5085 lines
175 KiB
JSON
5085 lines
175 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 3.0,
|
|
"eval_steps": 50,
|
|
"global_step": 2784,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.01078566854292359,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 2.990301724137931e-06,
|
|
"logits/chosen": -1.2338563203811646,
|
|
"logits/rejected": -1.2257438898086548,
|
|
"logps/chosen": -490.282470703125,
|
|
"logps/rejected": -482.52545166015625,
|
|
"loss": 0.6903,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": 0.0017994878580793738,
|
|
"rewards/margins": 0.006464972160756588,
|
|
"rewards/rejected": -0.004665483720600605,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.02157133708584718,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 2.979525862068966e-06,
|
|
"logits/chosen": -1.2723968029022217,
|
|
"logits/rejected": -1.2785792350769043,
|
|
"logps/chosen": -520.5911254882812,
|
|
"logps/rejected": -514.4813842773438,
|
|
"loss": 0.6856,
|
|
"rewards/accuracies": 0.6078125238418579,
|
|
"rewards/chosen": 0.0043027508072555065,
|
|
"rewards/margins": 0.0164328720420599,
|
|
"rewards/rejected": -0.01213012170046568,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.03235700562877077,
|
|
"grad_norm": 19.125,
|
|
"learning_rate": 2.96875e-06,
|
|
"logits/chosen": -1.2637840509414673,
|
|
"logits/rejected": -1.246565818786621,
|
|
"logps/chosen": -507.4403381347656,
|
|
"logps/rejected": -501.22479248046875,
|
|
"loss": 0.6789,
|
|
"rewards/accuracies": 0.6468750238418579,
|
|
"rewards/chosen": -0.0017711544642224908,
|
|
"rewards/margins": 0.030538281425833702,
|
|
"rewards/rejected": -0.032309435307979584,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.04314267417169436,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 2.9579741379310345e-06,
|
|
"logits/chosen": -1.2590439319610596,
|
|
"logits/rejected": -1.2330172061920166,
|
|
"logps/chosen": -485.248291015625,
|
|
"logps/rejected": -482.63458251953125,
|
|
"loss": 0.6765,
|
|
"rewards/accuracies": 0.6484375,
|
|
"rewards/chosen": -0.00601751497015357,
|
|
"rewards/margins": 0.03583287447690964,
|
|
"rewards/rejected": -0.041850391775369644,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.05392834271461795,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 2.947198275862069e-06,
|
|
"logits/chosen": -1.3426216840744019,
|
|
"logits/rejected": -1.3334815502166748,
|
|
"logps/chosen": -532.8280029296875,
|
|
"logps/rejected": -514.1970825195312,
|
|
"loss": 0.6684,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.00931160431355238,
|
|
"rewards/margins": 0.053057052195072174,
|
|
"rewards/rejected": -0.06236865371465683,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.05392834271461795,
|
|
"eval_logits/chosen": -1.3231499195098877,
|
|
"eval_logits/rejected": -1.3642845153808594,
|
|
"eval_logps/chosen": -549.2651977539062,
|
|
"eval_logps/rejected": -494.54583740234375,
|
|
"eval_loss": 0.6826924681663513,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -0.05200628936290741,
|
|
"eval_rewards/margins": 0.02476159855723381,
|
|
"eval_rewards/rejected": -0.07676788419485092,
|
|
"eval_runtime": 13.2928,
|
|
"eval_samples_per_second": 7.523,
|
|
"eval_steps_per_second": 0.978,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.06471401125754155,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 2.9364224137931035e-06,
|
|
"logits/chosen": -1.3449946641921997,
|
|
"logits/rejected": -1.343201756477356,
|
|
"logps/chosen": -548.1566772460938,
|
|
"logps/rejected": -531.6286010742188,
|
|
"loss": 0.6705,
|
|
"rewards/accuracies": 0.6499999761581421,
|
|
"rewards/chosen": -0.028984328731894493,
|
|
"rewards/margins": 0.04919930547475815,
|
|
"rewards/rejected": -0.07818363606929779,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.07549967980046514,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 2.925646551724138e-06,
|
|
"logits/chosen": -1.3576488494873047,
|
|
"logits/rejected": -1.3385895490646362,
|
|
"logps/chosen": -516.7993774414062,
|
|
"logps/rejected": -502.6502990722656,
|
|
"loss": 0.6624,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.03774970397353172,
|
|
"rewards/margins": 0.0672496110200882,
|
|
"rewards/rejected": -0.10499931871891022,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.08628534834338872,
|
|
"grad_norm": 19.5,
|
|
"learning_rate": 2.9148706896551725e-06,
|
|
"logits/chosen": -1.4393236637115479,
|
|
"logits/rejected": -1.4375989437103271,
|
|
"logps/chosen": -534.2485961914062,
|
|
"logps/rejected": -516.8245849609375,
|
|
"loss": 0.6547,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -0.05567573383450508,
|
|
"rewards/margins": 0.08416923880577087,
|
|
"rewards/rejected": -0.13984496891498566,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.09707101688631231,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 2.904094827586207e-06,
|
|
"logits/chosen": -1.4080358743667603,
|
|
"logits/rejected": -1.399465799331665,
|
|
"logps/chosen": -503.4671936035156,
|
|
"logps/rejected": -488.74383544921875,
|
|
"loss": 0.6565,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.07379738986492157,
|
|
"rewards/margins": 0.08224000781774521,
|
|
"rewards/rejected": -0.15603742003440857,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.1078566854292359,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 2.8933189655172415e-06,
|
|
"logits/chosen": -1.3748817443847656,
|
|
"logits/rejected": -1.366629958152771,
|
|
"logps/chosen": -518.0892944335938,
|
|
"logps/rejected": -499.7432556152344,
|
|
"loss": 0.6541,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.09199288487434387,
|
|
"rewards/margins": 0.08948174864053726,
|
|
"rewards/rejected": -0.18147462606430054,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.1078566854292359,
|
|
"eval_logits/chosen": -1.4332343339920044,
|
|
"eval_logits/rejected": -1.479236364364624,
|
|
"eval_logps/chosen": -551.7933349609375,
|
|
"eval_logps/rejected": -497.4646911621094,
|
|
"eval_loss": 0.6763917803764343,
|
|
"eval_rewards/accuracies": 0.5769230723381042,
|
|
"eval_rewards/chosen": -0.17841331660747528,
|
|
"eval_rewards/margins": 0.04429765045642853,
|
|
"eval_rewards/rejected": -0.222710981965065,
|
|
"eval_runtime": 12.8741,
|
|
"eval_samples_per_second": 7.768,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.1186423539721595,
|
|
"grad_norm": 19.25,
|
|
"learning_rate": 2.8825431034482758e-06,
|
|
"logits/chosen": -1.436632513999939,
|
|
"logits/rejected": -1.4507520198822021,
|
|
"logps/chosen": -508.725830078125,
|
|
"logps/rejected": -500.9781799316406,
|
|
"loss": 0.6421,
|
|
"rewards/accuracies": 0.729687511920929,
|
|
"rewards/chosen": -0.11871524900197983,
|
|
"rewards/margins": 0.11672022193670273,
|
|
"rewards/rejected": -0.23543548583984375,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.1294280225150831,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 2.8717672413793105e-06,
|
|
"logits/chosen": -1.4648942947387695,
|
|
"logits/rejected": -1.474923849105835,
|
|
"logps/chosen": -522.5699462890625,
|
|
"logps/rejected": -521.1087036132812,
|
|
"loss": 0.635,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -0.14662253856658936,
|
|
"rewards/margins": 0.13384078443050385,
|
|
"rewards/rejected": -0.2804633677005768,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.14021369105800668,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 2.860991379310345e-06,
|
|
"logits/chosen": -1.5117241144180298,
|
|
"logits/rejected": -1.5187674760818481,
|
|
"logps/chosen": -526.3902587890625,
|
|
"logps/rejected": -518.88525390625,
|
|
"loss": 0.6413,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.1973995864391327,
|
|
"rewards/margins": 0.12558409571647644,
|
|
"rewards/rejected": -0.32298368215560913,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.15099935960093028,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 2.8502155172413795e-06,
|
|
"logits/chosen": -1.5119014978408813,
|
|
"logits/rejected": -1.5130690336227417,
|
|
"logps/chosen": -533.3895263671875,
|
|
"logps/rejected": -527.9937744140625,
|
|
"loss": 0.6401,
|
|
"rewards/accuracies": 0.6703125238418579,
|
|
"rewards/chosen": -0.22728869318962097,
|
|
"rewards/margins": 0.13429734110832214,
|
|
"rewards/rejected": -0.3615860342979431,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.16178502814385384,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 2.839439655172414e-06,
|
|
"logits/chosen": -1.5527749061584473,
|
|
"logits/rejected": -1.5703589916229248,
|
|
"logps/chosen": -515.7926025390625,
|
|
"logps/rejected": -503.88360595703125,
|
|
"loss": 0.6397,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2661042809486389,
|
|
"rewards/margins": 0.13428668677806854,
|
|
"rewards/rejected": -0.40039095282554626,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.16178502814385384,
|
|
"eval_logits/chosen": -1.5442339181900024,
|
|
"eval_logits/rejected": -1.5949413776397705,
|
|
"eval_logps/chosen": -555.982666015625,
|
|
"eval_logps/rejected": -502.18585205078125,
|
|
"eval_loss": 0.6722157001495361,
|
|
"eval_rewards/accuracies": 0.5865384340286255,
|
|
"eval_rewards/chosen": -0.38787999749183655,
|
|
"eval_rewards/margins": 0.07088876515626907,
|
|
"eval_rewards/rejected": -0.4587687849998474,
|
|
"eval_runtime": 12.8342,
|
|
"eval_samples_per_second": 7.792,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.17257069668677744,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 2.8286637931034485e-06,
|
|
"logits/chosen": -1.5724695920944214,
|
|
"logits/rejected": -1.5730245113372803,
|
|
"logps/chosen": -506.13671875,
|
|
"logps/rejected": -495.46142578125,
|
|
"loss": 0.6699,
|
|
"rewards/accuracies": 0.659375011920929,
|
|
"rewards/chosen": -0.29719287157058716,
|
|
"rewards/margins": 0.10155584663152695,
|
|
"rewards/rejected": -0.3987486958503723,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.18335636522970103,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.817887931034483e-06,
|
|
"logits/chosen": -1.5566643476486206,
|
|
"logits/rejected": -1.555086612701416,
|
|
"logps/chosen": -557.1043701171875,
|
|
"logps/rejected": -539.7008666992188,
|
|
"loss": 0.6322,
|
|
"rewards/accuracies": 0.6656249761581421,
|
|
"rewards/chosen": -0.32251477241516113,
|
|
"rewards/margins": 0.1617726981639862,
|
|
"rewards/rejected": -0.48428741097450256,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.19414203377262462,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 2.807112068965517e-06,
|
|
"logits/chosen": -1.5289297103881836,
|
|
"logits/rejected": -1.5290358066558838,
|
|
"logps/chosen": -527.3707275390625,
|
|
"logps/rejected": -511.8185119628906,
|
|
"loss": 0.6347,
|
|
"rewards/accuracies": 0.667187511920929,
|
|
"rewards/chosen": -0.3478879928588867,
|
|
"rewards/margins": 0.1548273265361786,
|
|
"rewards/rejected": -0.5027152895927429,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.20492770231554822,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 2.796336206896552e-06,
|
|
"logits/chosen": -1.5106637477874756,
|
|
"logits/rejected": -1.4951450824737549,
|
|
"logps/chosen": -535.6061401367188,
|
|
"logps/rejected": -528.7681884765625,
|
|
"loss": 0.623,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.35832709074020386,
|
|
"rewards/margins": 0.20923912525177002,
|
|
"rewards/rejected": -0.5675662755966187,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.2157133708584718,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 2.785560344827586e-06,
|
|
"logits/chosen": -1.6122379302978516,
|
|
"logits/rejected": -1.6149158477783203,
|
|
"logps/chosen": -556.2003173828125,
|
|
"logps/rejected": -536.0861206054688,
|
|
"loss": 0.6237,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -0.40547609329223633,
|
|
"rewards/margins": 0.18066053092479706,
|
|
"rewards/rejected": -0.5861365795135498,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.2157133708584718,
|
|
"eval_logits/chosen": -1.6185290813446045,
|
|
"eval_logits/rejected": -1.671698808670044,
|
|
"eval_logps/chosen": -559.1907348632812,
|
|
"eval_logps/rejected": -505.73992919921875,
|
|
"eval_loss": 0.6703996062278748,
|
|
"eval_rewards/accuracies": 0.5769230723381042,
|
|
"eval_rewards/chosen": -0.5482814311981201,
|
|
"eval_rewards/margins": 0.08819277584552765,
|
|
"eval_rewards/rejected": -0.6364741921424866,
|
|
"eval_runtime": 12.8368,
|
|
"eval_samples_per_second": 7.79,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.2264990394013954,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 2.774784482758621e-06,
|
|
"logits/chosen": -1.602872610092163,
|
|
"logits/rejected": -1.6062767505645752,
|
|
"logps/chosen": -494.86566162109375,
|
|
"logps/rejected": -486.106689453125,
|
|
"loss": 0.6313,
|
|
"rewards/accuracies": 0.676562488079071,
|
|
"rewards/chosen": -0.41412362456321716,
|
|
"rewards/margins": 0.16503319144248962,
|
|
"rewards/rejected": -0.5791568160057068,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.237284707944319,
|
|
"grad_norm": 22.875,
|
|
"learning_rate": 2.764008620689655e-06,
|
|
"logits/chosen": -1.6498100757598877,
|
|
"logits/rejected": -1.6630780696868896,
|
|
"logps/chosen": -540.1375732421875,
|
|
"logps/rejected": -533.9187622070312,
|
|
"loss": 0.6307,
|
|
"rewards/accuracies": 0.6546875238418579,
|
|
"rewards/chosen": -0.4561203420162201,
|
|
"rewards/margins": 0.18029221892356873,
|
|
"rewards/rejected": -0.636412501335144,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.2480703764872426,
|
|
"grad_norm": 19.5,
|
|
"learning_rate": 2.75323275862069e-06,
|
|
"logits/chosen": -1.667067527770996,
|
|
"logits/rejected": -1.681429147720337,
|
|
"logps/chosen": -525.0696411132812,
|
|
"logps/rejected": -508.50067138671875,
|
|
"loss": 0.6261,
|
|
"rewards/accuracies": 0.6703125238418579,
|
|
"rewards/chosen": -0.4919908940792084,
|
|
"rewards/margins": 0.19147078692913055,
|
|
"rewards/rejected": -0.6834616661071777,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.2588560450301662,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 2.742456896551724e-06,
|
|
"logits/chosen": -1.646775484085083,
|
|
"logits/rejected": -1.6376352310180664,
|
|
"logps/chosen": -517.5604248046875,
|
|
"logps/rejected": -504.1005859375,
|
|
"loss": 0.6228,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.4839399755001068,
|
|
"rewards/margins": 0.1933339387178421,
|
|
"rewards/rejected": -0.6772739291191101,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.26964171357308975,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 2.7316810344827584e-06,
|
|
"logits/chosen": -1.630406379699707,
|
|
"logits/rejected": -1.6203149557113647,
|
|
"logps/chosen": -511.6064453125,
|
|
"logps/rejected": -504.46112060546875,
|
|
"loss": 0.6224,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.5028296709060669,
|
|
"rewards/margins": 0.19740687310695648,
|
|
"rewards/rejected": -0.700236439704895,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.26964171357308975,
|
|
"eval_logits/chosen": -1.671502947807312,
|
|
"eval_logits/rejected": -1.727898359298706,
|
|
"eval_logps/chosen": -561.3909912109375,
|
|
"eval_logps/rejected": -507.87896728515625,
|
|
"eval_loss": 0.6769301891326904,
|
|
"eval_rewards/accuracies": 0.567307710647583,
|
|
"eval_rewards/chosen": -0.6582961678504944,
|
|
"eval_rewards/margins": 0.08512917906045914,
|
|
"eval_rewards/rejected": -0.7434254288673401,
|
|
"eval_runtime": 12.8313,
|
|
"eval_samples_per_second": 7.793,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.28042738211601337,
|
|
"grad_norm": 22.75,
|
|
"learning_rate": 2.720905172413793e-06,
|
|
"logits/chosen": -1.6629798412322998,
|
|
"logits/rejected": -1.664310097694397,
|
|
"logps/chosen": -564.4110717773438,
|
|
"logps/rejected": -550.1747436523438,
|
|
"loss": 0.6344,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -0.51663738489151,
|
|
"rewards/margins": 0.17350056767463684,
|
|
"rewards/rejected": -0.6901379823684692,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.29121305065893693,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 2.7101293103448275e-06,
|
|
"logits/chosen": -1.6739919185638428,
|
|
"logits/rejected": -1.6669447422027588,
|
|
"logps/chosen": -562.1046752929688,
|
|
"logps/rejected": -553.902099609375,
|
|
"loss": 0.6256,
|
|
"rewards/accuracies": 0.6656249761581421,
|
|
"rewards/chosen": -0.5245037078857422,
|
|
"rewards/margins": 0.19063076376914978,
|
|
"rewards/rejected": -0.7151345014572144,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.30199871920186055,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 2.699353448275862e-06,
|
|
"logits/chosen": -1.6442272663116455,
|
|
"logits/rejected": -1.6505457162857056,
|
|
"logps/chosen": -529.9203491210938,
|
|
"logps/rejected": -513.3055419921875,
|
|
"loss": 0.6378,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -0.5445322394371033,
|
|
"rewards/margins": 0.16621330380439758,
|
|
"rewards/rejected": -0.710745632648468,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.3127843877447841,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.6885775862068965e-06,
|
|
"logits/chosen": -1.6472547054290771,
|
|
"logits/rejected": -1.6532646417617798,
|
|
"logps/chosen": -527.1898193359375,
|
|
"logps/rejected": -516.4867553710938,
|
|
"loss": 0.6132,
|
|
"rewards/accuracies": 0.675000011920929,
|
|
"rewards/chosen": -0.49309319257736206,
|
|
"rewards/margins": 0.21812161803245544,
|
|
"rewards/rejected": -0.7112148404121399,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.3235700562877077,
|
|
"grad_norm": 19.0,
|
|
"learning_rate": 2.677801724137931e-06,
|
|
"logits/chosen": -1.6490532159805298,
|
|
"logits/rejected": -1.6721302270889282,
|
|
"logps/chosen": -567.3697509765625,
|
|
"logps/rejected": -550.6924438476562,
|
|
"loss": 0.6216,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.5456413626670837,
|
|
"rewards/margins": 0.2058580368757248,
|
|
"rewards/rejected": -0.7514994144439697,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.3235700562877077,
|
|
"eval_logits/chosen": -1.6934152841567993,
|
|
"eval_logits/rejected": -1.7509509325027466,
|
|
"eval_logps/chosen": -562.0770874023438,
|
|
"eval_logps/rejected": -508.9422302246094,
|
|
"eval_loss": 0.6661080121994019,
|
|
"eval_rewards/accuracies": 0.5865384340286255,
|
|
"eval_rewards/chosen": -0.692603349685669,
|
|
"eval_rewards/margins": 0.10398232936859131,
|
|
"eval_rewards/rejected": -0.7965856194496155,
|
|
"eval_runtime": 12.8789,
|
|
"eval_samples_per_second": 7.765,
|
|
"eval_steps_per_second": 1.009,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.3343557248306313,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 2.6670258620689655e-06,
|
|
"logits/chosen": -1.7095540761947632,
|
|
"logits/rejected": -1.7263736724853516,
|
|
"logps/chosen": -530.0485229492188,
|
|
"logps/rejected": -530.2100830078125,
|
|
"loss": 0.6314,
|
|
"rewards/accuracies": 0.6468750238418579,
|
|
"rewards/chosen": -0.6013873815536499,
|
|
"rewards/margins": 0.17918090522289276,
|
|
"rewards/rejected": -0.7805682420730591,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.3451413933735549,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 2.6562499999999998e-06,
|
|
"logits/chosen": -1.7012121677398682,
|
|
"logits/rejected": -1.6956939697265625,
|
|
"logps/chosen": -520.4534301757812,
|
|
"logps/rejected": -511.6394958496094,
|
|
"loss": 0.6502,
|
|
"rewards/accuracies": 0.629687488079071,
|
|
"rewards/chosen": -0.6033864617347717,
|
|
"rewards/margins": 0.14426395297050476,
|
|
"rewards/rejected": -0.7476503252983093,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.3559270619164785,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 2.6454741379310345e-06,
|
|
"logits/chosen": -1.6911704540252686,
|
|
"logits/rejected": -1.6552823781967163,
|
|
"logps/chosen": -537.2453002929688,
|
|
"logps/rejected": -534.7839965820312,
|
|
"loss": 0.6255,
|
|
"rewards/accuracies": 0.6640625,
|
|
"rewards/chosen": -0.568359375,
|
|
"rewards/margins": 0.22105315327644348,
|
|
"rewards/rejected": -0.7894124388694763,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.36671273045940206,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 2.6346982758620688e-06,
|
|
"logits/chosen": -1.664223313331604,
|
|
"logits/rejected": -1.6563825607299805,
|
|
"logps/chosen": -565.0128784179688,
|
|
"logps/rejected": -551.3712158203125,
|
|
"loss": 0.6117,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.5529137849807739,
|
|
"rewards/margins": 0.23182833194732666,
|
|
"rewards/rejected": -0.7847420573234558,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.3774983990023257,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 2.6239224137931035e-06,
|
|
"logits/chosen": -1.6517295837402344,
|
|
