,
|
|
"logits/chosen": -2.3962643146514893,
|
|
"logits/rejected": -2.4296982288360596,
|
|
"logps/chosen": -171.71841430664062,
|
|
"logps/rejected": -189.31715393066406,
|
|
"loss": 0.7731,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5484548211097717,
|
|
"rewards/margins": 0.05871535837650299,
|
|
"rewards/rejected": -0.6071702241897583,
|
|
"step": 505
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.964315202329127e-05,
|
|
"logits/chosen": -2.1054232120513916,
|
|
"logits/rejected": -2.0613820552825928,
|
|
"logps/chosen": -196.4398956298828,
|
|
"logps/rejected": -197.6771240234375,
|
|
"loss": 0.6993,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4299621880054474,
|
|
"rewards/margins": 0.10146878659725189,
|
|
"rewards/rejected": -0.5314310193061829,
|
|
"step": 506
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.96413962064147e-05,
|
|
"logits/chosen": -2.4437365531921387,
|
|
"logits/rejected": -2.6009879112243652,
|
|
"logps/chosen": -218.52403259277344,
|
|
"logps/rejected": -253.1461639404297,
|
|
"loss": 0.5733,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.4768885374069214,
|
|
"rewards/margins": 0.29149848222732544,
|
|
"rewards/rejected": -0.7683870196342468,
|
|
"step": 507
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.963963611171957e-05,
|
|
"logits/chosen": -2.737818479537964,
|
|
"logits/rejected": -2.8296992778778076,
|
|
"logps/chosen": -194.77084350585938,
|
|
"logps/rejected": -196.74485778808594,
|
|
"loss": 0.6388,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4071928560733795,
|
|
"rewards/margins": 0.17611998319625854,
|
|
"rewards/rejected": -0.5833128690719604,
|
|
"step": 508
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.9637871739511454e-05,
|
|
"logits/chosen": -2.5671932697296143,
|
|
"logits/rejected": -2.4917585849761963,
|
|
"logps/chosen": -208.9228057861328,
|
|
"logps/rejected": -204.2896270751953,
|
|
"loss": 0.7151,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.38905590772628784,
|
|
"rewards/margins": 0.02341422438621521,
|
|
"rewards/rejected": -0.41247016191482544,
|
|
"step": 509
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.963610309009665e-05,
|
|
"logits/chosen": -2.3518598079681396,
|
|
"logits/rejected": -2.3721611499786377,
|
|
"logps/chosen": -154.779296875,
|
|
"logps/rejected": -190.75613403320312,
|
|
"loss": 0.6552,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.35764750838279724,
|
|
"rewards/margins": 0.1814635694026947,
|
|
"rewards/rejected": -0.5391110777854919,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.9634330163782205e-05,
|
|
"logits/chosen": -2.432107448577881,
|
|
"logits/rejected": -2.495102882385254,
|
|
"logps/chosen": -248.51341247558594,
|
|
"logps/rejected": -233.74266052246094,
|
|
"loss": 0.7034,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.45434871315956116,
|
|
"rewards/margins": 0.08803940564393997,
|
|
"rewards/rejected": -0.5423881411552429,
|
|
"step": 511
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.963255296087589e-05,
|
|
"logits/chosen": -2.4448506832122803,
|
|
"logits/rejected": -2.462013006210327,
|
|
"logps/chosen": -143.73773193359375,
|
|
"logps/rejected": -182.57113647460938,
|
|
"loss": 0.6068,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3137112855911255,
|
|
"rewards/margins": 0.21627672016620636,
|
|
"rewards/rejected": -0.5299879908561707,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.9630771481686244e-05,
|
|
"logits/chosen": -2.4147915840148926,
|
|
"logits/rejected": -2.368267297744751,
|
|
"logps/chosen": -189.82333374023438,
|
|
"logps/rejected": -175.15501403808594,
|
|
"loss": 0.6099,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.34441184997558594,
|
|
"rewards/margins": 0.29255932569503784,
|
|
"rewards/rejected": -0.636971116065979,
|
|
"step": 513
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.9628985726522535e-05,
|
|
"logits/chosen": -2.424818992614746,
|
|
"logits/rejected": -2.4481663703918457,
|
|
"logps/chosen": -204.38375854492188,
|
|
"logps/rejected": -204.50376892089844,
|
|
"loss": 0.6846,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4307776689529419,
|
|
"rewards/margins": 0.06390251964330673,
|
|
"rewards/rejected": -0.4946801960468292,
|
|
"step": 514
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.9627195695694774e-05,
|
|
"logits/chosen": -2.523521661758423,
|
|
"logits/rejected": -2.5060675144195557,
|
|
"logps/chosen": -163.70565795898438,
|
|
"logps/rejected": -207.99964904785156,
|
|
"loss": 0.6256,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.32005366683006287,
|
|
"rewards/margins": 0.24324071407318115,
|
|
"rewards/rejected": -0.5632944107055664,
|
|
"step": 515
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9625401389513715e-05,
|
|
"logits/chosen": -2.4181556701660156,
|
|
"logits/rejected": -2.369680881500244,
|
|
"logps/chosen": -205.77340698242188,
|
|
"logps/rejected": -155.3488311767578,
|
|
"loss": 0.8924,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.7356570363044739,
|
|
"rewards/margins": -0.26000097393989563,
|
|
"rewards/rejected": -0.47565609216690063,
|
|
"step": 516
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9623602808290855e-05,
|
|
"logits/chosen": -2.544684410095215,
|
|
"logits/rejected": -2.5888850688934326,
|
|
"logps/chosen": -229.81246948242188,
|
|
"logps/rejected": -251.07574462890625,
|
|
"loss": 0.6031,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4079195559024811,
|
|
"rewards/margins": 0.29860368371009827,
|
|
"rewards/rejected": -0.7065231800079346,
|
|
"step": 517
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.962179995233843e-05,
|
|
"logits/chosen": -2.397979259490967,
|
|
"logits/rejected": -2.5106217861175537,
|
|
"logps/chosen": -157.80418395996094,
|
|
"logps/rejected": -175.9936065673828,
|
|
"loss": 0.6093,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.44883906841278076,
|
|
"rewards/margins": 0.32764846086502075,
|
|
"rewards/rejected": -0.7764875888824463,
|
|
"step": 518
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.961999282196943e-05,
|
|
"logits/chosen": -2.4970529079437256,
|
|
"logits/rejected": -2.582210063934326,
|
|
"logps/chosen": -194.96694946289062,
|
|
"logps/rejected": -198.24681091308594,
|
|
"loss": 0.6883,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.37289613485336304,
|
|
"rewards/margins": 0.0580718033015728,
|
|
"rewards/rejected": -0.4309679865837097,
|
|
"step": 519
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9618181417497566e-05,
|
|
"logits/chosen": -2.469606399536133,
|
|
"logits/rejected": -2.54701566696167,
|
|
"logps/chosen": -215.56979370117188,
|
|
"logps/rejected": -233.58209228515625,
|
|
"loss": 0.691,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5863683819770813,
|
|
"rewards/margins": 0.03697293624281883,
|
|
"rewards/rejected": -0.6233413815498352,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.961636573923732e-05,
|
|
"logits/chosen": -2.4047884941101074,
|
|
"logits/rejected": -2.414797782897949,
|
|
"logps/chosen": -226.51625061035156,
|
|
"logps/rejected": -212.8037872314453,
|
|
"loss": 0.6559,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4074936807155609,
|
|
"rewards/margins": 0.16430170834064484,
|
|
"rewards/rejected": -0.5717953443527222,
|
|
"step": 521
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9614545787503886e-05,
|
|
"logits/chosen": -2.4261462688446045,
|
|
"logits/rejected": -2.482429027557373,
|
|
"logps/chosen": -197.58526611328125,
|
|
"logps/rejected": -215.66322326660156,
|
|
"loss": 0.6091,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4302670359611511,
|
|
"rewards/margins": 0.2678428888320923,
|
|
"rewards/rejected": -0.6981098651885986,
|
|
"step": 522
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.961272156261321e-05,
|
|
"logits/chosen": -2.22928786277771,
|
|
"logits/rejected": -2.3709988594055176,
|
|
"logps/chosen": -157.08120727539062,
|
|
"logps/rejected": -186.32533264160156,
|
|
"loss": 0.5613,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.32108116149902344,
|
|
"rewards/margins": 0.476359486579895,
|
|
"rewards/rejected": -0.7974405884742737,
|
|
"step": 523
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.961089306488199e-05,
|
|
"logits/chosen": -2.3764328956604004,
|
|
"logits/rejected": -2.423450231552124,
|
|
"logps/chosen": -225.89686584472656,
|
|
"logps/rejected": -250.8079376220703,
|
|
"loss": 0.6351,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4355972707271576,
|
|
"rewards/margins": 0.2791078984737396,
|
|
"rewards/rejected": -0.7147052884101868,
|
|
"step": 524
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.960906029462766e-05,
|
|
"logits/chosen": -2.338782548904419,
|
|
"logits/rejected": -2.423982858657837,
|
|
"logps/chosen": -210.87510681152344,
|
|
"logps/rejected": -207.5828857421875,
|
|
"loss": 0.7054,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5041165351867676,
|
|
"rewards/margins": 0.092525914311409,
|
|
"rewards/rejected": -0.5966423749923706,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.96072232521684e-05,
|
|
"logits/chosen": -2.4865126609802246,
|
|
"logits/rejected": -2.55895733833313,
|
|
"logps/chosen": -290.5992431640625,
|
|
"logps/rejected": -275.2317810058594,
|
|
"loss": 0.6608,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5548466444015503,
|
|
"rewards/margins": 0.17232416570186615,
|
|
"rewards/rejected": -0.7271708250045776,
|
|
"step": 526
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.96053819378231e-05,
|
|
"logits/chosen": -2.6567628383636475,
|
|
"logits/rejected": -2.5933961868286133,
|
|
"logps/chosen": -183.54367065429688,
|
|
"logps/rejected": -175.34194946289062,
|
|
"loss": 0.7362,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.7040901780128479,
|
|
"rewards/margins": -0.04544142633676529,
|
|
"rewards/rejected": -0.6586487889289856,
|
|
"step": 527
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.960353635191145e-05,
|
|
"logits/chosen": -2.514308452606201,
|
|
"logits/rejected": -2.5751612186431885,
|
|
"logps/chosen": -171.15457153320312,
|
|
"logps/rejected": -163.67311096191406,
|
|
"loss": 0.606,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.340698778629303,
|
|
"rewards/margins": 0.42078471183776855,
|
|
"rewards/rejected": -0.7614834904670715,
|
|
"step": 528
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9601686494753826e-05,
|
|
"logits/chosen": -2.318474292755127,
|
|
"logits/rejected": -2.2866580486297607,
|
|
"logps/chosen": -155.76071166992188,
|
|
"logps/rejected": -156.44679260253906,
|
|
"loss": 0.7173,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.30408087372779846,
|
|
"rewards/margins": -0.002507832832634449,
|
|
"rewards/rejected": -0.3015730381011963,
|
|
"step": 529
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.959983236667138e-05,
|
|
"logits/chosen": -2.577028274536133,
|
|
"logits/rejected": -2.6590182781219482,
|
|
"logps/chosen": -142.8551483154297,
|
|
"logps/rejected": -166.59149169921875,
|
|
"loss": 0.6867,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5840120315551758,
|
|
"rewards/margins": 0.07422871142625809,
|
|
"rewards/rejected": -0.6582406759262085,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.959797396798599e-05,
|
|
"logits/chosen": -2.409472703933716,
|
|
"logits/rejected": -2.4960474967956543,
|
|
"logps/chosen": -150.38006591796875,
|
|
"logps/rejected": -176.5684814453125,
|
|
"loss": 0.6631,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5444418787956238,
|
|
"rewards/margins": 0.12680965662002563,
|
|
"rewards/rejected": -0.6712515354156494,
|
|
"step": 531
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9596111299020284e-05,
|
|
"logits/chosen": -2.336719512939453,
|
|
"logits/rejected": -2.3440167903900146,
|
|
"logps/chosen": -184.50473022460938,
|
|
"logps/rejected": -187.92713928222656,
|
|
"loss": 0.627,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.33718907833099365,
|
|
"rewards/margins": 0.19500240683555603,
|
|
"rewards/rejected": -0.5321914553642273,
|
|
"step": 532
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.959424436009762e-05,
|
|
"logits/chosen": -2.1330745220184326,
|
|
"logits/rejected": -2.2025442123413086,
|
|
"logps/chosen": -255.2789306640625,
|
|
"logps/rejected": -287.85736083984375,
|
|
"loss": 0.5781,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.6389483213424683,
|
|
"rewards/margins": 0.5605921149253845,
|
|
"rewards/rejected": -1.199540376663208,
|
|
"step": 533
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9592373151542105e-05,
|
|
"logits/chosen": -2.783942699432373,
|
|
"logits/rejected": -2.766120672225952,
|
|
"logps/chosen": -185.41404724121094,
|
|
"logps/rejected": -203.95669555664062,
|
|
"loss": 0.7158,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4609447717666626,
|
|
"rewards/margins": 0.003200806677341461,
|
|
"rewards/rejected": -0.46414560079574585,
|
|
"step": 534
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.959049767367859e-05,
|
|
"logits/chosen": -2.4105114936828613,
|
|
"logits/rejected": -2.6107630729675293,
|
|
"logps/chosen": -172.59347534179688,
|
|
"logps/rejected": -212.30836486816406,
|
|
"loss": 0.6269,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5956838726997375,
|
|
"rewards/margins": 0.2615227699279785,
|
|
"rewards/rejected": -0.8572065830230713,
|
|
"step": 535
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.958861792683266e-05,
|
|
"logits/chosen": -2.1016757488250732,
|
|
"logits/rejected": -2.1774649620056152,
|
|
"logps/chosen": -204.3402862548828,
|
|
"logps/rejected": -215.18890380859375,
|
|
"loss": 0.6164,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.49696028232574463,
|
|
"rewards/margins": 0.2144584059715271,
|
|
"rewards/rejected": -0.7114187479019165,
|
|
"step": 536
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.958673391133065e-05,
|
|
"logits/chosen": -2.6260104179382324,
|
|
"logits/rejected": -2.648331880569458,
|
|
"logps/chosen": -180.17063903808594,
|
|
"logps/rejected": -175.27981567382812,
|
|
"loss": 0.7315,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7630718946456909,
|
|
"rewards/margins": 0.05031493306159973,
|
|
"rewards/rejected": -0.813386857509613,
|
|
"step": 537
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.958484562749963e-05,
|
|
"logits/chosen": -2.3528480529785156,
|
|
"logits/rejected": -2.35026216506958,
|
|
"logps/chosen": -240.7522735595703,
|
|
"logps/rejected": -284.01177978515625,
|
|
"loss": 0.6558,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5610861778259277,
|
|
"rewards/margins": 0.1676333099603653,
|
|
"rewards/rejected": -0.7287194132804871,
|
|
"step": 538
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9582953075667406e-05,
|
|
"logits/chosen": -2.7893500328063965,
|
|
"logits/rejected": -2.695694923400879,
|
|
"logps/chosen": -213.7672119140625,
|
|
"logps/rejected": -197.13304138183594,
|
|
"loss": 0.7034,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7025653719902039,
|
|
"rewards/margins": 0.08315228670835495,
|
|
"rewards/rejected": -0.7857176661491394,
|
|
"step": 539
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.958105625616253e-05,
|
|
"logits/chosen": -2.6460516452789307,
|
|
"logits/rejected": -2.6226911544799805,
|
|
"logps/chosen": -186.30870056152344,
|
|
"logps/rejected": -169.23074340820312,
|
|
"loss": 0.684,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4503140449523926,
|
|
"rewards/margins": 0.0906258150935173,
|
|
"rewards/rejected": -0.5409399271011353,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.95791551693143e-05,
|
|
"logits/chosen": -2.6068363189697266,
|
|
"logits/rejected": -2.645115852355957,
|
|
"logps/chosen": -225.66354370117188,
|
|
"logps/rejected": -248.78900146484375,
|
|
"loss": 0.7087,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.7523027062416077,
|
|
"rewards/margins": 0.06183048337697983,
|
|
"rewards/rejected": -0.8141331672668457,
|
|
"step": 541
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.957724981545275e-05,
|
|
"logits/chosen": -2.472942352294922,
|
|
"logits/rejected": -2.5449466705322266,
|
|
"logps/chosen": -165.9869384765625,
|
|
"logps/rejected": -195.7375946044922,
|
|
"loss": 0.6983,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5627427697181702,
|
|
"rewards/margins": 0.17754918336868286,
|
|
"rewards/rejected": -0.7402920126914978,
|
|
"step": 542
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.957534019490865e-05,
|
|
"logits/chosen": -2.6267778873443604,
|
|
"logits/rejected": -2.8627214431762695,
|
|
"logps/chosen": -208.47406005859375,
|
|
"logps/rejected": -230.64390563964844,
|
|
"loss": 0.5822,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.6202707290649414,
|
|
"rewards/margins": 0.35299497842788696,
|
|
"rewards/rejected": -0.9732657074928284,
|
|
"step": 543
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.957342630801353e-05,
|
|
"logits/chosen": -2.302680730819702,
|
|
"logits/rejected": -2.4394922256469727,
|
|
"logps/chosen": -169.56790161132812,
|
|
"logps/rejected": -198.37738037109375,
|
|
"loss": 0.6269,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.44470933079719543,
|
|
"rewards/margins": 0.24785885214805603,
|
|
"rewards/rejected": -0.6925681233406067,
|
|
"step": 544
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.957150815509963e-05,
|
|
"logits/chosen": -2.454188823699951,
|
|
"logits/rejected": -2.548893451690674,
|
|
"logps/chosen": -170.51040649414062,
|
|
"logps/rejected": -232.72727966308594,
|
|
"loss": 0.7024,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3639673590660095,
|
|
"rewards/margins": 0.021631799638271332,
|
|
"rewards/rejected": -0.38559919595718384,
|
|
"step": 545
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9569585736499945e-05,
|
|
"logits/chosen": -2.337388515472412,
|
|
"logits/rejected": -2.4017117023468018,
|
|
"logps/chosen": -177.1454620361328,
|
|
"logps/rejected": -191.3271484375,
|
|
"loss": 0.5661,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3925960958003998,
|
|
"rewards/margins": 0.34767282009124756,
|
|
"rewards/rejected": -0.740268886089325,
|
|
"step": 546
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9567659052548234e-05,
|
|
"logits/chosen": -2.653782367706299,
|
|
"logits/rejected": -2.6659772396087646,
|
|
"logps/chosen": -198.43539428710938,
|
|
"logps/rejected": -176.4965362548828,
|
|
"loss": 0.7005,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5790706872940063,
|
|
"rewards/margins": 0.11143729835748672,
|
|
"rewards/rejected": -0.6905080080032349,
|
|
"step": 547
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.956572810357896e-05,
|
|
"logits/chosen": -2.5503158569335938,
|
|
"logits/rejected": -2.520476818084717,
|
|
"logps/chosen": -190.98855590820312,
|
|
"logps/rejected": -173.04501342773438,
|
|
"loss": 0.7238,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5375619530677795,
|
|
"rewards/margins": 0.01934710703790188,
|
|
"rewards/rejected": -0.5569091439247131,
|
|
"step": 548
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.956379288992733e-05,
|
|
"logits/chosen": -2.2868709564208984,
|
|
"logits/rejected": -2.358032464981079,
|
|
"logps/chosen": -163.09713745117188,
|
|
"logps/rejected": -187.16030883789062,
|
|
"loss": 0.5837,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3862106502056122,
|
|
"rewards/margins": 0.4660491943359375,
|
|
"rewards/rejected": -0.8522598147392273,
|
|
"step": 549
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.956185341192933e-05,
|
|
"logits/chosen": -2.522181987762451,
|
|
"logits/rejected": -2.5097920894622803,
|
|
"logps/chosen": -167.14633178710938,
|
|
"logps/rejected": -205.06365966796875,
|
|
"loss": 0.7648,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5081343650817871,
|
|
"rewards/margins": -0.012615025043487549,
|
|
"rewards/rejected": -0.4955193102359772,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9559909669921635e-05,
|
|
"logits/chosen": -2.3636226654052734,
|
|
"logits/rejected": -2.438326120376587,
|
|
"logps/chosen": -197.59310913085938,
|
|
"logps/rejected": -186.26112365722656,
|
|
"loss": 0.7066,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5027388334274292,
|
|
"rewards/margins": 0.07569919526576996,
|
|
"rewards/rejected": -0.5784379839897156,
|
|
"step": 551
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9557961664241694e-05,
|
|
"logits/chosen": -2.425126552581787,
|
|
"logits/rejected": -2.5314579010009766,
|
|
"logps/chosen": -181.5507049560547,
|
|
"logps/rejected": -211.9708251953125,
|
|
"loss": 0.7133,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6780929565429688,
|
|
"rewards/margins": 0.19112253189086914,
|
|
"rewards/rejected": -0.8692154884338379,
|
|
"step": 552
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.955600939522769e-05,
|
|
"logits/chosen": -2.52880597114563,
|
|
"logits/rejected": -2.5222525596618652,
|
|
"logps/chosen": -178.28521728515625,
|
|
"logps/rejected": -210.90565490722656,
|
|
"loss": 0.587,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5996692180633545,
|
|
"rewards/margins": 0.25250738859176636,
|
|
"rewards/rejected": -0.8521764874458313,
|
|
"step": 553
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.955405286321854e-05,
|
|
"logits/chosen": -2.3732471466064453,
|
|
"logits/rejected": -2.4644832611083984,
|
|
"logps/chosen": -210.51051330566406,
|
|
"logps/rejected": -208.96826171875,
|
|
"loss": 0.5982,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.5088058114051819,
|
|
"rewards/margins": 0.2920345067977905,
|
|
"rewards/rejected": -0.8008402585983276,
|
|
"step": 554
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.95520920685539e-05,
|
|
"logits/chosen": -2.4812893867492676,
|
|
"logits/rejected": -2.5562007427215576,
|
|
"logps/chosen": -170.83322143554688,
|
|
"logps/rejected": -192.9698486328125,
|
|
"loss": 0.5716,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.6304159164428711,
|
|
"rewards/margins": 0.37207356095314026,
|
|
"rewards/rejected": -1.002489447593689,
|
|
"step": 555
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9550127011574176e-05,
|
|
"logits/chosen": -2.281599760055542,
|
|
"logits/rejected": -2.3443262577056885,
|
|
"logps/chosen": -188.46249389648438,
|
|
"logps/rejected": -179.5142364501953,
|
|
"loss": 0.6338,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.41307827830314636,
|
|
"rewards/margins": 0.17765894532203674,
|
|
"rewards/rejected": -0.5907372236251831,
|
|
"step": 556
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.95481576926205e-05,
|
|
"logits/chosen": -2.584645986557007,
|
|
"logits/rejected": -2.604243755340576,
|
|
"logps/chosen": -181.7239990234375,
|
|
"logps/rejected": -209.90145874023438,
|
|
"loss": 0.5821,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7002069354057312,
|
|
"rewards/margins": 0.43418359756469727,
|
|
"rewards/rejected": -1.1343904733657837,
|
|
"step": 557
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.954618411203476e-05,
|
|
"logits/chosen": -2.679819107055664,
|
|
"logits/rejected": -2.7520828247070312,
|
|
"logps/chosen": -217.54901123046875,
|
|
"logps/rejected": -236.81210327148438,
|
|
"loss": 0.6187,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.734368085861206,
|
|
"rewards/margins": 0.29621487855911255,
|
|
"rewards/rejected": -1.0305829048156738,
|
|
"step": 558
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.954420627015957e-05,
|
|
"logits/chosen": -2.530925750732422,
|
|
"logits/rejected": -2.5949485301971436,
|
|
"logps/chosen": -190.8072509765625,
|
|
"logps/rejected": -226.07455444335938,
|
|
"loss": 0.5481,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5543414950370789,
|
|
"rewards/margins": 0.42910295724868774,
|
|
"rewards/rejected": -0.9834444522857666,
|
|
"step": 559
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.954222416733829e-05,
|
|
"logits/chosen": -2.4444892406463623,
|
|
"logits/rejected": -2.4189417362213135,
|
|
"logps/chosen": -275.56109619140625,
|
|
"logps/rejected": -248.05233764648438,
|
|
"loss": 0.7178,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5780796408653259,
|
|
"rewards/margins": 0.0615503191947937,
|
|
"rewards/rejected": -0.6396299600601196,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.954023780391501e-05,
|
|
"logits/chosen": -2.360807180404663,
|
|
"logits/rejected": -2.308104991912842,
|
|
"logps/chosen": -172.66574096679688,
|
|
"logps/rejected": -191.79412841796875,
|
|
"loss": 0.5558,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.709125816822052,
|
|
"rewards/margins": 0.38629868626594543,
|
|
"rewards/rejected": -1.0954244136810303,
|
|
"step": 561
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953824718023459e-05,
|
|
"logits/chosen": -2.140808582305908,
|
|
"logits/rejected": -2.2450931072235107,
|
|
"logps/chosen": -135.81964111328125,
|
|
"logps/rejected": -176.7286376953125,
|
|
"loss": 0.7815,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.7152988910675049,
|
|
"rewards/margins": 0.006897151470184326,
|
|
"rewards/rejected": -0.722196102142334,
|
|
"step": 562
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953625229664259e-05,
|
|
"logits/chosen": -2.2745089530944824,
|
|
"logits/rejected": -2.2524397373199463,
|
|
"logps/chosen": -174.69879150390625,
|
|
"logps/rejected": -193.57516479492188,
|
|
"loss": 0.8268,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6095337867736816,
|
|
"rewards/margins": -0.15611900389194489,
|
|
"rewards/rejected": -0.45341476798057556,
|
|
"step": 563
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953425315348534e-05,
|
|
"logits/chosen": -2.4158096313476562,
|
|
"logits/rejected": -2.5616989135742188,
|
|
"logps/chosen": -161.41812133789062,
|
|
"logps/rejected": -195.5721893310547,
|
|
"loss": 0.5776,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4711029529571533,
|
|
"rewards/margins": 0.4865836799144745,
|
|
"rewards/rejected": -0.957686722278595,
|
|
"step": 564
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953224975110988e-05,
|
|
"logits/chosen": -2.3829874992370605,
|
|
"logits/rejected": -2.3937342166900635,
|
|
"logps/chosen": -168.71534729003906,
|
|
"logps/rejected": -178.2794189453125,
|
|
"loss": 0.7187,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5582124590873718,
|
|
"rewards/margins": -0.010660316795110703,
|
|
"rewards/rejected": -0.5475521087646484,
|
|
"step": 565
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9530242089864026e-05,
|
|
"logits/chosen": -2.3829197883605957,
|
|
"logits/rejected": -2.462663173675537,
|
|
"logps/chosen": -173.0697021484375,
|
|
"logps/rejected": -207.88348388671875,
|
|
"loss": 0.6547,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4435291290283203,
|
|
"rewards/margins": 0.13191349804401398,
|
|
"rewards/rejected": -0.5754426717758179,
|
|
"step": 566
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9528230170096305e-05,
|
|