"logits/rejected": -1.6636890172958374,
|
|
"logps/chosen": -499.03717041015625,
|
|
"logps/rejected": -469.7442932128906,
|
|
"loss": 0.6454,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.559511125087738,
|
|
"rewards/margins": 0.14985349774360657,
|
|
"rewards/rejected": -0.7093645930290222,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.3774983990023257,
|
|
"eval_logits/chosen": -1.7148549556732178,
|
|
"eval_logits/rejected": -1.7745919227600098,
|
|
"eval_logps/chosen": -562.5575561523438,
|
|
"eval_logps/rejected": -509.7618713378906,
|
|
"eval_loss": 0.6621462106704712,
|
|
"eval_rewards/accuracies": 0.625,
|
|
"eval_rewards/chosen": -0.7166208624839783,
|
|
"eval_rewards/margins": 0.12094759196043015,
|
|
"eval_rewards/rejected": -0.8375685811042786,
|
|
"eval_runtime": 12.8579,
|
|
"eval_samples_per_second": 7.777,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.38828406754524925,
|
|
"grad_norm": 18.625,
|
|
"learning_rate": 2.613146551724138e-06,
|
|
"logits/chosen": -1.6570751667022705,
|
|
"logits/rejected": -1.673710584640503,
|
|
"logps/chosen": -550.8926391601562,
|
|
"logps/rejected": -530.0281982421875,
|
|
"loss": 0.635,
|
|
"rewards/accuracies": 0.682812511920929,
|
|
"rewards/chosen": -0.5474696755409241,
|
|
"rewards/margins": 0.180132657289505,
|
|
"rewards/rejected": -0.7276023626327515,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.39906973608817287,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 2.6023706896551725e-06,
|
|
"logits/chosen": -1.7166540622711182,
|
|
"logits/rejected": -1.7119417190551758,
|
|
"logps/chosen": -535.0701904296875,
|
|
"logps/rejected": -521.47509765625,
|
|
"loss": 0.6346,
|
|
"rewards/accuracies": 0.6421874761581421,
|
|
"rewards/chosen": -0.6037973761558533,
|
|
"rewards/margins": 0.18038420379161835,
|
|
"rewards/rejected": -0.7841815948486328,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.40985540463109643,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 2.591594827586207e-06,
|
|
"logits/chosen": -1.711259126663208,
|
|
"logits/rejected": -1.7223602533340454,
|
|
"logps/chosen": -554.1033935546875,
|
|
"logps/rejected": -536.9190673828125,
|
|
"loss": 0.6215,
|
|
"rewards/accuracies": 0.667187511920929,
|
|
"rewards/chosen": -0.5765694379806519,
|
|
"rewards/margins": 0.21333327889442444,
|
|
"rewards/rejected": -0.7899028062820435,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.42064107317402,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 2.580818965517241e-06,
|
|
"logits/chosen": -1.674808144569397,
|
|
"logits/rejected": -1.7261720895767212,
|
|
"logps/chosen": -534.8923950195312,
|
|
"logps/rejected": -509.17437744140625,
|
|
"loss": 0.6234,
|
|
"rewards/accuracies": 0.645312488079071,
|
|
"rewards/chosen": -0.5851498246192932,
|
|
"rewards/margins": 0.20684213936328888,
|
|
"rewards/rejected": -0.7919918298721313,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.4314267417169436,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.5700431034482762e-06,
|
|
"logits/chosen": -1.7043625116348267,
|
|
"logits/rejected": -1.721663236618042,
|
|
"logps/chosen": -547.33935546875,
|
|
"logps/rejected": -535.8040771484375,
|
|
"loss": 0.6314,
|
|
"rewards/accuracies": 0.645312488079071,
|
|
"rewards/chosen": -0.6050974130630493,
|
|
"rewards/margins": 0.18475469946861267,
|
|
"rewards/rejected": -0.7898520827293396,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.4314267417169436,
|
|
"eval_logits/chosen": -1.7199591398239136,
|
|
"eval_logits/rejected": -1.7800374031066895,
|
|
"eval_logps/chosen": -562.4886474609375,
|
|
"eval_logps/rejected": -509.6477355957031,
|
|
"eval_loss": 0.6607488989830017,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -0.7131773829460144,
|
|
"eval_rewards/margins": 0.11868361383676529,
|
|
"eval_rewards/rejected": -0.831861138343811,
|
|
"eval_runtime": 12.8809,
|
|
"eval_samples_per_second": 7.763,
|
|
"eval_steps_per_second": 1.009,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.4422124102598672,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 2.5592672413793105e-06,
|
|
"logits/chosen": -1.7056515216827393,
|
|
"logits/rejected": -1.7032356262207031,
|
|
"logps/chosen": -544.6566162109375,
|
|
"logps/rejected": -526.3981323242188,
|
|
"loss": 0.6346,
|
|
"rewards/accuracies": 0.6640625,
|
|
"rewards/chosen": -0.6110819578170776,
|
|
"rewards/margins": 0.17850695550441742,
|
|
"rewards/rejected": -0.7895889282226562,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.4529980788027908,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.5484913793103452e-06,
|
|
"logits/chosen": -1.7077720165252686,
|
|
"logits/rejected": -1.7262470722198486,
|
|
"logps/chosen": -538.9916381835938,
|
|
"logps/rejected": -514.1024780273438,
|
|
"loss": 0.6319,
|
|
"rewards/accuracies": 0.6390625238418579,
|
|
"rewards/chosen": -0.6130255460739136,
|
|
"rewards/margins": 0.18954649567604065,
|
|
"rewards/rejected": -0.8025720715522766,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.46378374734571437,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.5377155172413795e-06,
|
|
"logits/chosen": -1.6869211196899414,
|
|
"logits/rejected": -1.7010374069213867,
|
|
"logps/chosen": -519.823486328125,
|
|
"logps/rejected": -514.0619506835938,
|
|
"loss": 0.6187,
|
|
"rewards/accuracies": 0.6734374761581421,
|
|
"rewards/chosen": -0.6134659647941589,
|
|
"rewards/margins": 0.21337684988975525,
|
|
"rewards/rejected": -0.8268427848815918,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.474569415888638,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 2.526939655172414e-06,
|
|
"logits/chosen": -1.7328685522079468,
|
|
"logits/rejected": -1.7307264804840088,
|
|
"logps/chosen": -519.9312744140625,
|
|
"logps/rejected": -513.1961669921875,
|
|
"loss": 0.6278,
|
|
"rewards/accuracies": 0.676562488079071,
|
|
"rewards/chosen": -0.6580926179885864,
|
|
"rewards/margins": 0.20139017701148987,
|
|
"rewards/rejected": -0.8594827651977539,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.48535508443156156,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 2.5161637931034486e-06,
|
|
"logits/chosen": -1.7335723638534546,
|
|
"logits/rejected": -1.7415554523468018,
|
|
"logps/chosen": -526.5194702148438,
|
|
"logps/rejected": -518.3421020507812,
|
|
"loss": 0.6219,
|
|
"rewards/accuracies": 0.659375011920929,
|
|
"rewards/chosen": -0.6824163198471069,
|
|
"rewards/margins": 0.2106122523546219,
|
|
"rewards/rejected": -0.8930285573005676,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.48535508443156156,
|
|
"eval_logits/chosen": -1.7433769702911377,
|
|
"eval_logits/rejected": -1.8047083616256714,
|
|
"eval_logps/chosen": -564.2274169921875,
|
|
"eval_logps/rejected": -511.63067626953125,
|
|
"eval_loss": 0.6593914031982422,
|
|
"eval_rewards/accuracies": 0.5769230723381042,
|
|
"eval_rewards/chosen": -0.8001139760017395,
|
|
"eval_rewards/margins": 0.13089530169963837,
|
|
"eval_rewards/rejected": -0.9310091733932495,
|
|
"eval_runtime": 12.8159,
|
|
"eval_samples_per_second": 7.803,
|
|
"eval_steps_per_second": 1.014,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.4961407529744852,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 2.505387931034483e-06,
|
|
"logits/chosen": -1.7393690347671509,
|
|
"logits/rejected": -1.7588695287704468,
|
|
"logps/chosen": -552.0422973632812,
|
|
"logps/rejected": -530.3653564453125,
|
|
"loss": 0.6202,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.686679482460022,
|
|
"rewards/margins": 0.22524280846118927,
|
|
"rewards/rejected": -0.9119223356246948,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.5069264215174087,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 2.4946120689655176e-06,
|
|
"logits/chosen": -1.7419532537460327,
|
|
"logits/rejected": -1.7394039630889893,
|
|
"logps/chosen": -526.1238403320312,
|
|
"logps/rejected": -506.7391052246094,
|
|
"loss": 0.6301,
|
|
"rewards/accuracies": 0.653124988079071,
|
|
"rewards/chosen": -0.6740504503250122,
|
|
"rewards/margins": 0.18307238817214966,
|
|
"rewards/rejected": -0.8571227788925171,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.5177120900603324,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 2.483836206896552e-06,
|
|
"logits/chosen": -1.6943247318267822,
|
|
"logits/rejected": -1.6783298254013062,
|
|
"logps/chosen": -568.21435546875,
|
|
"logps/rejected": -553.17822265625,
|
|
"loss": 0.6339,
|
|
"rewards/accuracies": 0.6390625238418579,
|
|
"rewards/chosen": -0.6776462197303772,
|
|
"rewards/margins": 0.1882401704788208,
|
|
"rewards/rejected": -0.865886390209198,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.5284977586032559,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.473060344827586e-06,
|
|
"logits/chosen": -1.7065378427505493,
|
|
"logits/rejected": -1.7249234914779663,
|
|
"logps/chosen": -529.4561767578125,
|
|
"logps/rejected": -506.04071044921875,
|
|
"loss": 0.6378,
|
|
"rewards/accuracies": 0.645312488079071,
|
|
"rewards/chosen": -0.6808103322982788,
|
|
"rewards/margins": 0.17550134658813477,
|
|
"rewards/rejected": -0.8563116788864136,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.5392834271461795,
|
|
"grad_norm": 23.75,
|
|
"learning_rate": 2.462284482758621e-06,
|
|
"logits/chosen": -1.6840463876724243,
|
|
"logits/rejected": -1.70773184299469,
|
|
"logps/chosen": -530.1900634765625,
|
|
"logps/rejected": -512.8983764648438,
|
|
"loss": 0.6382,
|
|
"rewards/accuracies": 0.6499999761581421,
|
|
"rewards/chosen": -0.6517452001571655,
|
|
"rewards/margins": 0.17097657918930054,
|
|
"rewards/rejected": -0.8227217793464661,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.5392834271461795,
|
|
"eval_logits/chosen": -1.7378157377243042,
|
|
"eval_logits/rejected": -1.7989575862884521,
|
|
"eval_logps/chosen": -563.543212890625,
|
|
"eval_logps/rejected": -510.96929931640625,
|
|
"eval_loss": 0.6553998589515686,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -0.7659082412719727,
|
|
"eval_rewards/margins": 0.1320342719554901,
|
|
"eval_rewards/rejected": -0.8979425430297852,
|
|
"eval_runtime": 12.8216,
|
|
"eval_samples_per_second": 7.799,
|
|
"eval_steps_per_second": 1.014,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.5500690956891031,
|
|
"grad_norm": 23.375,
|
|
"learning_rate": 2.451508620689655e-06,
|
|
"logits/chosen": -1.7287929058074951,
|
|
"logits/rejected": -1.747097373008728,
|
|
"logps/chosen": -568.22900390625,
|
|
"logps/rejected": -537.8626098632812,
|
|
"loss": 0.6234,
|
|
"rewards/accuracies": 0.660937488079071,
|
|
"rewards/chosen": -0.6510963439941406,
|
|
"rewards/margins": 0.20002727210521698,
|
|
"rewards/rejected": -0.8511236310005188,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.5608547642320267,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.44073275862069e-06,
|
|
"logits/chosen": -1.7096725702285767,
|
|
"logits/rejected": -1.713513970375061,
|
|
"logps/chosen": -578.6484375,
|
|
"logps/rejected": -552.06884765625,
|
|
"loss": 0.618,
|
|
"rewards/accuracies": 0.6734374761581421,
|
|
"rewards/chosen": -0.6893950700759888,
|
|
"rewards/margins": 0.22265811264514923,
|
|
"rewards/rejected": -0.9120532274246216,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.5716404327749502,
|
|
"grad_norm": 24.375,
|
|
"learning_rate": 2.429956896551724e-06,
|
|
"logits/chosen": -1.7773420810699463,
|
|
"logits/rejected": -1.7806062698364258,
|
|
"logps/chosen": -561.3190307617188,
|
|
"logps/rejected": -557.246826171875,
|
|
"loss": 0.6274,
|
|
"rewards/accuracies": 0.660937488079071,
|
|
"rewards/chosen": -0.6820749640464783,
|
|
"rewards/margins": 0.23882977664470673,
|
|
"rewards/rejected": -0.9209047555923462,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.5824261013178739,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 2.419181034482759e-06,
|
|
"logits/chosen": -1.7619960308074951,
|
|
"logits/rejected": -1.7642993927001953,
|
|
"logps/chosen": -520.9690551757812,
|
|
"logps/rejected": -509.5274353027344,
|
|
"loss": 0.6309,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -0.7083293795585632,
|
|
"rewards/margins": 0.1926591694355011,
|
|
"rewards/rejected": -0.9009885787963867,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.5932117698607975,
|
|
"grad_norm": 18.75,
|
|
"learning_rate": 2.408405172413793e-06,
|
|
"logits/chosen": -1.7466360330581665,
|
|
"logits/rejected": -1.746272325515747,
|
|
"logps/chosen": -513.3010864257812,
|
|
"logps/rejected": -490.46624755859375,
|
|
"loss": 0.6377,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.7287647128105164,
|
|
"rewards/margins": 0.16981419920921326,
|
|
"rewards/rejected": -0.8985790014266968,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.5932117698607975,
|
|
"eval_logits/chosen": -1.7363488674163818,
|
|
"eval_logits/rejected": -1.7969601154327393,
|
|
"eval_logps/chosen": -564.2831420898438,
|
|
"eval_logps/rejected": -511.74945068359375,
|
|
"eval_loss": 0.6561428904533386,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -0.80290687084198,
|
|
"eval_rewards/margins": 0.13403938710689545,
|
|
"eval_rewards/rejected": -0.9369462728500366,
|
|
"eval_runtime": 12.8373,
|
|
"eval_samples_per_second": 7.79,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.6039974384037211,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 2.3976293103448275e-06,
|
|
"logits/chosen": -1.7120717763900757,
|
|
"logits/rejected": -1.717246651649475,
|
|
"logps/chosen": -544.954833984375,
|
|
"logps/rejected": -519.8233032226562,
|
|
"loss": 0.6182,
|
|
"rewards/accuracies": 0.6703125238418579,
|
|
"rewards/chosen": -0.6614271402359009,
|
|
"rewards/margins": 0.22156758606433868,
|
|
"rewards/rejected": -0.8829947710037231,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.6147831069466446,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 2.386853448275862e-06,
|
|
"logits/chosen": -1.6905927658081055,
|
|
"logits/rejected": -1.709628701210022,
|
|
"logps/chosen": -529.6423950195312,
|
|
"logps/rejected": -513.8201904296875,
|
|
"loss": 0.626,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.6795625686645508,
|
|
"rewards/margins": 0.20236878097057343,
|
|
"rewards/rejected": -0.8819311857223511,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.6255687754895682,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.3760775862068965e-06,
|
|
"logits/chosen": -1.7218097448349,
|
|
"logits/rejected": -1.7233638763427734,
|
|
"logps/chosen": -566.1005249023438,
|
|
"logps/rejected": -550.3419799804688,
|
|
"loss": 0.6288,
|
|
"rewards/accuracies": 0.675000011920929,
|
|
"rewards/chosen": -0.7431906461715698,
|
|
"rewards/margins": 0.20129895210266113,
|
|
"rewards/rejected": -0.9444894790649414,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.6363544440324919,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 2.365301724137931e-06,
|
|
"logits/chosen": -1.718971848487854,
|
|
"logits/rejected": -1.7108612060546875,
|
|
"logps/chosen": -552.8258056640625,
|
|
"logps/rejected": -534.0645141601562,
|
|
"loss": 0.6376,
|
|
"rewards/accuracies": 0.6421874761581421,
|
|
"rewards/chosen": -0.7128755450248718,
|
|
"rewards/margins": 0.18357399106025696,
|
|
"rewards/rejected": -0.8964495658874512,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.6471401125754154,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 2.3545258620689655e-06,
|
|
"logits/chosen": -1.6846719980239868,
|
|
"logits/rejected": -1.6692850589752197,
|
|
"logps/chosen": -575.1969604492188,
|
|
"logps/rejected": -551.0662841796875,
|
|
"loss": 0.6428,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -0.7172713279724121,
|
|
"rewards/margins": 0.16418126225471497,
|
|
"rewards/rejected": -0.8814526796340942,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.6471401125754154,
|
|
"eval_logits/chosen": -1.7446637153625488,
|
|
"eval_logits/rejected": -1.805701494216919,
|
|
"eval_logps/chosen": -564.8399658203125,
|
|
"eval_logps/rejected": -512.3929443359375,
|
|
"eval_loss": 0.6552779674530029,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -0.8307470083236694,
|
|
"eval_rewards/margins": 0.13837800920009613,
|
|
"eval_rewards/rejected": -0.9691251516342163,
|
|
"eval_runtime": 12.8832,
|
|
"eval_samples_per_second": 7.762,
|
|
"eval_steps_per_second": 1.009,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.657925781118339,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 2.3437500000000002e-06,
|
|
"logits/chosen": -1.7127196788787842,
|
|
"logits/rejected": -1.722394585609436,
|
|
"logps/chosen": -534.408447265625,
|
|
"logps/rejected": -524.6024169921875,
|
|
"loss": 0.6382,
|
|
"rewards/accuracies": 0.6656249761581421,
|
|
"rewards/chosen": -0.7230153679847717,
|
|
"rewards/margins": 0.17467036843299866,
|
|
"rewards/rejected": -0.897685706615448,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.6687114496612626,
|
|
"grad_norm": 19.25,
|
|
"learning_rate": 2.3329741379310345e-06,
|
|
"logits/chosen": -1.731837511062622,
|
|
"logits/rejected": -1.7623554468154907,
|
|
"logps/chosen": -528.728515625,
|
|
"logps/rejected": -500.19500732421875,
|
|
"loss": 0.6294,
|
|
"rewards/accuracies": 0.6499999761581421,
|
|
"rewards/chosen": -0.695734441280365,
|
|
"rewards/margins": 0.19728422164916992,
|
|
"rewards/rejected": -0.8930186033248901,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.6794971182041862,
|
|
"grad_norm": 22.875,
|
|
"learning_rate": 2.322198275862069e-06,
|
|
"logits/chosen": -1.6935323476791382,
|
|
"logits/rejected": -1.7253952026367188,
|
|
"logps/chosen": -557.6375122070312,
|
|
"logps/rejected": -530.9302368164062,
|
|
"loss": 0.6425,
|
|
"rewards/accuracies": 0.6421874761581421,
|
|
"rewards/chosen": -0.7235587239265442,
|
|
"rewards/margins": 0.1692447066307068,
|
|
"rewards/rejected": -0.8928033709526062,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.6902827867471097,
|
|
"grad_norm": 22.75,
|
|
"learning_rate": 2.3114224137931035e-06,
|
|
"logits/chosen": -1.7822033166885376,
|
|
"logits/rejected": -1.7954456806182861,
|
|
"logps/chosen": -548.27294921875,
|
|
"logps/rejected": -529.1452026367188,
|
|
"loss": 0.6357,
|
|
"rewards/accuracies": 0.643750011920929,
|
|
"rewards/chosen": -0.733366847038269,
|
|
"rewards/margins": 0.18996267020702362,
|
|
"rewards/rejected": -0.9233294725418091,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.7010684552900334,
|
|
"grad_norm": 23.625,
|
|
"learning_rate": 2.300646551724138e-06,
|
|
"logits/chosen": -1.7504284381866455,
|
|
"logits/rejected": -1.7501709461212158,
|
|
"logps/chosen": -563.0472412109375,
|
|
"logps/rejected": -547.3125,
|
|
"loss": 0.6376,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -0.7303886413574219,
|
|
"rewards/margins": 0.17333553731441498,
|
|
"rewards/rejected": -0.903724193572998,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.7010684552900334,
|
|
"eval_logits/chosen": -1.7363975048065186,
|
|
"eval_logits/rejected": -1.7971389293670654,
|
|
"eval_logps/chosen": -564.5283813476562,
|
|
"eval_logps/rejected": -512.2813110351562,
|
|
"eval_loss": 0.6501194834709167,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -0.8151662349700928,
|
|
"eval_rewards/margins": 0.1483728587627411,
|
|
"eval_rewards/rejected": -0.9635391235351562,
|
|
"eval_runtime": 12.8651,
|
|
"eval_samples_per_second": 7.773,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.711854123832957,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 2.2898706896551725e-06,
|
|
"logits/chosen": -1.6629375219345093,
|
|
"logits/rejected": -1.686265230178833,
|
|
"logps/chosen": -579.2559814453125,
|
|
"logps/rejected": -550.0233764648438,
|
|
"loss": 0.6339,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -0.7081414461135864,
|
|
"rewards/margins": 0.19901154935359955,
|
|
"rewards/rejected": -0.9071530103683472,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.7226397923758805,
|
|
"grad_norm": 24.25,
|
|
"learning_rate": 2.279094827586207e-06,
|
|
"logits/chosen": -1.772698998451233,
|
|
"logits/rejected": -1.8019428253173828,
|
|
"logps/chosen": -505.48577880859375,
|
|
"logps/rejected": -490.10662841796875,
|
|
"loss": 0.6425,
|
|
"rewards/accuracies": 0.6078125238418579,
|
|
"rewards/chosen": -0.7496585249900818,
|
|
"rewards/margins": 0.17492921650409698,
|
|
"rewards/rejected": -0.9245878458023071,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.7334254609188041,
|
|
"grad_norm": 23.125,
|
|
"learning_rate": 2.2683189655172415e-06,
|
|
"logits/chosen": -1.7226699590682983,
|
|
"logits/rejected": -1.7439358234405518,
|
|
"logps/chosen": -551.814208984375,
|
|
"logps/rejected": -530.10888671875,
|
|
"loss": 0.6147,
|
|
"rewards/accuracies": 0.6703125238418579,
|
|
"rewards/chosen": -0.7403059601783752,
|
|
"rewards/margins": 0.22587330639362335,
|
|