"logits/chosen": -2.319032907485962,
|
|
"logits/rejected": -2.3326690196990967,
|
|
"logps/chosen": -169.28501892089844,
|
|
"logps/rejected": -194.19927978515625,
|
|
"loss": 0.7851,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7481305599212646,
|
|
"rewards/margins": 0.09495264291763306,
|
|
"rewards/rejected": -0.8430831432342529,
|
|
"step": 567
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.952621399215598e-05,
|
|
"logits/chosen": -2.630171537399292,
|
|
"logits/rejected": -2.617821455001831,
|
|
"logps/chosen": -192.14907836914062,
|
|
"logps/rejected": -193.33782958984375,
|
|
"loss": 0.687,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.47104740142822266,
|
|
"rewards/margins": 0.11374848335981369,
|
|
"rewards/rejected": -0.5847958922386169,
|
|
"step": 568
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9524193556393083e-05,
|
|
"logits/chosen": -2.3562679290771484,
|
|
"logits/rejected": -2.388498306274414,
|
|
"logps/chosen": -212.62057495117188,
|
|
"logps/rejected": -230.94476318359375,
|
|
"loss": 0.6921,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.575402557849884,
|
|
"rewards/margins": 0.19730643928050995,
|
|
"rewards/rejected": -0.7727090120315552,
|
|
"step": 569
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.952216886315837e-05,
|
|
"logits/chosen": -2.414254903793335,
|
|
"logits/rejected": -2.4329581260681152,
|
|
"logps/chosen": -175.03147888183594,
|
|
"logps/rejected": -214.68109130859375,
|
|
"loss": 0.6274,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6071965098381042,
|
|
"rewards/margins": 0.29113632440567017,
|
|
"rewards/rejected": -0.8983328342437744,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.952013991280332e-05,
|
|
"logits/chosen": -2.5487287044525146,
|
|
"logits/rejected": -2.563288688659668,
|
|
"logps/chosen": -226.3212890625,
|
|
"logps/rejected": -214.44525146484375,
|
|
"loss": 0.746,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6687456965446472,
|
|
"rewards/margins": -0.017180969938635826,
|
|
"rewards/rejected": -0.6515647172927856,
|
|
"step": 571
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.951810670568017e-05,
|
|
"logits/chosen": -2.445035219192505,
|
|
"logits/rejected": -2.3831865787506104,
|
|
"logps/chosen": -214.5970001220703,
|
|
"logps/rejected": -228.41064453125,
|
|
"loss": 0.7696,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.62672358751297,
|
|
"rewards/margins": -0.040074095129966736,
|
|
"rewards/rejected": -0.5866495370864868,
|
|
"step": 572
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9516069242141885e-05,
|
|
"logits/chosen": -2.321072578430176,
|
|
"logits/rejected": -2.4207139015197754,
|
|
"logps/chosen": -203.04151916503906,
|
|
"logps/rejected": -206.2476806640625,
|
|
"loss": 0.766,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6442697644233704,
|
|
"rewards/margins": -0.051033951342105865,
|
|
"rewards/rejected": -0.5932357907295227,
|
|
"step": 573
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.951402752254219e-05,
|
|
"logits/chosen": -2.3598363399505615,
|
|
"logits/rejected": -2.510725498199463,
|
|
"logps/chosen": -160.54183959960938,
|
|
"logps/rejected": -181.72743225097656,
|
|
"loss": 0.613,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.45199206471443176,
|
|
"rewards/margins": 0.4041283130645752,
|
|
"rewards/rejected": -0.8561204075813293,
|
|
"step": 574
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.951198154723552e-05,
|
|
"logits/chosen": -2.550727605819702,
|
|
"logits/rejected": -2.5166475772857666,
|
|
"logps/chosen": -150.58238220214844,
|
|
"logps/rejected": -157.48434448242188,
|
|
"loss": 0.6906,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3471527397632599,
|
|
"rewards/margins": 0.06844709068536758,
|
|
"rewards/rejected": -0.41559985280036926,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.950993131657706e-05,
|
|
"logits/chosen": -2.32822847366333,
|
|
"logits/rejected": -2.3608272075653076,
|
|
"logps/chosen": -183.2533721923828,
|
|
"logps/rejected": -155.5178680419922,
|
|
"loss": 0.8822,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.7413718700408936,
|
|
"rewards/margins": -0.2268693447113037,
|
|
"rewards/rejected": -0.5145025849342346,
|
|
"step": 576
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9507876830922755e-05,
|
|
"logits/chosen": -2.426633358001709,
|
|
"logits/rejected": -2.350067138671875,
|
|
"logps/chosen": -159.07473754882812,
|
|
"logps/rejected": -177.24722290039062,
|
|
"loss": 0.6643,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4699227511882782,
|
|
"rewards/margins": 0.37284964323043823,
|
|
"rewards/rejected": -0.842772364616394,
|
|
"step": 577
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.950581809062925e-05,
|
|
"logits/chosen": -2.3766255378723145,
|
|
"logits/rejected": -2.411029815673828,
|
|
"logps/chosen": -211.9445343017578,
|
|
"logps/rejected": -224.93450927734375,
|
|
"loss": 0.7549,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.5890372395515442,
|
|
"rewards/margins": 0.09359408915042877,
|
|
"rewards/rejected": -0.6826313138008118,
|
|
"step": 578
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.950375509605396e-05,
|
|
"logits/chosen": -2.3283326625823975,
|
|
"logits/rejected": -2.1340157985687256,
|
|
"logps/chosen": -159.19125366210938,
|
|
"logps/rejected": -161.9340362548828,
|
|
"loss": 0.7342,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4637477993965149,
|
|
"rewards/margins": -0.048650771379470825,
|
|
"rewards/rejected": -0.41509705781936646,
|
|
"step": 579
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9501687847555016e-05,
|
|
"logits/chosen": -2.4342098236083984,
|
|
"logits/rejected": -2.4008848667144775,
|
|
"logps/chosen": -214.58380126953125,
|
|
"logps/rejected": -184.8182373046875,
|
|
"loss": 0.7425,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.45641300082206726,
|
|
"rewards/margins": -0.036872051656246185,
|
|
"rewards/rejected": -0.4195409417152405,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.949961634549131e-05,
|
|
"logits/chosen": -2.4010324478149414,
|
|
"logits/rejected": -2.397801160812378,
|
|
"logps/chosen": -175.51223754882812,
|
|
"logps/rejected": -165.90208435058594,
|
|
"loss": 0.6852,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3178666830062866,
|
|
"rewards/margins": 0.06419295072555542,
|
|
"rewards/rejected": -0.38205963373184204,
|
|
"step": 581
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.949754059022246e-05,
|
|
"logits/chosen": -2.494157075881958,
|
|
"logits/rejected": -2.560157537460327,
|
|
"logps/chosen": -172.84835815429688,
|
|
"logps/rejected": -178.97540283203125,
|
|
"loss": 0.6419,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2800545394420624,
|
|
"rewards/margins": 0.1691555678844452,
|
|
"rewards/rejected": -0.44921010732650757,
|
|
"step": 582
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.949546058210881e-05,
|
|
"logits/chosen": -2.209298610687256,
|
|
"logits/rejected": -2.293645143508911,
|
|
"logps/chosen": -194.64381408691406,
|
|
"logps/rejected": -237.14974975585938,
|
|
"loss": 0.5891,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.257599413394928,
|
|
"rewards/margins": 0.3168815076351166,
|
|
"rewards/rejected": -0.5744808912277222,
|
|
"step": 583
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9493376321511464e-05,
|
|
"logits/chosen": -2.446950912475586,
|
|
"logits/rejected": -2.4006083011627197,
|
|
"logps/chosen": -199.42282104492188,
|
|
"logps/rejected": -196.51849365234375,
|
|
"loss": 0.6841,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.35894283652305603,
|
|
"rewards/margins": 0.1623423993587494,
|
|
"rewards/rejected": -0.5212852358818054,
|
|
"step": 584
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9491287808792265e-05,
|
|
"logits/chosen": -2.3545334339141846,
|
|
"logits/rejected": -2.452188491821289,
|
|
"logps/chosen": -198.55056762695312,
|
|
"logps/rejected": -195.9398956298828,
|
|
"loss": 0.6768,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4273378551006317,
|
|
"rewards/margins": 0.11289644986391068,
|
|
"rewards/rejected": -0.5402343273162842,
|
|
"step": 585
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.948919504431376e-05,
|
|
"logits/chosen": -2.687852144241333,
|
|
"logits/rejected": -2.6648173332214355,
|
|
"logps/chosen": -215.23638916015625,
|
|
"logps/rejected": -221.06588745117188,
|
|
"loss": 0.6981,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.30441591143608093,
|
|
"rewards/margins": 0.08797214180231094,
|
|
"rewards/rejected": -0.39238807559013367,
|
|
"step": 586
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.948709802843929e-05,
|
|
"logits/chosen": -2.2244153022766113,
|
|
"logits/rejected": -2.2272024154663086,
|
|
"logps/chosen": -197.98904418945312,
|
|
"logps/rejected": -188.39773559570312,
|
|
"loss": 0.6626,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.27749666571617126,
|
|
"rewards/margins": 0.12206709384918213,
|
|
"rewards/rejected": -0.3995637595653534,
|
|
"step": 587
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9484996761532886e-05,
|
|
"logits/chosen": -2.471318006515503,
|
|
"logits/rejected": -2.550818920135498,
|
|
"logps/chosen": -192.51039123535156,
|
|
"logps/rejected": -192.6507568359375,
|
|
"loss": 0.7235,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.27331510186195374,
|
|
"rewards/margins": -0.019602246582508087,
|
|
"rewards/rejected": -0.25371283292770386,
|
|
"step": 588
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9482891243959335e-05,
|
|
"logits/chosen": -2.4945497512817383,
|
|
"logits/rejected": -2.554481029510498,
|
|
"logps/chosen": -187.52774047851562,
|
|
"logps/rejected": -227.6110076904297,
|
|
"loss": 0.5713,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2509083151817322,
|
|
"rewards/margins": 0.3469720482826233,
|
|
"rewards/rejected": -0.5978803038597107,
|
|
"step": 589
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.948078147608416e-05,
|
|
"logits/chosen": -2.5424046516418457,
|
|
"logits/rejected": -2.58070707321167,
|
|
"logps/chosen": -181.62966918945312,
|
|
"logps/rejected": -219.62698364257812,
|
|
"loss": 0.6805,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4028276205062866,
|
|
"rewards/margins": 0.0835791528224945,
|
|
"rewards/rejected": -0.48640677332878113,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9478667458273624e-05,
|
|
"logits/chosen": -2.558363199234009,
|
|
"logits/rejected": -2.533869743347168,
|
|
"logps/chosen": -207.1812286376953,
|
|
"logps/rejected": -201.0562286376953,
|
|
"loss": 0.6612,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.24153894186019897,
|
|
"rewards/margins": 0.12561962008476257,
|
|
"rewards/rejected": -0.36715856194496155,
|
|
"step": 591
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9476549190894725e-05,
|
|
"logits/chosen": -2.4393560886383057,
|
|
"logits/rejected": -2.450369358062744,
|
|
"logps/chosen": -153.7821044921875,
|
|
"logps/rejected": -153.88775634765625,
|
|
"loss": 0.686,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.23368775844573975,
|
|
"rewards/margins": 0.0638914704322815,
|
|
"rewards/rejected": -0.29757922887802124,
|
|
"step": 592
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.947442667431522e-05,
|
|
"logits/chosen": -2.6815226078033447,
|
|
"logits/rejected": -2.506643056869507,
|
|
"logps/chosen": -229.4060516357422,
|
|
"logps/rejected": -240.65362548828125,
|
|
"loss": 0.7512,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3724709153175354,
|
|
"rewards/margins": -0.04746837168931961,
|
|
"rewards/rejected": -0.325002521276474,
|
|
"step": 593
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9472299908903555e-05,
|
|
"logits/chosen": -2.3556230068206787,
|
|
"logits/rejected": -2.4480998516082764,
|
|
"logps/chosen": -128.82077026367188,
|
|
"logps/rejected": -170.01109313964844,
|
|
"loss": 0.6209,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.1109066978096962,
|
|
"rewards/margins": 0.2432006597518921,
|
|
"rewards/rejected": -0.3541073799133301,
|
|
"step": 594
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.947016889502895e-05,
|
|
"logits/chosen": -2.5540051460266113,
|
|
"logits/rejected": -2.549333095550537,
|
|
"logps/chosen": -178.5426788330078,
|
|
"logps/rejected": -172.33694458007812,
|
|
"loss": 0.7123,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3693382441997528,
|
|
"rewards/margins": -0.0018721502274274826,
|
|
"rewards/rejected": -0.36746612191200256,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.946803363306137e-05,
|
|
"logits/chosen": -2.076066255569458,
|
|
"logits/rejected": -2.011690139770508,
|
|
"logps/chosen": -194.8577880859375,
|
|
"logps/rejected": -176.207763671875,
|
|
"loss": 0.7698,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.26399266719818115,
|
|
"rewards/margins": -0.08310166001319885,
|
|
"rewards/rejected": -0.1808909922838211,
|
|
"step": 596
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.946589412337149e-05,
|
|
"logits/chosen": -2.2557950019836426,
|
|
"logits/rejected": -2.3377816677093506,
|
|
"logps/chosen": -148.5780029296875,
|
|
"logps/rejected": -170.98593139648438,
|
|
"loss": 0.6012,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": 0.07843692600727081,
|
|
"rewards/margins": 0.26211532950401306,
|
|
"rewards/rejected": -0.18367838859558105,
|
|
"step": 597
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9463750366330734e-05,
|
|
"logits/chosen": -2.2877843379974365,
|
|
"logits/rejected": -2.359644889831543,
|
|
"logps/chosen": -187.41412353515625,
|
|
"logps/rejected": -194.01309204101562,
|
|
"loss": 0.734,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.3197644352912903,
|
|
"rewards/margins": -0.021138954907655716,
|
|
"rewards/rejected": -0.29862549901008606,
|
|
"step": 598
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9461602362311275e-05,
|
|
"logits/chosen": -2.493452548980713,
|
|
"logits/rejected": -2.5441408157348633,
|
|
"logps/chosen": -209.37081909179688,
|
|
"logps/rejected": -173.10861206054688,
|
|
"loss": 0.7112,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.324692964553833,
|
|
"rewards/margins": 0.05840418487787247,
|
|
"rewards/rejected": -0.3830971419811249,
|
|
"step": 599
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9459450111686e-05,
|
|
"logits/chosen": -2.4824540615081787,
|
|
"logits/rejected": -2.477165937423706,
|
|
"logps/chosen": -191.7369384765625,
|
|
"logps/rejected": -181.0056915283203,
|
|
"loss": 0.692,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.38107597827911377,
|
|
"rewards/margins": 0.05237661302089691,
|
|
"rewards/rejected": -0.4334526062011719,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.945729361482856e-05,
|
|
"logits/chosen": -2.450261116027832,
|
|
"logits/rejected": -2.500009536743164,
|
|
"logps/chosen": -232.3407440185547,
|
|
"logps/rejected": -212.10638427734375,
|
|
"loss": 0.5732,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.19914206862449646,
|
|
"rewards/margins": 0.3821813762187958,
|
|
"rewards/rejected": -0.5813234448432922,
|
|
"step": 601
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9455132872113316e-05,
|
|
"logits/chosen": -2.3420450687408447,
|
|
"logits/rejected": -2.3474817276000977,
|
|
"logps/chosen": -216.38914489746094,
|
|
"logps/rejected": -224.3979949951172,
|
|
"loss": 0.6047,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.1369827687740326,
|
|
"rewards/margins": 0.2419712096452713,
|
|
"rewards/rejected": -0.3789539933204651,
|
|
"step": 602
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.945296788391539e-05,
|
|
"logits/chosen": -2.3396308422088623,
|
|
"logits/rejected": -2.3148958683013916,
|
|
"logps/chosen": -169.25946044921875,
|
|
"logps/rejected": -197.1276397705078,
|
|
"loss": 0.7964,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.32238638401031494,
|
|
"rewards/margins": -0.1470273733139038,
|
|
"rewards/rejected": -0.17535902559757233,
|
|
"step": 603
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.945079865061062e-05,
|
|
"logits/chosen": -2.275747537612915,
|
|
"logits/rejected": -2.2709038257598877,
|
|
"logps/chosen": -139.6877899169922,
|
|
"logps/rejected": -148.5768585205078,
|
|
"loss": 0.6796,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.08058857917785645,
|
|
"rewards/margins": 0.04861634969711304,
|
|
"rewards/rejected": -0.12920492887496948,
|
|
"step": 604
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.94486251725756e-05,
|
|
"logits/chosen": -2.3698954582214355,
|
|
"logits/rejected": -2.453796625137329,
|
|
"logps/chosen": -229.4521484375,
|
|
"logps/rejected": -210.6240234375,
|
|
"loss": 0.6212,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.22786223888397217,
|
|
"rewards/margins": 0.20501334965229034,
|
|
"rewards/rejected": -0.4328756034374237,
|
|
"step": 605
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.944644745018765e-05,
|
|
"logits/chosen": -2.4110870361328125,
|
|
"logits/rejected": -2.423731565475464,
|
|
"logps/chosen": -219.2958221435547,
|
|
"logps/rejected": -202.20706176757812,
|
|
"loss": 0.7773,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.37566569447517395,
|
|
"rewards/margins": -0.0947473868727684,
|
|
"rewards/rejected": -0.28091832995414734,
|
|
"step": 606
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.944426548382482e-05,
|
|
"logits/chosen": -2.4861879348754883,
|
|
"logits/rejected": -2.5459907054901123,
|
|
"logps/chosen": -237.8024444580078,
|
|
"logps/rejected": -260.2302551269531,
|
|
"loss": 0.7621,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.42855891585350037,
|
|
"rewards/margins": -0.048416607081890106,
|
|
"rewards/rejected": -0.38014233112335205,
|
|
"step": 607
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.944207927386591e-05,
|
|
"logits/chosen": -2.478734254837036,
|
|
"logits/rejected": -2.5266644954681396,
|
|
"logps/chosen": -218.4932403564453,
|
|
"logps/rejected": -225.3731689453125,
|
|
"loss": 0.7075,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.23466593027114868,
|
|
"rewards/margins": 0.09529409557580948,
|
|
"rewards/rejected": -0.3299599885940552,
|
|
"step": 608
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9439888820690475e-05,
|
|
"logits/chosen": -2.5482990741729736,
|
|
"logits/rejected": -2.544922113418579,
|
|
"logps/chosen": -186.0775909423828,
|
|
"logps/rejected": -197.2762908935547,
|
|
"loss": 0.5692,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.01272206474095583,
|
|
"rewards/margins": 0.301639199256897,
|
|
"rewards/rejected": -0.3143612742424011,
|
|
"step": 609
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.943769412467875e-05,
|
|
"logits/chosen": -2.2644476890563965,
|
|
"logits/rejected": -2.3761229515075684,
|
|
"logps/chosen": -215.5855712890625,
|
|
"logps/rejected": -246.0447998046875,
|
|
"loss": 0.5599,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.19982381165027618,
|
|
"rewards/margins": 0.4122087359428406,
|
|
"rewards/rejected": -0.6120326519012451,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.943549518621176e-05,
|
|
"logits/chosen": -2.4406063556671143,
|
|
"logits/rejected": -2.529265880584717,
|
|
"logps/chosen": -175.3257598876953,
|
|
"logps/rejected": -188.4786376953125,
|
|
"loss": 0.6989,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.19547346234321594,
|
|
"rewards/margins": 0.05771362781524658,
|
|
"rewards/rejected": -0.2531871199607849,
|
|
"step": 611
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9433292005671236e-05,
|
|
"logits/chosen": -2.4887635707855225,
|
|
"logits/rejected": -2.48945951461792,
|
|
"logps/chosen": -187.1964111328125,
|
|
"logps/rejected": -231.7959747314453,
|
|
"loss": 0.6501,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.21680843830108643,
|
|
"rewards/margins": 0.15266810357570648,
|
|
"rewards/rejected": -0.3694765567779541,
|
|
"step": 612
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.943108458343968e-05,
|
|
"logits/chosen": -2.4516611099243164,
|
|
"logits/rejected": -2.4493842124938965,
|
|
"logps/chosen": -175.4194793701172,
|
|
"logps/rejected": -185.060546875,
|
|
"loss": 0.6172,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2488647699356079,
|
|
"rewards/margins": 0.22184765338897705,
|
|
"rewards/rejected": -0.47071248292922974,
|
|
"step": 613
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.942887291990028e-05,
|
|
"logits/chosen": -2.636701822280884,
|
|
"logits/rejected": -2.5966298580169678,
|
|
"logps/chosen": -194.91783142089844,
|
|
"logps/rejected": -182.28858947753906,
|
|
"loss": 0.6595,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.04664868861436844,
|
|
"rewards/margins": 0.11729076504707336,
|
|
"rewards/rejected": -0.1639394611120224,
|
|
"step": 614
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9426657015436994e-05,
|
|
"logits/chosen": -2.303603172302246,
|
|
"logits/rejected": -2.2422983646392822,
|
|
"logps/chosen": -144.08590698242188,
|
|
"logps/rejected": -148.51541137695312,
|
|
"loss": 0.7636,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.18376705050468445,
|
|
"rewards/margins": -0.09818767011165619,
|
|
"rewards/rejected": -0.08557939529418945,
|
|
"step": 615
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.9424436870434515e-05,
|
|
"logits/chosen": -2.7395613193511963,
|
|
"logits/rejected": -2.747368097305298,
|
|
"logps/chosen": -226.32354736328125,
|
|
"logps/rejected": -212.107421875,
|
|
"loss": 0.7233,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.19879278540611267,
|
|
"rewards/margins": 0.0006840275600552559,
|
|
"rewards/rejected": -0.19947680830955505,
|
|
"step": 616
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.942221248527826e-05,
|
|
"logits/chosen": -2.493027448654175,
|
|
"logits/rejected": -2.686164140701294,
|
|
"logps/chosen": -166.07200622558594,
|
|
"logps/rejected": -189.38475036621094,
|
|
"loss": 0.8368,
|
|
"rewards/accuracies": 0.1875,
|
|
"rewards/chosen": -0.28490307927131653,
|
|
"rewards/margins": -0.22959473729133606,
|
|
"rewards/rejected": -0.05530836060643196,
|
|
"step": 617
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.94199838603544e-05,
|
|
"logits/chosen": -2.3346376419067383,
|
|
"logits/rejected": -2.4675793647766113,
|
|
"logps/chosen": -177.6221160888672,
|
|
"logps/rejected": -186.67446899414062,
|
|
"loss": 0.4898,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": 0.1148318499326706,
|
|
"rewards/margins": 0.5469637513160706,
|
|
"rewards/rejected": -0.4321318566799164,
|
|
"step": 618
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.941775099604983e-05,
|
|
"logits/chosen": -2.5135788917541504,
|
|
"logits/rejected": -2.6071126461029053,
|
|
"logps/chosen": -186.59161376953125,
|
|
"logps/rejected": -205.1910400390625,
|
|
"loss": 0.6155,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.029235512018203735,
|
|
"rewards/margins": 0.20693793892860413,
|
|
"rewards/rejected": -0.23617346584796906,
|
|
"step": 619
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.941551389275217e-05,
|
|
"logits/chosen": -2.49117374420166,
|
|
"logits/rejected": -2.5229427814483643,
|
|
"logps/chosen": -208.4278106689453,
|
|
"logps/rejected": -191.97398376464844,
|
|
"loss": 0.6074,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.18853022158145905,
|
|
"rewards/margins": 0.2418961077928543,
|
|
"rewards/rejected": -0.43042635917663574,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.941327255084979e-05,
|
|
"logits/chosen": -2.4580295085906982,
|
|
"logits/rejected": -2.4623186588287354,
|
|
"logps/chosen": -191.52886962890625,
|
|
"logps/rejected": -169.84417724609375,
|
|
"loss": 0.7945,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.2514897584915161,
|
|
"rewards/margins": -0.04441142827272415,
|
|
"rewards/rejected": -0.20707830786705017,
|
|
"step": 621
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.9411026970731805e-05,
|
|
"logits/chosen": -2.5821356773376465,
|
|
"logits/rejected": -2.619976758956909,
|
|
"logps/chosen": -203.17149353027344,
|
|
"logps/rejected": -217.1163330078125,
|
|
"loss": 0.6965,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.33491429686546326,
|
|
"rewards/margins": 0.05494851991534233,
|
|
"rewards/rejected": -0.3898628354072571,
|
|
"step": 622
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.9408777152788044e-05,
|
|
"logits/chosen": -2.464843273162842,
|
|
"logits/rejected": -2.5318896770477295,
|
|
"logps/chosen": -169.26376342773438,
|
|
"logps/rejected": -182.11395263671875,
|
|
"loss": 0.6721,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.17076966166496277,
|
|
"rewards/margins": 0.10583178699016571,
|
|
"rewards/rejected": -0.2766014337539673,
|
|
"step": 623
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.9406523097409085e-05,
|
|
"logits/chosen": -2.262134075164795,
|
|
"logits/rejected": -2.329742670059204,
|
|
"logps/chosen": -153.96200561523438,
|
|
"logps/rejected": -157.85238647460938,
|
|
"loss": 0.6468,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.13661958277225494,
|
|
"rewards/margins": 0.20581209659576416,
|
|
"rewards/rejected": -0.3424316644668579,
|
|
"step": 624
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.940426480498623e-05,
|
|
"logits/chosen": -2.403775453567505,
|
|
"logits/rejected": -2.459458351135254,
|
|
"logps/chosen": -167.8171844482422,
|
|
"logps/rejected": -221.13282775878906,
|
|
"loss": 0.6038,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.2118593454360962,
|
|
"rewards/margins": 0.26167717576026917,
|
|
"rewards/rejected": -0.47353655099868774,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.940200227591154e-05,
|
|
"logits/chosen": -1.9581964015960693,
|
|
"logits/rejected": -2.0107665061950684,
|
|
"logps/chosen": -124.81986236572266,
|
|
"logps/rejected": -138.32220458984375,
|
|
"loss": 0.6216,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.05847623199224472,
|
|
"rewards/margins": 0.22099967300891876,
|
|
"rewards/rejected": -0.2794759273529053,
|
|
"step": 626
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.9399735510577796e-05,
|
|
"logits/chosen": -2.2749149799346924,
|
|
"logits/rejected": -2.283113718032837,
|
|
"logps/chosen": -155.7340087890625,
|
|
"logps/rejected": -150.0223388671875,
|
|
"loss": 0.6522,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.1051105409860611,
|
|
"rewards/margins": 0.16768980026245117,
|
|
"rewards/rejected": -0.27280035614967346,
|
|
"step": 627
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.93974645093785e-05,
|
|
"logits/chosen": -2.4072506427764893,
|
|
"logits/rejected": -2.4812541007995605,
|
|
"logps/chosen": -195.17230224609375,
|
|
"logps/rejected": -194.09622192382812,
|
|
"loss": 0.6298,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.06678425520658493,
|
|
"rewards/margins": 0.17194166779518127,
|
|
"rewards/rejected": -0.2387259304523468,
|
|
"step": 628
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.9395189272707916e-05,
|
|
"logits/chosen": -2.428178310394287,
|
|
"logits/rejected": -2.5200040340423584,
|
|
"logps/chosen": -194.95848083496094,
|
|
"logps/rejected": -215.4308319091797,
|
|
"loss": 0.699,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.37522369623184204,
|
|
"rewards/margins": 0.045931555330753326,
|
|
"rewards/rejected": -0.4211552143096924,
|
|
"step": 629
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 4.939290980096103e-05,
|
|
"logits/chosen": -2.243135690689087,
|
|
"logits/rejected": -2.25071120262146,
|
|
"logps/chosen": -188.89820861816406,
|
|
"logps/rejected": -219.1158447265625,
|
|
"loss": 0.7187,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2669258415699005,
|
|
"rewards/margins": 0.03279988840222359,
|
|
"rewards/rejected": -0.2997257113456726,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.9390626094533565e-05,
|
|
"logits/chosen": -2.496492385864258,
|
|
"logits/rejected": -2.6048879623413086,
|
|
"logps/chosen": -169.56320190429688,
|
|
"logps/rejected": -184.27169799804688,
|
|
"loss": 0.6584,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.31219586730003357,
|
|
"rewards/margins": 0.16650965809822083,
|
|
"rewards/rejected": -0.4787055253982544,
|
|
"step": 631
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.9388338153821976e-05,
|
|
"logits/chosen": -2.256747245788574,
|
|
"logits/rejected": -2.2907259464263916,
|
|
"logps/chosen": -163.09808349609375,
|
|
"logps/rejected": -231.26492309570312,
|
|
"loss": 0.5661,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.15426135063171387,
|
|
"rewards/margins": 0.48811718821525574,
|
|
"rewards/rejected": -0.6423785090446472,
|
|
"step": 632
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.938604597922346e-05,
|
|
"logits/chosen": -2.5338540077209473,
|
|
"logits/rejected": -2.580803871154785,
|
|
"logps/chosen": -221.49908447265625,
|
|
"logps/rejected": -222.09754943847656,
|
|
"loss": 0.7293,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.393398642539978,
|
|
"rewards/margins": 0.04478640481829643,
|
|
"rewards/rejected": -0.43818506598472595,
|
|
"step": 633
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.9383749571135946e-05,
|
|
"logits/chosen": -2.4923906326293945,
|
|
"logits/rejected": -2.552914619445801,
|
|
"logps/chosen": -188.54759216308594,
|
|
"logps/rejected": -201.86276245117188,
|
|
"loss": 0.6803,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.19801469147205353,
|
|
"rewards/margins": 0.08076779544353485,
|
|
"rewards/rejected": -0.2787824869155884,
|
|
"step": 634
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.938144892995809e-05,
|
|
"logits/chosen": -2.6368439197540283,
|
|
"logits/rejected": -2.681831121444702,
|
|
"logps/chosen": -172.92823791503906,
|
|
"logps/rejected": -153.6282501220703,
|
|
"loss": 0.6554,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.1948256492614746,
|
|
"rewards/margins": 0.14521902799606323,
|
|
"rewards/rejected": -0.34004467725753784,
|
|
"step": 635
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.93791440560893e-05,
|
|
"logits/chosen": -2.5344183444976807,
|
|
"logits/rejected": -2.5658974647521973,
|
|
"logps/chosen": -198.16390991210938,
|
|
"logps/rejected": -193.16055297851562,
|
|
"loss": 0.6981,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.06259515881538391,
|
|
"rewards/margins": 0.05056838318705559,
|
|
"rewards/rejected": -0.11316356062889099,
|
|
"step": 636
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.93768349499297e-05,
|
|
"logits/chosen": -2.6175780296325684,
|
|
"logits/rejected": -2.561044216156006,
|
|
"logps/chosen": -194.93130493164062,
|
|
"logps/rejected": -171.37437438964844,
|
|
"loss": 0.865,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.44466546177864075,
|
|
"rewards/margins": -0.2321683019399643,
|
|
"rewards/rejected": -0.21249720454216003,
|
|
"step": 637
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 4.9374521611880166e-05,
|
|
"logits/chosen": -2.552360773086548,
|
|
"logits/rejected": -2.585489273071289,
|
|
"logps/chosen": -157.82420349121094,
|
|
"logps/rejected": -171.7198028564453,
|
|
"loss": 0.6514,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.19169321656227112,
|
|
"rewards/margins": 0.18479710817337036,
|
|
"rewards/rejected": -0.3764903247356415,
|
|
"step": 638
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.9372204042342295e-05,
|
|
"logits/chosen": -2.55067777633667,
|
|
"logits/rejected": -2.6302952766418457,
|
|
"logps/chosen": -186.92031860351562,
|
|
"logps/rejected": -184.224365234375,
|
|
"loss": 0.676,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2839874029159546,
|
|
"rewards/margins": 0.20806796848773956,
|
|
"rewards/rejected": -0.49205532670021057,
|
|
"step": 639
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.936988224171842e-05,
|
|
"logits/chosen": -2.6061015129089355,
|
|
"logits/rejected": -2.609809637069702,
|
|
"logps/chosen": -209.86328125,
|
|
"logps/rejected": -199.8352813720703,
|
|
"loss": 0.6114,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.38792091608047485,
|
|
"rewards/margins": 0.29765814542770386,
|
|
"rewards/rejected": -0.6855790615081787,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.936755621041163e-05,
|
|
"logits/chosen": -2.600673198699951,
|
|
"logits/rejected": -2.6286747455596924,
|
|
"logps/chosen": -217.86569213867188,
|
|
"logps/rejected": -208.2232208251953,
|
|
"loss": 0.5999,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3269140124320984,
|
|
"rewards/margins": 0.2362305074930191,
|
|
"rewards/rejected": -0.5631444454193115,
|
|
"step": 641
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.93652259488257e-05,
|
|
"logits/chosen": -2.331010341644287,
|
|
"logits/rejected": -2.406320571899414,
|
|
"logps/chosen": -181.08682250976562,
|
|
"logps/rejected": -231.1434326171875,
|
|
"loss": 0.6473,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.18573978543281555,
|
|
"rewards/margins": 0.11375062167644501,
|
|
"rewards/rejected": -0.29949042201042175,
|
|
"step": 642
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.9362891457365194e-05,
|
|
"logits/chosen": -2.5639798641204834,
|
|
"logits/rejected": -2.511385917663574,
|
|
"logps/chosen": -167.39529418945312,
|
|
"logps/rejected": -190.17266845703125,
|
|
"loss": 0.6687,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.23445367813110352,
|
|
"rewards/margins": 0.10250186920166016,
|
|
"rewards/rejected": -0.33695557713508606,
|
|
"step": 643
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.9360552736435386e-05,
|
|
"logits/chosen": -2.484229326248169,
|
|
"logits/rejected": -2.5917751789093018,
|
|
"logps/chosen": -160.66372680664062,
|
|
"logps/rejected": -151.573974609375,
|
|
"loss": 0.7071,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3852943181991577,
|
|
"rewards/margins": 0.021035168319940567,
|
|
"rewards/rejected": -0.40632951259613037,
|
|
"step": 644
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.935820978644228e-05,
|
|
"logits/chosen": -2.4788904190063477,
|
|
"logits/rejected": -2.4254837036132812,
|
|
"logps/chosen": -161.3828887939453,
|
|
"logps/rejected": -179.1514129638672,
|
|
"loss": 0.7489,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.27134445309638977,
|
|
"rewards/margins": -0.01518234983086586,
|
|
"rewards/rejected": -0.2561620771884918,
|
|
"step": 645
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.9355862607792614e-05,
|
|
"logits/chosen": -2.3145663738250732,
|
|
"logits/rejected": -2.3181071281433105,
|
|
"logps/chosen": -175.7195281982422,
|
|
"logps/rejected": -180.3649139404297,
|
|
"loss": 0.7226,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.2857952117919922,
|
|
"rewards/margins": -0.0009603817015886307,
|
|
"rewards/rejected": -0.2848348617553711,
|
|
"step": 646
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.9353511200893865e-05,
|
|
"logits/chosen": -2.5312018394470215,
|
|
"logits/rejected": -2.5117855072021484,
|
|
"logps/chosen": -147.04232788085938,
|
|
"logps/rejected": -153.8798065185547,
|
|
"loss": 0.7482,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.2138766497373581,
|
|
"rewards/margins": -0.0894002690911293,
|
|
"rewards/rejected": -0.12447639554738998,
|
|
"step": 647
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.935115556615425e-05,
|
|
"logits/chosen": -2.5504438877105713,
|
|
"logits/rejected": -2.5526883602142334,
|
|
"logps/chosen": -198.34768676757812,
|
|
"logps/rejected": -190.35696411132812,
|
|
"loss": 0.6935,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.393574059009552,
|
|
"rewards/margins": 0.1438981145620346,
|
|
"rewards/rejected": -0.5374721884727478,
|
|
"step": 648
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.934879570398272e-05,
|
|
"logits/chosen": -2.450056314468384,
|
|
"logits/rejected": -2.440789222717285,
|
|
"logps/chosen": -198.56884765625,
|
|
"logps/rejected": -234.9272918701172,
|
|
"loss": 0.7365,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.5248376131057739,
|
|
"rewards/margins": -0.04359120875597,
|
|
"rewards/rejected": -0.4812464118003845,
|
|
"step": 649
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.9346431614788945e-05,
|
|
"logits/chosen": -2.481201648712158,
|
|
"logits/rejected": -2.540107250213623,
|
|
"logps/chosen": -193.42530822753906,
|
|
"logps/rejected": -196.03887939453125,
|
|
"loss": 0.7004,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.30591022968292236,
|
|
"rewards/margins": 0.07436452805995941,
|
|
"rewards/rejected": -0.3802747428417206,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.934406329898334e-05,
|
|
"logits/chosen": -2.4828920364379883,
|
|
"logits/rejected": -2.4486544132232666,
|
|
"logps/chosen": -174.25201416015625,
|
|
"logps/rejected": -167.93795776367188,
|
|
"loss": 0.8821,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.4259773790836334,
|
|
"rewards/margins": -0.30132219195365906,
|
|
"rewards/rejected": -0.12465521693229675,
|
|
"step": 651
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.934169075697704e-05,
|
|
"logits/chosen": -2.5756962299346924,
|
|
"logits/rejected": -2.53938364982605,
|
|
"logps/chosen": -261.4378356933594,
|
|
"logps/rejected": -217.41473388671875,
|
|
"loss": 0.7193,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.35280755162239075,
|
|
"rewards/margins": 0.009605936706066132,
|
|
"rewards/rejected": -0.36241352558135986,
|
|
"step": 652
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 4.9339313989181955e-05,
|
|
"logits/chosen": -2.479517936706543,
|
|
"logits/rejected": -2.4295461177825928,
|
|
"logps/chosen": -168.84185791015625,
|
|
"logps/rejected": -171.62991333007812,
|
|
"loss": 0.7111,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.4041588306427002,
|
|
"rewards/margins": 0.014433953911066055,
|
|
"rewards/rejected": -0.41859275102615356,
|
|
"step": 653
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.933693299601067e-05,
|
|
"logits/chosen": -2.2451162338256836,
|
|
"logits/rejected": -2.4006645679473877,
|
|
"logps/chosen": -178.17141723632812,
|
|
"logps/rejected": -210.85440063476562,
|
|
"loss": 0.5213,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": 0.0021903254091739655,
|
|
"rewards/margins": 0.5191294550895691,
|
|
"rewards/rejected": -0.516939103603363,
|
|
"step": 654
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.933454777787654e-05,
|
|
"logits/chosen": -2.6251702308654785,
|
|
"logits/rejected": -2.6193838119506836,
|
|
"logps/chosen": -261.71185302734375,
|
|
"logps/rejected": -258.5350646972656,
|
|
"loss": 0.6123,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3498530089855194,
|
|
"rewards/margins": 0.23792549967765808,
|
|
"rewards/rejected": -0.5877785086631775,
|
|
"step": 655
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.933215833519366e-05,
|
|
"logits/chosen": -2.515723466873169,
|
|
"logits/rejected": -2.4995017051696777,
|
|
"logps/chosen": -170.08506774902344,
|
|
"logps/rejected": -171.64215087890625,
|
|
"loss": 0.6167,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.03253903239965439,
|
|
"rewards/margins": 0.2270340621471405,
|
|
"rewards/rejected": -0.2595731019973755,
|
|
"step": 656
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.9329764668376824e-05,
|
|
"logits/chosen": -2.3417258262634277,
|
|
"logits/rejected": -2.5132431983947754,
|
|
"logps/chosen": -182.23594665527344,
|
|
"logps/rejected": -204.4223175048828,
|
|
"loss": 0.5153,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.08154846727848053,
|
|
"rewards/margins": 0.5844264626502991,
|
|
"rewards/rejected": -0.6659749746322632,
|
|
"step": 657
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.932736677784158e-05,
|
|
"logits/chosen": -2.406738758087158,
|
|
"logits/rejected": -2.4287314414978027,
|
|
"logps/chosen": -175.0724639892578,
|
|
"logps/rejected": -212.99569702148438,
|
|
"loss": 0.6727,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5249559879302979,
|
|
"rewards/margins": 0.10958628356456757,
|
|
"rewards/rejected": -0.6345422863960266,
|
|
"step": 658
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.9324964664004226e-05,
|
|
"logits/chosen": -2.4244885444641113,
|
|
"logits/rejected": -2.511654853820801,
|
|
"logps/chosen": -181.31353759765625,
|
|
"logps/rejected": -189.31832885742188,
|
|
"loss": 0.7071,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.27095070481300354,
|
|
"rewards/margins": -0.0028193406760692596,
|
|
"rewards/rejected": -0.2681313753128052,
|
|
"step": 659
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 4.9322558327281773e-05,
|
|
"logits/chosen": -2.5121233463287354,
|
|
"logits/rejected": -2.6510791778564453,
|
|
"logps/chosen": -230.39369201660156,
|
|
"logps/rejected": -241.02325439453125,
|
|
"loss": 0.4769,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.02516031824052334,
|
|
"rewards/margins": 0.551730751991272,
|
|
"rewards/rejected": -0.5768911242485046,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.9320147768091954e-05,
|
|
"logits/chosen": -2.076624631881714,
|
|
"logits/rejected": -2.117382287979126,
|
|
"logps/chosen": -228.64761352539062,
|
|
"logps/rejected": -211.5120849609375,
|
|
"loss": 0.696,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.37209880352020264,
|
|
"rewards/margins": 0.16731634736061096,
|
|
"rewards/rejected": -0.5394151210784912,
|
|
"step": 661
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.931773298685326e-05,
|
|
"logits/chosen": -2.3631484508514404,
|
|
"logits/rejected": -2.424973249435425,
|
|
"logps/chosen": -174.7392120361328,
|
|
"logps/rejected": -198.92578125,
|
|
"loss": 0.6354,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.23427049815654755,
|
|
"rewards/margins": 0.1624229997396469,
|
|
"rewards/rejected": -0.39669349789619446,
|
|
"step": 662
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.93153139839849e-05,
|
|
"logits/chosen": -2.271193742752075,
|
|
"logits/rejected": -2.2948765754699707,
|
|
"logps/chosen": -180.6439971923828,
|
|
"logps/rejected": -192.5905303955078,
|
|
"loss": 0.7982,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.23156273365020752,
|
|
"rewards/margins": -0.15015332400798798,
|
|
"rewards/rejected": -0.08140941709280014,
|
|
"step": 663
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.931289075990682e-05,
|
|
"logits/chosen": -2.6266207695007324,
|
|
"logits/rejected": -2.6377651691436768,
|
|
"logps/chosen": -214.2593536376953,
|
|
"logps/rejected": -202.11019897460938,
|
|
"loss": 0.7092,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.25713613629341125,
|
|
"rewards/margins": 0.02131093665957451,
|
|
"rewards/rejected": -0.2784470319747925,
|
|
"step": 664
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.93104633150397e-05,
|
|
"logits/chosen": -2.558809995651245,
|
|
"logits/rejected": -2.580552816390991,
|
|
"logps/chosen": -200.12879943847656,
|
|
"logps/rejected": -270.4982604980469,
|
|
"loss": 0.5893,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.21998968720436096,
|
|
"rewards/margins": 0.367271363735199,
|
|
"rewards/rejected": -0.5872610807418823,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.930803164980496e-05,
|
|
"logits/chosen": -2.324084520339966,
|
|
"logits/rejected": -2.4046638011932373,
|
|
"logps/chosen": -199.2420196533203,
|
|
"logps/rejected": -188.57501220703125,
|
|
"loss": 0.651,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.24593767523765564,
|
|
"rewards/margins": 0.17680732905864716,
|
|
"rewards/rejected": -0.4227449893951416,
|
|
"step": 666
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.930559576462473e-05,
|
|
"logits/chosen": -2.4203665256500244,
|
|
"logits/rejected": -2.5350735187530518,
|
|
"logps/chosen": -203.4146728515625,
|
|
"logps/rejected": -220.17127990722656,
|
|
"loss": 0.6676,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.36869651079177856,
|
|
"rewards/margins": 0.1714363694190979,
|
|
"rewards/rejected": -0.5401328802108765,
|
|
"step": 667
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 4.930315565992189e-05,
|
|
"logits/chosen": -2.5649983882904053,
|
|
"logits/rejected": -2.587256669998169,
|
|
"logps/chosen": -211.79006958007812,
|
|
"logps/rejected": -197.28616333007812,
|
|
"loss": 0.6443,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.41877490282058716,
|
|
"rewards/margins": 0.1472373902797699,
|
|
"rewards/rejected": -0.5660123229026794,
|
|
"step": 668
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.930071133612005e-05,
|
|
"logits/chosen": -2.4042892456054688,
|
|
"logits/rejected": -2.4121689796447754,
|
|
"logps/chosen": -163.4969940185547,
|
|
"logps/rejected": -214.19903564453125,
|
|
"loss": 0.5355,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.12853318452835083,
|
|
"rewards/margins": 0.4976612329483032,
|
|
"rewards/rejected": -0.626194417476654,
|
|
"step": 669
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.929826279364357e-05,
|
|
"logits/chosen": -2.6725361347198486,
|
|
"logits/rejected": -2.6962034702301025,
|
|
"logps/chosen": -206.29966735839844,
|
|
"logps/rejected": -225.02146911621094,
|
|
"loss": 0.7439,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.4318912625312805,
|
|
"rewards/margins": -0.017002558335661888,
|
|
"rewards/rejected": -0.4148887097835541,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.9295810032917486e-05,
|
|
"logits/chosen": -2.6312649250030518,
|
|
"logits/rejected": -2.677178144454956,
|
|
"logps/chosen": -233.12554931640625,
|
|
"logps/rejected": -249.84649658203125,
|
|
"loss": 0.6412,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3292105495929718,
|
|
"rewards/margins": 0.18580739200115204,
|
|
"rewards/rejected": -0.5150179266929626,
|
|
"step": 671
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.929335305436764e-05,
|
|
"logits/chosen": -2.350733995437622,
|
|
"logits/rejected": -2.4738504886627197,
|
|
"logps/chosen": -159.27565002441406,
|
|
"logps/rejected": -197.69430541992188,
|
|
"loss": 0.5996,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3497101664543152,
|
|
"rewards/margins": 0.35428282618522644,
|
|
"rewards/rejected": -0.7039930820465088,
|
|
"step": 672
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.9290891858420554e-05,
|
|
"logits/chosen": -2.341266632080078,
|
|
"logits/rejected": -2.412942409515381,
|
|
"logps/chosen": -221.41571044921875,
|
|
"logps/rejected": -226.8486328125,
|
|
"loss": 0.5985,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.34598851203918457,
|
|
"rewards/margins": 0.2551480233669281,
|
|
"rewards/rejected": -0.6011365652084351,
|
|
"step": 673
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.92884264455035e-05,
|
|
"logits/chosen": -2.302943706512451,
|
|
"logits/rejected": -2.3145551681518555,
|
|
"logps/chosen": -149.34109497070312,
|
|
"logps/rejected": -138.53506469726562,
|
|
"loss": 0.644,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.1495213508605957,
|
|
"rewards/margins": 0.15739065408706665,
|
|
"rewards/rejected": -0.30691200494766235,
|
|
"step": 674
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.9285956816044486e-05,
|
|
"logits/chosen": -2.228523015975952,
|
|
"logits/rejected": -2.173222064971924,
|
|
"logps/chosen": -177.70559692382812,
|
|
"logps/rejected": -210.69927978515625,
|
|
"loss": 0.7657,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3950153589248657,
|
|
"rewards/margins": -0.07809535413980484,
|
|
"rewards/rejected": -0.3169199824333191,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.9283482970472245e-05,
|
|
"logits/chosen": -2.3849382400512695,
|
|
"logits/rejected": -2.513273000717163,
|
|
"logps/chosen": -185.79124450683594,
|
|
"logps/rejected": -201.27066040039062,
|
|
"loss": 0.6207,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.13350968062877655,
|
|
"rewards/margins": 0.2174827605485916,
|
|
"rewards/rejected": -0.35099244117736816,
|
|
"step": 676
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.928100490921625e-05,
|
|
"logits/chosen": -2.458615779876709,
|
|
"logits/rejected": -2.4962515830993652,
|
|
"logps/chosen": -204.7447509765625,
|
|
"logps/rejected": -199.8260955810547,
|
|
"loss": 0.7519,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6041467189788818,
|
|
"rewards/margins": 0.038420408964157104,
|
|
"rewards/rejected": -0.6425671577453613,
|
|
"step": 677
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.927852263270667e-05,
|
|
"logits/chosen": -2.5404415130615234,
|
|
"logits/rejected": -2.5182065963745117,
|
|
"logps/chosen": -208.4690399169922,
|
|
"logps/rejected": -227.29690551757812,
|
|
"loss": 0.6729,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.28778883814811707,
|
|
"rewards/margins": 0.08314257860183716,
|
|
"rewards/rejected": -0.3709314167499542,
|
|
"step": 678
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.9276036141374473e-05,
|
|
"logits/chosen": -2.528585910797119,
|
|
"logits/rejected": -2.5861098766326904,
|
|
"logps/chosen": -211.7762908935547,
|
|
"logps/rejected": -236.7048797607422,
|
|
"loss": 0.7562,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4092783033847809,
|
|
"rewards/margins": -0.03306754678487778,
|
|
"rewards/rejected": -0.3762107491493225,
|
|
"step": 679
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.92735454356513e-05,
|
|
"logits/chosen": -2.172372341156006,
|
|
"logits/rejected": -2.1387619972229004,
|
|
"logps/chosen": -230.44168090820312,
|
|
"logps/rejected": -221.9505615234375,
|
|
"loss": 0.6126,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.0995088517665863,
|
|
"rewards/margins": 0.24330440163612366,
|
|
"rewards/rejected": -0.34281325340270996,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.927105051596956e-05,
|
|
"logits/chosen": -2.5018787384033203,
|
|
"logits/rejected": -2.6133289337158203,
|
|
"logps/chosen": -159.59141540527344,
|
|
"logps/rejected": -165.89492797851562,
|
|
"loss": 0.6031,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0614471361041069,
|
|
"rewards/margins": 0.24630111455917358,
|
|
"rewards/rejected": -0.3077481985092163,
|
|
"step": 681
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.926855138276236e-05,
|
|
"logits/chosen": -2.532803773880005,
|
|
"logits/rejected": -2.6221201419830322,
|
|
"logps/chosen": -192.40725708007812,
|
|
"logps/rejected": -176.5653839111328,
|
|
"loss": 0.7419,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4705914258956909,
|
|
"rewards/margins": 0.014578044414520264,
|
|
"rewards/rejected": -0.48516952991485596,
|
|
"step": 682
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 4.9266048036463566e-05,
|
|
"logits/chosen": -2.5754356384277344,
|
|
"logits/rejected": -2.668449878692627,
|
|
"logps/chosen": -223.462646484375,
|
|
"logps/rejected": -249.08425903320312,
|
|
"loss": 0.5664,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.4348244071006775,
|
|
"rewards/margins": 0.38006237149238586,
|
|
"rewards/rejected": -0.8148868680000305,
|
|
"step": 683
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.926354047750777e-05,
|
|
"logits/chosen": -2.148771047592163,
|
|
"logits/rejected": -2.174349546432495,
|
|
"logps/chosen": -191.0968780517578,
|
|
"logps/rejected": -201.15272521972656,
|
|
"loss": 0.7219,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.25711679458618164,
|
|
"rewards/margins": 0.0128050297498703,
|
|
"rewards/rejected": -0.26992183923721313,
|
|
"step": 684
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.926102870633029e-05,
|
|
"logits/chosen": -2.5497798919677734,
|
|
"logits/rejected": -2.5441908836364746,
|
|
"logps/chosen": -236.3572998046875,
|
|
"logps/rejected": -223.45065307617188,
|
|
"loss": 0.734,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.43852365016937256,
|
|
"rewards/margins": 0.028032146394252777,
|
|
"rewards/rejected": -0.46655580401420593,
|
|
"step": 685
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.925851272336717e-05,
|
|
"logits/chosen": -2.49284029006958,
|
|
"logits/rejected": -2.4962539672851562,
|
|
"logps/chosen": -187.33041381835938,
|
|
"logps/rejected": -161.27676391601562,
|
|
"loss": 0.7157,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4357530474662781,
|
|
"rewards/margins": 0.01825621724128723,
|
|
"rewards/rejected": -0.4540092647075653,
|
|
"step": 686
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.9255992529055195e-05,
|
|
"logits/chosen": -2.4232540130615234,
|
|
"logits/rejected": -2.4808590412139893,
|
|
"logps/chosen": -177.44873046875,
|
|
"logps/rejected": -173.36376953125,
|
|
"loss": 0.5234,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.000440012663602829,
|
|
"rewards/margins": 0.42676159739494324,
|
|
"rewards/rejected": -0.4272015690803528,
|
|
"step": 687
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.925346812383188e-05,
|
|
"logits/chosen": -2.5829415321350098,
|
|
"logits/rejected": -2.539314031600952,
|
|
"logps/chosen": -201.11561584472656,
|
|
"logps/rejected": -157.01829528808594,
|
|
"loss": 0.659,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.18970540165901184,
|
|
"rewards/margins": 0.1848244071006775,
|
|
"rewards/rejected": -0.37452977895736694,
|
|
"step": 688
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.925093950813547e-05,
|
|
"logits/chosen": -2.1912455558776855,
|
|
"logits/rejected": -2.235294818878174,
|
|
"logps/chosen": -158.46246337890625,
|
|
"logps/rejected": -182.1004180908203,
|
|
"loss": 0.7104,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.2667257487773895,
|
|
"rewards/margins": 0.0735367089509964,
|
|
"rewards/rejected": -0.3402624726295471,
|
|
"step": 689
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.924840668240495e-05,
|
|
"logits/chosen": -2.2375993728637695,
|
|
"logits/rejected": -2.270219564437866,
|
|
"logps/chosen": -206.09689331054688,
|
|
"logps/rejected": -185.82130432128906,
|
|
"loss": 0.8109,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.5142911672592163,
|
|
"rewards/margins": -0.055089227855205536,
|
|