"rewards/rejected": -0.966179370880127,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.7442111294617277,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 2.257543103448276e-06,
|
|
"logits/chosen": -1.7711864709854126,
|
|
"logits/rejected": -1.7668606042861938,
|
|
"logps/chosen": -540.2945556640625,
|
|
"logps/rejected": -520.48828125,
|
|
"loss": 0.6112,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.7231726050376892,
|
|
"rewards/margins": 0.24855723977088928,
|
|
"rewards/rejected": -0.9717298746109009,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.7549967980046514,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 2.24676724137931e-06,
|
|
"logits/chosen": -1.7296479940414429,
|
|
"logits/rejected": -1.7336111068725586,
|
|
"logps/chosen": -525.4596557617188,
|
|
"logps/rejected": -499.53826904296875,
|
|
"loss": 0.644,
|
|
"rewards/accuracies": 0.6578124761581421,
|
|
"rewards/chosen": -0.7867623567581177,
|
|
"rewards/margins": 0.18225276470184326,
|
|
"rewards/rejected": -0.9690152406692505,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.7549967980046514,
|
|
"eval_logits/chosen": -1.7320126295089722,
|
|
"eval_logits/rejected": -1.7923235893249512,
|
|
"eval_logps/chosen": -565.2008056640625,
|
|
"eval_logps/rejected": -512.877197265625,
|
|
"eval_loss": 0.6524822115898132,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -0.8487869501113892,
|
|
"eval_rewards/margins": 0.1445501744747162,
|
|
"eval_rewards/rejected": -0.9933372139930725,
|
|
"eval_runtime": 12.8132,
|
|
"eval_samples_per_second": 7.804,
|
|
"eval_steps_per_second": 1.015,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.7657824665475749,
|
|
"grad_norm": 36.25,
|
|
"learning_rate": 2.235991379310345e-06,
|
|
"logits/chosen": -1.6764347553253174,
|
|
"logits/rejected": -1.6931612491607666,
|
|
"logps/chosen": -555.44091796875,
|
|
"logps/rejected": -529.9513549804688,
|
|
"loss": 0.6177,
|
|
"rewards/accuracies": 0.667187511920929,
|
|
"rewards/chosen": -0.7847446203231812,
|
|
"rewards/margins": 0.23240776360034943,
|
|
"rewards/rejected": -1.0171524286270142,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.7765681350904985,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.225215517241379e-06,
|
|
"logits/chosen": -1.7237228155136108,
|
|
"logits/rejected": -1.7398170232772827,
|
|
"logps/chosen": -542.1912231445312,
|
|
"logps/rejected": -518.4722900390625,
|
|
"loss": 0.6354,
|
|
"rewards/accuracies": 0.6421874761581421,
|
|
"rewards/chosen": -0.7876893281936646,
|
|
"rewards/margins": 0.1978759765625,
|
|
"rewards/rejected": -0.9855653643608093,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.7873538036334221,
|
|
"grad_norm": 23.0,
|
|
"learning_rate": 2.214439655172414e-06,
|
|
"logits/chosen": -1.752096176147461,
|
|
"logits/rejected": -1.7941181659698486,
|
|
"logps/chosen": -548.2041015625,
|
|
"logps/rejected": -513.8216552734375,
|
|
"loss": 0.6555,
|
|
"rewards/accuracies": 0.628125011920929,
|
|
"rewards/chosen": -0.7717311978340149,
|
|
"rewards/margins": 0.14397189021110535,
|
|
"rewards/rejected": -0.9157029986381531,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.7981394721763457,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 2.203663793103448e-06,
|
|
"logits/chosen": -1.709242820739746,
|
|
"logits/rejected": -1.7145198583602905,
|
|
"logps/chosen": -539.0947265625,
|
|
"logps/rejected": -521.4993896484375,
|
|
"loss": 0.6394,
|
|
"rewards/accuracies": 0.6468750238418579,
|
|
"rewards/chosen": -0.7486387491226196,
|
|
"rewards/margins": 0.18467199802398682,
|
|
"rewards/rejected": -0.933310866355896,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.8089251407192692,
|
|
"grad_norm": 23.875,
|
|
"learning_rate": 2.192887931034483e-06,
|
|
"logits/chosen": -1.632927656173706,
|
|
"logits/rejected": -1.6515719890594482,
|
|
"logps/chosen": -559.2658081054688,
|
|
"logps/rejected": -526.0423583984375,
|
|
"loss": 0.6322,
|
|
"rewards/accuracies": 0.653124988079071,
|
|
"rewards/chosen": -0.7207778692245483,
|
|
"rewards/margins": 0.19092032313346863,
|
|
"rewards/rejected": -0.911698043346405,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.8089251407192692,
|
|
"eval_logits/chosen": -1.717968463897705,
|
|
"eval_logits/rejected": -1.7769808769226074,
|
|
"eval_logps/chosen": -564.5967407226562,
|
|
"eval_logps/rejected": -512.3129272460938,
|
|
"eval_loss": 0.6494570970535278,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -0.8185831308364868,
|
|
"eval_rewards/margins": 0.1465369611978531,
|
|
"eval_rewards/rejected": -0.9651200175285339,
|
|
"eval_runtime": 12.8208,
|
|
"eval_samples_per_second": 7.8,
|
|
"eval_steps_per_second": 1.014,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.8197108092621929,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 2.182112068965517e-06,
|
|
"logits/chosen": -1.7085649967193604,
|
|
"logits/rejected": -1.7567522525787354,
|
|
"logps/chosen": -565.7103271484375,
|
|
"logps/rejected": -539.7507934570312,
|
|
"loss": 0.626,
|
|
"rewards/accuracies": 0.675000011920929,
|
|
"rewards/chosen": -0.7772555351257324,
|
|
"rewards/margins": 0.20999836921691895,
|
|
"rewards/rejected": -0.9872539639472961,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.8304964778051165,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 2.1713362068965515e-06,
|
|
"logits/chosen": -1.7073513269424438,
|
|
"logits/rejected": -1.7328789234161377,
|
|
"logps/chosen": -536.3829345703125,
|
|
"logps/rejected": -514.23779296875,
|
|
"loss": 0.6285,
|
|
"rewards/accuracies": 0.6578124761581421,
|
|
"rewards/chosen": -0.7473222613334656,
|
|
"rewards/margins": 0.20041945576667786,
|
|
"rewards/rejected": -0.947741687297821,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 0.84128214634804,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 2.160560344827586e-06,
|
|
"logits/chosen": -1.6973340511322021,
|
|
"logits/rejected": -1.691319465637207,
|
|
"logps/chosen": -557.1148071289062,
|
|
"logps/rejected": -558.3483276367188,
|
|
"loss": 0.6268,
|
|
"rewards/accuracies": 0.6578124761581421,
|
|
"rewards/chosen": -0.7272000312805176,
|
|
"rewards/margins": 0.21510732173919678,
|
|
"rewards/rejected": -0.9423073530197144,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 0.8520678148909636,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 2.1497844827586205e-06,
|
|
"logits/chosen": -1.679863691329956,
|
|
"logits/rejected": -1.6904770135879517,
|
|
"logps/chosen": -543.8391723632812,
|
|
"logps/rejected": -531.5890502929688,
|
|
"loss": 0.625,
|
|
"rewards/accuracies": 0.6640625,
|
|
"rewards/chosen": -0.756318986415863,
|
|
"rewards/margins": 0.20582714676856995,
|
|
"rewards/rejected": -0.9621461629867554,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 0.8628534834338872,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 2.139008620689655e-06,
|
|
"logits/chosen": -1.7281272411346436,
|
|
"logits/rejected": -1.745347261428833,
|
|
"logps/chosen": -547.1482543945312,
|
|
"logps/rejected": -527.178466796875,
|
|
"loss": 0.6278,
|
|
"rewards/accuracies": 0.671875,
|
|
"rewards/chosen": -0.7604348063468933,
|
|
"rewards/margins": 0.19645504653453827,
|
|
"rewards/rejected": -0.9568899273872375,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.8628534834338872,
|
|
"eval_logits/chosen": -1.7295496463775635,
|
|
"eval_logits/rejected": -1.7896690368652344,
|
|
"eval_logps/chosen": -565.3712158203125,
|
|
"eval_logps/rejected": -513.1586303710938,
|
|
"eval_loss": 0.6495404243469238,
|
|
"eval_rewards/accuracies": 0.6730769276618958,
|
|
"eval_rewards/chosen": -0.8573046922683716,
|
|
"eval_rewards/margins": 0.15010415017604828,
|
|
"eval_rewards/rejected": -1.0074087381362915,
|
|
"eval_runtime": 12.8304,
|
|
"eval_samples_per_second": 7.794,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.8736391519768109,
|
|
"grad_norm": 24.0,
|
|
"learning_rate": 2.1282327586206895e-06,
|
|
"logits/chosen": -1.6808828115463257,
|
|
"logits/rejected": -1.7251535654067993,
|
|
"logps/chosen": -545.0057373046875,
|
|
"logps/rejected": -514.2200317382812,
|
|
"loss": 0.6473,
|
|
"rewards/accuracies": 0.6499999761581421,
|
|
"rewards/chosen": -0.7858768701553345,
|
|
"rewards/margins": 0.16710534691810608,
|
|
"rewards/rejected": -0.9529821276664734,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 0.8844248205197344,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 2.117456896551724e-06,
|
|
"logits/chosen": -1.7713664770126343,
|
|
"logits/rejected": -1.7678531408309937,
|
|
"logps/chosen": -544.8568115234375,
|
|
"logps/rejected": -516.0807495117188,
|
|
"loss": 0.6342,
|
|
"rewards/accuracies": 0.643750011920929,
|
|
"rewards/chosen": -0.8098779916763306,
|
|
"rewards/margins": 0.199488565325737,
|
|
"rewards/rejected": -1.009366512298584,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 0.895210489062658,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 2.106681034482759e-06,
|
|
"logits/chosen": -1.7336772680282593,
|
|
"logits/rejected": -1.7515900135040283,
|
|
"logps/chosen": -549.8850708007812,
|
|
"logps/rejected": -530.6832275390625,
|
|
"loss": 0.6074,
|
|
"rewards/accuracies": 0.703125,
|
|
"rewards/chosen": -0.7838916778564453,
|
|
"rewards/margins": 0.2659648060798645,
|
|
"rewards/rejected": -1.049856424331665,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 0.9059961576055816,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.0959051724137932e-06,
|
|
"logits/chosen": -1.7343952655792236,
|
|
"logits/rejected": -1.7316901683807373,
|
|
"logps/chosen": -574.8560791015625,
|
|
"logps/rejected": -536.2384033203125,
|
|
"loss": 0.6342,
|
|
"rewards/accuracies": 0.651562511920929,
|
|
"rewards/chosen": -0.7904583811759949,
|
|
"rewards/margins": 0.1979624629020691,
|
|
"rewards/rejected": -0.988420844078064,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 0.9167818261485051,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 2.085129310344828e-06,
|
|
"logits/chosen": -1.733007788658142,
|
|
"logits/rejected": -1.727246880531311,
|
|
"logps/chosen": -524.6848754882812,
|
|
"logps/rejected": -510.7845764160156,
|
|
"loss": 0.6307,
|
|
"rewards/accuracies": 0.6421874761581421,
|
|
"rewards/chosen": -0.8132136464118958,
|
|
"rewards/margins": 0.20207151770591736,
|
|
"rewards/rejected": -1.0152852535247803,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 0.9167818261485051,
|
|
"eval_logits/chosen": -1.724961757659912,
|
|
"eval_logits/rejected": -1.7844202518463135,
|
|
"eval_logps/chosen": -565.8350830078125,
|
|
"eval_logps/rejected": -513.8818969726562,
|
|
"eval_loss": 0.6429941058158875,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -0.8805010318756104,
|
|
"eval_rewards/margins": 0.16307112574577332,
|
|
"eval_rewards/rejected": -1.043572187423706,
|
|
"eval_runtime": 12.8678,
|
|
"eval_samples_per_second": 7.771,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 0.9275674946914287,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 2.0743534482758622e-06,
|
|
"logits/chosen": -1.678770661354065,
|
|
"logits/rejected": -1.6889280080795288,
|
|
"logps/chosen": -549.0493774414062,
|
|
"logps/rejected": -529.0235595703125,
|
|
"loss": 0.6311,
|
|
"rewards/accuracies": 0.6578124761581421,
|
|
"rewards/chosen": -0.8022142648696899,
|
|
"rewards/margins": 0.21385040879249573,
|
|
"rewards/rejected": -1.0160646438598633,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 0.9383531632343524,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 2.063577586206897e-06,
|
|
"logits/chosen": -1.677046775817871,
|
|
"logits/rejected": -1.715746521949768,
|
|
"logps/chosen": -558.4188842773438,
|
|
"logps/rejected": -526.5531616210938,
|
|
"loss": 0.6388,
|
|
"rewards/accuracies": 0.6499999761581421,
|
|
"rewards/chosen": -0.7892870903015137,
|
|
"rewards/margins": 0.1829388290643692,
|
|
"rewards/rejected": -0.9722259640693665,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 0.949138831777276,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 2.0528017241379312e-06,
|
|
"logits/chosen": -1.71084463596344,
|
|
"logits/rejected": -1.7499706745147705,
|
|
"logps/chosen": -547.9580078125,
|
|
"logps/rejected": -525.6466674804688,
|
|
"loss": 0.6215,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -0.7694340348243713,
|
|
"rewards/margins": 0.2342531681060791,
|
|
"rewards/rejected": -1.0036872625350952,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 0.9599245003201995,
|
|
"grad_norm": 22.5,
|
|
"learning_rate": 2.0420258620689655e-06,
|
|
"logits/chosen": -1.7067375183105469,
|
|
"logits/rejected": -1.7360172271728516,
|
|
"logps/chosen": -553.2872924804688,
|
|
"logps/rejected": -533.8543090820312,
|
|
"loss": 0.6409,
|
|
"rewards/accuracies": 0.6390625238418579,
|
|
"rewards/chosen": -0.7895435690879822,
|
|
"rewards/margins": 0.18897457420825958,
|
|
"rewards/rejected": -0.9785181879997253,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 0.9707101688631231,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 2.0312500000000002e-06,
|
|
"logits/chosen": -1.7305114269256592,
|
|
"logits/rejected": -1.733814001083374,
|
|
"logps/chosen": -544.6137084960938,
|
|
"logps/rejected": -523.6353149414062,
|
|
"loss": 0.6191,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -0.786180317401886,
|
|
"rewards/margins": 0.22723452746868134,
|
|
"rewards/rejected": -1.013414978981018,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.9707101688631231,
|
|
"eval_logits/chosen": -1.7138627767562866,
|
|
"eval_logits/rejected": -1.773409128189087,
|
|
"eval_logps/chosen": -565.8074340820312,
|
|
"eval_logps/rejected": -513.8848266601562,
|
|
"eval_loss": 0.6439433097839355,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -0.8791185617446899,
|
|
"eval_rewards/margins": 0.16459915041923523,
|
|
"eval_rewards/rejected": -1.043717622756958,
|
|
"eval_runtime": 12.8601,
|
|
"eval_samples_per_second": 7.776,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.9814958374060467,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 2.0204741379310345e-06,
|
|
"logits/chosen": -1.7137806415557861,
|
|
"logits/rejected": -1.714971899986267,
|
|
"logps/chosen": -546.196533203125,
|
|
"logps/rejected": -519.9130859375,
|
|
"loss": 0.6336,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -0.7919280529022217,
|
|
"rewards/margins": 0.189212828874588,
|
|
"rewards/rejected": -0.9811409115791321,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 0.9922815059489704,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.0096982758620693e-06,
|
|
"logits/chosen": -1.7518551349639893,
|
|
"logits/rejected": -1.789482831954956,
|
|
"logps/chosen": -539.8787841796875,
|
|
"logps/rejected": -518.9136962890625,
|
|
"loss": 0.6408,
|
|
"rewards/accuracies": 0.6187499761581421,
|
|
"rewards/chosen": -0.8202158808708191,
|
|
"rewards/margins": 0.1931016743183136,
|
|
"rewards/rejected": -1.0133177042007446,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 1.0021571337085846,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 1.9989224137931036e-06,
|
|
"logits/chosen": -1.667576789855957,
|
|
"logits/rejected": -1.6723417043685913,
|
|
"logps/chosen": -560.371826171875,
|
|
"logps/rejected": -535.334228515625,
|
|
"loss": 0.6217,
|
|
"rewards/accuracies": 0.6416382193565369,
|
|
"rewards/chosen": -0.7891966700553894,
|
|
"rewards/margins": 0.2126684933900833,
|
|
"rewards/rejected": -1.0018651485443115,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 1.0129428022515083,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.9881465517241383e-06,
|
|
"logits/chosen": -1.6589832305908203,
|
|
"logits/rejected": -1.6481475830078125,
|
|
"logps/chosen": -539.683837890625,
|
|
"logps/rejected": -524.0894165039062,
|
|
"loss": 0.6041,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -0.7246257662773132,
|
|
"rewards/margins": 0.25880277156829834,
|
|
"rewards/rejected": -0.9834285974502563,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 1.0237284707944319,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 1.9773706896551726e-06,
|
|
"logits/chosen": -1.7167028188705444,
|
|
"logits/rejected": -1.722948431968689,
|
|
"logps/chosen": -543.1702270507812,
|
|
"logps/rejected": -523.4526977539062,
|
|
"loss": 0.6061,
|
|
"rewards/accuracies": 0.6859375238418579,
|
|
"rewards/chosen": -0.7710289359092712,
|
|
"rewards/margins": 0.25660115480422974,
|
|
"rewards/rejected": -1.027630090713501,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.0237284707944319,
|
|
"eval_logits/chosen": -1.7156367301940918,
|
|
"eval_logits/rejected": -1.7743936777114868,
|
|
"eval_logps/chosen": -566.1229248046875,
|
|
"eval_logps/rejected": -514.0870971679688,
|
|
"eval_loss": 0.6457732319831848,
|
|
"eval_rewards/accuracies": 0.6730769276618958,
|
|
"eval_rewards/chosen": -0.8948926329612732,
|
|
"eval_rewards/margins": 0.15893827378749847,
|
|
"eval_rewards/rejected": -1.053830862045288,
|
|
"eval_runtime": 12.8266,
|
|
"eval_samples_per_second": 7.796,
|
|
"eval_steps_per_second": 1.014,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.0345141393373556,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.966594827586207e-06,
|
|
"logits/chosen": -1.7188589572906494,
|
|
"logits/rejected": -1.7054466009140015,
|
|
"logps/chosen": -549.7354736328125,
|
|
"logps/rejected": -536.328125,
|
|
"loss": 0.6156,
|
|
"rewards/accuracies": 0.684374988079071,
|
|
"rewards/chosen": -0.779514729976654,
|
|
"rewards/margins": 0.24889163672924042,
|
|
"rewards/rejected": -1.0284063816070557,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 1.045299807880279,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 1.9558189655172416e-06,
|
|
"logits/chosen": -1.6784546375274658,
|
|
"logits/rejected": -1.6850860118865967,
|
|
"logps/chosen": -554.5667724609375,
|
|
"logps/rejected": -547.0436401367188,
|
|
"loss": 0.5838,
|
|
"rewards/accuracies": 0.753125011920929,
|
|
"rewards/chosen": -0.7458912134170532,
|
|
"rewards/margins": 0.3095359802246094,
|
|
"rewards/rejected": -1.0554273128509521,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 1.0560854764232026,
|
|
"grad_norm": 19.25,
|
|
"learning_rate": 1.945043103448276e-06,
|
|
"logits/chosen": -1.7073291540145874,
|
|
"logits/rejected": -1.7111352682113647,
|
|
"logps/chosen": -538.4148559570312,
|
|
"logps/rejected": -526.5668334960938,
|
|
"loss": 0.6054,
|
|
"rewards/accuracies": 0.6890624761581421,
|
|
"rewards/chosen": -0.7842422127723694,
|
|
"rewards/margins": 0.26182490587234497,
|
|
"rewards/rejected": -1.0460671186447144,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 1.0668711449661263,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 1.9342672413793106e-06,
|
|
"logits/chosen": -1.6965501308441162,
|
|
"logits/rejected": -1.7186329364776611,
|
|
"logps/chosen": -569.547119140625,
|
|
"logps/rejected": -548.1654052734375,
|
|
"loss": 0.5954,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.7711578607559204,
|
|
"rewards/margins": 0.2901317775249481,
|
|
"rewards/rejected": -1.0612895488739014,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 1.0776568135090498,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 1.923491379310345e-06,
|
|
"logits/chosen": -1.7303965091705322,
|
|
"logits/rejected": -1.7479015588760376,
|
|
"logps/chosen": -535.7818603515625,
|
|
"logps/rejected": -525.8051147460938,
|
|
"loss": 0.6085,
|
|
"rewards/accuracies": 0.671875,
|
|
"rewards/chosen": -0.8089873194694519,
|
|
"rewards/margins": 0.2565574049949646,
|
|
"rewards/rejected": -1.065544605255127,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.0776568135090498,
|
|
"eval_logits/chosen": -1.7369959354400635,
|
|
"eval_logits/rejected": -1.796852946281433,
|
|
"eval_logps/chosen": -566.9774169921875,
|
|
"eval_logps/rejected": -515.3145751953125,
|
|
"eval_loss": 0.6404527425765991,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -0.9376153349876404,
|
|
"eval_rewards/margins": 0.1775885373353958,
|
|
"eval_rewards/rejected": -1.115203857421875,
|
|
"eval_runtime": 12.8545,
|
|
"eval_samples_per_second": 7.779,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.0884424820519734,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 1.9127155172413796e-06,
|
|
"logits/chosen": -1.7162755727767944,
|
|
"logits/rejected": -1.7419646978378296,
|
|
"logps/chosen": -533.6424560546875,
|
|
"logps/rejected": -518.5228881835938,
|
|
"loss": 0.5936,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -0.7938119173049927,
|
|
"rewards/margins": 0.28399401903152466,
|
|
"rewards/rejected": -1.0778058767318726,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 1.099228150594897,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 1.9019396551724139e-06,
|
|
"logits/chosen": -1.6942980289459229,
|
|
"logits/rejected": -1.707862138748169,
|
|
"logps/chosen": -572.4893798828125,
|
|