"rewards/rejected": -0.4592019319534302,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 4.924586964708e-05,
|
|
"logits/chosen": -2.683393716812134,
|
|
"logits/rejected": -2.745443344116211,
|
|
"logps/chosen": -211.4540252685547,
|
|
"logps/rejected": -204.97247314453125,
|
|
"loss": 0.6848,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.388435035943985,
|
|
"rewards/margins": 0.09198732674121857,
|
|
"rewards/rejected": -0.48042237758636475,
|
|
"step": 691
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.9243328402601075e-05,
|
|
"logits/chosen": -2.4001381397247314,
|
|
"logits/rejected": -2.446291208267212,
|
|
"logps/chosen": -173.99537658691406,
|
|
"logps/rejected": -170.11166381835938,
|
|
"loss": 0.6884,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3871830105781555,
|
|
"rewards/margins": 0.05776102468371391,
|
|
"rewards/rejected": -0.44494396448135376,
|
|
"step": 692
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.924078294940935e-05,
|
|
"logits/chosen": -2.3540093898773193,
|
|
"logits/rejected": -2.4053566455841064,
|
|
"logps/chosen": -238.71336364746094,
|
|
"logps/rejected": -232.11026000976562,
|
|
"loss": 0.6272,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.38184699416160583,
|
|
"rewards/margins": 0.2840040624141693,
|
|
"rewards/rejected": -0.6658511161804199,
|
|
"step": 693
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.92382332879467e-05,
|
|
"logits/chosen": -2.5318589210510254,
|
|
"logits/rejected": -2.570647954940796,
|
|
"logps/chosen": -182.59814453125,
|
|
"logps/rejected": -190.98745727539062,
|
|
"loss": 0.6351,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4041218161582947,
|
|
"rewards/margins": 0.2801179587841034,
|
|
"rewards/rejected": -0.6842397451400757,
|
|
"step": 694
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.923567941865577e-05,
|
|
"logits/chosen": -2.440756320953369,
|
|
"logits/rejected": -2.4445791244506836,
|
|
"logps/chosen": -174.62588500976562,
|
|
"logps/rejected": -186.00222778320312,
|
|
"loss": 0.6708,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3223167061805725,
|
|
"rewards/margins": 0.1542733609676361,
|
|
"rewards/rejected": -0.4765900671482086,
|
|
"step": 695
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.923312134197991e-05,
|
|
"logits/chosen": -2.3524816036224365,
|
|
"logits/rejected": -2.2868611812591553,
|
|
"logps/chosen": -174.77328491210938,
|
|
"logps/rejected": -193.611083984375,
|
|
"loss": 0.8216,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4883018434047699,
|
|
"rewards/margins": -0.1851535439491272,
|
|
"rewards/rejected": -0.3031483292579651,
|
|
"step": 696
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.9230559058363223e-05,
|
|
"logits/chosen": -2.2453904151916504,
|
|
"logits/rejected": -2.2618868350982666,
|
|
"logps/chosen": -204.50955200195312,
|
|
"logps/rejected": -213.33969116210938,
|
|
"loss": 0.6688,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2155238389968872,
|
|
"rewards/margins": 0.14622408151626587,
|
|
"rewards/rejected": -0.3617479205131531,
|
|
"step": 697
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.922799256825051e-05,
|
|
"logits/chosen": -2.2087559700012207,
|
|
"logits/rejected": -2.289468765258789,
|
|
"logps/chosen": -223.2991943359375,
|
|
"logps/rejected": -193.14785766601562,
|
|
"loss": 0.8451,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.45806586742401123,
|
|
"rewards/margins": -0.23903487622737885,
|
|
"rewards/rejected": -0.21903102099895477,
|
|
"step": 698
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.9225421872087344e-05,
|
|
"logits/chosen": -2.6118435859680176,
|
|
"logits/rejected": -2.5954737663269043,
|
|
"logps/chosen": -218.72201538085938,
|
|
"logps/rejected": -223.23770141601562,
|
|
"loss": 0.6797,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.33901646733283997,
|
|
"rewards/margins": 0.09220410138368607,
|
|
"rewards/rejected": -0.43122053146362305,
|
|
"step": 699
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.922284697031999e-05,
|
|
"logits/chosen": -2.532050371170044,
|
|
"logits/rejected": -2.5437393188476562,
|
|
"logps/chosen": -205.47463989257812,
|
|
"logps/rejected": -194.44210815429688,
|
|
"loss": 0.6153,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.293450266122818,
|
|
"rewards/margins": 0.32601144909858704,
|
|
"rewards/rejected": -0.619461715221405,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.922026786339545e-05,
|
|
"logits/chosen": -2.4742062091827393,
|
|
"logits/rejected": -2.507223606109619,
|
|
"logps/chosen": -249.90118408203125,
|
|
"logps/rejected": -243.16616821289062,
|
|
"loss": 0.6483,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.3557482063770294,
|
|
"rewards/margins": 0.16462744772434235,
|
|
"rewards/rejected": -0.5203756093978882,
|
|
"step": 701
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.921768455176149e-05,
|
|
"logits/chosen": -2.313598155975342,
|
|
"logits/rejected": -2.38657546043396,
|
|
"logps/chosen": -218.81178283691406,
|
|
"logps/rejected": -273.9117126464844,
|
|
"loss": 0.5901,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.16350752115249634,
|
|
"rewards/margins": 0.2863968014717102,
|
|
"rewards/rejected": -0.44990429282188416,
|
|
"step": 702
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.921509703586657e-05,
|
|
"logits/chosen": -2.4725236892700195,
|
|
"logits/rejected": -2.46490216255188,
|
|
"logps/chosen": -204.388916015625,
|
|
"logps/rejected": -202.4940948486328,
|
|
"loss": 0.6274,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.11353873461484909,
|
|
"rewards/margins": 0.18249502778053284,
|
|
"rewards/rejected": -0.2960337698459625,
|
|
"step": 703
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.9212505316159876e-05,
|
|
"logits/chosen": -2.4417524337768555,
|
|
"logits/rejected": -2.4327855110168457,
|
|
"logps/chosen": -185.76559448242188,
|
|
"logps/rejected": -185.82675170898438,
|
|
"loss": 0.6754,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.16719910502433777,
|
|
"rewards/margins": 0.131285160779953,
|
|
"rewards/rejected": -0.29848429560661316,
|
|
"step": 704
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.920990939309135e-05,
|
|
"logits/chosen": -2.405703544616699,
|
|
"logits/rejected": -2.4366328716278076,
|
|
"logps/chosen": -182.18052673339844,
|
|
"logps/rejected": -177.89529418945312,
|
|
"loss": 0.5939,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.0767878070473671,
|
|
"rewards/margins": 0.26920846104621887,
|
|
"rewards/rejected": -0.34599626064300537,
|
|
"step": 705
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 4.9207309267111654e-05,
|
|
"logits/chosen": -2.267641067504883,
|
|
"logits/rejected": -2.359164237976074,
|
|
"logps/chosen": -167.9193115234375,
|
|
"logps/rejected": -198.30853271484375,
|
|
"loss": 0.5458,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.17353487014770508,
|
|
"rewards/margins": 0.4786756932735443,
|
|
"rewards/rejected": -0.6522104144096375,
|
|
"step": 706
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.920470493867217e-05,
|
|
"logits/chosen": -2.3301897048950195,
|
|
"logits/rejected": -2.287158489227295,
|
|
"logps/chosen": -179.1868133544922,
|
|
"logps/rejected": -173.25205993652344,
|
|
"loss": 0.8096,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.3885132968425751,
|
|
"rewards/margins": -0.15448781847953796,
|
|
"rewards/rejected": -0.2340254783630371,
|
|
"step": 707
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.920209640822503e-05,
|
|
"logits/chosen": -2.3286001682281494,
|
|
"logits/rejected": -2.372384786605835,
|
|
"logps/chosen": -183.89163208007812,
|
|
"logps/rejected": -171.33837890625,
|
|
"loss": 0.6243,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.200511634349823,
|
|
"rewards/margins": 0.1888035386800766,
|
|
"rewards/rejected": -0.3893151581287384,
|
|
"step": 708
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.919948367622307e-05,
|
|
"logits/chosen": -2.049145221710205,
|
|
"logits/rejected": -2.1091737747192383,
|
|
"logps/chosen": -165.6485595703125,
|
|
"logps/rejected": -196.3398895263672,
|
|
"loss": 0.6972,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.21507728099822998,
|
|
"rewards/margins": 0.1645798683166504,
|
|
"rewards/rejected": -0.37965714931488037,
|
|
"step": 709
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.919686674311987e-05,
|
|
"logits/chosen": -2.2241785526275635,
|
|
"logits/rejected": -2.1966073513031006,
|
|
"logps/chosen": -197.89120483398438,
|
|
"logps/rejected": -222.09132385253906,
|
|
"loss": 0.6355,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.12498195469379425,
|
|
"rewards/margins": 0.2979465425014496,
|
|
"rewards/rejected": -0.42292848229408264,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.9194245609369746e-05,
|
|
"logits/chosen": -2.3444929122924805,
|
|
"logits/rejected": -2.3332207202911377,
|
|
"logps/chosen": -240.3154296875,
|
|
"logps/rejected": -229.0690460205078,
|
|
"loss": 0.6236,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.17825375497341156,
|
|
"rewards/margins": 0.23597893118858337,
|
|
"rewards/rejected": -0.41423267126083374,
|
|
"step": 711
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.9191620275427725e-05,
|
|
"logits/chosen": -2.472026824951172,
|
|
"logits/rejected": -2.424983501434326,
|
|
"logps/chosen": -212.58364868164062,
|
|
"logps/rejected": -186.08990478515625,
|
|
"loss": 0.684,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.1622922122478485,
|
|
"rewards/margins": 0.10522370040416718,
|
|
"rewards/rejected": -0.2675159275531769,
|
|
"step": 712
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.9188990741749576e-05,
|
|
"logits/chosen": -2.481387138366699,
|
|
"logits/rejected": -2.4912960529327393,
|
|
"logps/chosen": -259.65753173828125,
|
|
"logps/rejected": -245.88864135742188,
|
|
"loss": 0.6385,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.3589114546775818,
|
|
"rewards/margins": 0.13590216636657715,
|
|
"rewards/rejected": -0.49481362104415894,
|
|
"step": 713
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 4.918635700879179e-05,
|
|
"logits/chosen": -2.4197511672973633,
|
|
"logits/rejected": -2.442139148712158,
|
|
"logps/chosen": -180.3101806640625,
|
|
"logps/rejected": -201.78469848632812,
|
|
"loss": 0.696,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.27619680762290955,
|
|
"rewards/margins": 0.04660003259778023,
|
|
"rewards/rejected": -0.32279688119888306,
|
|
"step": 714
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.918371907701159e-05,
|
|
"logits/chosen": -2.333615303039551,
|
|
"logits/rejected": -2.3523952960968018,
|
|
"logps/chosen": -176.4866485595703,
|
|
"logps/rejected": -190.32089233398438,
|
|
"loss": 0.755,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.353998064994812,
|
|
"rewards/margins": -0.033618077635765076,
|
|
"rewards/rejected": -0.32037997245788574,
|
|
"step": 715
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.9181076946866944e-05,
|
|
"logits/chosen": -2.45198655128479,
|
|
"logits/rejected": -2.432290554046631,
|
|
"logps/chosen": -178.83279418945312,
|
|
"logps/rejected": -188.7766876220703,
|
|
"loss": 0.7776,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.4141733944416046,
|
|
"rewards/margins": -0.06833554804325104,
|
|
"rewards/rejected": -0.3458378314971924,
|
|
"step": 716
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.917843061881652e-05,
|
|
"logits/chosen": -2.4485764503479004,
|
|
"logits/rejected": -2.412081241607666,
|
|
"logps/chosen": -229.4910430908203,
|
|
"logps/rejected": -225.20999145507812,
|
|
"loss": 0.7713,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4986688494682312,
|
|
"rewards/margins": -0.1049906462430954,
|
|
"rewards/rejected": -0.3936781883239746,
|
|
"step": 717
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.9175780093319723e-05,
|
|
"logits/chosen": -2.619525194168091,
|
|
"logits/rejected": -2.5976409912109375,
|
|
"logps/chosen": -152.8652801513672,
|
|
"logps/rejected": -190.03179931640625,
|
|
"loss": 0.6384,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.044289495795965195,
|
|
"rewards/margins": 0.1848900467157364,
|
|
"rewards/rejected": -0.2291795164346695,
|
|
"step": 718
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.91731253708367e-05,
|
|
"logits/chosen": -2.2807188034057617,
|
|
"logits/rejected": -2.338017463684082,
|
|
"logps/chosen": -182.47259521484375,
|
|
"logps/rejected": -196.39678955078125,
|
|
"loss": 0.6927,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.26287615299224854,
|
|
"rewards/margins": 0.050617627799510956,
|
|
"rewards/rejected": -0.3134937882423401,
|
|
"step": 719
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.9170466451828326e-05,
|
|
"logits/chosen": -2.2561631202697754,
|
|
"logits/rejected": -2.3567490577697754,
|
|
"logps/chosen": -242.33407592773438,
|
|
"logps/rejected": -263.4303283691406,
|
|
"loss": 0.7338,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.2756967544555664,
|
|
"rewards/margins": 0.13067740201950073,
|
|
"rewards/rejected": -0.40637409687042236,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.916780333675618e-05,
|
|
"logits/chosen": -2.5038459300994873,
|
|
"logits/rejected": -2.500086545944214,
|
|
"logps/chosen": -187.3552703857422,
|
|
"logps/rejected": -187.51268005371094,
|
|
"loss": 0.6628,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.2900952398777008,
|
|
"rewards/margins": 0.16025833785533905,
|
|
"rewards/rejected": -0.4503535330295563,
|
|
"step": 721
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.9165136026082604e-05,
|
|
"logits/chosen": -2.2002265453338623,
|
|
"logits/rejected": -2.226296901702881,
|
|
"logps/chosen": -200.2354278564453,
|
|
"logps/rejected": -195.80328369140625,
|
|
"loss": 0.6676,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.4443664848804474,
|
|
"rewards/margins": 0.14572381973266602,
|
|
"rewards/rejected": -0.5900903344154358,
|
|
"step": 722
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.916246452027063e-05,
|
|
"logits/chosen": -2.1249847412109375,
|
|
"logits/rejected": -2.116081953048706,
|
|
"logps/chosen": -214.81161499023438,
|
|
"logps/rejected": -220.05429077148438,
|
|
"loss": 0.664,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.25689569115638733,
|
|
"rewards/margins": 0.1331162005662918,
|
|
"rewards/rejected": -0.39001190662384033,
|
|
"step": 723
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.9159788819784064e-05,
|
|
"logits/chosen": -2.3239493370056152,
|
|
"logits/rejected": -2.3413610458374023,
|
|
"logps/chosen": -209.00692749023438,
|
|
"logps/rejected": -198.1204376220703,
|
|
"loss": 0.627,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.2650049328804016,
|
|
"rewards/margins": 0.20938965678215027,
|
|
"rewards/rejected": -0.47439464926719666,
|
|
"step": 724
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.9157108925087405e-05,
|
|
"logits/chosen": -2.420534610748291,
|
|
"logits/rejected": -2.442646026611328,
|
|
"logps/chosen": -211.38424682617188,
|
|
"logps/rejected": -207.95278930664062,
|
|
"loss": 0.5418,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.22766509652137756,
|
|
"rewards/margins": 0.3782108724117279,
|
|
"rewards/rejected": -0.6058759689331055,
|
|
"step": 725
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.915442483664588e-05,
|
|
"logits/chosen": -2.6422438621520996,
|
|
"logits/rejected": -2.4694361686706543,
|
|
"logps/chosen": -173.2271728515625,
|
|
"logps/rejected": -145.02798461914062,
|
|
"loss": 0.8003,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.25257498025894165,
|
|
"rewards/margins": -0.11290948837995529,
|
|
"rewards/rejected": -0.13966546952724457,
|
|
"step": 726
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.915173655492547e-05,
|
|
"logits/chosen": -2.5392398834228516,
|
|
"logits/rejected": -2.4417877197265625,
|
|
"logps/chosen": -228.2000732421875,
|
|
"logps/rejected": -199.39288330078125,
|
|
"loss": 0.609,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.23776192963123322,
|
|
"rewards/margins": 0.23510348796844482,
|
|
"rewards/rejected": -0.47286540269851685,
|
|
"step": 727
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.914904408039286e-05,
|
|
"logits/chosen": -2.4822921752929688,
|
|
"logits/rejected": -2.5780792236328125,
|
|
"logps/chosen": -194.31045532226562,
|
|
"logps/rejected": -235.24607849121094,
|
|
"loss": 0.5732,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3551061451435089,
|
|
"rewards/margins": 0.39131835103034973,
|
|
"rewards/rejected": -0.7464244961738586,
|
|
"step": 728
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 4.914634741351547e-05,
|
|
"logits/chosen": -2.4005560874938965,
|
|
"logits/rejected": -2.4636473655700684,
|
|
"logps/chosen": -222.8491668701172,
|
|
"logps/rejected": -229.69638061523438,
|
|
"loss": 0.7059,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.444613516330719,
|
|
"rewards/margins": 0.05394814535975456,
|
|
"rewards/rejected": -0.49856165051460266,
|
|
"step": 729
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.914364655476146e-05,
|
|
"logits/chosen": -2.541283369064331,
|
|
"logits/rejected": -2.4821250438690186,
|
|
"logps/chosen": -199.74075317382812,
|
|
"logps/rejected": -244.3222198486328,
|
|
"loss": 0.5754,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.24406713247299194,
|
|
"rewards/margins": 0.44718241691589355,
|
|
"rewards/rejected": -0.6912496089935303,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.914094150459969e-05,
|
|
"logits/chosen": -2.3721487522125244,
|
|
"logits/rejected": -2.4787163734436035,
|
|
"logps/chosen": -185.42625427246094,
|
|
"logps/rejected": -214.8418426513672,
|
|
"loss": 0.741,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.5116318464279175,
|
|
"rewards/margins": -0.03801153600215912,
|
|
"rewards/rejected": -0.47362035512924194,
|
|
"step": 731
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.9138232263499784e-05,
|
|
"logits/chosen": -2.5538036823272705,
|
|
"logits/rejected": -2.467003583908081,
|
|
"logps/chosen": -197.03834533691406,
|
|
"logps/rejected": -214.0052947998047,
|
|
"loss": 0.812,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5602855682373047,
|
|
"rewards/margins": -0.166466623544693,
|
|
"rewards/rejected": -0.3938189446926117,
|
|
"step": 732
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.913551883193206e-05,
|
|
"logits/chosen": -2.4558639526367188,
|
|
"logits/rejected": -2.5085341930389404,
|
|
"logps/chosen": -225.220947265625,
|
|
"logps/rejected": -227.97213745117188,
|
|
"loss": 0.6471,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.2800041437149048,
|
|
"rewards/margins": 0.34519922733306885,
|
|
"rewards/rejected": -0.6252033114433289,
|
|
"step": 733
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.9132801210367586e-05,
|
|
"logits/chosen": -2.405069351196289,
|
|
"logits/rejected": -2.4901866912841797,
|
|
"logps/chosen": -152.56527709960938,
|
|
"logps/rejected": -152.64671325683594,
|
|
"loss": 0.6334,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.3601643145084381,
|
|
"rewards/margins": 0.2201770395040512,
|
|
"rewards/rejected": -0.5803413987159729,
|
|
"step": 734
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.913007939927814e-05,
|
|
"logits/chosen": -2.3362531661987305,
|
|
"logits/rejected": -2.3917181491851807,
|
|
"logps/chosen": -193.826904296875,
|
|
"logps/rejected": -231.73898315429688,
|
|
"loss": 0.6171,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5505282282829285,
|
|
"rewards/margins": 0.3396502137184143,
|
|
"rewards/rejected": -0.890178382396698,
|
|
"step": 735
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.912735339913625e-05,
|
|
"logits/chosen": -2.132725715637207,
|
|
"logits/rejected": -2.2681026458740234,
|
|
"logps/chosen": -198.4554443359375,
|
|
"logps/rejected": -205.1267852783203,
|
|
"loss": 0.6419,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.34412312507629395,
|
|
"rewards/margins": 0.1572694331407547,
|
|
"rewards/rejected": -0.5013926029205322,
|
|
"step": 736
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 4.912462321041513e-05,
|
|
"logits/chosen": -2.352013349533081,
|
|
"logits/rejected": -2.474015235900879,
|
|
"logps/chosen": -216.76731872558594,
|
|
"logps/rejected": -208.0941162109375,
|
|
"loss": 0.7258,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.49095839262008667,
|
|
"rewards/margins": -0.004371330142021179,
|
|
"rewards/rejected": -0.4865871071815491,
|
|
"step": 737
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.9121888833588795e-05,
|
|
"logits/chosen": -2.395151138305664,
|
|
"logits/rejected": -2.40535306930542,
|
|
"logps/chosen": -163.3991241455078,
|
|
"logps/rejected": -184.1988525390625,
|
|
"loss": 0.6083,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.35962533950805664,
|
|
"rewards/margins": 0.20414502918720245,
|
|
"rewards/rejected": -0.5637703537940979,
|
|
"step": 738
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.91191502691319e-05,
|
|
"logits/chosen": -2.3076629638671875,
|
|
"logits/rejected": -2.460707426071167,
|
|
"logps/chosen": -163.59388732910156,
|
|
"logps/rejected": -190.1605987548828,
|
|
"loss": 0.6092,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2011553794145584,
|
|
"rewards/margins": 0.24963317811489105,
|
|
"rewards/rejected": -0.45078855752944946,
|
|
"step": 739
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.911640751751988e-05,
|
|
"logits/chosen": -2.419591188430786,
|
|
"logits/rejected": -2.537069797515869,
|
|
"logps/chosen": -250.05996704101562,
|
|
"logps/rejected": -225.0284881591797,
|
|
"loss": 0.7242,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6081552505493164,
|
|
"rewards/margins": 0.06761530041694641,
|
|
"rewards/rejected": -0.6757705807685852,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.9113660579228886e-05,
|
|
"logits/chosen": -2.4080655574798584,
|
|
"logits/rejected": -2.4866342544555664,
|
|
"logps/chosen": -168.5004119873047,
|
|
"logps/rejected": -216.32675170898438,
|
|
"loss": 0.6338,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5959919691085815,
|
|
"rewards/margins": 0.22804662585258484,
|
|
"rewards/rejected": -0.8240386247634888,
|
|
"step": 741
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.91109094547358e-05,
|
|
"logits/chosen": -2.2880287170410156,
|
|
"logits/rejected": -2.4074809551239014,
|
|
"logps/chosen": -184.07696533203125,
|
|
"logps/rejected": -215.08717346191406,
|
|
"loss": 0.6027,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.49041178822517395,
|
|
"rewards/margins": 0.23776140809059143,
|
|
"rewards/rejected": -0.7281732559204102,
|
|
"step": 742
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.910815414451821e-05,
|
|
"logits/chosen": -2.4181840419769287,
|
|
"logits/rejected": -2.4892990589141846,
|
|
"logps/chosen": -215.46221923828125,
|
|
"logps/rejected": -229.75799560546875,
|
|
"loss": 0.608,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.7831647992134094,
|
|
"rewards/margins": 0.38174957036972046,
|
|
"rewards/rejected": -1.1649143695831299,
|
|
"step": 743
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.910539464905447e-05,
|
|
"logits/chosen": -2.2858383655548096,
|
|
"logits/rejected": -2.3867931365966797,
|
|
"logps/chosen": -257.6205749511719,
|
|
"logps/rejected": -225.37481689453125,
|
|
"loss": 0.6948,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5961012840270996,
|
|
"rewards/margins": 0.15920449793338776,
|
|
"rewards/rejected": -0.7553057670593262,
|
|
"step": 744
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.910263096882362e-05,
|
|
"logits/chosen": -2.2314653396606445,
|
|
"logits/rejected": -2.242586851119995,
|
|
"logps/chosen": -157.20448303222656,
|
|
"logps/rejected": -189.92567443847656,
|
|
"loss": 0.7938,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5152949690818787,
|
|
"rewards/margins": -0.07010670006275177,
|
|
"rewards/rejected": -0.4451882541179657,
|
|
"step": 745
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.9099863104305446e-05,
|
|
"logits/chosen": -2.34870982170105,
|
|
"logits/rejected": -2.3678641319274902,
|
|
"logps/chosen": -157.9035186767578,
|
|
"logps/rejected": -194.49575805664062,
|
|
"loss": 0.6431,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.39278292655944824,
|
|
"rewards/margins": 0.1728055626153946,
|
|
"rewards/rejected": -0.5655884742736816,
|
|
"step": 746
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.909709105598045e-05,
|
|
"logits/chosen": -2.290536403656006,
|
|
"logits/rejected": -2.316704034805298,
|
|
"logps/chosen": -173.55892944335938,
|
|
"logps/rejected": -190.94125366210938,
|
|
"loss": 0.6078,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.3803100883960724,
|
|
"rewards/margins": 0.2535950839519501,
|
|
"rewards/rejected": -0.6339051723480225,
|
|
"step": 747
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.9094314824329876e-05,
|
|
"logits/chosen": -2.5316176414489746,
|
|
"logits/rejected": -2.4976158142089844,
|
|
"logps/chosen": -206.0629119873047,
|
|
"logps/rejected": -219.75418090820312,
|
|
"loss": 0.8139,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.5901838541030884,
|
|
"rewards/margins": -0.1481025069952011,
|
|
"rewards/rejected": -0.4420813322067261,
|
|
"step": 748
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.9091534409835694e-05,
|
|
"logits/chosen": -2.500584363937378,
|
|
"logits/rejected": -2.5805132389068604,
|
|
"logps/chosen": -174.99212646484375,
|
|
"logps/rejected": -207.73011779785156,
|
|
"loss": 0.574,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.4852156937122345,
|
|
"rewards/margins": 0.3404516577720642,
|
|
"rewards/rejected": -0.8256674408912659,
|
|
"step": 749
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.908874981298057e-05,
|
|
"logits/chosen": -2.5304508209228516,
|
|
"logits/rejected": -2.6482512950897217,
|
|
"logps/chosen": -160.60971069335938,
|
|
"logps/rejected": -178.6311492919922,
|
|
"loss": 0.7716,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.4607134461402893,
|
|
"rewards/margins": -0.038665421307086945,
|
|
"rewards/rejected": -0.42204800248146057,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.908596103424794e-05,
|
|
"logits/chosen": -2.247166633605957,
|
|
"logits/rejected": -2.2175076007843018,
|
|
"logps/chosen": -185.73583984375,
|
|
"logps/rejected": -191.45596313476562,
|
|
"loss": 0.8569,
|
|
"rewards/accuracies": 0.25,
|
|
"rewards/chosen": -0.6400306820869446,
|
|
"rewards/margins": -0.2516973912715912,
|
|
"rewards/rejected": -0.3883333206176758,
|
|
"step": 751
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 4.908316807412194e-05,
|
|
"logits/chosen": -2.492708206176758,
|
|
"logits/rejected": -2.585394859313965,
|
|
"logps/chosen": -213.05731201171875,
|
|
"logps/rejected": -260.92462158203125,
|
|
"loss": 0.6273,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.5991235971450806,
|
|
"rewards/margins": 0.2069501429796219,
|
|
"rewards/rejected": -0.806073784828186,
|
|
"step": 752
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.908037093308742e-05,
|
|
"logits/chosen": -2.5198185443878174,
|
|
"logits/rejected": -2.5527148246765137,
|
|
"logps/chosen": -238.94021606445312,
|
|
"logps/rejected": -262.2608337402344,
|
|
"loss": 0.699,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.38716763257980347,
|
|
"rewards/margins": 0.033374398946762085,
|
|
"rewards/rejected": -0.42054203152656555,
|
|
"step": 753
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.907756961162999e-05,
|
|
"logits/chosen": -2.26259708404541,
|
|
"logits/rejected": -2.2545177936553955,
|
|
"logps/chosen": -150.26882934570312,
|
|
"logps/rejected": -227.8256072998047,
|
|
"loss": 0.579,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.3753708302974701,
|
|
"rewards/margins": 0.4033656418323517,
|
|
"rewards/rejected": -0.7787365317344666,
|
|
"step": 754
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.907476411023596e-05,
|
|
"logits/chosen": -2.3281123638153076,
|
|
"logits/rejected": -2.330186128616333,
|
|
"logps/chosen": -149.7628936767578,
|
|
"logps/rejected": -193.10093688964844,
|
|
"loss": 0.7339,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.6727653741836548,
|
|
"rewards/margins": 0.06607560813426971,
|
|
"rewards/rejected": -0.7388409376144409,
|
|
"step": 755
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.907195442939237e-05,
|
|
"logits/chosen": -2.4469387531280518,
|
|
"logits/rejected": -2.509931802749634,
|
|
"logps/chosen": -174.05670166015625,
|
|
"logps/rejected": -189.08465576171875,
|
|
"loss": 0.701,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.5176682472229004,
|
|
"rewards/margins": 0.04994421452283859,
|
|
"rewards/rejected": -0.5676124095916748,
|
|
"step": 756
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.906914056958699e-05,
|
|
"logits/chosen": -2.6258902549743652,
|
|
"logits/rejected": -2.6027467250823975,
|
|
"logps/chosen": -208.80203247070312,
|
|
"logps/rejected": -200.4151611328125,
|
|
"loss": 0.7588,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6569511294364929,
|
|
"rewards/margins": -0.044889748096466064,
|
|
"rewards/rejected": -0.6120614409446716,
|
|
"step": 757
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.906632253130833e-05,
|
|
"logits/chosen": -2.5387301445007324,
|
|
"logits/rejected": -2.496183156967163,
|
|
"logps/chosen": -193.3511199951172,
|
|
"logps/rejected": -199.80130004882812,
|
|
"loss": 0.7945,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6182142496109009,
|
|
"rewards/margins": -0.13901755213737488,
|
|
"rewards/rejected": -0.4791966676712036,
|
|
"step": 758
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.9063500315045586e-05,
|
|
"logits/chosen": -2.5884640216827393,
|
|
"logits/rejected": -2.641512393951416,
|
|
"logps/chosen": -186.569091796875,
|
|
"logps/rejected": -200.59988403320312,
|
|
"loss": 0.7447,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.6641149520874023,
|
|
"rewards/margins": -0.048228248953819275,
|
|
"rewards/rejected": -0.6158866882324219,
|
|
"step": 759
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 4.9060673921288716e-05,
|
|
"logits/chosen": -2.633413791656494,
|
|
"logits/rejected": -2.598926305770874,
|
|
"logps/chosen": -161.99928283691406,
|
|
"logps/rejected": -154.25979614257812,
|
|
"loss": 0.7324,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.37947404384613037,
|
|
"rewards/margins": -0.021064819768071175,
|
|
"rewards/rejected": -0.35840916633605957,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.905784335052839e-05,
|
|
"logits/chosen": -2.534295082092285,
|
|
"logits/rejected": -2.5714004039764404,
|
|
"logps/chosen": -213.0132598876953,
|
|
"logps/rejected": -223.10958862304688,
|
|
"loss": 0.685,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.7191438674926758,
|
|
"rewards/margins": 0.12883953750133514,
|
|
"rewards/rejected": -0.8479833602905273,
|
|
"step": 761
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.905500860325599e-05,
|
|
"logits/chosen": -2.3805532455444336,
|
|
"logits/rejected": -2.502260208129883,
|
|
"logps/chosen": -187.92947387695312,
|
|
"logps/rejected": -244.21273803710938,
|
|
"loss": 0.5568,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.30298084020614624,
|
|
"rewards/margins": 0.5082517862319946,
|
|
"rewards/rejected": -0.8112326264381409,
|
|
"step": 762
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.905216967996367e-05,
|
|
"logits/chosen": -2.494596242904663,
|
|
"logits/rejected": -2.5522210597991943,
|
|
"logps/chosen": -193.31663513183594,
|
|
"logps/rejected": -253.28793334960938,
|
|
"loss": 0.5326,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.31040558218955994,
|
|
"rewards/margins": 0.41559797525405884,
|
|
"rewards/rejected": -0.7260035872459412,
|
|
"step": 763
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.904932658114423e-05,
|
|
"logits/chosen": -2.4357709884643555,
|
|
"logits/rejected": -2.402021884918213,
|
|
"logps/chosen": -207.14736938476562,
|
|
"logps/rejected": -224.48757934570312,
|
|
"loss": 0.7383,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.6394506692886353,
|
|
"rewards/margins": 0.015030726790428162,
|
|
"rewards/rejected": -0.6544813513755798,
|
|
"step": 764
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.904647930729128e-05,
|
|
"logits/chosen": -2.492413282394409,
|
|
"logits/rejected": -2.5191187858581543,
|
|
"logps/chosen": -209.2412872314453,
|
|
"logps/rejected": -204.59518432617188,
|
|
"loss": 0.2419,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.42869845032691956,
|
|
"rewards/margins": 2.3007590770721436,
|
|
"rewards/rejected": -1.872060775756836,
|
|
"step": 765
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.9043627858899086e-05,
|
|
"logits/chosen": -2.616361618041992,
|
|
"logits/rejected": -2.640256404876709,
|
|
"logps/chosen": -180.7271270751953,
|
|
"logps/rejected": -231.54254150390625,
|
|
"loss": 0.1455,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.1313096433877945,
|
|
"rewards/margins": 2.220848321914673,
|
|
"rewards/rejected": -2.08953857421875,
|
|
"step": 766
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.9040772236462695e-05,
|
|
"logits/chosen": -2.3850975036621094,
|
|
"logits/rejected": -2.354506254196167,
|
|
"logps/chosen": -176.04693603515625,
|
|
"logps/rejected": -191.62985229492188,
|
|
"loss": 0.2206,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.09096850454807281,
|
|
"rewards/margins": 1.8953343629837036,
|
|
"rewards/rejected": -1.8043657541275024,
|
|
"step": 767
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.903791244047783e-05,
|
|
"logits/chosen": -2.448336362838745,
|
|
"logits/rejected": -2.4442625045776367,
|
|
"logps/chosen": -159.71731567382812,
|
|
"logps/rejected": -198.8458251953125,
|
|
"loss": 0.2542,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.14124417304992676,
|
|
"rewards/margins": 1.9044493436813354,
|
|
"rewards/rejected": -1.7632049322128296,
|
|
"step": 768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.9035048471440966e-05,
|
|
"logits/chosen": -2.39510178565979,
|
|
"logits/rejected": -2.3210549354553223,
|
|
"logps/chosen": -146.8027801513672,
|
|
"logps/rejected": -182.59580993652344,
|
|
"loss": 0.2789,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.44124722480773926,
|
|
"rewards/margins": 2.291222095489502,
|
|
"rewards/rejected": -1.8499748706817627,
|
|
"step": 769
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.90321803298493e-05,
|
|
"logits/chosen": -2.610016107559204,
|
|
"logits/rejected": -2.5617635250091553,
|
|
"logps/chosen": -191.46194458007812,
|
|
"logps/rejected": -232.915283203125,
|
|
"loss": 0.1781,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.716643214225769,
|
|
"rewards/margins": 2.4686214923858643,
|
|
"rewards/rejected": -1.7519782781600952,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.9029308016200735e-05,
|
|
"logits/chosen": -2.512070655822754,
|
|
"logits/rejected": -2.522047758102417,
|
|
"logps/chosen": -228.5362548828125,
|
|
"logps/rejected": -267.5350341796875,
|
|
"loss": 0.1084,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.7526074051856995,
|
|
"rewards/margins": 3.0486981868743896,
|
|
"rewards/rejected": -2.296090602874756,
|
|
"step": 771
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.902643153099394e-05,
|
|
"logits/chosen": -2.7084665298461914,
|
|
"logits/rejected": -2.702622413635254,
|
|
"logps/chosen": -214.2321319580078,
|
|
"logps/rejected": -234.52236938476562,
|
|
"loss": 0.2319,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": 0.38715142011642456,
|
|
"rewards/margins": 2.1679859161376953,
|
|
"rewards/rejected": -1.780834436416626,
|
|
"step": 772
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.902355087472826e-05,
|
|
"logits/chosen": -2.4830422401428223,
|
|
"logits/rejected": -2.449504852294922,
|
|
"logps/chosen": -157.030517578125,
|
|
"logps/rejected": -182.37347412109375,
|
|
"loss": 0.2089,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.3868522644042969,
|
|
"rewards/margins": 2.223161220550537,
|
|
"rewards/rejected": -1.8363089561462402,
|
|
"step": 773
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.902066604790379e-05,
|
|
"logits/chosen": -2.387758731842041,
|
|
"logits/rejected": -2.3118813037872314,
|
|
"logps/chosen": -216.08274841308594,
|
|
"logps/rejected": -232.153076171875,
|
|
"loss": 0.3762,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": 0.1578804850578308,
|
|
"rewards/margins": 1.8973913192749023,
|
|
"rewards/rejected": -1.7395107746124268,
|
|
"step": 774
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.901777705102135e-05,
|
|
"logits/chosen": -2.4494662284851074,
|
|
"logits/rejected": -2.5127110481262207,
|
|
"logps/chosen": -173.77740478515625,
|
|
"logps/rejected": -214.7391357421875,
|
|
"loss": 0.1939,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.14900563657283783,
|
|
"rewards/margins": 2.133729934692383,
|
|
"rewards/rejected": -1.9847242832183838,
|
|
"step": 775
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.901488388458247e-05,
|
|
"logits/chosen": -2.4321305751800537,
|
|
"logits/rejected": -2.449230432510376,
|
|
"logps/chosen": -173.65748596191406,
|
|
"logps/rejected": -176.95252990722656,
|
|
"loss": 0.218,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": 0.25331249833106995,
|
|
"rewards/margins": 2.11264967918396,
|
|
"rewards/rejected": -1.8593370914459229,
|
|
"step": 776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.901198654908942e-05,
|
|
"logits/chosen": -2.3953897953033447,
|
|
"logits/rejected": -2.4313783645629883,
|
|
"logps/chosen": -207.082275390625,
|
|
"logps/rejected": -261.56427001953125,
|
|
"loss": 0.1348,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4747293293476105,
|
|
"rewards/margins": 3.3515007495880127,
|
|
"rewards/rejected": -2.8767709732055664,
|
|
"step": 777
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.9009085045045175e-05,
|
|
"logits/chosen": -2.613166093826294,
|
|
"logits/rejected": -2.6309146881103516,
|
|
"logps/chosen": -218.4897003173828,
|
|
"logps/rejected": -231.17636108398438,
|
|
"loss": 0.2987,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.09049064666032791,
|
|
"rewards/margins": 1.798804521560669,
|
|
"rewards/rejected": -1.8892953395843506,
|
|
"step": 778
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.9006179372953466e-05,
|
|
"logits/chosen": -2.4088118076324463,
|
|
"logits/rejected": -2.3893849849700928,
|
|
"logps/chosen": -231.118408203125,
|
|
"logps/rejected": -272.7298278808594,
|
|
"loss": 0.1272,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.22063946723937988,
|
|
"rewards/margins": 3.302457332611084,
|
|
"rewards/rejected": -3.0818183422088623,
|
|
"step": 779
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.9003269533318704e-05,
|
|
"logits/chosen": -2.716017961502075,
|
|
"logits/rejected": -2.5699970722198486,
|
|
"logps/chosen": -216.4443817138672,
|
|
"logps/rejected": -191.3568115234375,
|
|
"loss": 0.1168,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5278897285461426,
|
|
"rewards/margins": 2.6199638843536377,
|
|
"rewards/rejected": -2.092073917388916,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.900035552664605e-05,
|
|
"logits/chosen": -2.5695183277130127,
|
|
"logits/rejected": -2.6082301139831543,
|
|
"logps/chosen": -200.3180389404297,
|
|
"logps/rejected": -233.2794189453125,
|
|
"loss": 0.1978,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": 0.10721993446350098,
|
|
"rewards/margins": 2.7741177082061768,
|
|
"rewards/rejected": -2.666897773742676,
|
|
"step": 781
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.899743735344139e-05,
|
|
"logits/chosen": -2.1258223056793213,
|
|
"logits/rejected": -2.2032220363616943,
|
|
"logps/chosen": -201.60614013671875,
|
|
"logps/rejected": -282.1663818359375,
|
|
"loss": 0.1137,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.04111824184656143,
|
|
"rewards/margins": 3.605041027069092,
|
|
"rewards/rejected": -3.56392240524292,
|
|
"step": 782
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.899451501421133e-05,
|
|
"logits/chosen": -2.280001163482666,
|
|
"logits/rejected": -2.454638957977295,
|
|
"logps/chosen": -185.27694702148438,
|
|
"logps/rejected": -216.79037475585938,
|
|
"loss": 0.181,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.409013569355011,
|
|
"rewards/margins": 2.352353572845459,
|
|
"rewards/rejected": -2.7613673210144043,
|
|
"step": 783
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.899158850946319e-05,
|
|
"logits/chosen": -2.4744930267333984,
|
|
"logits/rejected": -2.5907466411590576,
|
|
"logps/chosen": -216.58287048339844,
|
|
"logps/rejected": -261.81353759765625,
|
|
"loss": 0.1848,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.11592866480350494,
|
|
"rewards/margins": 2.6159138679504395,
|
|
"rewards/rejected": -2.4999852180480957,
|
|
"step": 784
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.8988657839705024e-05,
|
|
"logits/chosen": -2.6295738220214844,
|
|
"logits/rejected": -2.565519332885742,
|
|
"logps/chosen": -175.5509033203125,
|
|
"logps/rejected": -192.17001342773438,
|
|
"loss": 0.2344,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.13090485334396362,
|
|
"rewards/margins": 2.3108062744140625,
|
|
"rewards/rejected": -2.44171142578125,
|
|
"step": 785
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.89857230054456e-05,
|
|
"logits/chosen": -2.4025466442108154,
|
|
"logits/rejected": -2.478341817855835,
|
|
"logps/chosen": -160.97872924804688,
|
|
"logps/rejected": -221.78729248046875,
|
|
"loss": 0.2185,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.18785348534584045,
|
|
"rewards/margins": 2.667365074157715,
|
|
"rewards/rejected": -2.8552186489105225,
|
|
"step": 786
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.898278400719442e-05,
|
|
"logits/chosen": -2.309408664703369,
|
|
"logits/rejected": -2.336895227432251,
|
|
"logps/chosen": -194.12857055664062,
|
|
"logps/rejected": -229.64393615722656,
|
|
"loss": 0.1647,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.5826263427734375,
|
|
"rewards/margins": 2.673534870147705,
|
|
"rewards/rejected": -3.2561612129211426,
|
|
"step": 787
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.897984084546169e-05,
|
|
"logits/chosen": -2.5822911262512207,
|
|
"logits/rejected": -2.594147205352783,
|
|
"logps/chosen": -189.2248992919922,
|
|
"logps/rejected": -217.4370880126953,
|
|
"loss": 0.1947,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.18159398436546326,
|
|
"rewards/margins": 2.9417152404785156,
|
|
"rewards/rejected": -3.123309373855591,
|
|
"step": 788
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.897689352075837e-05,
|
|
"logits/chosen": -2.395965576171875,
|
|
"logits/rejected": -2.3766961097717285,
|
|
"logps/chosen": -234.74583435058594,
|
|
"logps/rejected": -283.2368469238281,
|
|
"loss": 0.2304,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.37718701362609863,
|
|
"rewards/margins": 3.130066394805908,
|
|
"rewards/rejected": -3.507253408432007,
|
|
"step": 789
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.897394203359611e-05,
|
|
"logits/chosen": -2.223119020462036,
|
|
"logits/rejected": -2.1850533485412598,
|
|
"logps/chosen": -163.25404357910156,
|
|
"logps/rejected": -190.954833984375,
|
|
"loss": 0.1986,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.48350968956947327,
|
|
"rewards/margins": 2.4246490001678467,
|
|
"rewards/rejected": -2.908158779144287,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.897098638448731e-05,
|
|
"logits/chosen": -2.617220878601074,
|
|
"logits/rejected": -2.6898252964019775,
|
|
"logps/chosen": -191.36790466308594,
|
|
"logps/rejected": -270.26287841796875,
|
|
"loss": 0.1789,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7169384360313416,
|
|
"rewards/margins": 2.659780979156494,
|
|
"rewards/rejected": -3.3767194747924805,
|
|
"step": 791
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.896802657394506e-05,
|
|
"logits/chosen": -2.3435750007629395,
|
|
"logits/rejected": -2.2688729763031006,
|
|
"logps/chosen": -154.95065307617188,
|
|
"logps/rejected": -174.447021484375,
|
|
"loss": 0.2578,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.7484995722770691,
|
|
"rewards/margins": 2.27825927734375,
|
|
"rewards/rejected": -3.0267586708068848,
|
|
"step": 792
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.8965062602483205e-05,
|
|
"logits/chosen": -2.4353139400482178,
|
|
"logits/rejected": -2.381152629852295,
|
|
"logps/chosen": -270.47259521484375,
|
|
"logps/rejected": -281.2034606933594,
|
|
"loss": 0.16,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.4860559105873108,
|
|
"rewards/margins": 3.1953186988830566,
|
|
"rewards/rejected": -3.6813745498657227,
|
|
"step": 793
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.89620944706163e-05,
|
|
"logits/chosen": -2.391406536102295,
|
|
"logits/rejected": -2.426130533218384,
|
|
"logps/chosen": -167.80072021484375,
|
|
"logps/rejected": -207.9481964111328,
|
|
"loss": 0.1768,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.36457765102386475,
|
|
"rewards/margins": 3.225442409515381,
|
|
"rewards/rejected": -3.590019941329956,
|
|
"step": 794
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.8959122178859616e-05,
|
|
"logits/chosen": -2.5002002716064453,
|
|
"logits/rejected": -2.584379196166992,
|
|
"logps/chosen": -194.40185546875,
|
|
"logps/rejected": -240.38218688964844,
|
|
"loss": 0.1416,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7205836176872253,
|
|
"rewards/margins": 3.2785701751708984,
|
|
"rewards/rejected": -3.9991536140441895,
|
|
"step": 795
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.8956145727729156e-05,
|
|
"logits/chosen": -2.7242343425750732,
|
|
"logits/rejected": -2.7420852184295654,
|
|
"logps/chosen": -193.07855224609375,
|
|
"logps/rejected": -213.01675415039062,
|
|
"loss": 0.2132,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.5593734979629517,
|
|
"rewards/margins": 2.7595760822296143,
|
|
"rewards/rejected": -3.3189496994018555,
|
|
"step": 796
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.895316511774164e-05,
|
|
"logits/chosen": -2.5653488636016846,
|
|
"logits/rejected": -2.6561145782470703,
|
|
"logps/chosen": -187.71011352539062,
|
|
"logps/rejected": -234.84353637695312,
|
|
"loss": 0.2078,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.685555100440979,
|
|
"rewards/margins": 2.5933103561401367,
|
|
"rewards/rejected": -3.278865337371826,
|
|
"step": 797
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 4.895018034941451e-05,
|
|
"logits/chosen": -2.6024954319000244,
|
|
"logits/rejected": -2.519937753677368,
|
|
"logps/chosen": -168.91847229003906,
|
|
"logps/rejected": -224.6294708251953,
|
|
"loss": 0.1913,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.535056471824646,
|
|
"rewards/margins": 3.0812528133392334,
|
|
"rewards/rejected": -3.61630916595459,
|
|
"step": 798
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.894719142326593e-05,
|
|
"logits/chosen": -2.21370005607605,
|
|
"logits/rejected": -2.3001461029052734,
|
|
"logps/chosen": -166.51718139648438,
|
|
"logps/rejected": -224.1083526611328,
|
|
"loss": 0.3661,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.7483687996864319,
|
|
"rewards/margins": 2.128541946411133,
|
|
"rewards/rejected": -2.8769102096557617,
|
|
"step": 799
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.894419833981478e-05,
|
|
"logits/chosen": -2.4892656803131104,
|
|
"logits/rejected": -2.568337917327881,
|
|
"logps/chosen": -247.6711883544922,
|
|
"logps/rejected": -264.4457092285156,
|
|
"loss": 0.1815,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.5068062543869019,
|
|
"rewards/margins": 3.0614163875579834,
|
|
"rewards/rejected": -3.5682225227355957,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.894120109958068e-05,
|
|
"logits/chosen": -2.5446887016296387,
|
|
"logits/rejected": -2.5595788955688477,
|
|
"logps/chosen": -148.94400024414062,
|
|
"logps/rejected": -197.58807373046875,
|
|
"loss": 0.2073,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.3782866895198822,
|
|
"rewards/margins": 2.5527689456939697,
|
|
"rewards/rejected": -2.931055784225464,
|
|
"step": 801
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.893819970308394e-05,
|
|
"logits/chosen": -2.6223721504211426,
|
|
"logits/rejected": -2.6008565425872803,
|
|
"logps/chosen": -223.73875427246094,
|
|
"logps/rejected": -270.4544982910156,
|
|
"loss": 0.131,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.582880973815918,
|
|
"rewards/margins": 2.6789181232452393,
|
|
"rewards/rejected": -3.2617993354797363,
|
|
"step": 802
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.893519415084564e-05,
|
|
"logits/chosen": -2.576484203338623,
|
|
"logits/rejected": -2.459780216217041,
|
|
"logps/chosen": -176.4852294921875,
|
|
"logps/rejected": -229.3186492919922,
|
|
"loss": 0.1817,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.5410618782043457,
|
|
"rewards/margins": 2.697970390319824,
|
|
"rewards/rejected": -3.239032030105591,
|
|
"step": 803
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.8932184443387536e-05,
|
|
"logits/chosen": -2.32121205329895,
|
|
"logits/rejected": -2.379164695739746,
|
|
"logps/chosen": -186.08949279785156,
|
|
"logps/rejected": -203.67755126953125,
|
|
"loss": 0.3161,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.43141627311706543,
|
|
"rewards/margins": 2.3219566345214844,
|
|
"rewards/rejected": -2.7533726692199707,
|
|
"step": 804
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.892917058123212e-05,
|
|
"logits/chosen": -2.4327099323272705,
|
|
"logits/rejected": -2.4539988040924072,
|
|
"logps/chosen": -223.7473602294922,
|
|
"logps/rejected": -243.0752410888672,
|
|
"loss": 0.2913,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5450294017791748,
|
|
"rewards/margins": 1.9214046001434326,
|
|
"rewards/rejected": -2.4664340019226074,
|
|
"step": 805
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 4.892615256490262e-05,
|
|
"logits/chosen": -2.549992084503174,
|
|
"logits/rejected": -2.67326283454895,
|
|
"logps/chosen": -192.02581787109375,
|
|
"logps/rejected": -212.2666778564453,
|
|
"loss": 0.2081,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.1347290426492691,
|
|
"rewards/margins": 2.5369441509246826,
|
|
"rewards/rejected": -2.671673059463501,
|
|
"step": 806
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.8923130394922947e-05,
|
|
"logits/chosen": -2.503438949584961,
|
|
"logits/rejected": -2.5232760906219482,
|
|
"logps/chosen": -202.92604064941406,
|
|
"logps/rejected": -230.6805877685547,
|
|
"loss": 0.2341,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.3751166760921478,
|
|
"rewards/margins": 2.6738672256469727,
|
|
"rewards/rejected": -3.0489840507507324,
|
|
"step": 807
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.8920104071817786e-05,
|
|
"logits/chosen": -2.3755786418914795,
|
|
"logits/rejected": -2.5557618141174316,
|
|
"logps/chosen": -137.71795654296875,
|
|
"logps/rejected": -216.0382080078125,
|
|
"loss": 0.1602,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.13777978718280792,
|
|
"rewards/margins": 3.1972692012786865,
|
|
"rewards/rejected": -3.3350489139556885,
|
|
"step": 808
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.891707359611251e-05,
|
|
"logits/chosen": -2.4458518028259277,
|
|
"logits/rejected": -2.485722303390503,
|
|
"logps/chosen": -232.0552520751953,
|
|
"logps/rejected": -290.8924255371094,
|
|
"loss": 0.112,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.5124866962432861,
|
|
"rewards/margins": 3.1787357330322266,
|
|
"rewards/rejected": -3.6912224292755127,
|
|
"step": 809
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.89140389683332e-05,
|
|
"logits/chosen": -2.6487104892730713,
|
|
"logits/rejected": -2.691654920578003,
|
|
"logps/chosen": -200.54293823242188,
|
|
"logps/rejected": -267.951416015625,
|
|
"loss": 0.0792,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.20664259791374207,
|
|
"rewards/margins": 3.448357582092285,
|
|
"rewards/rejected": -3.6550002098083496,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.8911000189006694e-05,
|
|
"logits/chosen": -2.47886323928833,
|
|
"logits/rejected": -2.5354814529418945,
|
|
"logps/chosen": -155.83856201171875,
|
|
"logps/rejected": -193.6571502685547,
|
|
"loss": 0.1333,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.1021997407078743,
|
|
"rewards/margins": 3.2138075828552246,
|
|
"rewards/rejected": -3.111607789993286,
|
|
"step": 811
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.8907957258660534e-05,
|
|
"logits/chosen": -2.288719654083252,
|
|
"logits/rejected": -2.2621853351593018,
|
|
"logps/chosen": -219.19277954101562,
|
|
"logps/rejected": -233.79931640625,
|
|
"loss": 0.291,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.2972843647003174,
|
|
"rewards/margins": 2.0650277137756348,
|
|
"rewards/rejected": -2.3623123168945312,
|
|
"step": 812
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.890491017782296e-05,
|
|
"logits/chosen": -2.5341038703918457,
|
|
"logits/rejected": -2.636235475540161,
|
|
"logps/chosen": -166.26109313964844,
|
|
"logps/rejected": -217.66732788085938,
|
|
"loss": 0.2438,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.7079929709434509,
|
|
"rewards/margins": 2.6824710369110107,
|
|
"rewards/rejected": -3.3904640674591064,
|
|
"step": 813
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.890185894702298e-05,
|
|
"logits/chosen": -2.5911717414855957,
|
|
"logits/rejected": -2.6073203086853027,
|
|
"logps/chosen": -208.44723510742188,
|
|
"logps/rejected": -287.8471374511719,
|
|
"loss": 0.0952,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.16351568698883057,
|
|