"logps/rejected": -546.4171142578125,
|
|
"loss": 0.5999,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.8284515142440796,
|
|
"rewards/margins": 0.2772662043571472,
|
|
"rewards/rejected": -1.1057178974151611,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 1.1100138191378206,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 1.8911637931034484e-06,
|
|
"logits/chosen": -1.670231580734253,
|
|
"logits/rejected": -1.6834871768951416,
|
|
"logps/chosen": -534.8599243164062,
|
|
"logps/rejected": -513.09375,
|
|
"loss": 0.618,
|
|
"rewards/accuracies": 0.645312488079071,
|
|
"rewards/chosen": -0.8424933552742004,
|
|
"rewards/margins": 0.24433858692646027,
|
|
"rewards/rejected": -1.08683180809021,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 1.120799487680744,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 1.880387931034483e-06,
|
|
"logits/chosen": -1.6783840656280518,
|
|
"logits/rejected": -1.6901963949203491,
|
|
"logps/chosen": -579.30322265625,
|
|
"logps/rejected": -550.1730346679688,
|
|
"loss": 0.5868,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -0.7912231087684631,
|
|
"rewards/margins": 0.30552273988723755,
|
|
"rewards/rejected": -1.0967457294464111,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 1.1315851562236678,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 1.8696120689655172e-06,
|
|
"logits/chosen": -1.7412052154541016,
|
|
"logits/rejected": -1.7302591800689697,
|
|
"logps/chosen": -552.0925903320312,
|
|
"logps/rejected": -544.4852294921875,
|
|
"loss": 0.5885,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -0.780128002166748,
|
|
"rewards/margins": 0.3037031292915344,
|
|
"rewards/rejected": -1.0838311910629272,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 1.1315851562236678,
|
|
"eval_logits/chosen": -1.7486636638641357,
|
|
"eval_logits/rejected": -1.8092385530471802,
|
|
"eval_logps/chosen": -567.4676513671875,
|
|
"eval_logps/rejected": -515.6458740234375,
|
|
"eval_loss": 0.6456284523010254,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -0.9621263742446899,
|
|
"eval_rewards/margins": 0.16964147984981537,
|
|
"eval_rewards/rejected": -1.1317678689956665,
|
|
"eval_runtime": 12.8512,
|
|
"eval_samples_per_second": 7.781,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 1.1423708247665914,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 1.8588362068965517e-06,
|
|
"logits/chosen": -1.6765000820159912,
|
|
"logits/rejected": -1.6997390985488892,
|
|
"logps/chosen": -557.3540649414062,
|
|
"logps/rejected": -541.9246826171875,
|
|
"loss": 0.5946,
|
|
"rewards/accuracies": 0.7328125238418579,
|
|
"rewards/chosen": -0.7725203037261963,
|
|
"rewards/margins": 0.28744369745254517,
|
|
"rewards/rejected": -1.0599639415740967,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 1.153156493309515,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 1.8480603448275862e-06,
|
|
"logits/chosen": -1.7148462533950806,
|
|
"logits/rejected": -1.6918646097183228,
|
|
"logps/chosen": -542.8770751953125,
|
|
"logps/rejected": -535.9141845703125,
|
|
"loss": 0.6051,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -0.818956196308136,
|
|
"rewards/margins": 0.269645631313324,
|
|
"rewards/rejected": -1.08860182762146,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 1.1639421618524386,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.8372844827586207e-06,
|
|
"logits/chosen": -1.736419677734375,
|
|
"logits/rejected": -1.751819372177124,
|
|
"logps/chosen": -533.5783081054688,
|
|
"logps/rejected": -524.98046875,
|
|
"loss": 0.5886,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.8092219233512878,
|
|
"rewards/margins": 0.3041132390499115,
|
|
"rewards/rejected": -1.1133352518081665,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 1.174727830395362,
|
|
"grad_norm": 24.875,
|
|
"learning_rate": 1.8265086206896552e-06,
|
|
"logits/chosen": -1.794930100440979,
|
|
"logits/rejected": -1.8180984258651733,
|
|
"logps/chosen": -542.23583984375,
|
|
"logps/rejected": -527.9422607421875,
|
|
"loss": 0.5897,
|
|
"rewards/accuracies": 0.7093750238418579,
|
|
"rewards/chosen": -0.8398362994194031,
|
|
"rewards/margins": 0.3197947144508362,
|
|
"rewards/rejected": -1.1596310138702393,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 1.1855134989382858,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 1.8157327586206897e-06,
|
|
"logits/chosen": -1.7432777881622314,
|
|
"logits/rejected": -1.7690292596817017,
|
|
"logps/chosen": -536.7589721679688,
|
|
"logps/rejected": -513.2578735351562,
|
|
"loss": 0.5885,
|
|
"rewards/accuracies": 0.699999988079071,
|
|
"rewards/chosen": -0.8450536727905273,
|
|
"rewards/margins": 0.3091221749782562,
|
|
"rewards/rejected": -1.154175877571106,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 1.1855134989382858,
|
|
"eval_logits/chosen": -1.764207124710083,
|
|
"eval_logits/rejected": -1.8264764547348022,
|
|
"eval_logps/chosen": -568.4033203125,
|
|
"eval_logps/rejected": -516.8681030273438,
|
|
"eval_loss": 0.6412051320075989,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -1.0089114904403687,
|
|
"eval_rewards/margins": 0.18396909534931183,
|
|
"eval_rewards/rejected": -1.192880630493164,
|
|
"eval_runtime": 12.9704,
|
|
"eval_samples_per_second": 7.71,
|
|
"eval_steps_per_second": 1.002,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 1.1962991674812093,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 1.8049568965517242e-06,
|
|
"logits/chosen": -1.739747405052185,
|
|
"logits/rejected": -1.7718337774276733,
|
|
"logps/chosen": -540.0535888671875,
|
|
"logps/rejected": -523.0402221679688,
|
|
"loss": 0.5792,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -0.8350652456283569,
|
|
"rewards/margins": 0.3389832377433777,
|
|
"rewards/rejected": -1.1740485429763794,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 1.2070848360241329,
|
|
"grad_norm": 23.75,
|
|
"learning_rate": 1.7941810344827585e-06,
|
|
"logits/chosen": -1.7355035543441772,
|
|
"logits/rejected": -1.740517258644104,
|
|
"logps/chosen": -553.8098754882812,
|
|
"logps/rejected": -549.6728515625,
|
|
"loss": 0.5757,
|
|
"rewards/accuracies": 0.760937511920929,
|
|
"rewards/chosen": -0.8573840856552124,
|
|
"rewards/margins": 0.35040709376335144,
|
|
"rewards/rejected": -1.2077910900115967,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 1.2178705045670566,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 1.783405172413793e-06,
|
|
"logits/chosen": -1.743051528930664,
|
|
"logits/rejected": -1.7518497705459595,
|
|
"logps/chosen": -539.1353759765625,
|
|
"logps/rejected": -531.3056640625,
|
|
"loss": 0.5989,
|
|
"rewards/accuracies": 0.7203124761581421,
|
|
"rewards/chosen": -0.8813650012016296,
|
|
"rewards/margins": 0.28817206621170044,
|
|
"rewards/rejected": -1.16953706741333,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 1.22865617310998,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.7726293103448275e-06,
|
|
"logits/chosen": -1.780400037765503,
|
|
"logits/rejected": -1.797412633895874,
|
|
"logps/chosen": -538.9906005859375,
|
|
"logps/rejected": -524.0303955078125,
|
|
"loss": 0.6014,
|
|
"rewards/accuracies": 0.7093750238418579,
|
|
"rewards/chosen": -0.8999722599983215,
|
|
"rewards/margins": 0.29530972242355347,
|
|
"rewards/rejected": -1.1952821016311646,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 1.2394418416529036,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.761853448275862e-06,
|
|
"logits/chosen": -1.7877986431121826,
|
|
"logits/rejected": -1.7923310995101929,
|
|
"logps/chosen": -525.5618286132812,
|
|
"logps/rejected": -501.60028076171875,
|
|
"loss": 0.591,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -0.9110702276229858,
|
|
"rewards/margins": 0.30747610330581665,
|
|
"rewards/rejected": -1.2185462713241577,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 1.2394418416529036,
|
|
"eval_logits/chosen": -1.7822145223617554,
|
|
"eval_logits/rejected": -1.844849705696106,
|
|
"eval_logps/chosen": -569.3731689453125,
|
|
"eval_logps/rejected": -517.9542846679688,
|
|
"eval_loss": 0.6419017314910889,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -1.0574066638946533,
|
|
"eval_rewards/margins": 0.18978391587734222,
|
|
"eval_rewards/rejected": -1.2471905946731567,
|
|
"eval_runtime": 12.8369,
|
|
"eval_samples_per_second": 7.79,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 1.2502275101958273,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 1.7510775862068965e-06,
|
|
"logits/chosen": -1.7190322875976562,
|
|
"logits/rejected": -1.730328917503357,
|
|
"logps/chosen": -524.3182373046875,
|
|
"logps/rejected": -510.26251220703125,
|
|
"loss": 0.5975,
|
|
"rewards/accuracies": 0.682812511920929,
|
|
"rewards/chosen": -0.8995493650436401,
|
|
"rewards/margins": 0.28964298963546753,
|
|
"rewards/rejected": -1.189192295074463,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 1.2610131787387509,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 1.740301724137931e-06,
|
|
"logits/chosen": -1.7682945728302002,
|
|
"logits/rejected": -1.7626434564590454,
|
|
"logps/chosen": -545.8931884765625,
|
|
"logps/rejected": -537.7764892578125,
|
|
"loss": 0.584,
|
|
"rewards/accuracies": 0.7265625,
|
|
"rewards/chosen": -0.8533053398132324,
|
|
"rewards/margins": 0.31390124559402466,
|
|
"rewards/rejected": -1.1672066450119019,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 1.2717988472816746,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 1.7295258620689656e-06,
|
|
"logits/chosen": -1.7791554927825928,
|
|
"logits/rejected": -1.8204240798950195,
|
|
"logps/chosen": -569.207275390625,
|
|
"logps/rejected": -549.9439697265625,
|
|
"loss": 0.6009,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.8925979733467102,
|
|
"rewards/margins": 0.3236784338951111,
|
|
"rewards/rejected": -1.2162764072418213,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 1.282584515824598,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.7187499999999998e-06,
|
|
"logits/chosen": -1.7999378442764282,
|
|
"logits/rejected": -1.8201545476913452,
|
|
"logps/chosen": -541.1659545898438,
|
|
"logps/rejected": -517.34765625,
|
|
"loss": 0.5867,
|
|
"rewards/accuracies": 0.729687511920929,
|
|
"rewards/chosen": -0.9191481471061707,
|
|
"rewards/margins": 0.32020530104637146,
|
|
"rewards/rejected": -1.2393534183502197,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 1.2933701843675216,
|
|
"grad_norm": 24.125,
|
|
"learning_rate": 1.7079741379310344e-06,
|
|
"logits/chosen": -1.7037147283554077,
|
|
"logits/rejected": -1.7427594661712646,
|
|
"logps/chosen": -560.2366943359375,
|
|
"logps/rejected": -532.6011962890625,
|
|
"loss": 0.599,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -0.9538860321044922,
|
|
"rewards/margins": 0.2922009825706482,
|
|
"rewards/rejected": -1.2460869550704956,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 1.2933701843675216,
|
|
"eval_logits/chosen": -1.790303349494934,
|
|
"eval_logits/rejected": -1.853431224822998,
|
|
"eval_logps/chosen": -569.9891967773438,
|
|
"eval_logps/rejected": -518.47216796875,
|
|
"eval_loss": 0.6446880102157593,
|
|
"eval_rewards/accuracies": 0.682692289352417,
|
|
"eval_rewards/chosen": -1.0882103443145752,
|
|
"eval_rewards/margins": 0.18487460911273956,
|
|
"eval_rewards/rejected": -1.273085117340088,
|
|
"eval_runtime": 12.8511,
|
|
"eval_samples_per_second": 7.781,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 1.3041558529104451,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.6971982758620689e-06,
|
|
"logits/chosen": -1.737642526626587,
|
|
"logits/rejected": -1.7478208541870117,
|
|
"logps/chosen": -555.0159301757812,
|
|
"logps/rejected": -539.2918701171875,
|
|
"loss": 0.6072,
|
|
"rewards/accuracies": 0.6703125238418579,
|
|
"rewards/chosen": -0.9134560823440552,
|
|
"rewards/margins": 0.280630499124527,
|
|
"rewards/rejected": -1.1940864324569702,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 1.3149415214533688,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 1.6864224137931034e-06,
|
|
"logits/chosen": -1.7710367441177368,
|
|
"logits/rejected": -1.7797397375106812,
|
|
"logps/chosen": -581.2069702148438,
|
|
"logps/rejected": -565.8582153320312,
|
|
"loss": 0.5837,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -0.9405378103256226,
|
|
"rewards/margins": 0.3353542387485504,
|
|
"rewards/rejected": -1.2758920192718506,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 1.3257271899962924,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.6756465517241379e-06,
|
|
"logits/chosen": -1.7992254495620728,
|
|
"logits/rejected": -1.8066284656524658,
|
|
"logps/chosen": -534.38037109375,
|
|
"logps/rejected": -526.4951782226562,
|
|
"loss": 0.6098,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.9164705276489258,
|
|
"rewards/margins": 0.28371772170066833,
|
|
"rewards/rejected": -1.200188398361206,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 1.336512858539216,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.6648706896551724e-06,
|
|
"logits/chosen": -1.718231201171875,
|
|
"logits/rejected": -1.7256206274032593,
|
|
"logps/chosen": -565.9593505859375,
|
|
"logps/rejected": -551.340576171875,
|
|
"loss": 0.5799,
|
|
"rewards/accuracies": 0.7250000238418579,
|
|
"rewards/chosen": -0.9390344619750977,
|
|
"rewards/margins": 0.33844050765037537,
|
|
"rewards/rejected": -1.2774749994277954,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 1.3472985270821396,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 1.6540948275862069e-06,
|
|
"logits/chosen": -1.8084981441497803,
|
|
"logits/rejected": -1.8036025762557983,
|
|
"logps/chosen": -536.2025756835938,
|
|
"logps/rejected": -532.8880615234375,
|
|
"loss": 0.5941,
|
|
"rewards/accuracies": 0.698437511920929,
|
|
"rewards/chosen": -0.9396790266036987,
|
|
"rewards/margins": 0.3215208649635315,
|
|
"rewards/rejected": -1.261199712753296,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 1.3472985270821396,
|
|
"eval_logits/chosen": -1.7861320972442627,
|
|
"eval_logits/rejected": -1.8498179912567139,
|
|
"eval_logps/chosen": -569.7782592773438,
|
|
"eval_logps/rejected": -518.2344970703125,
|
|
"eval_loss": 0.6447553038597107,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.0776572227478027,
|
|
"eval_rewards/margins": 0.1835414469242096,
|
|
"eval_rewards/rejected": -1.2611987590789795,
|
|
"eval_runtime": 12.8276,
|
|
"eval_samples_per_second": 7.796,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 1.358084195625063,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 1.6433189655172412e-06,
|
|
"logits/chosen": -1.674708604812622,
|
|
"logits/rejected": -1.6811631917953491,
|
|
"logps/chosen": -543.1943359375,
|
|
"logps/rejected": -522.18505859375,
|
|
"loss": 0.5817,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -0.8475159406661987,
|
|
"rewards/margins": 0.33499693870544434,
|
|
"rewards/rejected": -1.182512879371643,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 1.3688698641679868,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 1.6325431034482761e-06,
|
|
"logits/chosen": -1.7509996891021729,
|
|
"logits/rejected": -1.7384611368179321,
|
|
"logps/chosen": -561.0062255859375,
|
|
"logps/rejected": -544.8842163085938,
|
|
"loss": 0.5948,
|
|
"rewards/accuracies": 0.737500011920929,
|
|
"rewards/chosen": -0.8584550619125366,
|
|
"rewards/margins": 0.2931164801120758,
|
|
"rewards/rejected": -1.15157151222229,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 1.3796555327109103,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 1.6217672413793106e-06,
|
|
"logits/chosen": -1.7544962167739868,
|
|
"logits/rejected": -1.7836391925811768,
|
|
"logps/chosen": -558.3255615234375,
|
|
"logps/rejected": -549.2784423828125,
|
|
"loss": 0.594,
|
|
"rewards/accuracies": 0.699999988079071,
|
|
"rewards/chosen": -0.887876033782959,
|
|
"rewards/margins": 0.2937227785587311,
|
|
"rewards/rejected": -1.1815989017486572,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 1.390441201253834,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 1.6109913793103451e-06,
|
|
"logits/chosen": -1.766278862953186,
|
|
"logits/rejected": -1.765802025794983,
|
|
"logps/chosen": -527.0892333984375,
|
|
"logps/rejected": -512.3470458984375,
|
|
"loss": 0.5855,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -0.9098323583602905,
|
|
"rewards/margins": 0.3318841755390167,
|
|
"rewards/rejected": -1.2417166233062744,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 1.4012268697967576,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 1.6002155172413794e-06,
|
|
"logits/chosen": -1.767046332359314,
|
|
"logits/rejected": -1.7728359699249268,
|
|
"logps/chosen": -531.3649291992188,
|
|
"logps/rejected": -521.9031982421875,
|
|
"loss": 0.5806,
|
|
"rewards/accuracies": 0.699999988079071,
|
|
"rewards/chosen": -0.9428930282592773,
|
|
"rewards/margins": 0.34324660897254944,
|
|
"rewards/rejected": -1.2861396074295044,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 1.4012268697967576,
|
|
"eval_logits/chosen": -1.7951500415802002,
|
|
"eval_logits/rejected": -1.859129786491394,
|
|
"eval_logps/chosen": -570.1704711914062,
|
|
"eval_logps/rejected": -518.6593017578125,
|
|
"eval_loss": 0.645327091217041,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -1.0972670316696167,
|
|
"eval_rewards/margins": 0.18517184257507324,
|
|
"eval_rewards/rejected": -1.2824387550354004,
|
|
"eval_runtime": 12.8396,
|
|
"eval_samples_per_second": 7.788,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 1.412012538339681,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 1.589439655172414e-06,
|
|
"logits/chosen": -1.7911393642425537,
|
|
"logits/rejected": -1.798229455947876,
|
|
"logps/chosen": -542.2491455078125,
|
|
"logps/rejected": -538.056640625,
|
|
"loss": 0.5839,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -0.9245232343673706,
|
|
"rewards/margins": 0.33389586210250854,
|
|
"rewards/rejected": -1.2584190368652344,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 1.4227982068826046,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 1.5786637931034484e-06,
|
|
"logits/chosen": -1.773233413696289,
|
|
"logits/rejected": -1.7785921096801758,
|
|
"logps/chosen": -529.9383544921875,
|
|
"logps/rejected": -518.8634643554688,
|
|
"loss": 0.6115,
|
|
"rewards/accuracies": 0.684374988079071,
|
|
"rewards/chosen": -0.9675337672233582,
|
|
"rewards/margins": 0.279288113117218,
|
|
"rewards/rejected": -1.2468219995498657,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 1.4335838754255283,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 1.567887931034483e-06,
|
|
"logits/chosen": -1.7141765356063843,
|
|
"logits/rejected": -1.7172152996063232,
|
|
"logps/chosen": -541.616943359375,
|
|
"logps/rejected": -534.3272705078125,
|
|
"loss": 0.5887,
|
|
"rewards/accuracies": 0.729687511920929,
|
|
"rewards/chosen": -0.8925272226333618,
|
|
"rewards/margins": 0.3137276768684387,
|
|
"rewards/rejected": -1.2062549591064453,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 1.4443695439684519,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 1.5571120689655174e-06,
|
|
"logits/chosen": -1.7413345575332642,
|
|
"logits/rejected": -1.762304663658142,
|
|
"logps/chosen": -568.6903076171875,
|
|
"logps/rejected": -551.1282348632812,
|
|
"loss": 0.5889,
|
|
"rewards/accuracies": 0.7265625,
|
|
"rewards/chosen": -0.9233890771865845,
|
|
"rewards/margins": 0.32237863540649414,
|
|
"rewards/rejected": -1.2457677125930786,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 1.4551552125113756,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 1.546336206896552e-06,
|
|
"logits/chosen": -1.7776886224746704,
|
|
"logits/rejected": -1.7672054767608643,
|
|
"logps/chosen": -537.7005004882812,
|
|
"logps/rejected": -535.3060302734375,
|
|
"loss": 0.6075,
|
|
"rewards/accuracies": 0.690625011920929,
|
|
"rewards/chosen": -0.9527568817138672,
|
|
"rewards/margins": 0.28692346811294556,
|
|
"rewards/rejected": -1.239680528640747,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 1.4551552125113756,
|
|
"eval_logits/chosen": -1.799996018409729,
|
|
"eval_logits/rejected": -1.8631552457809448,
|
|
"eval_logps/chosen": -570.1870727539062,
|
|
"eval_logps/rejected": -518.773681640625,
|
|
"eval_loss": 0.6434743404388428,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.0980980396270752,
|
|
"eval_rewards/margins": 0.19006212055683136,
|
|
"eval_rewards/rejected": -1.2881600856781006,
|
|
"eval_runtime": 12.8518,
|
|
"eval_samples_per_second": 7.781,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 1.465940881054299,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 1.5355603448275862e-06,
|
|
"logits/chosen": -1.7860710620880127,
|
|