"rewards/margins": 3.8280396461486816,
|
|
"rewards/rejected": -3.6645243167877197,
|
|
"step": 814
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.8898803566790296e-05,
|
|
"logits/chosen": -2.3474974632263184,
|
|
"logits/rejected": -2.4600653648376465,
|
|
"logps/chosen": -225.6314697265625,
|
|
"logps/rejected": -247.96067810058594,
|
|
"loss": 0.1813,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.11835913360118866,
|
|
"rewards/margins": 3.3629634380340576,
|
|
"rewards/rejected": -3.481323003768921,
|
|
"step": 815
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.889574403765531e-05,
|
|
"logits/chosen": -2.314152717590332,
|
|
"logits/rejected": -2.2856571674346924,
|
|
"logps/chosen": -222.77545166015625,
|
|
"logps/rejected": -240.05615234375,
|
|
"loss": 0.2903,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5903599858283997,
|
|
"rewards/margins": 2.8621745109558105,
|
|
"rewards/rejected": -3.4525346755981445,
|
|
"step": 816
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.889268036014918e-05,
|
|
"logits/chosen": -2.2042934894561768,
|
|
"logits/rejected": -2.252229928970337,
|
|
"logps/chosen": -168.29946899414062,
|
|
"logps/rejected": -220.2432403564453,
|
|
"loss": 0.181,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.6510707139968872,
|
|
"rewards/margins": 2.5488698482513428,
|
|
"rewards/rejected": -3.1999404430389404,
|
|
"step": 817
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.888961253480376e-05,
|
|
"logits/chosen": -2.330993890762329,
|
|
"logits/rejected": -2.439598321914673,
|
|
"logps/chosen": -191.20274353027344,
|
|
"logps/rejected": -240.98287963867188,
|
|
"loss": 0.0927,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.08680248260498047,
|
|
"rewards/margins": 3.754706382751465,
|
|
"rewards/rejected": -3.8415091037750244,
|
|
"step": 818
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.8886540562151653e-05,
|
|
"logits/chosen": -2.60939884185791,
|
|
"logits/rejected": -2.64247989654541,
|
|
"logps/chosen": -186.75680541992188,
|
|
"logps/rejected": -217.1247100830078,
|
|
"loss": 0.1396,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.13618728518486023,
|
|
"rewards/margins": 2.6276023387908936,
|
|
"rewards/rejected": -2.7637898921966553,
|
|
"step": 819
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.8883464442726146e-05,
|
|
"logits/chosen": -2.451847791671753,
|
|
"logits/rejected": -2.550415277481079,
|
|
"logps/chosen": -194.68426513671875,
|
|
"logps/rejected": -221.12335205078125,
|
|
"loss": 0.2368,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.38820838928222656,
|
|
"rewards/margins": 2.107210159301758,
|
|
"rewards/rejected": -2.4954187870025635,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 4.888038417706126e-05,
|
|
"logits/chosen": -2.4475693702697754,
|
|
"logits/rejected": -2.4619081020355225,
|
|
"logps/chosen": -185.6713409423828,
|
|
"logps/rejected": -246.03526306152344,
|
|
"loss": 0.1785,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.18588271737098694,
|
|
"rewards/margins": 3.4940121173858643,
|
|
"rewards/rejected": -3.6798949241638184,
|
|
"step": 821
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.887729976569174e-05,
|
|
"logits/chosen": -2.6383659839630127,
|
|
"logits/rejected": -2.641087293624878,
|
|
"logps/chosen": -220.7265625,
|
|
"logps/rejected": -246.0904083251953,
|
|
"loss": 0.1677,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5302943587303162,
|
|
"rewards/margins": 3.091730833053589,
|
|
"rewards/rejected": -3.6220250129699707,
|
|
"step": 822
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.8874211209153066e-05,
|
|
"logits/chosen": -2.7357144355773926,
|
|
"logits/rejected": -2.756209373474121,
|
|
"logps/chosen": -228.9771270751953,
|
|
"logps/rejected": -244.65692138671875,
|
|
"loss": 0.1871,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.2874932289123535,
|
|
"rewards/margins": 3.040170669555664,
|
|
"rewards/rejected": -3.3276634216308594,
|
|
"step": 823
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.8871118507981396e-05,
|
|
"logits/chosen": -2.541903257369995,
|
|
"logits/rejected": -2.642256259918213,
|
|
"logps/chosen": -172.7350311279297,
|
|
"logps/rejected": -263.6834716796875,
|
|
"loss": 0.1574,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.31187328696250916,
|
|
"rewards/margins": 3.909829616546631,
|
|
"rewards/rejected": -4.221703052520752,
|
|
"step": 824
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.886802166271364e-05,
|
|
"logits/chosen": -2.5496389865875244,
|
|
"logits/rejected": -2.632146120071411,
|
|
"logps/chosen": -189.05809020996094,
|
|
"logps/rejected": -210.13885498046875,
|
|
"loss": 0.15,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.08079969137907028,
|
|
"rewards/margins": 3.155087947845459,
|
|
"rewards/rejected": -3.2358880043029785,
|
|
"step": 825
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.8864920673887434e-05,
|
|
"logits/chosen": -2.508222818374634,
|
|
"logits/rejected": -2.495868682861328,
|
|
"logps/chosen": -178.13548278808594,
|
|
"logps/rejected": -204.01275634765625,
|
|
"loss": 0.1071,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.022821810096502304,
|
|
"rewards/margins": 3.6802027225494385,
|
|
"rewards/rejected": -3.703024387359619,
|
|
"step": 826
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.88618155420411e-05,
|
|
"logits/chosen": -2.3088462352752686,
|
|
"logits/rejected": -2.442979574203491,
|
|
"logps/chosen": -207.34799194335938,
|
|
"logps/rejected": -256.6688232421875,
|
|
"loss": 0.1586,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.63365638256073,
|
|
"rewards/margins": 2.7970571517944336,
|
|
"rewards/rejected": -3.430713415145874,
|
|
"step": 827
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.8858706267713704e-05,
|
|
"logits/chosen": -2.532752275466919,
|
|
"logits/rejected": -2.47641658782959,
|
|
"logps/chosen": -181.96983337402344,
|
|
"logps/rejected": -179.16371154785156,
|
|
"loss": 0.2413,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7146714329719543,
|
|
"rewards/margins": 2.151493787765503,
|
|
"rewards/rejected": -2.8661651611328125,
|
|
"step": 828
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 4.885559285144503e-05,
|
|
"logits/chosen": -2.4036006927490234,
|
|
"logits/rejected": -2.451720952987671,
|
|
"logps/chosen": -179.5876922607422,
|
|
"logps/rejected": -236.59486389160156,
|
|
"loss": 0.1633,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.3670279383659363,
|
|
"rewards/margins": 3.0567290782928467,
|
|
"rewards/rejected": -3.4237565994262695,
|
|
"step": 829
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.885247529377557e-05,
|
|
"logits/chosen": -2.4562742710113525,
|
|
"logits/rejected": -2.516852855682373,
|
|
"logps/chosen": -139.80506896972656,
|
|
"logps/rejected": -169.3076171875,
|
|
"loss": 0.2494,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.586427628993988,
|
|
"rewards/margins": 1.9348812103271484,
|
|
"rewards/rejected": -2.521308660507202,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.884935359524655e-05,
|
|
"logits/chosen": -2.433465003967285,
|
|
"logits/rejected": -2.401857852935791,
|
|
"logps/chosen": -181.75381469726562,
|
|
"logps/rejected": -226.99884033203125,
|
|
"loss": 0.1503,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.24370934069156647,
|
|
"rewards/margins": 3.303581953048706,
|
|
"rewards/rejected": -3.5472910404205322,
|
|
"step": 831
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.884622775639989e-05,
|
|
"logits/chosen": -2.4640562534332275,
|
|
"logits/rejected": -2.4529128074645996,
|
|
"logps/chosen": -228.76239013671875,
|
|
"logps/rejected": -277.97503662109375,
|
|
"loss": 0.1871,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.40687793493270874,
|
|
"rewards/margins": 2.9247701168060303,
|
|
"rewards/rejected": -3.331648111343384,
|
|
"step": 832
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.884309777777826e-05,
|
|
"logits/chosen": -2.600520372390747,
|
|
"logits/rejected": -2.6682722568511963,
|
|
"logps/chosen": -239.098388671875,
|
|
"logps/rejected": -298.6829528808594,
|
|
"loss": 0.0985,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.40839752554893494,
|
|
"rewards/margins": 3.9475584030151367,
|
|
"rewards/rejected": -4.355956077575684,
|
|
"step": 833
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.883996365992502e-05,
|
|
"logits/chosen": -2.470278024673462,
|
|
"logits/rejected": -2.446775197982788,
|
|
"logps/chosen": -225.15782165527344,
|
|
"logps/rejected": -273.9099426269531,
|
|
"loss": 0.1943,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.9289353489875793,
|
|
"rewards/margins": 3.2997019290924072,
|
|
"rewards/rejected": -4.228637218475342,
|
|
"step": 834
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.883682540338428e-05,
|
|
"logits/chosen": -2.265939474105835,
|
|
"logits/rejected": -2.3391735553741455,
|
|
"logps/chosen": -203.70504760742188,
|
|
"logps/rejected": -275.56427001953125,
|
|
"loss": 0.1327,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.4642513394355774,
|
|
"rewards/margins": 4.042669296264648,
|
|
"rewards/rejected": -4.506920337677002,
|
|
"step": 835
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.883368300870083e-05,
|
|
"logits/chosen": -2.431640148162842,
|
|
"logits/rejected": -2.453253984451294,
|
|
"logps/chosen": -199.0352020263672,
|
|
"logps/rejected": -255.3938751220703,
|
|
"loss": 0.1519,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.729556679725647,
|
|
"rewards/margins": 3.61323881149292,
|
|
"rewards/rejected": -4.3427958488464355,
|
|
"step": 836
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.883053647642022e-05,
|
|
"logits/chosen": -2.531667947769165,
|
|
"logits/rejected": -2.473618507385254,
|
|
"logps/chosen": -258.14215087890625,
|
|
"logps/rejected": -253.9119110107422,
|
|
"loss": 0.05,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.14007891714572906,
|
|
"rewards/margins": 4.125094890594482,
|
|
"rewards/rejected": -4.26517391204834,
|
|
"step": 837
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.8827385807088674e-05,
|
|
"logits/chosen": -2.368967056274414,
|
|
"logits/rejected": -2.3205454349517822,
|
|
"logps/chosen": -237.3386688232422,
|
|
"logps/rejected": -306.61737060546875,
|
|
"loss": 0.1212,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8273277282714844,
|
|
"rewards/margins": 3.641183614730835,
|
|
"rewards/rejected": -4.468511581420898,
|
|
"step": 838
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.8824231001253186e-05,
|
|
"logits/chosen": -2.4554178714752197,
|
|
"logits/rejected": -2.457733631134033,
|
|
"logps/chosen": -219.16522216796875,
|
|
"logps/rejected": -298.85003662109375,
|
|
"loss": 0.2137,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.3690099716186523,
|
|
"rewards/margins": 2.852839231491089,
|
|
"rewards/rejected": -4.221848964691162,
|
|
"step": 839
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.882107205946142e-05,
|
|
"logits/chosen": -2.19034743309021,
|
|
"logits/rejected": -2.1936280727386475,
|
|
"logps/chosen": -196.35073852539062,
|
|
"logps/rejected": -266.54638671875,
|
|
"loss": 0.2458,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5001093149185181,
|
|
"rewards/margins": 3.14082670211792,
|
|
"rewards/rejected": -3.6409358978271484,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.8817908982261765e-05,
|
|
"logits/chosen": -2.3098065853118896,
|
|
"logits/rejected": -2.4067697525024414,
|
|
"logps/chosen": -197.4673614501953,
|
|
"logps/rejected": -242.01669311523438,
|
|
"loss": 0.1775,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.7165354490280151,
|
|
"rewards/margins": 3.300830841064453,
|
|
"rewards/rejected": -4.017366409301758,
|
|
"step": 841
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.881474177020336e-05,
|
|
"logits/chosen": -2.541229009628296,
|
|
"logits/rejected": -2.5617356300354004,
|
|
"logps/chosen": -219.83966064453125,
|
|
"logps/rejected": -247.85601806640625,
|
|
"loss": 0.0592,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8507525324821472,
|
|
"rewards/margins": 3.5395710468292236,
|
|
"rewards/rejected": -4.390323638916016,
|
|
"step": 842
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.881157042383605e-05,
|
|
"logits/chosen": -2.2904212474823,
|
|
"logits/rejected": -2.3347108364105225,
|
|
"logps/chosen": -231.440673828125,
|
|
"logps/rejected": -303.37939453125,
|
|
"loss": 0.2497,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.1560659408569336,
|
|
"rewards/margins": 2.958097457885742,
|
|
"rewards/rejected": -4.114163398742676,
|
|
"step": 843
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 4.8808394943710376e-05,
|
|
"logits/chosen": -2.279738187789917,
|
|
"logits/rejected": -2.3323707580566406,
|
|
"logps/chosen": -213.64846801757812,
|
|
"logps/rejected": -239.2349395751953,
|
|
"loss": 0.243,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.1719560623168945,
|
|
"rewards/margins": 2.3868768215179443,
|
|
"rewards/rejected": -3.558833122253418,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.880521533037762e-05,
|
|
"logits/chosen": -2.3850831985473633,
|
|
"logits/rejected": -2.4819254875183105,
|
|
"logps/chosen": -211.81105041503906,
|
|
"logps/rejected": -253.8992919921875,
|
|
"loss": 0.1005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6735836267471313,
|
|
"rewards/margins": 3.870424509048462,
|
|
"rewards/rejected": -4.544007778167725,
|
|
"step": 845
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.880203158438975e-05,
|
|
"logits/chosen": -2.4267005920410156,
|
|
"logits/rejected": -2.4583656787872314,
|
|
"logps/chosen": -185.1043243408203,
|
|
"logps/rejected": -223.85244750976562,
|
|
"loss": 0.1452,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.911338746547699,
|
|
"rewards/margins": 3.077333688735962,
|
|
"rewards/rejected": -3.9886722564697266,
|
|
"step": 846
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.87988437062995e-05,
|
|
"logits/chosen": -2.0992445945739746,
|
|
"logits/rejected": -2.171211004257202,
|
|
"logps/chosen": -159.3900909423828,
|
|
"logps/rejected": -242.703369140625,
|
|
"loss": 0.25,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7375649213790894,
|
|
"rewards/margins": 3.8496501445770264,
|
|
"rewards/rejected": -4.587214946746826,
|
|
"step": 847
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.879565169666028e-05,
|
|
"logits/chosen": -2.2952377796173096,
|
|
"logits/rejected": -2.3588685989379883,
|
|
"logps/chosen": -201.49908447265625,
|
|
"logps/rejected": -241.18902587890625,
|
|
"loss": 0.2027,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5940706729888916,
|
|
"rewards/margins": 3.3175528049468994,
|
|
"rewards/rejected": -3.91162371635437,
|
|
"step": 848
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.879245555602624e-05,
|
|
"logits/chosen": -1.9072455167770386,
|
|
"logits/rejected": -2.0710573196411133,
|
|
"logps/chosen": -167.665283203125,
|
|
"logps/rejected": -231.74786376953125,
|
|
"loss": 0.2365,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.0657278299331665,
|
|
"rewards/margins": 3.313453435897827,
|
|
"rewards/rejected": -4.379181385040283,
|
|
"step": 849
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.878925528495223e-05,
|
|
"logits/chosen": -2.076167106628418,
|
|
"logits/rejected": -2.0585241317749023,
|
|
"logps/chosen": -160.9062957763672,
|
|
"logps/rejected": -198.1348419189453,
|
|
"loss": 0.2318,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.9388778209686279,
|
|
"rewards/margins": 2.9717299938201904,
|
|
"rewards/rejected": -3.9106080532073975,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 4.878605088399384e-05,
|
|
"logits/chosen": -2.1791696548461914,
|
|
"logits/rejected": -2.2479426860809326,
|
|
"logps/chosen": -191.9034881591797,
|
|
"logps/rejected": -257.7882995605469,
|
|
"loss": 0.0744,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.5156456828117371,
|
|
"rewards/margins": 3.969588041305542,
|
|
"rewards/rejected": -4.485233783721924,
|
|
"step": 851
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.878284235370735e-05,
|
|
"logits/chosen": -2.1952590942382812,
|
|
"logits/rejected": -2.3678793907165527,
|
|
"logps/chosen": -195.3539276123047,
|
|
"logps/rejected": -240.06231689453125,
|
|
"loss": 0.1083,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.9977157115936279,
|
|
"rewards/margins": 3.5119776725769043,
|
|
"rewards/rejected": -4.509693622589111,
|
|
"step": 852
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.877962969464978e-05,
|
|
"logits/chosen": -2.0996241569519043,
|
|
"logits/rejected": -2.2190651893615723,
|
|
"logps/chosen": -184.02914428710938,
|
|
"logps/rejected": -213.2720489501953,
|
|
"loss": 0.2316,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.107424020767212,
|
|
"rewards/margins": 2.9243013858795166,
|
|
"rewards/rejected": -4.0317254066467285,
|
|
"step": 853
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.877641290737884e-05,
|
|
"logits/chosen": -2.398710012435913,
|
|
"logits/rejected": -2.455815315246582,
|
|
"logps/chosen": -196.89927673339844,
|
|
"logps/rejected": -228.26943969726562,
|
|
"loss": 0.2229,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.7392411828041077,
|
|
"rewards/margins": 3.144235134124756,
|
|
"rewards/rejected": -3.8834760189056396,
|
|
"step": 854
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.8773191992453e-05,
|
|
"logits/chosen": -2.585550546646118,
|
|
"logits/rejected": -2.6128385066986084,
|
|
"logps/chosen": -221.81942749023438,
|
|
"logps/rejected": -235.17910766601562,
|
|
"loss": 0.1117,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.5257879495620728,
|
|
"rewards/margins": 3.225809335708618,
|
|
"rewards/rejected": -3.7515969276428223,
|
|
"step": 855
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.87699669504314e-05,
|
|
"logits/chosen": -2.3643717765808105,
|
|
"logits/rejected": -2.397857427597046,
|
|
"logps/chosen": -220.61892700195312,
|
|
"logps/rejected": -270.6251220703125,
|
|
"loss": 0.1944,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.2166273593902588,
|
|
"rewards/margins": 3.119513511657715,
|
|
"rewards/rejected": -4.3361406326293945,
|
|
"step": 856
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.876673778187393e-05,
|
|
"logits/chosen": -2.182027578353882,
|
|
"logits/rejected": -2.232729434967041,
|
|
"logps/chosen": -172.77215576171875,
|
|
"logps/rejected": -192.00790405273438,
|
|
"loss": 0.2663,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.9920349717140198,
|
|
"rewards/margins": 3.2848072052001953,
|
|
"rewards/rejected": -4.2768425941467285,
|
|
"step": 857
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.876350448734117e-05,
|
|
"logits/chosen": -2.317965507507324,
|
|
"logits/rejected": -2.4213790893554688,
|
|
"logps/chosen": -170.614990234375,
|
|
"logps/rejected": -179.1256866455078,
|
|
"loss": 0.3067,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.353677749633789,
|
|
"rewards/margins": 1.795522928237915,
|
|
"rewards/rejected": -3.149200916290283,
|
|
"step": 858
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 4.876026706739444e-05,
|
|
"logits/chosen": -2.362269401550293,
|
|
"logits/rejected": -2.385591506958008,
|
|
"logps/chosen": -277.1317138671875,
|
|
"logps/rejected": -302.71575927734375,
|
|
"loss": 0.2337,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.7664923667907715,
|
|
"rewards/margins": 2.5697085857391357,
|
|
"rewards/rejected": -4.336201190948486,
|
|
"step": 859
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.8757025522595756e-05,
|
|
"logits/chosen": -2.5295486450195312,
|
|
"logits/rejected": -2.5323615074157715,
|
|
"logps/chosen": -205.54013061523438,
|
|
"logps/rejected": -218.6345977783203,
|
|
"loss": 0.1368,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.2794644236564636,
|
|
"rewards/margins": 2.9989418983459473,
|
|
"rewards/rejected": -3.2784063816070557,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.8753779853507874e-05,
|
|
"logits/chosen": -2.117382764816284,
|
|
"logits/rejected": -2.120049238204956,
|
|
"logps/chosen": -196.75279235839844,
|
|
"logps/rejected": -217.56846618652344,
|
|
"loss": 0.1823,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5300887227058411,
|
|
"rewards/margins": 3.763706922531128,
|
|
"rewards/rejected": -4.293795585632324,
|
|
"step": 861
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.8750530060694236e-05,
|
|
"logits/chosen": -1.7907865047454834,
|
|
"logits/rejected": -1.8680351972579956,
|
|
"logps/chosen": -134.9779510498047,
|
|
"logps/rejected": -194.37901306152344,
|
|
"loss": 0.1639,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8513069152832031,
|
|
"rewards/margins": 2.7600948810577393,
|
|
"rewards/rejected": -3.6114020347595215,
|
|
"step": 862
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.874727614471903e-05,
|
|
"logits/chosen": -2.110868453979492,
|
|
"logits/rejected": -2.1362719535827637,
|
|
"logps/chosen": -194.04483032226562,
|
|
"logps/rejected": -234.01123046875,
|
|
"loss": 0.2074,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.6307097673416138,
|
|
"rewards/margins": 3.406496047973633,
|
|
"rewards/rejected": -5.037205696105957,
|
|
"step": 863
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.8744018106147135e-05,
|
|
"logits/chosen": -2.32252836227417,
|
|
"logits/rejected": -2.344344139099121,
|
|
"logps/chosen": -224.74435424804688,
|
|
"logps/rejected": -252.48287963867188,
|
|
"loss": 0.1992,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.4851446151733398,
|
|
"rewards/margins": 3.0550522804260254,
|
|
"rewards/rejected": -4.540196418762207,
|
|
"step": 864
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.8740755945544156e-05,
|
|
"logits/chosen": -2.2035303115844727,
|
|
"logits/rejected": -2.198829174041748,
|
|
"logps/chosen": -147.6013641357422,
|
|
"logps/rejected": -178.75584411621094,
|
|
"loss": 0.218,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -1.2339822053909302,
|
|
"rewards/margins": 2.6723711490631104,
|
|
"rewards/rejected": -3.90635347366333,
|
|
"step": 865
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.873748966347642e-05,
|
|
"logits/chosen": -2.1133739948272705,
|
|
"logits/rejected": -2.088212251663208,
|
|
"logps/chosen": -183.02053833007812,
|
|
"logps/rejected": -214.74398803710938,
|
|
"loss": 0.1746,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.064955234527588,
|
|
"rewards/margins": 3.2688629627227783,
|
|
"rewards/rejected": -4.333818435668945,
|
|
"step": 866
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.873421926051095e-05,
|
|
"logits/chosen": -2.3783798217773438,
|
|
"logits/rejected": -2.4469428062438965,
|
|
"logps/chosen": -175.8081817626953,
|
|
"logps/rejected": -228.7090301513672,
|
|
"loss": 0.2164,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.9884763956069946,
|
|
"rewards/margins": 3.5400943756103516,
|
|
"rewards/rejected": -4.528570652008057,
|
|
"step": 867
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.873094473721552e-05,
|
|
"logits/chosen": -2.122225284576416,
|
|
"logits/rejected": -2.229064464569092,
|
|
"logps/chosen": -163.4876708984375,
|
|
"logps/rejected": -226.0997314453125,
|
|
"loss": 0.1045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.584133267402649,
|
|
"rewards/margins": 3.3204777240753174,
|
|
"rewards/rejected": -4.904611110687256,
|
|
"step": 868
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.872766609415857e-05,
|
|
"logits/chosen": -2.342576265335083,
|
|
"logits/rejected": -2.389909029006958,
|
|
"logps/chosen": -161.72445678710938,
|
|
"logps/rejected": -186.76344299316406,
|
|
"loss": 0.2182,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.8539198040962219,
|
|
"rewards/margins": 2.9423582553863525,
|
|
"rewards/rejected": -3.7962779998779297,
|
|
"step": 869
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.872438333190931e-05,
|
|
"logits/chosen": -2.137856960296631,
|
|
"logits/rejected": -2.179835081100464,
|
|
"logps/chosen": -191.9931182861328,
|
|
"logps/rejected": -224.08697509765625,
|
|
"loss": 0.1825,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.108015775680542,
|
|
"rewards/margins": 3.190267562866211,
|
|
"rewards/rejected": -4.298283100128174,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.872109645103762e-05,
|
|
"logits/chosen": -2.0209460258483887,
|
|
"logits/rejected": -2.05436372756958,
|
|
"logps/chosen": -242.2396697998047,
|
|
"logps/rejected": -256.2543640136719,
|
|
"loss": 0.233,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.9021006226539612,
|
|
"rewards/margins": 2.7395806312561035,
|
|
"rewards/rejected": -3.64168119430542,
|
|
"step": 871
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.871780545211411e-05,
|
|
"logits/chosen": -2.396819829940796,
|
|
"logits/rejected": -2.3696892261505127,
|
|
"logps/chosen": -262.4898376464844,
|
|
"logps/rejected": -300.5354919433594,
|
|
"loss": 0.3388,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.824455976486206,
|
|
"rewards/margins": 3.6746246814727783,
|
|
"rewards/rejected": -4.499080657958984,
|
|
"step": 872
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.8714510335710114e-05,
|
|
"logits/chosen": -2.2704551219940186,
|
|
"logits/rejected": -2.289527416229248,
|
|
"logps/chosen": -190.1211395263672,
|
|
"logps/rejected": -210.47433471679688,
|
|
"loss": 0.2625,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.8054894208908081,
|
|
"rewards/margins": 3.2580995559692383,
|
|
"rewards/rejected": -4.063589096069336,
|
|
"step": 873
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 4.871121110239768e-05,
|
|
"logits/chosen": -2.0636115074157715,
|
|
"logits/rejected": -2.149376392364502,
|
|
"logps/chosen": -178.03355407714844,
|
|
"logps/rejected": -191.28359985351562,
|
|
"loss": 0.1928,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7658480405807495,
|
|
"rewards/margins": 3.3343939781188965,
|
|
"rewards/rejected": -4.100241661071777,
|
|
"step": 874
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.870790775274955e-05,
|
|
"logits/chosen": -2.3600223064422607,
|
|
"logits/rejected": -2.313652276992798,
|
|
"logps/chosen": -202.82212829589844,
|
|
"logps/rejected": -241.97523498535156,
|
|
"loss": 0.1404,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.3490352630615234,
|
|
"rewards/margins": 3.479907512664795,
|
|
"rewards/rejected": -4.828942775726318,
|
|
"step": 875
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.8704600287339205e-05,
|
|
"logits/chosen": -2.272904396057129,
|
|
"logits/rejected": -2.3197202682495117,
|
|
"logps/chosen": -195.68267822265625,
|
|
"logps/rejected": -230.81982421875,
|
|
"loss": 0.1807,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.507330060005188,
|
|
"rewards/margins": 3.294886827468872,
|
|
"rewards/rejected": -4.802216529846191,
|
|
"step": 876
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.870128870674082e-05,
|
|
"logits/chosen": -2.062432289123535,
|
|
"logits/rejected": -2.095883846282959,
|
|
"logps/chosen": -177.28150939941406,
|
|
"logps/rejected": -238.36048889160156,
|
|
"loss": 0.0998,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.5119123458862305,
|
|
"rewards/margins": 3.5114619731903076,
|
|
"rewards/rejected": -5.023374080657959,
|
|
"step": 877
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.869797301152931e-05,
|
|
"logits/chosen": -2.322573184967041,
|
|
"logits/rejected": -2.352694511413574,
|
|
"logps/chosen": -171.1867218017578,
|
|
"logps/rejected": -209.19857788085938,
|
|
"loss": 0.1745,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.4840812683105469,
|
|
"rewards/margins": 3.0589983463287354,
|
|
"rewards/rejected": -4.543079376220703,
|
|
"step": 878
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.869465320228028e-05,
|
|
"logits/chosen": -1.5731091499328613,
|
|
"logits/rejected": -1.5635374784469604,
|
|
"logps/chosen": -162.3439483642578,
|
|
"logps/rejected": -218.9515380859375,
|
|
"loss": 0.1648,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.3453707695007324,
|
|
"rewards/margins": 3.3265371322631836,
|
|
"rewards/rejected": -4.671907901763916,
|
|
"step": 879
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.869132927957007e-05,
|
|
"logits/chosen": -2.057164192199707,
|
|
"logits/rejected": -2.169400930404663,
|
|
"logps/chosen": -146.08116149902344,
|
|
"logps/rejected": -217.79507446289062,
|
|
"loss": 0.1879,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.3128044605255127,
|
|
"rewards/margins": 4.181766033172607,
|
|
"rewards/rejected": -5.494570732116699,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.86880012439757e-05,
|
|
"logits/chosen": -2.09201979637146,
|
|
"logits/rejected": -2.123706817626953,
|
|
"logps/chosen": -205.8859100341797,
|
|
"logps/rejected": -238.71531677246094,
|
|
"loss": 0.2388,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.5409486293792725,
|
|
"rewards/margins": 2.8045778274536133,
|
|
"rewards/rejected": -4.345526695251465,
|
|
"step": 881
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 4.868466909607494e-05,
|
|
"logits/chosen": -1.9597307443618774,
|
|
"logits/rejected": -2.0664074420928955,
|
|
"logps/chosen": -208.81736755371094,
|
|
"logps/rejected": -236.41000366210938,
|
|
"loss": 0.0614,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.3827075958251953,
|
|
"rewards/margins": 3.953075647354126,
|
|
"rewards/rejected": -5.3357834815979,
|
|
"step": 882
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.868133283644627e-05,
|
|
"logits/chosen": -2.1879220008850098,
|
|
"logits/rejected": -2.2641048431396484,
|
|
"logps/chosen": -195.26947021484375,
|
|
"logps/rejected": -233.18475341796875,
|
|
"loss": 0.1049,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.485203742980957,
|
|
"rewards/margins": 3.5149152278900146,
|
|
"rewards/rejected": -5.000119209289551,
|
|
"step": 883
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.867799246566887e-05,
|
|
"logits/chosen": -1.873902440071106,
|
|
"logits/rejected": -1.9271022081375122,
|
|
"logps/chosen": -218.30856323242188,
|
|
"logps/rejected": -250.31251525878906,
|
|
"loss": 0.1605,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.570319652557373,
|
|
"rewards/margins": 4.087669372558594,
|
|
"rewards/rejected": -5.657989025115967,
|
|
"step": 884
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.867464798432262e-05,
|
|
"logits/chosen": -2.293184995651245,
|
|
"logits/rejected": -2.348799705505371,
|
|
"logps/chosen": -204.55638122558594,
|
|
"logps/rejected": -239.99803161621094,
|
|
"loss": 0.1635,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.4229328632354736,
|
|
"rewards/margins": 4.076297283172607,
|
|
"rewards/rejected": -5.499229907989502,
|
|
"step": 885
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.8671299392988146e-05,
|
|
"logits/chosen": -1.9167914390563965,
|
|
"logits/rejected": -1.835463285446167,
|
|
"logps/chosen": -228.05908203125,
|
|
"logps/rejected": -258.6400146484375,
|
|
"loss": 0.2577,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.6862237453460693,
|
|
"rewards/margins": 2.3948185443878174,
|
|
"rewards/rejected": -4.081042289733887,
|
|
"step": 886
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.866794669224678e-05,
|
|
"logits/chosen": -2.263591766357422,
|
|
"logits/rejected": -2.2648370265960693,
|
|
"logps/chosen": -323.7992248535156,
|
|
"logps/rejected": -340.5594177246094,
|
|
"loss": 0.0843,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.0585577487945557,
|
|
"rewards/margins": 3.989744186401367,
|
|
"rewards/rejected": -5.048301696777344,
|
|
"step": 887
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.8664589882680544e-05,
|
|
"logits/chosen": -2.3918349742889404,
|
|
"logits/rejected": -2.4805490970611572,
|
|
"logps/chosen": -153.11036682128906,
|
|
"logps/rejected": -231.31317138671875,
|
|
"loss": 0.2822,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.8393661975860596,
|
|
"rewards/margins": 3.1571884155273438,
|
|
"rewards/rejected": -3.9965548515319824,
|
|
"step": 888
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.866122896487221e-05,
|
|
"logits/chosen": -2.1840920448303223,
|
|
"logits/rejected": -2.304276466369629,
|
|
"logps/chosen": -167.58294677734375,
|
|
"logps/rejected": -200.28634643554688,
|
|
"loss": 0.1857,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.8503859043121338,
|
|
"rewards/margins": 3.1673898696899414,
|
|
"rewards/rejected": -4.017776012420654,
|
|
"step": 889
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.865786393940522e-05,
|
|
"logits/chosen": -2.069397211074829,
|
|
"logits/rejected": -2.1860339641571045,
|
|
"logps/chosen": -179.6785125732422,
|
|
"logps/rejected": -228.13262939453125,
|
|
"loss": 0.0932,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.1931277513504028,
|
|
"rewards/margins": 3.2066426277160645,
|
|
"rewards/rejected": -4.399770736694336,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.8654494806863775e-05,
|
|
"logits/chosen": -2.1783037185668945,
|
|
"logits/rejected": -2.2106220722198486,
|
|
"logps/chosen": -193.16891479492188,
|
|
"logps/rejected": -243.27835083007812,
|
|
"loss": 0.2,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.0907084941864014,
|
|
"rewards/margins": 3.496738910675049,
|
|
"rewards/rejected": -4.587447643280029,
|
|
"step": 891
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.865112156783275e-05,
|
|
"logits/chosen": -2.294567584991455,
|
|
"logits/rejected": -2.1514720916748047,
|
|
"logps/chosen": -214.3270263671875,
|
|
"logps/rejected": -187.45822143554688,
|
|
"loss": 0.214,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.0721594095230103,
|
|
"rewards/margins": 2.2023167610168457,
|
|
"rewards/rejected": -3.2744765281677246,
|
|
"step": 892
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.864774422289776e-05,
|
|
"logits/chosen": -2.0853824615478516,
|
|
"logits/rejected": -2.1526715755462646,
|
|
"logps/chosen": -204.90089416503906,
|
|
"logps/rejected": -246.01055908203125,
|
|
"loss": 0.112,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.11933372169733047,
|
|
"rewards/margins": 4.104992866516113,
|
|
"rewards/rejected": -4.2243266105651855,
|
|
"step": 893
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.864436277264511e-05,
|
|
"logits/chosen": -2.1137301921844482,
|
|
"logits/rejected": -2.1245357990264893,
|
|
"logps/chosen": -159.5576171875,
|
|
"logps/rejected": -167.99693298339844,
|
|
"loss": 0.3612,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.1939194202423096,
|
|
"rewards/margins": 2.4540228843688965,
|
|
"rewards/rejected": -3.647942543029785,
|
|
"step": 894
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.864097721766184e-05,
|
|
"logits/chosen": -2.0749716758728027,
|
|
"logits/rejected": -2.199066162109375,
|
|
"logps/chosen": -196.0050048828125,
|
|
"logps/rejected": -234.677001953125,
|
|
"loss": 0.1027,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.997223973274231,
|
|
"rewards/margins": 3.582221269607544,
|
|
"rewards/rejected": -4.5794453620910645,
|
|
"step": 895
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.8637587558535695e-05,
|
|
"logits/chosen": -2.2920432090759277,
|
|
"logits/rejected": -2.3417181968688965,
|
|
"logps/chosen": -236.25619506835938,
|
|
"logps/rejected": -282.4396057128906,
|
|
"loss": 0.2044,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.38354048132896423,
|
|
"rewards/margins": 2.915595531463623,
|
|
"rewards/rejected": -3.29913592338562,
|
|
"step": 896
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 4.863419379585512e-05,
|
|
"logits/chosen": -2.2515666484832764,
|
|
"logits/rejected": -2.2877609729766846,
|
|
"logps/chosen": -189.43197631835938,
|
|
"logps/rejected": -230.85414123535156,
|
|
"loss": 0.1691,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.3132818937301636,
|
|
"rewards/margins": 3.6810238361358643,
|
|
"rewards/rejected": -3.9943056106567383,
|
|
"step": 897
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.863079593020928e-05,
|
|
"logits/chosen": -2.3358492851257324,
|
|
"logits/rejected": -2.2418711185455322,
|
|
"logps/chosen": -200.99757385253906,
|
|
"logps/rejected": -208.28480529785156,
|
|
"loss": 0.1435,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.17593544721603394,
|
|
"rewards/margins": 3.3050811290740967,
|
|
"rewards/rejected": -3.4810166358947754,
|
|
"step": 898
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.8627393962188075e-05,
|
|
"logits/chosen": -2.2615466117858887,
|
|
"logits/rejected": -2.26615047454834,
|
|
"logps/chosen": -265.6878662109375,
|
|
"logps/rejected": -265.2508239746094,
|
|
"loss": 0.2,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.6857917904853821,
|
|
"rewards/margins": 3.3371822834014893,
|
|
"rewards/rejected": -4.022973537445068,
|
|
"step": 899
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.8623987892382066e-05,
|
|
"logits/chosen": -2.3570547103881836,
|
|
"logits/rejected": -2.3362598419189453,
|
|
"logps/chosen": -199.10662841796875,
|
|
"logps/rejected": -217.95907592773438,
|
|
"loss": 0.111,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.27905502915382385,
|
|
"rewards/margins": 3.4710071086883545,
|
|
"rewards/rejected": -3.7500622272491455,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.862057772138258e-05,
|
|
"logits/chosen": -2.2032244205474854,
|
|
"logits/rejected": -2.2657198905944824,
|
|
"logps/chosen": -213.52206420898438,
|
|
"logps/rejected": -225.04454040527344,
|
|
"loss": 0.1846,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8243767023086548,
|
|
"rewards/margins": 3.046542167663574,
|
|
"rewards/rejected": -3.8709182739257812,
|
|
"step": 901
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.861716344978162e-05,
|
|
"logits/chosen": -2.394089698791504,
|
|
"logits/rejected": -2.292032241821289,
|
|
"logps/chosen": -234.7174835205078,
|
|
"logps/rejected": -255.44989013671875,
|
|
"loss": 0.1544,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5697602033615112,
|
|
"rewards/margins": 3.2541327476501465,
|
|
"rewards/rejected": -3.8238930702209473,
|
|
"step": 902
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.861374507817193e-05,
|
|
"logits/chosen": -2.3428890705108643,
|
|
"logits/rejected": -2.3085713386535645,
|
|
"logps/chosen": -221.3970489501953,
|
|
"logps/rejected": -264.0530090332031,
|
|
"loss": 0.0761,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.48718422651290894,
|
|
"rewards/margins": 3.7521519660949707,
|
|
"rewards/rejected": -4.239336013793945,
|
|
"step": 903
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.8610322607146926e-05,
|
|
"logits/chosen": -2.417067766189575,
|
|
"logits/rejected": -2.393913745880127,
|
|
"logps/chosen": -178.41981506347656,
|
|
"logps/rejected": -186.7463836669922,
|
|
"loss": 0.277,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.6322096586227417,
|
|
"rewards/margins": 1.9191558361053467,
|
|
"rewards/rejected": -2.551365375518799,
|
|
"step": 904
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 4.8606896037300776e-05,
|
|
"logits/chosen": -2.2859747409820557,
|
|
"logits/rejected": -2.3659892082214355,
|
|
"logps/chosen": -210.5970458984375,
|
|
"logps/rejected": -241.645263671875,
|
|
"loss": 0.0721,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.2545009255409241,
|
|
"rewards/margins": 3.632350444793701,
|
|
"rewards/rejected": -3.8868515491485596,
|
|
"step": 905
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.860346536922834e-05,
|
|
"logits/chosen": -2.369377374649048,
|
|
"logits/rejected": -2.4897830486297607,
|
|
"logps/chosen": -254.58274841308594,
|
|
"logps/rejected": -319.5621643066406,
|
|
"loss": 0.1598,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.3183284401893616,
|
|
"rewards/margins": 3.658812999725342,
|
|
"rewards/rejected": -3.9771416187286377,
|
|
"step": 906
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.860003060352518e-05,
|
|
"logits/chosen": -2.3488306999206543,
|
|
"logits/rejected": -2.3815548419952393,
|
|
"logps/chosen": -171.2317352294922,
|
|
"logps/rejected": -204.7757568359375,
|
|
"loss": 0.1461,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7481203675270081,
|
|
"rewards/margins": 2.8177552223205566,
|
|
"rewards/rejected": -3.56587553024292,
|
|
"step": 907
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.85965917407876e-05,
|
|
"logits/chosen": -2.2379674911499023,
|
|
"logits/rejected": -2.3881618976593018,
|
|
"logps/chosen": -165.37860107421875,
|
|
"logps/rejected": -203.4194793701172,
|
|
"loss": 0.2971,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.6805337071418762,
|
|
"rewards/margins": 2.6565561294555664,
|
|
"rewards/rejected": -3.337089776992798,
|
|
"step": 908
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.8593148781612587e-05,
|
|
"logits/chosen": -2.630882740020752,
|
|
"logits/rejected": -2.544029951095581,
|
|
"logps/chosen": -201.74972534179688,
|
|
"logps/rejected": -221.31356811523438,
|
|
"loss": 0.1897,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.4801185429096222,
|
|
"rewards/margins": 3.1227259635925293,
|
|
"rewards/rejected": -3.602844715118408,
|
|
"step": 909
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.858970172659785e-05,
|
|
"logits/chosen": -2.405344247817993,
|
|
"logits/rejected": -2.471182107925415,
|
|
"logps/chosen": -232.18885803222656,
|
|
"logps/rejected": -307.023681640625,
|
|
"loss": 0.0831,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.4412127435207367,
|
|
"rewards/margins": 3.6795432567596436,
|
|
"rewards/rejected": -4.12075662612915,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.85862505763418e-05,
|
|
"logits/chosen": -1.9770020246505737,
|
|
"logits/rejected": -1.8537812232971191,
|
|
"logps/chosen": -236.51913452148438,
|
|
"logps/rejected": -222.99220275878906,
|
|
"loss": 0.2417,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.4903448224067688,
|
|
"rewards/margins": 2.971850872039795,
|
|
"rewards/rejected": -3.46219539642334,
|
|
"step": 911
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.858279533144358e-05,
|
|
"logits/chosen": -2.5417568683624268,
|
|
"logits/rejected": -2.6035280227661133,
|
|
"logps/chosen": -172.0741729736328,
|
|
"logps/rejected": -214.9192657470703,
|
|
"loss": 0.1949,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.9094291925430298,
|
|
"rewards/margins": 2.6431045532226562,
|
|
"rewards/rejected": -3.5525336265563965,
|
|
"step": 912
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.857933599250303e-05,
|
|
"logits/chosen": -2.342681407928467,
|
|
"logits/rejected": -2.256180763244629,
|
|
"logps/chosen": -196.65240478515625,
|
|
"logps/rejected": -195.11648559570312,
|
|
"loss": 0.1316,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.5412741899490356,
|
|
"rewards/margins": 3.1819822788238525,
|
|
"rewards/rejected": -3.7232565879821777,
|
|
"step": 913
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.857587256012068e-05,
|
|
"logits/chosen": -2.2213358879089355,
|
|
"logits/rejected": -2.223132371902466,
|
|
"logps/chosen": -217.80227661132812,
|
|
"logps/rejected": -258.8191833496094,
|
|
"loss": 0.2183,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.33522820472717285,
|
|
"rewards/margins": 3.4243385791778564,
|
|
"rewards/rejected": -3.759566307067871,
|
|
"step": 914
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.857240503489782e-05,
|
|
"logits/chosen": -2.633331060409546,
|
|
"logits/rejected": -2.5645124912261963,
|
|
"logps/chosen": -200.99314880371094,
|
|
"logps/rejected": -216.61944580078125,
|
|
"loss": 0.0838,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.03604121133685112,
|
|
"rewards/margins": 3.366603136062622,
|
|
"rewards/rejected": -3.402644157409668,
|
|
"step": 915
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.8568933417436416e-05,
|
|
"logits/chosen": -2.60661244392395,
|
|
"logits/rejected": -2.568995475769043,
|
|
"logps/chosen": -229.74319458007812,
|
|
"logps/rejected": -272.7928161621094,
|
|
"loss": 0.0913,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.10647392272949219,
|
|
"rewards/margins": 4.324003219604492,
|
|
"rewards/rejected": -4.430477142333984,
|
|
"step": 916
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.856545770833913e-05,
|
|
"logits/chosen": -2.1530046463012695,
|
|
"logits/rejected": -2.075482130050659,
|
|
"logps/chosen": -218.10240173339844,
|
|
"logps/rejected": -234.6465301513672,
|
|
"loss": 0.166,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": 0.3659510612487793,
|
|
"rewards/margins": 4.201175212860107,
|
|
"rewards/rejected": -3.8352243900299072,
|
|
"step": 917
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.8561977908209384e-05,
|
|
"logits/chosen": -2.322742462158203,
|
|
"logits/rejected": -2.2862117290496826,
|
|
"logps/chosen": -188.68533325195312,
|
|
"logps/rejected": -230.53916931152344,
|
|
"loss": 0.1539,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.2586498260498047,
|
|
"rewards/margins": 3.89760684967041,
|
|
"rewards/rejected": -4.156256198883057,
|
|
"step": 918
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.855849401765127e-05,
|
|
"logits/chosen": -2.245067834854126,
|
|
"logits/rejected": -2.264173984527588,
|
|
"logps/chosen": -163.19442749023438,
|
|
"logps/rejected": -244.39859008789062,
|
|
"loss": 0.2118,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5528127551078796,
|
|
"rewards/margins": 2.983473539352417,
|
|
"rewards/rejected": -3.5362865924835205,
|
|
"step": 919
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 4.8555006037269594e-05,
|
|
"logits/chosen": -2.4369189739227295,
|
|
"logits/rejected": -2.4961891174316406,
|
|
"logps/chosen": -199.59515380859375,
|
|
"logps/rejected": -218.68148803710938,
|
|
"loss": 0.2001,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.507201611995697,
|
|
"rewards/margins": 2.782557487487793,
|
|
"rewards/rejected": -3.2897589206695557,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.855151396766988e-05,
|
|
"logits/chosen": -2.557804584503174,
|
|
"logits/rejected": -2.554647922515869,
|
|
"logps/chosen": -195.2904815673828,
|
|
"logps/rejected": -226.03387451171875,
|
|
"loss": 0.1858,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.013599276542663574,
|
|
"rewards/margins": 4.105103492736816,
|
|
"rewards/rejected": -4.091504096984863,
|
|
"step": 921
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.854801780945837e-05,
|
|
"logits/chosen": -1.9899704456329346,
|
|
"logits/rejected": -1.9653894901275635,
|
|
"logps/chosen": -201.0773468017578,
|
|
"logps/rejected": -210.57366943359375,
|
|
"loss": 0.247,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.5728083252906799,
|
|
"rewards/margins": 2.8485476970672607,
|
|
"rewards/rejected": -3.421356201171875,
|
|
"step": 922
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.854451756324201e-05,
|
|
"logits/chosen": -2.334390878677368,
|
|
"logits/rejected": -2.2994754314422607,
|
|
"logps/chosen": -231.74093627929688,
|
|
"logps/rejected": -252.7148895263672,
|
|
"loss": 0.2625,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.6329984664916992,
|
|
"rewards/margins": 2.753272533416748,
|
|
"rewards/rejected": -3.386270523071289,
|
|
"step": 923
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.854101322962845e-05,
|
|
"logits/chosen": -2.4350640773773193,
|
|
"logits/rejected": -2.4160242080688477,
|
|
"logps/chosen": -202.79510498046875,
|
|
"logps/rejected": -196.56761169433594,
|
|
"loss": 0.1783,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7683102488517761,
|
|
"rewards/margins": 2.6973066329956055,
|
|
"rewards/rejected": -3.4656169414520264,
|
|
"step": 924
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.853750480922604e-05,
|
|
"logits/chosen": -2.4578471183776855,
|
|
"logits/rejected": -2.4201016426086426,
|
|
"logps/chosen": -236.7185516357422,
|
|
"logps/rejected": -225.40760803222656,
|
|
"loss": 0.1667,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.3481481373310089,
|
|
"rewards/margins": 3.050295352935791,
|
|
"rewards/rejected": -3.3984434604644775,
|
|
"step": 925
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.853399230264387e-05,
|
|
"logits/chosen": -2.5239510536193848,
|
|
"logits/rejected": -2.5775818824768066,
|
|
"logps/chosen": -139.86842346191406,
|
|
"logps/rejected": -190.04673767089844,
|
|
"loss": 0.3665,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.7791432738304138,
|
|
"rewards/margins": 1.9640579223632812,
|
|
"rewards/rejected": -2.74320125579834,
|
|
"step": 926
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.853047571049171e-05,
|
|
"logits/chosen": -2.3366544246673584,
|
|
"logits/rejected": -2.2847678661346436,
|
|
"logps/chosen": -184.1388702392578,
|
|
"logps/rejected": -197.55262756347656,
|
|
"loss": 0.1811,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.3248026967048645,
|
|
"rewards/margins": 2.802187919616699,
|
|
"rewards/rejected": -3.126990795135498,
|
|
"step": 927
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 4.852695503338005e-05,
|
|
"logits/chosen": -2.2322957515716553,
|
|
"logits/rejected": -2.232264518737793,
|
|
"logps/chosen": -218.7916259765625,
|
|
"logps/rejected": -239.33642578125,
|
|
"loss": 0.2817,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.7828483581542969,
|
|
"rewards/margins": 2.7246387004852295,
|
|
"rewards/rejected": -3.5074868202209473,
|
|
"step": 928
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.85234302719201e-05,
|
|
"logits/chosen": -2.3104605674743652,
|
|
"logits/rejected": -2.292299509048462,
|
|
"logps/chosen": -223.07472229003906,
|
|
"logps/rejected": -220.38296508789062,
|
|
"loss": 0.1399,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.13662558794021606,
|
|
"rewards/margins": 3.366118907928467,
|
|
"rewards/rejected": -3.502744436264038,
|
|
"step": 929
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.851990142672376e-05,
|
|
"logits/chosen": -2.27694034576416,
|
|
"logits/rejected": -2.416929244995117,
|
|
"logps/chosen": -163.55599975585938,
|
|
"logps/rejected": -187.94540405273438,
|
|
"loss": 0.1647,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.6064982414245605,
|
|
"rewards/margins": 2.823979616165161,
|
|
"rewards/rejected": -3.4304776191711426,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.8516368498403654e-05,
|
|
"logits/chosen": -2.1894800662994385,
|
|
"logits/rejected": -2.198641777038574,
|
|
"logps/chosen": -192.32850646972656,
|
|
"logps/rejected": -232.83926391601562,
|
|
"loss": 0.2586,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.7106565237045288,
|
|
"rewards/margins": 2.401294231414795,
|
|
"rewards/rejected": -3.1119508743286133,
|
|
"step": 931
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.85128314875731e-05,
|
|
"logits/chosen": -2.2995123863220215,
|
|
"logits/rejected": -2.458733320236206,
|
|
"logps/chosen": -184.31101989746094,
|
|
"logps/rejected": -263.00048828125,
|
|
"loss": 0.0955,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.1651136875152588,
|
|
"rewards/margins": 3.4938466548919678,
|
|
"rewards/rejected": -3.6589601039886475,
|
|
"step": 932
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.850929039484614e-05,
|
|
"logits/chosen": -2.3589913845062256,
|
|
"logits/rejected": -2.4086785316467285,
|
|
"logps/chosen": -147.282470703125,
|
|
"logps/rejected": -214.50318908691406,
|
|
"loss": 0.1216,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.45500433444976807,
|
|
"rewards/margins": 3.236088275909424,
|
|
"rewards/rejected": -3.6910927295684814,
|
|
"step": 933
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.850574522083753e-05,
|
|
"logits/chosen": -2.222720146179199,
|
|
"logits/rejected": -2.3172354698181152,
|
|
"logps/chosen": -180.91482543945312,
|
|
"logps/rejected": -231.47238159179688,
|
|
"loss": 0.1583,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7558724284172058,
|
|
"rewards/margins": 2.776618480682373,
|
|
"rewards/rejected": -3.5324912071228027,
|
|
"step": 934
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.8502195966162694e-05,
|
|
"logits/chosen": -2.328801393508911,
|
|
"logits/rejected": -2.4272334575653076,
|
|
"logps/chosen": -154.9918212890625,
|
|
"logps/rejected": -246.27830505371094,
|
|
"loss": 0.0796,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6289770007133484,
|
|
"rewards/margins": 3.6553382873535156,
|
|
"rewards/rejected": -4.28431510925293,
|
|
"step": 935
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.849864263143781e-05,
|
|
"logits/chosen": -2.335493564605713,
|
|
"logits/rejected": -2.3662097454071045,
|
|
"logps/chosen": -163.90878295898438,
|
|
"logps/rejected": -192.3372802734375,
|
|
"loss": 0.2689,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.5199804306030273,
|
|
"rewards/margins": 2.6915531158447266,
|
|
"rewards/rejected": -3.211533546447754,
|
|
"step": 936
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.8495085217279755e-05,
|
|
"logits/chosen": -2.3108789920806885,
|
|
"logits/rejected": -2.411635637283325,
|
|
"logps/chosen": -178.46237182617188,
|
|
"logps/rejected": -252.6321563720703,
|
|
"loss": 0.0912,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.3370632529258728,
|
|
"rewards/margins": 3.8149256706237793,
|
|
"rewards/rejected": -4.151988506317139,
|
|
"step": 937
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.849152372430609e-05,
|
|
"logits/chosen": -2.3515238761901855,
|
|
"logits/rejected": -2.4710798263549805,
|
|
"logps/chosen": -186.52601623535156,
|
|
"logps/rejected": -236.93490600585938,
|
|
"loss": 0.1544,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.4036705791950226,
|
|
"rewards/margins": 3.555159091949463,
|
|
"rewards/rejected": -3.958829402923584,
|
|
"step": 938
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.848795815313511e-05,
|
|
"logits/chosen": -2.0485763549804688,