"logits/rejected": -1.7912073135375977,
|
|
"logps/chosen": -555.429931640625,
|
|
"logps/rejected": -541.3480224609375,
|
|
"loss": 0.5877,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -0.9607505798339844,
|
|
"rewards/margins": 0.33639881014823914,
|
|
"rewards/rejected": -1.297149419784546,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 1.4767265495972226,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.5247844827586207e-06,
|
|
"logits/chosen": -1.7754265069961548,
|
|
"logits/rejected": -1.7940123081207275,
|
|
"logps/chosen": -540.836181640625,
|
|
"logps/rejected": -531.376953125,
|
|
"loss": 0.5871,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.8988661766052246,
|
|
"rewards/margins": 0.32094043493270874,
|
|
"rewards/rejected": -1.2198066711425781,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 1.4875122181401463,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 1.5140086206896552e-06,
|
|
"logits/chosen": -1.7347028255462646,
|
|
"logits/rejected": -1.7507202625274658,
|
|
"logps/chosen": -550.8851928710938,
|
|
"logps/rejected": -533.6559448242188,
|
|
"loss": 0.6083,
|
|
"rewards/accuracies": 0.690625011920929,
|
|
"rewards/chosen": -0.8865949511528015,
|
|
"rewards/margins": 0.271259605884552,
|
|
"rewards/rejected": -1.157854437828064,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 1.4982978866830698,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 1.5032327586206897e-06,
|
|
"logits/chosen": -1.815341591835022,
|
|
"logits/rejected": -1.816318154335022,
|
|
"logps/chosen": -524.8699951171875,
|
|
"logps/rejected": -511.8518981933594,
|
|
"loss": 0.5777,
|
|
"rewards/accuracies": 0.7437499761581421,
|
|
"rewards/chosen": -0.9125308990478516,
|
|
"rewards/margins": 0.3498198091983795,
|
|
"rewards/rejected": -1.2623505592346191,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 1.5090835552259936,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 1.4924568965517243e-06,
|
|
"logits/chosen": -1.7852647304534912,
|
|
"logits/rejected": -1.799068808555603,
|
|
"logps/chosen": -533.41455078125,
|
|
"logps/rejected": -519.7373657226562,
|
|
"loss": 0.5829,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -0.9632998704910278,
|
|
"rewards/margins": 0.3437105119228363,
|
|
"rewards/rejected": -1.3070104122161865,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 1.5090835552259936,
|
|
"eval_logits/chosen": -1.7990481853485107,
|
|
"eval_logits/rejected": -1.8625297546386719,
|
|
"eval_logps/chosen": -570.4349975585938,
|
|
"eval_logps/rejected": -519.1015014648438,
|
|
"eval_loss": 0.6433591246604919,
|
|
"eval_rewards/accuracies": 0.6346153616905212,
|
|
"eval_rewards/chosen": -1.1104950904846191,
|
|
"eval_rewards/margins": 0.19405566155910492,
|
|
"eval_rewards/rejected": -1.3045506477355957,
|
|
"eval_runtime": 12.8758,
|
|
"eval_samples_per_second": 7.767,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 1.519869223768917,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 1.4816810344827588e-06,
|
|
"logits/chosen": -1.7966508865356445,
|
|
"logits/rejected": -1.8273112773895264,
|
|
"logps/chosen": -548.8226318359375,
|
|
"logps/rejected": -528.164306640625,
|
|
"loss": 0.5799,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -0.9004064798355103,
|
|
"rewards/margins": 0.3501775860786438,
|
|
"rewards/rejected": -1.2505838871002197,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 1.5306548923118406,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 1.4709051724137933e-06,
|
|
"logits/chosen": -1.7765086889266968,
|
|
"logits/rejected": -1.8015928268432617,
|
|
"logps/chosen": -549.3746948242188,
|
|
"logps/rejected": -529.9730224609375,
|
|
"loss": 0.5816,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -0.9200440645217896,
|
|
"rewards/margins": 0.33418479561805725,
|
|
"rewards/rejected": -1.2542288303375244,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 1.541440560854764,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 1.4601293103448276e-06,
|
|
"logits/chosen": -1.8374097347259521,
|
|
"logits/rejected": -1.8128893375396729,
|
|
"logps/chosen": -553.1683349609375,
|
|
"logps/rejected": -540.2759399414062,
|
|
"loss": 0.5902,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.964668869972229,
|
|
"rewards/margins": 0.3389004170894623,
|
|
"rewards/rejected": -1.3035693168640137,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 1.5522262293976878,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 1.449353448275862e-06,
|
|
"logits/chosen": -1.828120231628418,
|
|
"logits/rejected": -1.8493430614471436,
|
|
"logps/chosen": -569.1422119140625,
|
|
"logps/rejected": -556.3524780273438,
|
|
"loss": 0.5997,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -1.006941795349121,
|
|
"rewards/margins": 0.31565380096435547,
|
|
"rewards/rejected": -1.3225957155227661,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 1.5630118979406116,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 1.4385775862068966e-06,
|
|
"logits/chosen": -1.7910667657852173,
|
|
"logits/rejected": -1.8110182285308838,
|
|
"logps/chosen": -550.4522094726562,
|
|
"logps/rejected": -535.9798583984375,
|
|
"loss": 0.5846,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -0.9523816108703613,
|
|
"rewards/margins": 0.34083276987075806,
|
|
"rewards/rejected": -1.2932144403457642,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 1.5630118979406116,
|
|
"eval_logits/chosen": -1.8040180206298828,
|
|
"eval_logits/rejected": -1.8676948547363281,
|
|
"eval_logps/chosen": -570.4541015625,
|
|
"eval_logps/rejected": -519.1150512695312,
|
|
"eval_loss": 0.6397081613540649,
|
|
"eval_rewards/accuracies": 0.682692289352417,
|
|
"eval_rewards/chosen": -1.1114522218704224,
|
|
"eval_rewards/margins": 0.19377943873405457,
|
|
"eval_rewards/rejected": -1.3052315711975098,
|
|
"eval_runtime": 12.8296,
|
|
"eval_samples_per_second": 7.794,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 1.573797566483535,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 1.427801724137931e-06,
|
|
"logits/chosen": -1.7791173458099365,
|
|
"logits/rejected": -1.78426992893219,
|
|
"logps/chosen": -552.9260864257812,
|
|
"logps/rejected": -532.0748291015625,
|
|
"loss": 0.5938,
|
|
"rewards/accuracies": 0.6734374761581421,
|
|
"rewards/chosen": -0.9482347369194031,
|
|
"rewards/margins": 0.3171328008174896,
|
|
"rewards/rejected": -1.2653675079345703,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 1.5845832350264586,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.4170258620689656e-06,
|
|
"logits/chosen": -1.8079183101654053,
|
|
"logits/rejected": -1.8390960693359375,
|
|
"logps/chosen": -544.3607177734375,
|
|
"logps/rejected": -525.732421875,
|
|
"loss": 0.5943,
|
|
"rewards/accuracies": 0.698437511920929,
|
|
"rewards/chosen": -0.9570525288581848,
|
|
"rewards/margins": 0.3199313282966614,
|
|
"rewards/rejected": -1.2769839763641357,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 1.595368903569382,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.40625e-06,
|
|
"logits/chosen": -1.8212658166885376,
|
|
"logits/rejected": -1.8325313329696655,
|
|
"logps/chosen": -531.6943969726562,
|
|
"logps/rejected": -520.8181762695312,
|
|
"loss": 0.5909,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.9884265065193176,
|
|
"rewards/margins": 0.3185192048549652,
|
|
"rewards/rejected": -1.30694580078125,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 1.6061545721123056,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 1.3954741379310346e-06,
|
|
"logits/chosen": -1.7863876819610596,
|
|
"logits/rejected": -1.7843122482299805,
|
|
"logps/chosen": -527.3268432617188,
|
|
"logps/rejected": -513.3717041015625,
|
|
"loss": 0.5917,
|
|
"rewards/accuracies": 0.6968749761581421,
|
|
"rewards/chosen": -0.9644559025764465,
|
|
"rewards/margins": 0.3113047182559967,
|
|
"rewards/rejected": -1.2757607698440552,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 1.6169402406552293,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 1.3846982758620689e-06,
|
|
"logits/chosen": -1.7883723974227905,
|
|
"logits/rejected": -1.7981631755828857,
|
|
"logps/chosen": -559.0615234375,
|
|
"logps/rejected": -548.7954711914062,
|
|
"loss": 0.5805,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -0.943580150604248,
|
|
"rewards/margins": 0.36219388246536255,
|
|
"rewards/rejected": -1.3057740926742554,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 1.6169402406552293,
|
|
"eval_logits/chosen": -1.805317759513855,
|
|
"eval_logits/rejected": -1.8692779541015625,
|
|
"eval_logps/chosen": -570.7977905273438,
|
|
"eval_logps/rejected": -519.4142456054688,
|
|
"eval_loss": 0.6438877582550049,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1286320686340332,
|
|
"eval_rewards/margins": 0.19155457615852356,
|
|
"eval_rewards/rejected": -1.320186734199524,
|
|
"eval_runtime": 12.8603,
|
|
"eval_samples_per_second": 7.776,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 1.627725909198153,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.3739224137931034e-06,
|
|
"logits/chosen": -1.7962541580200195,
|
|
"logits/rejected": -1.8288599252700806,
|
|
"logps/chosen": -543.5225219726562,
|
|
"logps/rejected": -517.6636962890625,
|
|
"loss": 0.6067,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -0.9828208088874817,
|
|
"rewards/margins": 0.28606343269348145,
|
|
"rewards/rejected": -1.2688841819763184,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 1.6385115777410766,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 1.363146551724138e-06,
|
|
"logits/chosen": -1.7885265350341797,
|
|
"logits/rejected": -1.8046941757202148,
|
|
"logps/chosen": -537.9309692382812,
|
|
"logps/rejected": -522.3575439453125,
|
|
"loss": 0.6148,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -0.9945136904716492,
|
|
"rewards/margins": 0.2747846841812134,
|
|
"rewards/rejected": -1.2692983150482178,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 1.649297246284,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.3523706896551724e-06,
|
|
"logits/chosen": -1.8596391677856445,
|
|
"logits/rejected": -1.880578637123108,
|
|
"logps/chosen": -543.6522827148438,
|
|
"logps/rejected": -529.7686767578125,
|
|
"loss": 0.5866,
|
|
"rewards/accuracies": 0.698437511920929,
|
|
"rewards/chosen": -0.9741571545600891,
|
|
"rewards/margins": 0.33137011528015137,
|
|
"rewards/rejected": -1.3055272102355957,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 1.6600829148269236,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.341594827586207e-06,
|
|
"logits/chosen": -1.7963918447494507,
|
|
"logits/rejected": -1.8010103702545166,
|
|
"logps/chosen": -547.7319946289062,
|
|
"logps/rejected": -522.956298828125,
|
|
"loss": 0.5891,
|
|
"rewards/accuracies": 0.703125,
|
|
"rewards/chosen": -0.9565073251724243,
|
|
"rewards/margins": 0.31486618518829346,
|
|
"rewards/rejected": -1.2713735103607178,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 1.6708685833698473,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 1.3308189655172414e-06,
|
|
"logits/chosen": -1.8648087978363037,
|
|
"logits/rejected": -1.8721832036972046,
|
|
"logps/chosen": -559.5469360351562,
|
|
"logps/rejected": -548.2471313476562,
|
|
"loss": 0.5717,
|
|
"rewards/accuracies": 0.745312511920929,
|
|
"rewards/chosen": -0.9510286450386047,
|
|
"rewards/margins": 0.36009079217910767,
|
|
"rewards/rejected": -1.3111193180084229,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 1.6708685833698473,
|
|
"eval_logits/chosen": -1.806795597076416,
|
|
"eval_logits/rejected": -1.871440052986145,
|
|
"eval_logps/chosen": -570.8436889648438,
|
|
"eval_logps/rejected": -519.555908203125,
|
|
"eval_loss": 0.641202986240387,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.130926489830017,
|
|
"eval_rewards/margins": 0.19634580612182617,
|
|
"eval_rewards/rejected": -1.3272722959518433,
|
|
"eval_runtime": 12.8672,
|
|
"eval_samples_per_second": 7.772,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 1.681654251912771,
|
|
"grad_norm": 23.25,
|
|
"learning_rate": 1.320043103448276e-06,
|
|
"logits/chosen": -1.7812159061431885,
|
|
"logits/rejected": -1.7934643030166626,
|
|
"logps/chosen": -552.8800048828125,
|
|
"logps/rejected": -549.5223999023438,
|
|
"loss": 0.5768,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -0.9710870981216431,
|
|
"rewards/margins": 0.3766544461250305,
|
|
"rewards/rejected": -1.3477414846420288,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 1.6924399204556946,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 1.3092672413793102e-06,
|
|
"logits/chosen": -1.8353354930877686,
|
|
"logits/rejected": -1.8481401205062866,
|
|
"logps/chosen": -519.0048217773438,
|
|
"logps/rejected": -509.49591064453125,
|
|
"loss": 0.5817,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -0.9764666557312012,
|
|
"rewards/margins": 0.3405064642429352,
|
|
"rewards/rejected": -1.316973090171814,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 1.703225588998618,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 1.2984913793103447e-06,
|
|
"logits/chosen": -1.7921619415283203,
|
|
"logits/rejected": -1.804537057876587,
|
|
"logps/chosen": -537.7366943359375,
|
|
"logps/rejected": -510.7959899902344,
|
|
"loss": 0.5992,
|
|
"rewards/accuracies": 0.676562488079071,
|
|
"rewards/chosen": -0.9829401969909668,
|
|
"rewards/margins": 0.31215885281562805,
|
|
"rewards/rejected": -1.2950990200042725,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 1.7140112575415416,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.2877155172413794e-06,
|
|
"logits/chosen": -1.8133258819580078,
|
|
"logits/rejected": -1.8023666143417358,
|
|
"logps/chosen": -543.982666015625,
|
|
"logps/rejected": -542.6021728515625,
|
|
"loss": 0.5781,
|
|
"rewards/accuracies": 0.7281249761581421,
|
|
"rewards/chosen": -0.9853167533874512,
|
|
"rewards/margins": 0.36080607771873474,
|
|
"rewards/rejected": -1.3461229801177979,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 1.724796926084465,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 1.276939655172414e-06,
|
|
"logits/chosen": -1.8179728984832764,
|
|
"logits/rejected": -1.8313579559326172,
|
|
"logps/chosen": -551.9962158203125,
|
|
"logps/rejected": -533.030029296875,
|
|
"loss": 0.5797,
|
|
"rewards/accuracies": 0.698437511920929,
|
|
"rewards/chosen": -1.0232875347137451,
|
|
"rewards/margins": 0.3661686182022095,
|
|
"rewards/rejected": -1.3894561529159546,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 1.724796926084465,
|
|
"eval_logits/chosen": -1.80728280544281,
|
|
"eval_logits/rejected": -1.871204137802124,
|
|
"eval_logps/chosen": -571.1751098632812,
|
|
"eval_logps/rejected": -519.9603881835938,
|
|
"eval_loss": 0.6392231583595276,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -1.1474987268447876,
|
|
"eval_rewards/margins": 0.19999684393405914,
|
|
"eval_rewards/rejected": -1.347495436668396,
|
|
"eval_runtime": 12.8417,
|
|
"eval_samples_per_second": 7.787,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 1.7355825946273888,
|
|
"grad_norm": 23.75,
|
|
"learning_rate": 1.2661637931034484e-06,
|
|
"logits/chosen": -1.7909305095672607,
|
|
"logits/rejected": -1.8273541927337646,
|
|
"logps/chosen": -543.5694580078125,
|
|
"logps/rejected": -525.0535888671875,
|
|
"loss": 0.6002,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -0.9681240320205688,
|
|
"rewards/margins": 0.2939276099205017,
|
|
"rewards/rejected": -1.2620514631271362,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 1.7463682631703126,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 1.255387931034483e-06,
|
|
"logits/chosen": -1.8298015594482422,
|
|
"logits/rejected": -1.8496816158294678,
|
|
"logps/chosen": -559.7938232421875,
|
|
"logps/rejected": -542.3406982421875,
|
|
"loss": 0.5936,
|
|
"rewards/accuracies": 0.6968749761581421,
|
|
"rewards/chosen": -1.017887830734253,
|
|
"rewards/margins": 0.32336899638175964,
|
|
"rewards/rejected": -1.341256856918335,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 1.757153931713236,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 1.2446120689655172e-06,
|
|
"logits/chosen": -1.795771598815918,
|
|
"logits/rejected": -1.8096462488174438,
|
|
"logps/chosen": -550.1193237304688,
|
|
"logps/rejected": -526.5523071289062,
|
|
"loss": 0.5825,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -0.9442169070243835,
|
|
"rewards/margins": 0.3580732047557831,
|
|
"rewards/rejected": -1.3022902011871338,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 1.7679396002561596,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 1.2338362068965518e-06,
|
|
"logits/chosen": -1.7974588871002197,
|
|
"logits/rejected": -1.820927381515503,
|
|
"logps/chosen": -539.9898681640625,
|
|
"logps/rejected": -533.06689453125,
|
|
"loss": 0.5873,
|
|
"rewards/accuracies": 0.7203124761581421,
|
|
"rewards/chosen": -0.9383485913276672,
|
|
"rewards/margins": 0.336906760931015,
|
|
"rewards/rejected": -1.2752554416656494,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 1.778725268799083,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 1.2230603448275863e-06,
|
|
"logits/chosen": -1.7889001369476318,
|
|
"logits/rejected": -1.7788540124893188,
|
|
"logps/chosen": -549.4357299804688,
|
|
"logps/rejected": -537.9326171875,
|
|
"loss": 0.5948,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -0.9701086282730103,
|
|
"rewards/margins": 0.31890323758125305,
|
|
"rewards/rejected": -1.289011836051941,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 1.778725268799083,
|
|
"eval_logits/chosen": -1.808042049407959,
|
|
"eval_logits/rejected": -1.8723417520523071,
|
|
"eval_logps/chosen": -570.9359741210938,
|
|
"eval_logps/rejected": -519.9014892578125,
|
|
"eval_loss": 0.6358182430267334,
|
|
"eval_rewards/accuracies": 0.6730769276618958,
|
|
"eval_rewards/chosen": -1.1355457305908203,
|
|
"eval_rewards/margins": 0.2090054303407669,
|
|
"eval_rewards/rejected": -1.3445510864257812,
|
|
"eval_runtime": 12.9132,
|
|
"eval_samples_per_second": 7.744,
|
|
"eval_steps_per_second": 1.007,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 1.7895109373420068,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 1.2122844827586208e-06,
|
|
"logits/chosen": -1.8406860828399658,
|
|
"logits/rejected": -1.8749334812164307,
|
|
"logps/chosen": -565.1243896484375,
|
|
"logps/rejected": -564.97119140625,
|
|
"loss": 0.5715,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -1.0176526308059692,
|
|
"rewards/margins": 0.35605573654174805,
|
|
"rewards/rejected": -1.3737082481384277,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 1.8002966058849303,
|
|
"grad_norm": 23.125,
|
|
"learning_rate": 1.2015086206896553e-06,
|
|
"logits/chosen": -1.8156760931015015,
|
|
"logits/rejected": -1.821239709854126,
|
|
"logps/chosen": -532.1543579101562,
|
|
"logps/rejected": -519.3363037109375,
|
|
"loss": 0.5788,
|
|
"rewards/accuracies": 0.699999988079071,
|
|
"rewards/chosen": -1.0197910070419312,
|
|
"rewards/margins": 0.367012083530426,
|
|
"rewards/rejected": -1.3868030309677124,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 1.811082274427854,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 1.1907327586206898e-06,
|
|
"logits/chosen": -1.7746264934539795,
|
|
"logits/rejected": -1.7720403671264648,
|
|
"logps/chosen": -531.2168579101562,
|
|
"logps/rejected": -520.6815185546875,
|
|
"loss": 0.594,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -1.0364903211593628,
|
|
"rewards/margins": 0.32573047280311584,
|
|
"rewards/rejected": -1.3622207641601562,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 1.8218679429707776,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 1.179956896551724e-06,
|
|
"logits/chosen": -1.8178173303604126,
|
|
"logits/rejected": -1.8478190898895264,
|
|
"logps/chosen": -531.7720947265625,
|
|
"logps/rejected": -516.5214233398438,
|
|
"loss": 0.6073,
|
|
"rewards/accuracies": 0.684374988079071,
|
|
"rewards/chosen": -1.040016531944275,
|
|
"rewards/margins": 0.2878865599632263,
|
|
"rewards/rejected": -1.3279030323028564,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 1.832653611513701,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.1691810344827586e-06,
|
|
"logits/chosen": -1.8465070724487305,
|
|
"logits/rejected": -1.8316357135772705,
|
|
"logps/chosen": -533.6699829101562,
|
|
"logps/rejected": -520.6783447265625,
|
|
"loss": 0.5861,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -0.9806226491928101,
|
|
"rewards/margins": 0.33271270990371704,
|
|
"rewards/rejected": -1.313335657119751,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 1.832653611513701,
|
|
"eval_logits/chosen": -1.8129793405532837,
|
|
"eval_logits/rejected": -1.8767609596252441,
|
|
"eval_logps/chosen": -571.2252197265625,
|
|
"eval_logps/rejected": -519.8920288085938,
|
|
"eval_loss": 0.6430386900901794,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1500080823898315,
|
|
"eval_rewards/margins": 0.19407115876674652,
|
|
"eval_rewards/rejected": -1.3440792560577393,
|
|
"eval_runtime": 12.8277,
|
|
"eval_samples_per_second": 7.796,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 1.8434392800566246,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.158405172413793e-06,
|
|
"logits/chosen": -1.769142508506775,
|
|
"logits/rejected": -1.7703624963760376,
|
|
"logps/chosen": -527.4718627929688,
|
|
"logps/rejected": -509.33013916015625,
|
|
"loss": 0.5793,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.9673894643783569,
|
|
"rewards/margins": 0.36447519063949585,
|
|
"rewards/rejected": -1.3318647146224976,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 1.8542249485995483,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 1.1476293103448276e-06,
|
|
"logits/chosen": -1.7493419647216797,
|
|
"logits/rejected": -1.740152359008789,
|
|
"logps/chosen": -551.5736083984375,
|
|
"logps/rejected": -546.8021240234375,
|
|
"loss": 0.5859,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -1.000551700592041,
|
|
"rewards/margins": 0.3465580344200134,
|
|
"rewards/rejected": -1.3471099138259888,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 1.865010617142472,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 1.136853448275862e-06,
|
|
"logits/chosen": -1.795250654220581,
|
|
"logits/rejected": -1.8162511587142944,
|
|
"logps/chosen": -535.1650390625,
|
|
"logps/rejected": -517.9396362304688,
|
|
"loss": 0.5948,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -1.001084566116333,
|
|
"rewards/margins": 0.31550517678260803,
|
|
"rewards/rejected": -1.3165898323059082,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 1.8757962856853956,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 1.1260775862068966e-06,
|
|
"logits/chosen": -1.8540292978286743,
|
|
"logits/rejected": -1.866869330406189,
|
|
"logps/chosen": -534.6907958984375,
|
|
"logps/rejected": -514.3424072265625,
|
|
"loss": 0.5831,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -1.0445291996002197,
|
|
"rewards/margins": 0.34380587935447693,
|
|
"rewards/rejected": -1.3883349895477295,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 1.886581954228319,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 1.115301724137931e-06,
|
|
"logits/chosen": -1.8096578121185303,
|
|
"logits/rejected": -1.8162269592285156,
|
|
"logps/chosen": -558.2589721679688,
|
|
"logps/rejected": -544.1400146484375,
|
|
"loss": 0.594,
|
|
"rewards/accuracies": 0.703125,
|
|
"rewards/chosen": -1.0290067195892334,
|
|
"rewards/margins": 0.3109704852104187,
|
|
"rewards/rejected": -1.3399772644042969,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 1.886581954228319,
|
|
"eval_logits/chosen": -1.8141746520996094,
|
|
"eval_logits/rejected": -1.8783446550369263,
|
|
"eval_logps/chosen": -571.4691772460938,
|
|
"eval_logps/rejected": -520.206787109375,
|
|
"eval_loss": 0.6430192589759827,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1622072458267212,
|
|
"eval_rewards/margins": 0.19761131703853607,
|
|
"eval_rewards/rejected": -1.359818696975708,
|
|
"eval_runtime": 12.8695,
|
|
"eval_samples_per_second": 7.77,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 1.8973676227712426,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 1.1045258620689654e-06,
|
|
"logits/chosen": -1.8399471044540405,
|
|
"logits/rejected": -1.8462409973144531,
|
|
"logps/chosen": -528.4866943359375,
|
|
"logps/rejected": -521.9978637695312,
|
|
"loss": 0.5739,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -0.996717095375061,
|
|
"rewards/margins": 0.3896409571170807,
|
|
"rewards/rejected": -1.3863580226898193,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 1.9081532913141663,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 1.09375e-06,
|
|
"logits/chosen": -1.824342131614685,
|
|
"logits/rejected": -1.851354956626892,
|
|
"logps/chosen": -541.2124633789062,
|
|
"logps/rejected": -521.5025024414062,
|
|
"loss": 0.5939,
|
|
"rewards/accuracies": 0.6734374761581421,
|
|
"rewards/chosen": -1.0108633041381836,
|
|
"rewards/margins": 0.32563039660453796,
|
|
"rewards/rejected": -1.3364938497543335,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 1.9189389598570898,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.0829741379310344e-06,
|
|
"logits/chosen": -1.7720905542373657,
|
|
"logits/rejected": -1.7949879169464111,
|
|
"logps/chosen": -558.3751220703125,
|
|
"logps/rejected": -534.5128173828125,
|
|
"loss": 0.604,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.9803739786148071,
|
|
"rewards/margins": 0.30562636256217957,
|
|
"rewards/rejected": -1.2860002517700195,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 1.9297246284000136,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 1.072198275862069e-06,
|
|
"logits/chosen": -1.8202836513519287,
|
|
"logits/rejected": -1.830665946006775,
|
|
"logps/chosen": -528.1467895507812,
|
|
"logps/rejected": -520.9862060546875,
|
|
"loss": 0.6036,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -1.015037178993225,
|
|
"rewards/margins": 0.2926950752735138,
|
|
"rewards/rejected": -1.307732343673706,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 1.940510296942937,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 1.0614224137931034e-06,
|
|
"logits/chosen": -1.8530925512313843,
|
|
"logits/rejected": -1.8568578958511353,
|
|
"logps/chosen": -514.7476196289062,
|
|
"logps/rejected": -507.7210998535156,
|
|
"loss": 0.6026,
|
|
"rewards/accuracies": 0.682812511920929,
|
|
"rewards/chosen": -1.0476781129837036,
|
|
"rewards/margins": 0.30604255199432373,
|
|
"rewards/rejected": -1.3537206649780273,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 1.940510296942937,
|
|
"eval_logits/chosen": -1.8167260885238647,
|
|
"eval_logits/rejected": -1.881202220916748,
|
|
"eval_logps/chosen": -571.269775390625,
|
|
"eval_logps/rejected": -520.2171630859375,
|
|
"eval_loss": 0.6376249194145203,
|
|
"eval_rewards/accuracies": 0.6346153616905212,
|
|
"eval_rewards/chosen": -1.1522341966629028,
|
|
"eval_rewards/margins": 0.20809604227542877,
|
|
"eval_rewards/rejected": -1.3603301048278809,
|
|
"eval_runtime": 12.8279,
|
|
"eval_samples_per_second": 7.796,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 1.9512959654858606,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 1.0506465517241381e-06,
|
|
"logits/chosen": -1.7945997714996338,
|
|
"logits/rejected": -1.8242772817611694,
|
|
"logps/chosen": -558.9090576171875,
|
|
"logps/rejected": -544.4629516601562,
|
|
"loss": 0.6057,
|
|
"rewards/accuracies": 0.6953125,
|
|
"rewards/chosen": -1.033698558807373,
|
|
"rewards/margins": 0.2901691198348999,
|
|
"rewards/rejected": -1.323867678642273,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 1.962081634028784,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 1.0398706896551724e-06,
|
|
"logits/chosen": -1.763892412185669,
|
|
"logits/rejected": -1.7766506671905518,
|
|
"logps/chosen": -549.2996826171875,
|
|
"logps/rejected": -538.8092041015625,
|
|
"loss": 0.5675,
|
|
"rewards/accuracies": 0.746874988079071,
|
|
"rewards/chosen": -0.9804860353469849,
|
|
"rewards/margins": 0.3770001232624054,
|
|
"rewards/rejected": -1.3574860095977783,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 1.9728673025717078,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 1.029094827586207e-06,
|
|
"logits/chosen": -1.8451900482177734,
|
|
"logits/rejected": -1.8786299228668213,
|
|
"logps/chosen": -567.0401611328125,
|
|
"logps/rejected": -549.7001953125,
|
|
"loss": 0.5843,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -0.9965025782585144,
|
|
"rewards/margins": 0.3378158509731293,
|
|
"rewards/rejected": -1.3343183994293213,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 1.9836529711146316,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 1.0183189655172414e-06,
|
|
"logits/chosen": -1.8494704961776733,
|
|
"logits/rejected": -1.8769400119781494,
|
|
"logps/chosen": -552.8043823242188,
|
|
"logps/rejected": -537.8701782226562,
|
|
"loss": 0.5857,
|
|
"rewards/accuracies": 0.703125,
|
|
"rewards/chosen": -1.03322172164917,
|
|
"rewards/margins": 0.4004359841346741,
|
|
"rewards/rejected": -1.4336576461791992,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 1.994438639657555,
|
|
"grad_norm": 24.25,
|
|
"learning_rate": 1.007543103448276e-06,
|
|
"logits/chosen": -1.8407787084579468,
|
|
"logits/rejected": -1.8548295497894287,
|
|
"logps/chosen": -557.3915405273438,
|
|
"logps/rejected": -536.4610595703125,
|
|
"loss": 0.6012,
|
|
"rewards/accuracies": 0.684374988079071,
|
|
"rewards/chosen": -1.0534279346466064,
|
|
"rewards/margins": 0.3128618001937866,
|
|
"rewards/rejected": -1.3662898540496826,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 1.994438639657555,
|
|
"eval_logits/chosen": -1.8164806365966797,
|
|
"eval_logits/rejected": -1.8804469108581543,
|
|
"eval_logps/chosen": -571.3651123046875,
|
|
"eval_logps/rejected": -520.392333984375,
|
|
"eval_loss": 0.6372777819633484,
|
|
"eval_rewards/accuracies": 0.6730769276618958,
|
|
"eval_rewards/chosen": -1.1570004224777222,
|
|
"eval_rewards/margins": 0.2120911180973053,
|
|
"eval_rewards/rejected": -1.3690916299819946,
|
|
"eval_runtime": 12.854,
|
|
"eval_samples_per_second": 7.78,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 2.0043142674171692,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 9.967672413793105e-07,
|
|
"logits/chosen": -1.8243948221206665,
|
|
"logits/rejected": -1.8106485605239868,
|
|
"logps/chosen": -563.4216918945312,
|
|
"logps/rejected": -554.7511596679688,
|
|
"loss": 0.5964,
|
|
"rewards/accuracies": 0.7047781348228455,
|
|
"rewards/chosen": -1.0186046361923218,
|
|
"rewards/margins": 0.33347877860069275,
|
|
"rewards/rejected": -1.3520833253860474,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 2.015099935960093,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 9.85991379310345e-07,
|
|
"logits/chosen": -1.816082239151001,
|
|
"logits/rejected": -1.821375846862793,
|
|
"logps/chosen": -525.9388427734375,
|
|
"logps/rejected": -514.32861328125,
|
|
"loss": 0.5994,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -1.0036041736602783,
|
|
"rewards/margins": 0.3070666193962097,
|
|
"rewards/rejected": -1.3106707334518433,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 2.0258856045030167,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 9.752155172413795e-07,
|
|
"logits/chosen": -1.8075281381607056,
|
|
"logits/rejected": -1.8388919830322266,
|
|
"logps/chosen": -552.760009765625,
|
|
"logps/rejected": -546.732421875,
|
|
"loss": 0.5852,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -1.0357027053833008,
|
|
"rewards/margins": 0.33780843019485474,
|
|
"rewards/rejected": -1.3735110759735107,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 2.03667127304594,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 9.644396551724138e-07,
|
|
"logits/chosen": -1.767423391342163,
|
|
"logits/rejected": -1.7957046031951904,
|
|
"logps/chosen": -553.3221435546875,
|
|
"logps/rejected": -534.2494506835938,
|
|
"loss": 0.574,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -0.977568507194519,
|
|
"rewards/margins": 0.36738520860671997,
|
|
"rewards/rejected": -1.3449536561965942,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 2.0474569415888637,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 9.536637931034483e-07,
|
|
"logits/chosen": -1.8196423053741455,
|
|
"logits/rejected": -1.8127524852752686,
|
|
"logps/chosen": -534.7515258789062,
|
|
"logps/rejected": -519.46484375,
|
|
"loss": 0.5833,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -0.9940136075019836,
|
|
"rewards/margins": 0.34578242897987366,
|
|
"rewards/rejected": -1.3397960662841797,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 2.0474569415888637,
|
|
"eval_logits/chosen": -1.8176302909851074,
|
|
"eval_logits/rejected": -1.8824341297149658,
|
|
"eval_logps/chosen": -571.3641967773438,
|
|
"eval_logps/rejected": -520.0811157226562,
|
|
"eval_loss": 0.6425002813339233,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1569527387619019,
|
|
"eval_rewards/margins": 0.19657853245735168,
|
|
"eval_rewards/rejected": -1.3535313606262207,
|
|
"eval_runtime": 12.834,
|
|
"eval_samples_per_second": 7.792,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 2.058242610131787,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 9.428879310344828e-07,
|
|
"logits/chosen": -1.7818291187286377,
|
|
"logits/rejected": -1.7981443405151367,
|
|
"logps/chosen": -541.3946533203125,
|
|
"logps/rejected": -527.3668823242188,
|
|
"loss": 0.5742,
|
|
"rewards/accuracies": 0.734375,
|
|
"rewards/chosen": -0.9855313301086426,
|
|
"rewards/margins": 0.37073832750320435,
|
|
"rewards/rejected": -1.3562697172164917,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"epoch": 2.069028278674711,
|
|
"grad_norm": 18.5,
|
|
"learning_rate": 9.321120689655173e-07,
|
|
"logits/chosen": -1.833970069885254,
|
|
"logits/rejected": -1.8262455463409424,
|
|
"logps/chosen": -546.7005004882812,
|
|
"logps/rejected": -538.780029296875,
|
|
"loss": 0.5723,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -0.9842655062675476,
|
|
"rewards/margins": 0.3666042685508728,
|
|
"rewards/rejected": -1.3508696556091309,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"epoch": 2.0798139472176347,
|
|
"grad_norm": 19.625,
|
|
"learning_rate": 9.213362068965518e-07,
|
|
"logits/chosen": -1.7877027988433838,
|
|
"logits/rejected": -1.7972294092178345,
|
|
"logps/chosen": -541.6177978515625,
|
|
"logps/rejected": -527.7945556640625,
|
|
"loss": 0.5934,
|
|
"rewards/accuracies": 0.7250000238418579,
|
|
"rewards/chosen": -0.9702165722846985,
|
|
"rewards/margins": 0.3309406340122223,
|
|
"rewards/rejected": -1.3011572360992432,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"epoch": 2.090599615760558,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 9.105603448275862e-07,
|
|
"logits/chosen": -1.7990188598632812,
|
|
"logits/rejected": -1.8132622241973877,
|
|
"logps/chosen": -550.9437255859375,
|
|
"logps/rejected": -540.9758911132812,
|
|
"loss": 0.5811,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -0.9730231165885925,
|
|
"rewards/margins": 0.35077935457229614,
|
|
"rewards/rejected": -1.3238024711608887,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"epoch": 2.1013852843034817,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 8.997844827586207e-07,
|
|
"logits/chosen": -1.8408838510513306,
|
|
"logits/rejected": -1.860565423965454,
|
|
"logps/chosen": -528.8717041015625,
|
|
"logps/rejected": -515.1477661132812,
|
|
"loss": 0.591,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -1.0306185483932495,
|
|
"rewards/margins": 0.3226325213909149,
|
|
"rewards/rejected": -1.3532510995864868,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 2.1013852843034817,
|
|
"eval_logits/chosen": -1.8168575763702393,
|
|
"eval_logits/rejected": -1.8818013668060303,
|
|
"eval_logps/chosen": -571.19580078125,
|
|
"eval_logps/rejected": -520.1525268554688,
|
|
"eval_loss": 0.6379492282867432,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1485364437103271,
|
|
"eval_rewards/margins": 0.20856307446956635,
|
|
"eval_rewards/rejected": -1.3570995330810547,
|
|
"eval_runtime": 12.8416,
|
|
"eval_samples_per_second": 7.787,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 2.112170952846405,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 8.890086206896552e-07,
|
|
"logits/chosen": -1.8283774852752686,
|
|
"logits/rejected": -1.8340866565704346,
|
|
"logps/chosen": -553.7230224609375,
|
|
"logps/rejected": -539.7496337890625,
|
|
"loss": 0.5792,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.9757739901542664,
|
|
"rewards/margins": 0.35080114006996155,
|
|
"rewards/rejected": -1.3265750408172607,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"epoch": 2.1229566213893287,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 8.782327586206896e-07,
|
|
"logits/chosen": -1.8246581554412842,
|
|
"logits/rejected": -1.8176138401031494,
|
|
"logps/chosen": -556.3699951171875,
|
|
"logps/rejected": -536.9313354492188,
|
|
"loss": 0.5713,
|
|
"rewards/accuracies": 0.7203124761581421,
|
|
"rewards/chosen": -0.939971923828125,
|
|
"rewards/margins": 0.3758017420768738,
|
|
"rewards/rejected": -1.3157737255096436,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"epoch": 2.1337422899322527,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 8.674568965517241e-07,
|
|
"logits/chosen": -1.7579801082611084,
|
|
"logits/rejected": -1.792454719543457,
|
|
"logps/chosen": -506.5731506347656,
|
|
"logps/rejected": -492.03741455078125,
|
|
"loss": 0.5807,
|
|
"rewards/accuracies": 0.7406250238418579,
|
|
"rewards/chosen": -0.9504076242446899,
|
|
"rewards/margins": 0.3443571925163269,
|
|
"rewards/rejected": -1.294764757156372,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"epoch": 2.144527958475176,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 8.566810344827586e-07,
|
|
"logits/chosen": -1.8132164478302002,
|
|
"logits/rejected": -1.8298476934432983,
|
|
"logps/chosen": -543.1507568359375,
|
|
"logps/rejected": -532.009033203125,
|
|
"loss": 0.5807,
|
|
"rewards/accuracies": 0.707812488079071,
|
|
"rewards/chosen": -0.953994631767273,
|
|
"rewards/margins": 0.35666000843048096,
|
|
"rewards/rejected": -1.310654640197754,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"epoch": 2.1553136270180997,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 8.459051724137931e-07,
|
|
"logits/chosen": -1.8091182708740234,
|
|
"logits/rejected": -1.8285820484161377,
|
|
"logps/chosen": -507.40240478515625,
|
|
"logps/rejected": -491.46307373046875,
|
|
"loss": 0.5944,
|
|
"rewards/accuracies": 0.692187488079071,
|
|
"rewards/chosen": -0.9950447082519531,
|
|
"rewards/margins": 0.3224475085735321,
|
|
"rewards/rejected": -1.3174922466278076,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 2.1553136270180997,
|
|
"eval_logits/chosen": -1.8191330432891846,
|
|
"eval_logits/rejected": -1.8835722208023071,
|
|
"eval_logps/chosen": -571.5982666015625,
|
|
"eval_logps/rejected": -520.4741821289062,
|
|
"eval_loss": 0.6388756036758423,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1686609983444214,
|
|
"eval_rewards/margins": 0.20452477037906647,
|
|
"eval_rewards/rejected": -1.3731858730316162,
|
|
"eval_runtime": 12.8754,
|
|
"eval_samples_per_second": 7.767,
|
|
"eval_steps_per_second": 1.01,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 2.166099295561023,
|
|
"grad_norm": 22.875,
|
|
"learning_rate": 8.351293103448275e-07,
|
|
"logits/chosen": -1.734623670578003,
|
|
"logits/rejected": -1.738992691040039,
|
|
"logps/chosen": -557.7481689453125,
|
|
"logps/rejected": -540.2991943359375,
|
|
"loss": 0.6125,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -1.022360920906067,
|
|
"rewards/margins": 0.27458736300468445,
|
|
"rewards/rejected": -1.2969481945037842,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"epoch": 2.1768849641039467,
|
|
"grad_norm": 22.875,
|
|
"learning_rate": 8.24353448275862e-07,
|
|
"logits/chosen": -1.8071972131729126,
|
|
"logits/rejected": -1.8338969945907593,
|
|
"logps/chosen": -535.3201293945312,
|
|
"logps/rejected": -517.41796875,
|
|
"loss": 0.5833,
|
|
"rewards/accuracies": 0.7093750238418579,
|
|
"rewards/chosen": -1.0038800239562988,
|
|
"rewards/margins": 0.34552377462387085,
|
|
"rewards/rejected": -1.3494038581848145,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"epoch": 2.1876706326468707,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 8.135775862068966e-07,
|
|
"logits/chosen": -1.8387718200683594,
|
|
"logits/rejected": -1.84176766872406,
|
|
"logps/chosen": -546.7960205078125,
|
|
"logps/rejected": -534.9925537109375,
|
|
"loss": 0.5817,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -1.0042190551757812,
|
|
"rewards/margins": 0.3391868472099304,
|
|
"rewards/rejected": -1.3434059619903564,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"epoch": 2.198456301189794,
|
|
"grad_norm": 21.75,
|
|
"learning_rate": 8.028017241379311e-07,
|
|
"logits/chosen": -1.840580701828003,
|
|
"logits/rejected": -1.8485066890716553,
|
|
"logps/chosen": -529.6387939453125,
|
|
"logps/rejected": -517.8297119140625,
|
|
"loss": 0.6042,
|
|
"rewards/accuracies": 0.6812499761581421,
|
|
"rewards/chosen": -1.0427838563919067,
|
|
"rewards/margins": 0.29375189542770386,
|
|
"rewards/rejected": -1.336535930633545,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"epoch": 2.2092419697327177,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 7.920258620689656e-07,