|
|
"logits/rejected": -2.0782063007354736,
|
|
"logps/chosen": -190.405517578125,
|
|
"logps/rejected": -239.02328491210938,
|
|
"loss": 0.1048,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.12277688086032867,
|
|
"rewards/margins": 3.6550817489624023,
|
|
"rewards/rejected": -3.7778589725494385,
|
|
"step": 939
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.8484388504385806e-05,
|
|
"logits/chosen": -2.2379536628723145,
|
|
"logits/rejected": -2.215625762939453,
|
|
"logps/chosen": -169.50054931640625,
|
|
"logps/rejected": -222.1637420654297,
|
|
"loss": 0.2537,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.6230670809745789,
|
|
"rewards/margins": 2.660698413848877,
|
|
"rewards/rejected": -3.2837653160095215,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.8480814778677886e-05,
|
|
"logits/chosen": -2.2735724449157715,
|
|
"logits/rejected": -2.31823468208313,
|
|
"logps/chosen": -181.47183227539062,
|
|
"logps/rejected": -204.40936279296875,
|
|
"loss": 0.2299,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.45316997170448303,
|
|
"rewards/margins": 3.5822412967681885,
|
|
"rewards/rejected": -4.035411357879639,
|
|
"step": 941
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.847723697663175e-05,
|
|
"logits/chosen": -2.504059076309204,
|
|
"logits/rejected": -2.55910325050354,
|
|
"logps/chosen": -228.25503540039062,
|
|
"logps/rejected": -302.8750305175781,
|
|
"loss": 0.0823,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8190544843673706,
|
|
"rewards/margins": 3.961775541305542,
|
|
"rewards/rejected": -4.780829906463623,
|
|
"step": 942
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 4.847365509886851e-05,
|
|
"logits/chosen": -2.3396451473236084,
|
|
"logits/rejected": -2.4200096130371094,
|
|
"logps/chosen": -202.72042846679688,
|
|
"logps/rejected": -239.9745635986328,
|
|
"loss": 0.0795,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.4111427068710327,
|
|
"rewards/margins": 3.3438093662261963,
|
|
"rewards/rejected": -3.7549519538879395,
|
|
"step": 943
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.847006914601001e-05,
|
|
"logits/chosen": -2.4212615489959717,
|
|
"logits/rejected": -2.48210072517395,
|
|
"logps/chosen": -213.84280395507812,
|
|
"logps/rejected": -260.6656188964844,
|
|
"loss": 0.1249,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8076397776603699,
|
|
"rewards/margins": 3.792834997177124,
|
|
"rewards/rejected": -4.600475311279297,
|
|
"step": 944
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.8466479118678766e-05,
|
|
"logits/chosen": -2.2285969257354736,
|
|
"logits/rejected": -2.315225124359131,
|
|
"logps/chosen": -187.538330078125,
|
|
"logps/rejected": -268.7898254394531,
|
|
"loss": 0.2034,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.2221934795379639,
|
|
"rewards/margins": 3.516084671020508,
|
|
"rewards/rejected": -4.738278388977051,
|
|
"step": 945
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.846288501749802e-05,
|
|
"logits/chosen": -2.5156943798065186,
|
|
"logits/rejected": -2.5410516262054443,
|
|
"logps/chosen": -222.49359130859375,
|
|
"logps/rejected": -282.3955993652344,
|
|
"loss": 0.1176,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.7577163577079773,
|
|
"rewards/margins": 4.017005920410156,
|
|
"rewards/rejected": -4.774722576141357,
|
|
"step": 946
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.8459286843091724e-05,
|
|
"logits/chosen": -2.2465834617614746,
|
|
"logits/rejected": -2.3958494663238525,
|
|
"logps/chosen": -205.92015075683594,
|
|
"logps/rejected": -250.27149963378906,
|
|
"loss": 0.1785,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.3019455671310425,
|
|
"rewards/margins": 3.5620269775390625,
|
|
"rewards/rejected": -4.8639726638793945,
|
|
"step": 947
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.845568459608453e-05,
|
|
"logits/chosen": -2.2198760509490967,
|
|
"logits/rejected": -2.3069236278533936,
|
|
"logps/chosen": -164.4407196044922,
|
|
"logps/rejected": -227.1414794921875,
|
|
"loss": 0.19,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.6457600593566895,
|
|
"rewards/margins": 3.604132652282715,
|
|
"rewards/rejected": -4.249892711639404,
|
|
"step": 948
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.8452078277101783e-05,
|
|
"logits/chosen": -2.1330924034118652,
|
|
"logits/rejected": -2.2624382972717285,
|
|
"logps/chosen": -141.55337524414062,
|
|
"logps/rejected": -193.1168670654297,
|
|
"loss": 0.18,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.4278246760368347,
|
|
"rewards/margins": 3.105529308319092,
|
|
"rewards/rejected": -3.5333540439605713,
|
|
"step": 949
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.8448467886769585e-05,
|
|
"logits/chosen": -1.9669086933135986,
|
|
"logits/rejected": -2.0849437713623047,
|
|
"logps/chosen": -181.0906982421875,
|
|
"logps/rejected": -237.8538055419922,
|
|
"loss": 0.1504,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.0563127994537354,
|
|
"rewards/margins": 3.4143903255462646,
|
|
"rewards/rejected": -4.470703125,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 4.844485342571468e-05,
|
|
"logits/chosen": -1.7133499383926392,
|
|
"logits/rejected": -1.701385259628296,
|
|
"logps/chosen": -169.70852661132812,
|
|
"logps/rejected": -198.37237548828125,
|
|
"loss": 0.4077,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.4771201610565186,
|
|
"rewards/margins": 2.8859434127807617,
|
|
"rewards/rejected": -4.363063335418701,
|
|
"step": 951
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.844123489456456e-05,
|
|
"logits/chosen": -2.262361526489258,
|
|
"logits/rejected": -2.4248523712158203,
|
|
"logps/chosen": -198.08021545410156,
|
|
"logps/rejected": -303.7059631347656,
|
|
"loss": 0.1379,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.432203769683838,
|
|
"rewards/margins": 3.657777786254883,
|
|
"rewards/rejected": -5.089981555938721,
|
|
"step": 952
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.843761229394741e-05,
|
|
"logits/chosen": -2.3017332553863525,
|
|
"logits/rejected": -2.291661500930786,
|
|
"logps/chosen": -189.6435089111328,
|
|
"logps/rejected": -246.8826141357422,
|
|
"loss": 0.0758,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8683138489723206,
|
|
"rewards/margins": 4.264334678649902,
|
|
"rewards/rejected": -5.132648468017578,
|
|
"step": 953
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.843398562449212e-05,
|
|
"logits/chosen": -1.9980626106262207,
|
|
"logits/rejected": -1.996732473373413,
|
|
"logps/chosen": -190.13116455078125,
|
|
"logps/rejected": -222.26028442382812,
|
|
"loss": 0.2367,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.423780918121338,
|
|
"rewards/margins": 2.6461732387542725,
|
|
"rewards/rejected": -4.069953918457031,
|
|
"step": 954
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.84303548868283e-05,
|
|
"logits/chosen": -2.3841123580932617,
|
|
"logits/rejected": -2.439812660217285,
|
|
"logps/chosen": -231.7444610595703,
|
|
"logps/rejected": -276.234619140625,
|
|
"loss": 0.1215,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.5399219989776611,
|
|
"rewards/margins": 3.503429651260376,
|
|
"rewards/rejected": -5.043351650238037,
|
|
"step": 955
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.842672008158625e-05,
|
|
"logits/chosen": -2.028545379638672,
|
|
"logits/rejected": -2.0214412212371826,
|
|
"logps/chosen": -144.1702880859375,
|
|
"logps/rejected": -193.11929321289062,
|
|
"loss": 0.3515,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -1.5417052507400513,
|
|
"rewards/margins": 2.0481746196746826,
|
|
"rewards/rejected": -3.5898799896240234,
|
|
"step": 956
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.842308120939699e-05,
|
|
"logits/chosen": -2.440342426300049,
|
|
"logits/rejected": -2.4507193565368652,
|
|
"logps/chosen": -215.12884521484375,
|
|
"logps/rejected": -226.1699981689453,
|
|
"loss": 0.1973,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.6147558689117432,
|
|
"rewards/margins": 3.0818982124328613,
|
|
"rewards/rejected": -4.696654319763184,
|
|
"step": 957
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.8419438270892226e-05,
|
|
"logits/chosen": -2.179013252258301,
|
|
"logits/rejected": -2.0757102966308594,
|
|
"logps/chosen": -157.2231903076172,
|
|
"logps/rejected": -176.61666870117188,
|
|
"loss": 0.2506,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.5769833326339722,
|
|
"rewards/margins": 2.5216851234436035,
|
|
"rewards/rejected": -4.098668098449707,
|
|
"step": 958
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.8415791266704404e-05,
|
|
"logits/chosen": -2.1697206497192383,
|
|
"logits/rejected": -2.1985995769500732,
|
|
"logps/chosen": -197.36570739746094,
|
|
"logps/rejected": -242.4941864013672,
|
|
"loss": 0.217,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.0622727870941162,
|
|
"rewards/margins": 3.3413608074188232,
|
|
"rewards/rejected": -4.4036335945129395,
|
|
"step": 959
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.841214019746663e-05,
|
|
"logits/chosen": -2.132883310317993,
|
|
"logits/rejected": -2.2006301879882812,
|
|
"logps/chosen": -195.13247680664062,
|
|
"logps/rejected": -226.83316040039062,
|
|
"loss": 0.2752,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.512275218963623,
|
|
"rewards/margins": 2.76702880859375,
|
|
"rewards/rejected": -4.279304027557373,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.840848506381276e-05,
|
|
"logits/chosen": -2.286079168319702,
|
|
"logits/rejected": -2.3662781715393066,
|
|
"logps/chosen": -201.76133728027344,
|
|
"logps/rejected": -226.46603393554688,
|
|
"loss": 0.1871,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.301385521888733,
|
|
"rewards/margins": 3.0335183143615723,
|
|
"rewards/rejected": -4.334904193878174,
|
|
"step": 961
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.840482586637732e-05,
|
|
"logits/chosen": -2.354762554168701,
|
|
"logits/rejected": -2.3351876735687256,
|
|
"logps/chosen": -213.7999267578125,
|
|
"logps/rejected": -268.6205139160156,
|
|
"loss": 0.3769,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.8948153257369995,
|
|
"rewards/margins": 2.8223719596862793,
|
|
"rewards/rejected": -4.71718692779541,
|
|
"step": 962
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.840116260579557e-05,
|
|
"logits/chosen": -2.3253631591796875,
|
|
"logits/rejected": -2.4139952659606934,
|
|
"logps/chosen": -196.48281860351562,
|
|
"logps/rejected": -258.4675598144531,
|
|
"loss": 0.0903,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.2855195999145508,
|
|
"rewards/margins": 4.241164684295654,
|
|
"rewards/rejected": -5.526684284210205,
|
|
"step": 963
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.839749528270345e-05,
|
|
"logits/chosen": -2.1117494106292725,
|
|
"logits/rejected": -2.233508586883545,
|
|
"logps/chosen": -162.23635864257812,
|
|
"logps/rejected": -216.20143127441406,
|
|
"loss": 0.1694,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.412933349609375,
|
|
"rewards/margins": 3.0781798362731934,
|
|
"rewards/rejected": -4.49111270904541,
|
|
"step": 964
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.8393823897737634e-05,
|
|
"logits/chosen": -1.8761091232299805,
|
|
"logits/rejected": -1.8636384010314941,
|
|
"logps/chosen": -202.1988525390625,
|
|
"logps/rejected": -255.47442626953125,
|
|
"loss": 0.1544,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.0524611473083496,
|
|
"rewards/margins": 3.7740285396575928,
|
|
"rewards/rejected": -4.8264899253845215,
|
|
"step": 965
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 4.839014845153548e-05,
|
|
"logits/chosen": -2.455805778503418,
|
|
"logits/rejected": -2.4408648014068604,
|
|
"logps/chosen": -265.5133056640625,
|
|
"logps/rejected": -322.55694580078125,
|
|
"loss": 0.1826,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.1125422716140747,
|
|
"rewards/margins": 3.8850979804992676,
|
|
"rewards/rejected": -4.997640132904053,
|
|
"step": 966
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.838646894473505e-05,
|
|
"logits/chosen": -1.7058061361312866,
|
|
"logits/rejected": -1.7635576725006104,
|
|
"logps/chosen": -178.57357788085938,
|
|
"logps/rejected": -224.11141967773438,
|
|
"loss": 0.2088,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.7308789491653442,
|
|
"rewards/margins": 2.972151279449463,
|
|
"rewards/rejected": -4.703030586242676,
|
|
"step": 967
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.8382785377975116e-05,
|
|
"logits/chosen": -2.187974452972412,
|
|
"logits/rejected": -2.2810630798339844,
|
|
"logps/chosen": -189.5592803955078,
|
|
"logps/rejected": -282.3250732421875,
|
|
"loss": 0.2646,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -1.5101513862609863,
|
|
"rewards/margins": 3.0083746910095215,
|
|
"rewards/rejected": -4.518526077270508,
|
|
"step": 968
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.8379097751895166e-05,
|
|
"logits/chosen": -2.3393805027008057,
|
|
"logits/rejected": -2.3446669578552246,
|
|
"logps/chosen": -280.5021667480469,
|
|
"logps/rejected": -287.1352233886719,
|
|
"loss": 0.1423,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8774653077125549,
|
|
"rewards/margins": 3.9873814582824707,
|
|
"rewards/rejected": -4.864847183227539,
|
|
"step": 969
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.837540606713538e-05,
|
|
"logits/chosen": -2.3770768642425537,
|
|
"logits/rejected": -2.3758389949798584,
|
|
"logps/chosen": -198.17864990234375,
|
|
"logps/rejected": -225.19105529785156,
|
|
"loss": 0.1787,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.3526650667190552,
|
|
"rewards/margins": 2.9090161323547363,
|
|
"rewards/rejected": -4.26168155670166,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.837171032433663e-05,
|
|
"logits/chosen": -2.226538896560669,
|
|
"logits/rejected": -2.3716554641723633,
|
|
"logps/chosen": -181.35006713867188,
|
|
"logps/rejected": -279.17645263671875,
|
|
"loss": 0.0983,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6791074275970459,
|
|
"rewards/margins": 4.5800275802612305,
|
|
"rewards/rejected": -5.259134292602539,
|
|
"step": 971
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.836801052414053e-05,
|
|
"logits/chosen": -1.950883388519287,
|
|
"logits/rejected": -2.080726146697998,
|
|
"logps/chosen": -170.36874389648438,
|
|
"logps/rejected": -215.67941284179688,
|
|
"loss": 0.2373,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.109346866607666,
|
|
"rewards/margins": 2.8225221633911133,
|
|
"rewards/rejected": -3.9318690299987793,
|
|
"step": 972
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.8364306667189364e-05,
|
|
"logits/chosen": -2.23186993598938,
|
|
"logits/rejected": -2.1386542320251465,
|
|
"logps/chosen": -180.2755126953125,
|
|
"logps/rejected": -230.13412475585938,
|
|
"loss": 0.112,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -0.8742038011550903,
|
|
"rewards/margins": 3.692150115966797,
|
|
"rewards/rejected": -4.566353797912598,
|
|
"step": 973
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 4.836059875412613e-05,
|
|
"logits/chosen": -2.152693748474121,
|
|
"logits/rejected": -2.1664419174194336,
|
|
"logps/chosen": -197.43896484375,
|
|
"logps/rejected": -206.69210815429688,
|
|
"loss": 0.2823,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.0961377620697021,
|
|
"rewards/margins": 3.136141538619995,
|
|
"rewards/rejected": -4.2322797775268555,
|
|
"step": 974
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.8356886785594544e-05,
|
|
"logits/chosen": -2.3830044269561768,
|
|
"logits/rejected": -2.392028331756592,
|
|
"logps/chosen": -191.5318603515625,
|
|
"logps/rejected": -222.13458251953125,
|
|
"loss": 0.082,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.9227029085159302,
|
|
"rewards/margins": 4.220012664794922,
|
|
"rewards/rejected": -5.142715930938721,
|
|
"step": 975
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.835317076223901e-05,
|
|
"logits/chosen": -2.298764228820801,
|
|
"logits/rejected": -2.360914468765259,
|
|
"logps/chosen": -238.4506378173828,
|
|
"logps/rejected": -276.4215087890625,
|
|
"loss": 0.1926,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.0809506177902222,
|
|
"rewards/margins": 3.495990037918091,
|
|
"rewards/rejected": -4.576941013336182,
|
|
"step": 976
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.834945068470463e-05,
|
|
"logits/chosen": -2.3851685523986816,
|
|
"logits/rejected": -2.402043342590332,
|
|
"logps/chosen": -215.42071533203125,
|
|
"logps/rejected": -260.2996520996094,
|
|
"loss": 0.1705,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.8322378396987915,
|
|
"rewards/margins": 3.276536703109741,
|
|
"rewards/rejected": -4.1087751388549805,
|
|
"step": 977
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.834572655363723e-05,
|
|
"logits/chosen": -2.3521459102630615,
|
|
"logits/rejected": -2.359070301055908,
|
|
"logps/chosen": -172.63397216796875,
|
|
"logps/rejected": -249.02813720703125,
|
|
"loss": 0.1009,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.6295977830886841,
|
|
"rewards/margins": 4.736910343170166,
|
|
"rewards/rejected": -5.3665080070495605,
|
|
"step": 978
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.834199836968334e-05,
|
|
"logits/chosen": -2.232576847076416,
|
|
"logits/rejected": -2.2787835597991943,
|
|
"logps/chosen": -212.65316772460938,
|
|
"logps/rejected": -277.7938537597656,
|
|
"loss": 0.1771,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.9311662912368774,
|
|
"rewards/margins": 4.264695644378662,
|
|
"rewards/rejected": -5.19586181640625,
|
|
"step": 979
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.833826613349016e-05,
|
|
"logits/chosen": -2.2517361640930176,
|
|
"logits/rejected": -2.284072160720825,
|
|
"logps/chosen": -220.1957550048828,
|
|
"logps/rejected": -214.6461181640625,
|
|
"loss": 0.2048,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.8580664396286011,
|
|
"rewards/margins": 3.004542589187622,
|
|
"rewards/rejected": -3.8626086711883545,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.8334529845705635e-05,
|
|
"logits/chosen": -2.3734512329101562,
|
|
"logits/rejected": -2.3852882385253906,
|
|
"logps/chosen": -185.3246612548828,
|
|
"logps/rejected": -235.97036743164062,
|
|
"loss": 0.0987,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8764241337776184,
|
|
"rewards/margins": 3.519378900527954,
|
|
"rewards/rejected": -4.395802974700928,
|
|
"step": 981
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.833078950697839e-05,
|
|
"logits/chosen": -2.318061113357544,
|
|
"logits/rejected": -2.3952810764312744,
|
|
"logps/chosen": -203.9101104736328,
|
|
"logps/rejected": -264.8881530761719,
|
|
"loss": 0.1093,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.5848926305770874,
|
|
"rewards/margins": 3.398397207260132,
|
|
"rewards/rejected": -4.98328971862793,
|
|
"step": 982
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.832704511795776e-05,
|
|
"logits/chosen": -2.257204294204712,
|
|
"logits/rejected": -2.243340015411377,
|
|
"logps/chosen": -220.27609252929688,
|
|
"logps/rejected": -302.30224609375,
|
|
"loss": 0.0464,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.7187144756317139,
|
|
"rewards/margins": 5.5636396408081055,
|
|
"rewards/rejected": -6.282354354858398,
|
|
"step": 983
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.8323296679293774e-05,
|
|
"logits/chosen": -2.0718610286712646,
|
|
"logits/rejected": -2.163663148880005,
|
|
"logps/chosen": -201.58529663085938,
|
|
"logps/rejected": -284.6639099121094,
|
|
"loss": 0.186,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.4365129470825195,
|
|
"rewards/margins": 3.119493007659912,
|
|
"rewards/rejected": -4.55600643157959,
|
|
"step": 984
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.831954419163719e-05,
|
|
"logits/chosen": -2.378993034362793,
|
|
"logits/rejected": -2.4145779609680176,
|
|
"logps/chosen": -194.3405303955078,
|
|
"logps/rejected": -258.2578430175781,
|
|
"loss": 0.1569,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.0772855281829834,
|
|
"rewards/margins": 3.5224766731262207,
|
|
"rewards/rejected": -4.599761962890625,
|
|
"step": 985
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.8315787655639436e-05,
|
|
"logits/chosen": -2.158259391784668,
|
|
"logits/rejected": -2.242321729660034,
|
|
"logps/chosen": -172.10662841796875,
|
|
"logps/rejected": -238.54212951660156,
|
|
"loss": 0.1116,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.0470584630966187,
|
|
"rewards/margins": 3.594759225845337,
|
|
"rewards/rejected": -4.641817569732666,
|
|
"step": 986
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.8312027071952645e-05,
|
|
"logits/chosen": -2.3166654109954834,
|
|
"logits/rejected": -2.371119499206543,
|
|
"logps/chosen": -201.59706115722656,
|
|
"logps/rejected": -208.94171142578125,
|
|
"loss": 0.1019,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.161280632019043,
|
|
"rewards/margins": 3.300464630126953,
|
|
"rewards/rejected": -4.461745262145996,
|
|
"step": 987
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.8308262441229704e-05,
|
|
"logits/chosen": -2.368316888809204,
|
|
"logits/rejected": -2.464874029159546,
|
|
"logps/chosen": -227.5435333251953,
|
|
"logps/rejected": -269.635009765625,
|
|
"loss": 0.1186,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.2794337272644043,
|
|
"rewards/margins": 3.7162654399871826,
|
|
"rewards/rejected": -4.995699882507324,
|
|
"step": 988
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 4.830449376412413e-05,
|
|
"logits/chosen": -2.4263885021209717,
|
|
"logits/rejected": -2.330914258956909,
|
|
"logps/chosen": -174.01873779296875,
|
|
"logps/rejected": -176.23687744140625,
|
|
"loss": 0.2376,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.1324949264526367,
|
|
"rewards/margins": 2.7193217277526855,
|
|
"rewards/rejected": -3.8518168926239014,
|
|
"step": 989
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.8300721041290194e-05,
|
|
"logits/chosen": -2.482351303100586,
|
|
"logits/rejected": -2.511679172515869,
|
|
"logps/chosen": -258.9258117675781,
|
|
"logps/rejected": -289.4918212890625,
|
|
"loss": 0.0891,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.4783955812454224,
|
|
"rewards/margins": 3.5242300033569336,
|
|
"rewards/rejected": -5.002625465393066,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.829694427338284e-05,
|
|
"logits/chosen": -2.210787773132324,
|
|
"logits/rejected": -2.2695932388305664,
|
|
"logps/chosen": -242.7936553955078,
|
|
"logps/rejected": -285.85845947265625,
|
|
"loss": 0.0737,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.1135588884353638,
|
|
"rewards/margins": 4.644796371459961,
|
|
"rewards/rejected": -5.758354663848877,
|
|
"step": 991
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.829316346105773e-05,
|
|
"logits/chosen": -2.5152206420898438,
|
|
"logits/rejected": -2.598931312561035,
|
|
"logps/chosen": -233.4760284423828,
|
|
"logps/rejected": -257.0856628417969,
|
|
"loss": 0.1722,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.4175981283187866,
|
|
"rewards/margins": 3.751023530960083,
|
|
"rewards/rejected": -5.168622016906738,
|
|
"step": 992
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.828937860497124e-05,
|
|
"logits/chosen": -1.9849010705947876,
|
|
"logits/rejected": -2.02589750289917,
|
|
"logps/chosen": -202.5492706298828,
|
|
"logps/rejected": -240.67250061035156,
|
|
"loss": 0.2435,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.1498560905456543,
|
|
"rewards/margins": 2.9630770683288574,
|
|
"rewards/rejected": -4.112933158874512,
|
|
"step": 993
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.82855897057804e-05,
|
|
"logits/chosen": -2.2394447326660156,
|
|
"logits/rejected": -2.2644007205963135,
|
|
"logps/chosen": -200.177978515625,
|
|
"logps/rejected": -232.31613159179688,
|
|
"loss": 0.1656,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.2112644910812378,
|
|
"rewards/margins": 3.5165750980377197,
|
|
"rewards/rejected": -4.727839469909668,
|
|
"step": 994
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.8281796764143e-05,
|
|
"logits/chosen": -2.1232497692108154,
|
|
"logits/rejected": -2.1741788387298584,
|
|
"logps/chosen": -187.11965942382812,
|
|
"logps/rejected": -211.57391357421875,
|
|
"loss": 0.0946,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.9103628396987915,
|
|
"rewards/margins": 3.7156426906585693,
|
|
"rewards/rejected": -4.62600564956665,
|
|
"step": 995
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.8277999780717496e-05,
|
|
"logits/chosen": -2.3636045455932617,
|
|
"logits/rejected": -2.3993918895721436,
|
|
"logps/chosen": -209.7058868408203,
|
|
"logps/rejected": -273.6556396484375,
|
|
"loss": 0.1253,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.197162389755249,
|
|
"rewards/margins": 4.630073547363281,
|
|
"rewards/rejected": -5.827235698699951,
|
|
"step": 996
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 4.827419875616306e-05,
|
|
"logits/chosen": -2.4287328720092773,
|
|
"logits/rejected": -2.444946765899658,
|
|
"logps/chosen": -187.06781005859375,
|
|
"logps/rejected": -227.0997314453125,
|
|
"loss": 0.3696,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -1.5417091846466064,
|
|
"rewards/margins": 3.166154623031616,
|
|
"rewards/rejected": -4.707863807678223,
|
|
"step": 997
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 4.827039369113956e-05,
|
|
"logits/chosen": -2.3056039810180664,
|
|
"logits/rejected": -2.3643782138824463,
|
|
"logps/chosen": -218.290283203125,
|
|
"logps/rejected": -289.4818420410156,
|
|
"loss": 0.0622,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.1654748916625977,
|
|
"rewards/margins": 4.145327091217041,
|
|
"rewards/rejected": -5.310801982879639,
|
|
"step": 998
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 4.826658458630756e-05,
|
|
"logits/chosen": -2.3559858798980713,
|
|
"logits/rejected": -2.4298417568206787,
|
|
"logps/chosen": -279.0400695800781,
|
|
"logps/rejected": -276.64044189453125,
|
|
"loss": 0.1911,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.6909762620925903,
|
|
"rewards/margins": 3.3049442768096924,
|
|
"rewards/rejected": -4.995920658111572,
|
|
"step": 999
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 4.826277144232834e-05,
|
|
"logits/chosen": -2.3466007709503174,
|
|
"logits/rejected": -2.376455068588257,
|
|
"logps/chosen": -215.84930419921875,
|
|
"logps/rejected": -249.32769775390625,
|
|
"loss": 0.1875,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.2421891689300537,
|
|
"rewards/margins": 4.225651264190674,
|
|
"rewards/rejected": -5.467840194702148,
|
|
"step": 1000
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 7640,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 10,
|
|
"save_steps": 500,
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|