|
|
"logits/chosen": -1.7776069641113281,
|
|
"logits/rejected": -1.7974965572357178,
|
|
"logps/chosen": -557.3050537109375,
|
|
"logps/rejected": -541.1644897460938,
|
|
"loss": 0.5813,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -1.0097219944000244,
|
|
"rewards/margins": 0.3600006699562073,
|
|
"rewards/rejected": -1.369722843170166,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 2.2092419697327177,
|
|
"eval_logits/chosen": -1.8171120882034302,
|
|
"eval_logits/rejected": -1.8817297220230103,
|
|
"eval_logps/chosen": -571.6415405273438,
|
|
"eval_logps/rejected": -520.3262939453125,
|
|
"eval_loss": 0.6439670324325562,
|
|
"eval_rewards/accuracies": 0.6153846383094788,
|
|
"eval_rewards/chosen": -1.1708241701126099,
|
|
"eval_rewards/margins": 0.19496536254882812,
|
|
"eval_rewards/rejected": -1.365789532661438,
|
|
"eval_runtime": 12.8352,
|
|
"eval_samples_per_second": 7.791,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 2.220027638275641,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 7.8125e-07,
|
|
"logits/chosen": -1.8244011402130127,
|
|
"logits/rejected": -1.8032004833221436,
|
|
"logps/chosen": -533.1487426757812,
|
|
"logps/rejected": -523.3587646484375,
|
|
"loss": 0.5793,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -1.0152502059936523,
|
|
"rewards/margins": 0.3824854791164398,
|
|
"rewards/rejected": -1.397735834121704,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"epoch": 2.2308133068185647,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 7.704741379310345e-07,
|
|
"logits/chosen": -1.840766191482544,
|
|
"logits/rejected": -1.8367950916290283,
|
|
"logps/chosen": -548.8405151367188,
|
|
"logps/rejected": -544.241943359375,
|
|
"loss": 0.5811,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -0.9654202461242676,
|
|
"rewards/margins": 0.3421328663825989,
|
|
"rewards/rejected": -1.3075530529022217,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"epoch": 2.241598975361488,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 7.59698275862069e-07,
|
|
"logits/chosen": -1.7699668407440186,
|
|
"logits/rejected": -1.7745425701141357,
|
|
"logps/chosen": -566.4268188476562,
|
|
"logps/rejected": -563.216796875,
|
|
"loss": 0.5914,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -0.9823424220085144,
|
|
"rewards/margins": 0.3274309039115906,
|
|
"rewards/rejected": -1.309773325920105,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"epoch": 2.252384643904412,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 7.489224137931034e-07,
|
|
"logits/chosen": -1.761265516281128,
|
|
"logits/rejected": -1.7772560119628906,
|
|
"logps/chosen": -553.53564453125,
|
|
"logps/rejected": -528.5107421875,
|
|
"loss": 0.5898,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -0.9453617930412292,
|
|
"rewards/margins": 0.3203446567058563,
|
|
"rewards/rejected": -1.2657064199447632,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"epoch": 2.2631703124473357,
|
|
"grad_norm": 18.875,
|
|
"learning_rate": 7.38146551724138e-07,
|
|
"logits/chosen": -1.769878625869751,
|
|
"logits/rejected": -1.7828420400619507,
|
|
"logps/chosen": -575.3273315429688,
|
|
"logps/rejected": -545.9244995117188,
|
|
"loss": 0.5906,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -1.0056273937225342,
|
|
"rewards/margins": 0.33718934655189514,
|
|
"rewards/rejected": -1.342816710472107,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 2.2631703124473357,
|
|
"eval_logits/chosen": -1.8217333555221558,
|
|
"eval_logits/rejected": -1.8864943981170654,
|
|
"eval_logps/chosen": -571.5655517578125,
|
|
"eval_logps/rejected": -520.4244384765625,
|
|
"eval_loss": 0.6396628618240356,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1670244932174683,
|
|
"eval_rewards/margins": 0.2036767601966858,
|
|
"eval_rewards/rejected": -1.3707013130187988,
|
|
"eval_runtime": 12.8554,
|
|
"eval_samples_per_second": 7.779,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 2.273955980990259,
|
|
"grad_norm": 23.25,
|
|
"learning_rate": 7.273706896551725e-07,
|
|
"logits/chosen": -1.816007375717163,
|
|
"logits/rejected": -1.8105392456054688,
|
|
"logps/chosen": -547.5933837890625,
|
|
"logps/rejected": -546.60888671875,
|
|
"loss": 0.5834,
|
|
"rewards/accuracies": 0.6890624761581421,
|
|
"rewards/chosen": -0.9704920649528503,
|
|
"rewards/margins": 0.33989059925079346,
|
|
"rewards/rejected": -1.310382604598999,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 2.2847416495331827,
|
|
"grad_norm": 22.25,
|
|
"learning_rate": 7.16594827586207e-07,
|
|
"logits/chosen": -1.8106517791748047,
|
|
"logits/rejected": -1.7941129207611084,
|
|
"logps/chosen": -539.07568359375,
|
|
"logps/rejected": -527.458740234375,
|
|
"loss": 0.5906,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -1.0233216285705566,
|
|
"rewards/margins": 0.33753374218940735,
|
|
"rewards/rejected": -1.3608553409576416,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"epoch": 2.295527318076106,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 7.058189655172414e-07,
|
|
"logits/chosen": -1.8167377710342407,
|
|
"logits/rejected": -1.8257455825805664,
|
|
"logps/chosen": -563.2874755859375,
|
|
"logps/rejected": -553.4884643554688,
|
|
"loss": 0.5733,
|
|
"rewards/accuracies": 0.7203124761581421,
|
|
"rewards/chosen": -0.9766351580619812,
|
|
"rewards/margins": 0.4000927805900574,
|
|
"rewards/rejected": -1.3767280578613281,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"epoch": 2.30631298661903,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 6.950431034482759e-07,
|
|
"logits/chosen": -1.786746621131897,
|
|
"logits/rejected": -1.7818912267684937,
|
|
"logps/chosen": -545.81640625,
|
|
"logps/rejected": -537.4017333984375,
|
|
"loss": 0.5852,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -0.9767181277275085,
|
|
"rewards/margins": 0.3548073172569275,
|
|
"rewards/rejected": -1.331525444984436,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"epoch": 2.3170986551619537,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 6.842672413793104e-07,
|
|
"logits/chosen": -1.8164615631103516,
|
|
"logits/rejected": -1.8195968866348267,
|
|
"logps/chosen": -524.2017822265625,
|
|
"logps/rejected": -513.7794799804688,
|
|
"loss": 0.5909,
|
|
"rewards/accuracies": 0.7015625238418579,
|
|
"rewards/chosen": -0.9479745626449585,
|
|
"rewards/margins": 0.31994181871414185,
|
|
"rewards/rejected": -1.2679163217544556,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 2.3170986551619537,
|
|
"eval_logits/chosen": -1.819827914237976,
|
|
"eval_logits/rejected": -1.8852392435073853,
|
|
"eval_logps/chosen": -571.2940063476562,
|
|
"eval_logps/rejected": -520.1119995117188,
|
|
"eval_loss": 0.6401126384735107,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1534450054168701,
|
|
"eval_rewards/margins": 0.20163044333457947,
|
|
"eval_rewards/rejected": -1.3550753593444824,
|
|
"eval_runtime": 12.8257,
|
|
"eval_samples_per_second": 7.797,
|
|
"eval_steps_per_second": 1.014,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 2.327884323704877,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 6.734913793103448e-07,
|
|
"logits/chosen": -1.7366435527801514,
|
|
"logits/rejected": -1.7245728969573975,
|
|
"logps/chosen": -527.2462768554688,
|
|
"logps/rejected": -515.2891235351562,
|
|
"loss": 0.5881,
|
|
"rewards/accuracies": 0.699999988079071,
|
|
"rewards/chosen": -0.951647937297821,
|
|
"rewards/margins": 0.32886141538619995,
|
|
"rewards/rejected": -1.2805094718933105,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"epoch": 2.3386699922478007,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 6.627155172413793e-07,
|
|
"logits/chosen": -1.75906240940094,
|
|
"logits/rejected": -1.772367238998413,
|
|
"logps/chosen": -567.78759765625,
|
|
"logps/rejected": -553.91162109375,
|
|
"loss": 0.5751,
|
|
"rewards/accuracies": 0.7359374761581421,
|
|
"rewards/chosen": -0.9418633580207825,
|
|
"rewards/margins": 0.3597813546657562,
|
|
"rewards/rejected": -1.3016446828842163,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"epoch": 2.349455660790724,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 6.519396551724138e-07,
|
|
"logits/chosen": -1.8051865100860596,
|
|
"logits/rejected": -1.8138115406036377,
|
|
"logps/chosen": -551.5945434570312,
|
|
"logps/rejected": -533.2476806640625,
|
|
"loss": 0.5676,
|
|
"rewards/accuracies": 0.7328125238418579,
|
|
"rewards/chosen": -0.9801983833312988,
|
|
"rewards/margins": 0.39750877022743225,
|
|
"rewards/rejected": -1.3777072429656982,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"epoch": 2.3602413293336477,
|
|
"grad_norm": 22.5,
|
|
"learning_rate": 6.411637931034483e-07,
|
|
"logits/chosen": -1.8182309865951538,
|
|
"logits/rejected": -1.8295986652374268,
|
|
"logps/chosen": -559.96435546875,
|
|
"logps/rejected": -546.2293701171875,
|
|
"loss": 0.5932,
|
|
"rewards/accuracies": 0.6859375238418579,
|
|
"rewards/chosen": -0.997664749622345,
|
|
"rewards/margins": 0.3369225859642029,
|
|
"rewards/rejected": -1.3345873355865479,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"epoch": 2.3710269978765717,
|
|
"grad_norm": 18.375,
|
|
"learning_rate": 6.303879310344828e-07,
|
|
"logits/chosen": -1.7372815608978271,
|
|
"logits/rejected": -1.7619171142578125,
|
|
"logps/chosen": -533.6831665039062,
|
|
"logps/rejected": -516.8760375976562,
|
|
"loss": 0.5755,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -0.933220386505127,
|
|
"rewards/margins": 0.3638560175895691,
|
|
"rewards/rejected": -1.2970764636993408,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 2.3710269978765717,
|
|
"eval_logits/chosen": -1.815535068511963,
|
|
"eval_logits/rejected": -1.880157470703125,
|
|
"eval_logps/chosen": -571.4099731445312,
|
|
"eval_logps/rejected": -520.0239868164062,
|
|
"eval_loss": 0.6433662176132202,
|
|
"eval_rewards/accuracies": 0.6346153616905212,
|
|
"eval_rewards/chosen": -1.1592445373535156,
|
|
"eval_rewards/margins": 0.1914299577474594,
|
|
"eval_rewards/rejected": -1.3506745100021362,
|
|
"eval_runtime": 12.8533,
|
|
"eval_samples_per_second": 7.78,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 2.381812666419495,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 6.196120689655173e-07,
|
|
"logits/chosen": -1.8025963306427002,
|
|
"logits/rejected": -1.7924509048461914,
|
|
"logps/chosen": -551.2867431640625,
|
|
"logps/rejected": -542.8524169921875,
|
|
"loss": 0.5904,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -0.9992715716362,
|
|
"rewards/margins": 0.32568755745887756,
|
|
"rewards/rejected": -1.3249592781066895,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"epoch": 2.3925983349624187,
|
|
"grad_norm": 21.5,
|
|
"learning_rate": 6.088362068965517e-07,
|
|
"logits/chosen": -1.812631368637085,
|
|
"logits/rejected": -1.8184080123901367,
|
|
"logps/chosen": -559.1375732421875,
|
|
"logps/rejected": -542.5402221679688,
|
|
"loss": 0.5766,
|
|
"rewards/accuracies": 0.753125011920929,
|
|
"rewards/chosen": -0.9521558880805969,
|
|
"rewards/margins": 0.3602093458175659,
|
|
"rewards/rejected": -1.3123652935028076,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"epoch": 2.403384003505342,
|
|
"grad_norm": 23.25,
|
|
"learning_rate": 5.980603448275862e-07,
|
|
"logits/chosen": -1.8015377521514893,
|
|
"logits/rejected": -1.808902382850647,
|
|
"logps/chosen": -552.8629150390625,
|
|
"logps/rejected": -537.6948852539062,
|
|
"loss": 0.5718,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -0.932092547416687,
|
|
"rewards/margins": 0.37704476714134216,
|
|
"rewards/rejected": -1.3091373443603516,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"epoch": 2.4141696720482657,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 5.872844827586207e-07,
|
|
"logits/chosen": -1.8348724842071533,
|
|
"logits/rejected": -1.861788034439087,
|
|
"logps/chosen": -537.8258056640625,
|
|
"logps/rejected": -525.0621948242188,
|
|
"loss": 0.5866,
|
|
"rewards/accuracies": 0.703125,
|
|
"rewards/chosen": -1.0409669876098633,
|
|
"rewards/margins": 0.33658367395401,
|
|
"rewards/rejected": -1.3775508403778076,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"epoch": 2.4249553405911897,
|
|
"grad_norm": 23.5,
|
|
"learning_rate": 5.765086206896552e-07,
|
|
"logits/chosen": -1.760894775390625,
|
|
"logits/rejected": -1.772892951965332,
|
|
"logps/chosen": -585.08544921875,
|
|
"logps/rejected": -567.151123046875,
|
|
"loss": 0.5984,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.9715090990066528,
|
|
"rewards/margins": 0.30948516726493835,
|
|
"rewards/rejected": -1.280994176864624,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 2.4249553405911897,
|
|
"eval_logits/chosen": -1.8186331987380981,
|
|
"eval_logits/rejected": -1.8828355073928833,
|
|
"eval_logps/chosen": -571.3914794921875,
|
|
"eval_logps/rejected": -520.0492553710938,
|
|
"eval_loss": 0.6430271863937378,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1583155393600464,
|
|
"eval_rewards/margins": 0.19362236559391022,
|
|
"eval_rewards/rejected": -1.351938009262085,
|
|
"eval_runtime": 12.8415,
|
|
"eval_samples_per_second": 7.787,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 2.435741009134113,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 5.657327586206896e-07,
|
|
"logits/chosen": -1.8033708333969116,
|
|
"logits/rejected": -1.7999566793441772,
|
|
"logps/chosen": -550.396728515625,
|
|
"logps/rejected": -535.7592163085938,
|
|
"loss": 0.6,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -1.0161781311035156,
|
|
"rewards/margins": 0.3219712972640991,
|
|
"rewards/rejected": -1.3381493091583252,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"epoch": 2.4465266776770367,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 5.549568965517241e-07,
|
|
"logits/chosen": -1.8223384618759155,
|
|
"logits/rejected": -1.831163763999939,
|
|
"logps/chosen": -535.5626220703125,
|
|
"logps/rejected": -530.4205322265625,
|
|
"loss": 0.5904,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -1.0250645875930786,
|
|
"rewards/margins": 0.32182878255844116,
|
|
"rewards/rejected": -1.346893310546875,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"epoch": 2.45731234621996,
|
|
"grad_norm": 19.5,
|
|
"learning_rate": 5.441810344827586e-07,
|
|
"logits/chosen": -1.793067216873169,
|
|
"logits/rejected": -1.8330637216567993,
|
|
"logps/chosen": -552.1552124023438,
|
|
"logps/rejected": -533.558349609375,
|
|
"loss": 0.5834,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -1.0002939701080322,
|
|
"rewards/margins": 0.34119078516960144,
|
|
"rewards/rejected": -1.341484785079956,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"epoch": 2.4680980147628837,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 5.33405172413793e-07,
|
|
"logits/chosen": -1.7839548587799072,
|
|
"logits/rejected": -1.7871309518814087,
|
|
"logps/chosen": -559.6815185546875,
|
|
"logps/rejected": -536.8385009765625,
|
|
"loss": 0.6067,
|
|
"rewards/accuracies": 0.6781250238418579,
|
|
"rewards/chosen": -1.0037622451782227,
|
|
"rewards/margins": 0.2857220768928528,
|
|
"rewards/rejected": -1.2894842624664307,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"epoch": 2.478883683305807,
|
|
"grad_norm": 18.5,
|
|
"learning_rate": 5.226293103448276e-07,
|
|
"logits/chosen": -1.8063087463378906,
|
|
"logits/rejected": -1.8256887197494507,
|
|
"logps/chosen": -540.8095703125,
|
|
"logps/rejected": -519.8458251953125,
|
|
"loss": 0.5698,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -0.9602404832839966,
|
|
"rewards/margins": 0.3675882816314697,
|
|
"rewards/rejected": -1.3278287649154663,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 2.478883683305807,
|
|
"eval_logits/chosen": -1.820350170135498,
|
|
"eval_logits/rejected": -1.8840982913970947,
|
|
"eval_logps/chosen": -571.4730224609375,
|
|
"eval_logps/rejected": -520.2723999023438,
|
|
"eval_loss": 0.6414951086044312,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.16239595413208,
|
|
"eval_rewards/margins": 0.20070229470729828,
|
|
"eval_rewards/rejected": -1.3630982637405396,
|
|
"eval_runtime": 12.8591,
|
|
"eval_samples_per_second": 7.777,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 2.489669351848731,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 5.118534482758621e-07,
|
|
"logits/chosen": -1.8190813064575195,
|
|
"logits/rejected": -1.823015809059143,
|
|
"logps/chosen": -526.2591552734375,
|
|
"logps/rejected": -516.8885498046875,
|
|
"loss": 0.5822,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -0.9837486147880554,
|
|
"rewards/margins": 0.3439069986343384,
|
|
"rewards/rejected": -1.327655553817749,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"epoch": 2.5004550203916547,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 5.010775862068965e-07,
|
|
"logits/chosen": -1.8792375326156616,
|
|
"logits/rejected": -1.8921849727630615,
|
|
"logps/chosen": -548.6787719726562,
|
|
"logps/rejected": -541.21435546875,
|
|
"loss": 0.5713,
|
|
"rewards/accuracies": 0.739062488079071,
|
|
"rewards/chosen": -1.0175832509994507,
|
|
"rewards/margins": 0.37862181663513184,
|
|
"rewards/rejected": -1.396205186843872,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"epoch": 2.511240688934578,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 4.903017241379311e-07,
|
|
"logits/chosen": -1.7593591213226318,
|
|
"logits/rejected": -1.772139549255371,
|
|
"logps/chosen": -526.41650390625,
|
|
"logps/rejected": -535.5494384765625,
|
|
"loss": 0.5849,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -0.9334033727645874,
|
|
"rewards/margins": 0.32959261536598206,
|
|
"rewards/rejected": -1.2629960775375366,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"epoch": 2.5220263574775017,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 4.795258620689656e-07,
|
|
"logits/chosen": -1.7565523386001587,
|
|
"logits/rejected": -1.7509338855743408,
|
|
"logps/chosen": -560.8541870117188,
|
|
"logps/rejected": -542.8634643554688,
|
|
"loss": 0.5863,
|
|
"rewards/accuracies": 0.6968749761581421,
|
|
"rewards/chosen": -0.9780709147453308,
|
|
"rewards/margins": 0.33096015453338623,
|
|
"rewards/rejected": -1.3090311288833618,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"epoch": 2.532812026020425,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 4.6875e-07,
|
|
"logits/chosen": -1.8536609411239624,
|
|
"logits/rejected": -1.890777349472046,
|
|
"logps/chosen": -559.2564697265625,
|
|
"logps/rejected": -546.4683837890625,
|
|
"loss": 0.5739,
|
|
"rewards/accuracies": 0.734375,
|
|
"rewards/chosen": -1.012340784072876,
|
|
"rewards/margins": 0.37036722898483276,
|
|
"rewards/rejected": -1.382707953453064,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 2.532812026020425,
|
|
"eval_logits/chosen": -1.8193501234054565,
|
|
"eval_logits/rejected": -1.8834855556488037,
|
|
"eval_logps/chosen": -571.5364379882812,
|
|
"eval_logps/rejected": -520.3049926757812,
|
|
"eval_loss": 0.6403661966323853,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.16556715965271,
|
|
"eval_rewards/margins": 0.19915804266929626,
|
|
"eval_rewards/rejected": -1.3647252321243286,
|
|
"eval_runtime": 12.8276,
|
|
"eval_samples_per_second": 7.796,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 2.543597694563349,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 4.5797413793103446e-07,
|
|
"logits/chosen": -1.8258733749389648,
|
|
"logits/rejected": -1.8301451206207275,
|
|
"logps/chosen": -536.6988525390625,
|
|
"logps/rejected": -524.0502319335938,
|
|
"loss": 0.5685,
|
|
"rewards/accuracies": 0.729687511920929,
|
|
"rewards/chosen": -0.9772807359695435,
|
|
"rewards/margins": 0.3694283366203308,
|
|
"rewards/rejected": -1.3467090129852295,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"epoch": 2.5543833631062727,
|
|
"grad_norm": 22.375,
|
|
"learning_rate": 4.4719827586206897e-07,
|
|
"logits/chosen": -1.8070919513702393,
|
|
"logits/rejected": -1.7797702550888062,
|
|
"logps/chosen": -574.8311767578125,
|
|
"logps/rejected": -568.0702514648438,
|
|
"loss": 0.575,
|
|
"rewards/accuracies": 0.7484375238418579,
|
|
"rewards/chosen": -0.9659520983695984,
|
|
"rewards/margins": 0.3521731197834015,
|
|
"rewards/rejected": -1.3181252479553223,
|
|
"step": 2370
|
|
},
|
|
{
|
|
"epoch": 2.565169031649196,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 4.364224137931034e-07,
|
|
"logits/chosen": -1.8053117990493774,
|
|
"logits/rejected": -1.8097355365753174,
|
|
"logps/chosen": -553.0218505859375,
|
|
"logps/rejected": -546.5211181640625,
|
|
"loss": 0.5767,
|
|
"rewards/accuracies": 0.745312511920929,
|
|
"rewards/chosen": -0.9683555364608765,
|
|
"rewards/margins": 0.35642895102500916,
|
|
"rewards/rejected": -1.324784517288208,
|
|
"step": 2380
|
|
},
|
|
{
|
|
"epoch": 2.5759547001921197,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 4.2564655172413793e-07,
|
|
"logits/chosen": -1.805479645729065,
|
|
"logits/rejected": -1.838136911392212,
|
|
"logps/chosen": -561.8768920898438,
|
|
"logps/rejected": -540.8943481445312,
|
|
"loss": 0.5959,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -1.0204946994781494,
|
|
"rewards/margins": 0.3259061276912689,
|
|
"rewards/rejected": -1.3464009761810303,
|
|
"step": 2390
|
|
},
|
|
{
|
|
"epoch": 2.586740368735043,
|
|
"grad_norm": 21.625,
|
|
"learning_rate": 4.148706896551724e-07,
|
|
"logits/chosen": -1.8230018615722656,
|
|
"logits/rejected": -1.8241417407989502,
|
|
"logps/chosen": -530.9923706054688,
|
|
"logps/rejected": -517.3659057617188,
|
|
"loss": 0.5808,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -0.9844582676887512,
|
|
"rewards/margins": 0.3568059206008911,
|
|
"rewards/rejected": -1.341264247894287,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 2.586740368735043,
|
|
"eval_logits/chosen": -1.8191014528274536,
|
|
"eval_logits/rejected": -1.8839582204818726,
|
|
"eval_logps/chosen": -571.337646484375,
|
|
"eval_logps/rejected": -520.2830200195312,
|
|
"eval_loss": 0.6378647089004517,
|
|
"eval_rewards/accuracies": 0.6346153616905212,
|
|
"eval_rewards/chosen": -1.15562903881073,
|
|
"eval_rewards/margins": 0.20799769461154938,
|
|
"eval_rewards/rejected": -1.3636267185211182,
|
|
"eval_runtime": 12.8404,
|
|
"eval_samples_per_second": 7.788,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 2.5975260372779667,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 4.0409482758620694e-07,
|
|
"logits/chosen": -1.8546650409698486,
|
|
"logits/rejected": -1.8732141256332397,
|
|
"logps/chosen": -525.1890869140625,
|
|
"logps/rejected": -511.1866760253906,
|
|
"loss": 0.5827,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -1.0207396745681763,
|
|
"rewards/margins": 0.34293699264526367,
|
|
"rewards/rejected": -1.36367666721344,
|
|
"step": 2410
|
|
},
|
|
{
|
|
"epoch": 2.6083117058208902,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 3.9331896551724145e-07,
|
|
"logits/chosen": -1.7957582473754883,
|
|
"logits/rejected": -1.8112951517105103,
|
|
"logps/chosen": -539.9984130859375,
|
|
"logps/rejected": -546.8135375976562,
|
|
"loss": 0.5695,
|
|
"rewards/accuracies": 0.745312511920929,
|
|
"rewards/chosen": -0.9470235109329224,
|
|
"rewards/margins": 0.41715818643569946,
|
|
"rewards/rejected": -1.364181637763977,
|
|
"step": 2420
|
|
},
|
|
{
|
|
"epoch": 2.619097374363814,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 3.825431034482759e-07,
|
|
"logits/chosen": -1.8335113525390625,
|
|
"logits/rejected": -1.8350120782852173,
|
|
"logps/chosen": -557.8736572265625,
|
|
"logps/rejected": -546.7265014648438,
|
|
"loss": 0.5721,
|
|
"rewards/accuracies": 0.731249988079071,
|
|
"rewards/chosen": -0.944877028465271,
|
|
"rewards/margins": 0.3734719753265381,
|
|
"rewards/rejected": -1.318349003791809,
|
|
"step": 2430
|
|
},
|
|
{
|
|
"epoch": 2.6298830429067377,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 3.7176724137931035e-07,
|
|
"logits/chosen": -1.8570324182510376,
|
|
"logits/rejected": -1.875575065612793,
|
|
"logps/chosen": -540.6766967773438,
|
|
"logps/rejected": -514.4322509765625,
|
|
"loss": 0.5742,
|
|
"rewards/accuracies": 0.7203124761581421,
|
|
"rewards/chosen": -1.010191559791565,
|
|
"rewards/margins": 0.363949716091156,
|
|
"rewards/rejected": -1.3741410970687866,
|
|
"step": 2440
|
|
},
|
|
{
|
|
"epoch": 2.640668711449661,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 3.6099137931034486e-07,
|
|
"logits/chosen": -1.8072179555892944,
|
|
"logits/rejected": -1.8312402963638306,
|
|
"logps/chosen": -570.2026977539062,
|
|
"logps/rejected": -541.2198486328125,
|
|
"loss": 0.5743,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -0.9453884959220886,
|
|
"rewards/margins": 0.36282581090927124,
|
|
"rewards/rejected": -1.3082144260406494,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 2.640668711449661,
|
|
"eval_logits/chosen": -1.8189289569854736,
|
|
"eval_logits/rejected": -1.882326602935791,
|
|
"eval_logps/chosen": -571.3778076171875,
|
|
"eval_logps/rejected": -520.1837158203125,
|
|
"eval_loss": 0.6420935988426208,
|
|
"eval_rewards/accuracies": 0.6538461446762085,
|
|
"eval_rewards/chosen": -1.1576350927352905,
|
|
"eval_rewards/margins": 0.2010272890329361,
|
|
"eval_rewards/rejected": -1.3586623668670654,
|
|
"eval_runtime": 12.8126,
|
|
"eval_samples_per_second": 7.805,
|
|
"eval_steps_per_second": 1.015,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 2.6514543799925847,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 3.502155172413793e-07,
|
|
"logits/chosen": -1.8071362972259521,
|
|
"logits/rejected": -1.8451793193817139,
|
|
"logps/chosen": -558.2564086914062,
|
|
"logps/rejected": -535.6451416015625,
|
|
"loss": 0.5825,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -1.0327677726745605,
|
|
"rewards/margins": 0.3471210300922394,
|
|
"rewards/rejected": -1.379888892173767,
|
|
"step": 2460
|
|
},
|
|
{
|
|
"epoch": 2.6622400485355087,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 3.394396551724138e-07,
|
|
"logits/chosen": -1.7823054790496826,
|
|
"logits/rejected": -1.800473928451538,
|
|
"logps/chosen": -574.7456665039062,
|
|
"logps/rejected": -547.697265625,
|
|
"loss": 0.5809,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -1.008268117904663,
|
|
"rewards/margins": 0.3359683156013489,
|
|
"rewards/rejected": -1.3442364931106567,
|
|
"step": 2470
|
|
},
|
|
{
|
|
"epoch": 2.673025717078432,
|
|
"grad_norm": 19.0,
|
|
"learning_rate": 3.2866379310344827e-07,
|
|
"logits/chosen": -1.8167625665664673,
|
|
"logits/rejected": -1.8581584692001343,
|
|
"logps/chosen": -546.1006469726562,
|
|
"logps/rejected": -524.2261962890625,
|
|
"loss": 0.5784,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -0.9644547700881958,
|
|
"rewards/margins": 0.3466854691505432,
|
|
"rewards/rejected": -1.3111401796340942,
|
|
"step": 2480
|
|
},
|
|
{
|
|
"epoch": 2.6838113856213557,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 3.1788793103448277e-07,
|
|
"logits/chosen": -1.8069427013397217,
|
|
"logits/rejected": -1.826890230178833,
|
|
"logps/chosen": -549.4500122070312,
|
|
"logps/rejected": -528.8165283203125,
|
|
"loss": 0.5792,
|
|
"rewards/accuracies": 0.706250011920929,
|
|
"rewards/chosen": -1.0348377227783203,
|
|
"rewards/margins": 0.3798985183238983,
|
|
"rewards/rejected": -1.414736270904541,
|
|
"step": 2490
|
|
},
|
|
{
|
|
"epoch": 2.694597054164279,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 3.071120689655173e-07,
|
|
"logits/chosen": -1.8492558002471924,
|
|
"logits/rejected": -1.9055678844451904,
|
|
"logps/chosen": -531.1982421875,
|
|
"logps/rejected": -509.7709045410156,
|
|
"loss": 0.5947,
|
|
"rewards/accuracies": 0.6968749761581421,
|
|
"rewards/chosen": -0.9890526533126831,
|
|
"rewards/margins": 0.31767064332962036,
|
|
"rewards/rejected": -1.3067232370376587,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 2.694597054164279,
|
|
"eval_logits/chosen": -1.8213615417480469,
|
|
"eval_logits/rejected": -1.8862541913986206,
|
|
"eval_logps/chosen": -571.5040283203125,
|
|
"eval_logps/rejected": -520.3077392578125,
|
|
"eval_loss": 0.6428215503692627,
|
|
"eval_rewards/accuracies": 0.625,
|
|
"eval_rewards/chosen": -1.163952350616455,
|
|
"eval_rewards/margins": 0.2009093016386032,
|
|
"eval_rewards/rejected": -1.3648617267608643,
|
|
"eval_runtime": 12.8298,
|
|
"eval_samples_per_second": 7.794,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 2.7053827227072027,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 2.9633620689655173e-07,
|
|
"logits/chosen": -1.8131103515625,
|
|
"logits/rejected": -1.8387792110443115,
|
|
"logps/chosen": -550.4927978515625,
|
|
"logps/rejected": -543.66162109375,
|
|
"loss": 0.5763,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.9916373491287231,
|
|
"rewards/margins": 0.38830363750457764,
|
|
"rewards/rejected": -1.3799409866333008,
|
|
"step": 2510
|
|
},
|
|
{
|
|
"epoch": 2.716168391250126,
|
|
"grad_norm": 23.375,
|
|
"learning_rate": 2.855603448275862e-07,
|
|
"logits/chosen": -1.8427207469940186,
|
|
"logits/rejected": -1.8411712646484375,
|
|
"logps/chosen": -552.4677734375,
|
|
"logps/rejected": -533.5294799804688,
|
|
"loss": 0.5717,
|
|
"rewards/accuracies": 0.6890624761581421,
|
|
"rewards/chosen": -1.0246634483337402,
|
|
"rewards/margins": 0.3778466582298279,
|
|
"rewards/rejected": -1.402510166168213,
|
|
"step": 2520
|
|
},
|
|
{
|
|
"epoch": 2.7269540597930497,
|
|
"grad_norm": 21.125,
|
|
"learning_rate": 2.747844827586207e-07,
|
|
"logits/chosen": -1.7936853170394897,
|
|
"logits/rejected": -1.7962579727172852,
|
|
"logps/chosen": -556.09521484375,
|
|
"logps/rejected": -547.2576904296875,
|
|
"loss": 0.5785,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -1.012605905532837,
|
|
"rewards/margins": 0.35908856987953186,
|
|
"rewards/rejected": -1.3716943264007568,
|
|
"step": 2530
|
|
},
|
|
{
|
|
"epoch": 2.7377397283359737,
|
|
"grad_norm": 20.375,
|
|
"learning_rate": 2.6400862068965514e-07,
|
|
"logits/chosen": -1.7601362466812134,
|
|
"logits/rejected": -1.7817468643188477,
|
|
"logps/chosen": -525.7601928710938,
|
|
"logps/rejected": -519.701171875,
|
|
"loss": 0.5731,
|
|
"rewards/accuracies": 0.7328125238418579,
|
|
"rewards/chosen": -0.9400823712348938,
|
|
"rewards/margins": 0.3647005558013916,
|
|
"rewards/rejected": -1.3047831058502197,
|
|
"step": 2540
|
|
},
|
|
{
|
|
"epoch": 2.748525396878897,
|
|
"grad_norm": 19.875,
|
|
"learning_rate": 2.532327586206897e-07,
|
|
"logits/chosen": -1.8128465414047241,
|
|
"logits/rejected": -1.8134959936141968,
|
|
"logps/chosen": -549.0126342773438,
|
|
"logps/rejected": -539.8706665039062,
|
|
"loss": 0.5836,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -1.0344898700714111,
|
|
"rewards/margins": 0.35030892491340637,
|
|
"rewards/rejected": -1.3847988843917847,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 2.748525396878897,
|
|
"eval_logits/chosen": -1.821547269821167,
|
|
"eval_logits/rejected": -1.8859292268753052,
|
|
"eval_logps/chosen": -571.573974609375,
|
|
"eval_logps/rejected": -520.504150390625,
|
|
"eval_loss": 0.6378675103187561,
|
|
"eval_rewards/accuracies": 0.6634615659713745,
|
|
"eval_rewards/chosen": -1.167443037033081,
|
|
"eval_rewards/margins": 0.20724283158779144,
|
|
"eval_rewards/rejected": -1.3746860027313232,
|
|
"eval_runtime": 12.8565,
|
|
"eval_samples_per_second": 7.778,
|
|
"eval_steps_per_second": 1.011,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 2.7593110654218207,
|
|
"grad_norm": 24.875,
|
|
"learning_rate": 2.4245689655172415e-07,
|
|
"logits/chosen": -1.8031337261199951,
|
|
"logits/rejected": -1.7945318222045898,
|
|
"logps/chosen": -555.7886962890625,
|
|
"logps/rejected": -544.8328857421875,
|
|
"loss": 0.594,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.9998787045478821,
|
|
"rewards/margins": 0.3179456293582916,
|
|
"rewards/rejected": -1.3178244829177856,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 2.770096733964744,
|
|
"grad_norm": 22.0,
|
|
"learning_rate": 2.3168103448275863e-07,
|
|
"logits/chosen": -1.7943198680877686,
|
|
"logits/rejected": -1.8033891916275024,
|
|
"logps/chosen": -543.3651733398438,
|
|
"logps/rejected": -532.5569458007812,
|
|
"loss": 0.5897,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -0.997797966003418,
|
|
"rewards/margins": 0.326891154050827,
|
|
"rewards/rejected": -1.3246891498565674,
|
|
"step": 2570
|
|
},
|
|
{
|
|
"epoch": 2.780882402507668,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 2.209051724137931e-07,
|
|
"logits/chosen": -1.8031666278839111,
|
|
"logits/rejected": -1.8120911121368408,
|
|
"logps/chosen": -544.0496826171875,
|
|
"logps/rejected": -541.0577392578125,
|
|
"loss": 0.5774,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.9857531785964966,
|
|
"rewards/margins": 0.35895711183547974,
|
|
"rewards/rejected": -1.3447102308273315,
|
|
"step": 2580
|
|
},
|
|
{
|
|
"epoch": 2.7916680710505917,
|
|
"grad_norm": 19.75,
|
|
"learning_rate": 2.1012931034482756e-07,
|
|
"logits/chosen": -1.8335663080215454,
|
|
"logits/rejected": -1.8373979330062866,
|
|
"logps/chosen": -548.4537963867188,
|
|
"logps/rejected": -540.0748291015625,
|
|
"loss": 0.5892,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -1.0308494567871094,
|
|
"rewards/margins": 0.3219280242919922,
|
|
"rewards/rejected": -1.3527776002883911,
|
|
"step": 2590
|
|
},
|
|
{
|
|
"epoch": 2.802453739593515,
|
|
"grad_norm": 20.875,
|
|
"learning_rate": 1.993534482758621e-07,
|
|
"logits/chosen": -1.8183438777923584,
|
|
"logits/rejected": -1.81350576877594,
|
|
"logps/chosen": -543.9989624023438,
|
|
"logps/rejected": -523.77490234375,
|
|
"loss": 0.5911,
|
|
"rewards/accuracies": 0.7124999761581421,
|
|
"rewards/chosen": -1.045743703842163,
|
|
"rewards/margins": 0.32820892333984375,
|
|
"rewards/rejected": -1.3739526271820068,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 2.802453739593515,
|
|
"eval_logits/chosen": -1.8199844360351562,
|
|
"eval_logits/rejected": -1.8842284679412842,
|
|
"eval_logps/chosen": -571.6898193359375,
|
|
"eval_logps/rejected": -520.4623413085938,
|
|
"eval_loss": 0.6425648331642151,
|
|
"eval_rewards/accuracies": 0.625,
|
|
"eval_rewards/chosen": -1.1732372045516968,
|
|
"eval_rewards/margins": 0.19935639202594757,
|
|
"eval_rewards/rejected": -1.3725935220718384,
|
|
"eval_runtime": 12.8299,
|
|
"eval_samples_per_second": 7.794,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 2.8132394081364387,
|
|
"grad_norm": 22.875,
|
|
"learning_rate": 1.8857758620689658e-07,
|
|
"logits/chosen": -1.7655651569366455,
|
|
"logits/rejected": -1.7627290487289429,
|
|
"logps/chosen": -573.8090209960938,
|
|
"logps/rejected": -579.2223510742188,
|
|
"loss": 0.588,
|
|
"rewards/accuracies": 0.7109375,
|
|
"rewards/chosen": -0.9776386022567749,
|
|
"rewards/margins": 0.37593865394592285,
|
|
"rewards/rejected": -1.3535772562026978,
|
|
"step": 2610
|
|
},
|
|
{
|
|
"epoch": 2.824025076679362,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 1.7780172413793103e-07,
|
|
"logits/chosen": -1.8132917881011963,
|
|
"logits/rejected": -1.833608865737915,
|
|
"logps/chosen": -539.3812866210938,
|
|
"logps/rejected": -532.1210327148438,
|
|
"loss": 0.5694,
|
|
"rewards/accuracies": 0.734375,
|
|
"rewards/chosen": -0.993198037147522,
|
|
"rewards/margins": 0.3927215039730072,
|
|
"rewards/rejected": -1.3859195709228516,
|
|
"step": 2620
|
|
},
|
|
{
|
|
"epoch": 2.8348107452222857,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.670258620689655e-07,
|
|
"logits/chosen": -1.8571910858154297,
|
|
"logits/rejected": -1.8688180446624756,
|
|
"logps/chosen": -533.7222900390625,
|
|
"logps/rejected": -527.7282104492188,
|
|
"loss": 0.5942,
|
|
"rewards/accuracies": 0.6937500238418579,
|
|
"rewards/chosen": -1.0575587749481201,
|
|
"rewards/margins": 0.3217419385910034,
|
|
"rewards/rejected": -1.379300832748413,
|
|
"step": 2630
|
|
},
|
|
{
|
|
"epoch": 2.845596413765209,
|
|
"grad_norm": 22.125,
|
|
"learning_rate": 1.5625e-07,
|
|
"logits/chosen": -1.801645278930664,
|
|
"logits/rejected": -1.8315293788909912,
|
|
"logps/chosen": -531.2138061523438,
|
|
"logps/rejected": -502.26953125,
|
|
"loss": 0.5914,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -0.9891794323921204,
|
|
"rewards/margins": 0.3226587176322937,
|
|
"rewards/rejected": -1.3118380308151245,
|
|
"step": 2640
|
|
},
|
|
{
|
|
"epoch": 2.856382082308133,
|
|
"grad_norm": 19.0,
|
|
"learning_rate": 1.454741379310345e-07,
|
|
"logits/chosen": -1.8319976329803467,
|
|
"logits/rejected": -1.8414185047149658,
|
|
"logps/chosen": -547.1237182617188,
|
|
"logps/rejected": -527.4207763671875,
|
|
"loss": 0.5862,
|
|
"rewards/accuracies": 0.7046874761581421,
|
|
"rewards/chosen": -1.0397682189941406,
|
|
"rewards/margins": 0.3290941119194031,
|
|
"rewards/rejected": -1.3688621520996094,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 2.856382082308133,
|
|
"eval_logits/chosen": -1.8245118856430054,
|
|
"eval_logits/rejected": -1.8896102905273438,
|
|
"eval_logps/chosen": -571.779052734375,
|
|
"eval_logps/rejected": -520.575439453125,
|
|
"eval_loss": 0.6429222822189331,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1776981353759766,
|
|
"eval_rewards/margins": 0.20054802298545837,
|
|
"eval_rewards/rejected": -1.3782461881637573,
|
|
"eval_runtime": 12.8292,
|
|
"eval_samples_per_second": 7.795,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 2.8671677508510567,
|
|
"grad_norm": 20.75,
|
|
"learning_rate": 1.3469827586206897e-07,
|
|
"logits/chosen": -1.8030239343643188,
|
|
"logits/rejected": -1.8243894577026367,
|
|
"logps/chosen": -543.3869018554688,
|
|
"logps/rejected": -528.6986083984375,
|
|
"loss": 0.582,
|
|
"rewards/accuracies": 0.7328125238418579,
|
|
"rewards/chosen": -0.964773952960968,
|
|
"rewards/margins": 0.3342544436454773,
|
|
"rewards/rejected": -1.2990283966064453,
|
|
"step": 2660
|
|
},
|
|
{
|
|
"epoch": 2.87795341939398,
|
|
"grad_norm": 21.0,
|
|
"learning_rate": 1.2392241379310345e-07,
|
|
"logits/chosen": -1.7584861516952515,
|
|
"logits/rejected": -1.7730573415756226,
|
|
"logps/chosen": -561.669677734375,
|
|
"logps/rejected": -548.2999267578125,
|
|
"loss": 0.6069,
|
|
"rewards/accuracies": 0.671875,
|
|
"rewards/chosen": -1.0046155452728271,
|
|
"rewards/margins": 0.30270934104919434,
|
|
"rewards/rejected": -1.3073248863220215,
|
|
"step": 2670
|
|
},
|
|
{
|
|
"epoch": 2.8887390879369037,
|
|
"grad_norm": 20.25,
|
|
"learning_rate": 1.1314655172413793e-07,
|
|
"logits/chosen": -1.7781982421875,
|
|
"logits/rejected": -1.7862926721572876,
|
|
"logps/chosen": -554.6754150390625,
|
|
"logps/rejected": -538.4164428710938,
|
|
"loss": 0.5843,
|
|
"rewards/accuracies": 0.715624988079071,
|
|
"rewards/chosen": -1.047992467880249,
|
|
"rewards/margins": 0.3425953984260559,
|
|
"rewards/rejected": -1.3905879259109497,
|
|
"step": 2680
|
|
},
|
|
{
|
|
"epoch": 2.8995247564798277,
|
|
"grad_norm": 21.25,
|
|
"learning_rate": 1.0237068965517242e-07,
|
|
"logits/chosen": -1.8572750091552734,
|
|
"logits/rejected": -1.8793262243270874,
|
|
"logps/chosen": -549.9866943359375,
|
|
"logps/rejected": -524.3643188476562,
|
|
"loss": 0.5601,
|
|
"rewards/accuracies": 0.746874988079071,
|
|
"rewards/chosen": -1.0121921300888062,
|
|
"rewards/margins": 0.3896443247795105,
|
|
"rewards/rejected": -1.401836633682251,
|
|
"step": 2690
|
|
},
|
|
{
|
|
"epoch": 2.910310425022751,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 9.15948275862069e-08,
|
|
"logits/chosen": -1.80484938621521,
|
|
"logits/rejected": -1.80769944190979,
|
|
"logps/chosen": -524.7546997070312,
|
|
"logps/rejected": -520.0099487304688,
|
|
"loss": 0.5821,
|
|
"rewards/accuracies": 0.7281249761581421,
|
|
"rewards/chosen": -0.9889782071113586,
|
|
"rewards/margins": 0.34463515877723694,
|
|
"rewards/rejected": -1.333613395690918,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 2.910310425022751,
|
|
"eval_logits/chosen": -1.8193469047546387,
|
|
"eval_logits/rejected": -1.8841547966003418,
|
|
"eval_logps/chosen": -571.4869384765625,
|
|
"eval_logps/rejected": -520.3854370117188,
|
|
"eval_loss": 0.6395397782325745,
|
|
"eval_rewards/accuracies": 0.6730769276618958,
|
|
"eval_rewards/chosen": -1.163095235824585,
|
|
"eval_rewards/margins": 0.2056487500667572,
|
|
"eval_rewards/rejected": -1.368743896484375,
|
|
"eval_runtime": 12.8307,
|
|
"eval_samples_per_second": 7.794,
|
|
"eval_steps_per_second": 1.013,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 2.9210960935656747,
|
|
"grad_norm": 22.625,
|
|
"learning_rate": 8.081896551724138e-08,
|
|
"logits/chosen": -1.8104597330093384,
|
|
"logits/rejected": -1.8475234508514404,
|
|
"logps/chosen": -556.8909912109375,
|
|
"logps/rejected": -538.2855224609375,
|
|
"loss": 0.5742,
|
|
"rewards/accuracies": 0.737500011920929,
|
|
"rewards/chosen": -0.9970060586929321,
|
|
"rewards/margins": 0.3584115207195282,
|
|
"rewards/rejected": -1.3554174900054932,
|
|
"step": 2710
|
|
},
|
|
{
|
|
"epoch": 2.931881762108598,
|
|
"grad_norm": 20.125,
|
|
"learning_rate": 7.004310344827586e-08,
|
|
"logits/chosen": -1.8974506855010986,
|
|
"logits/rejected": -1.9189106225967407,
|
|
"logps/chosen": -536.3151245117188,
|
|
"logps/rejected": -511.659912109375,
|
|
"loss": 0.5842,
|
|
"rewards/accuracies": 0.721875011920929,
|
|
"rewards/chosen": -1.0638625621795654,
|
|
"rewards/margins": 0.34429508447647095,
|
|
"rewards/rejected": -1.4081575870513916,
|
|
"step": 2720
|
|
},
|
|
{
|
|
"epoch": 2.9426674306515217,
|
|
"grad_norm": 23.0,
|
|
"learning_rate": 5.9267241379310345e-08,
|
|
"logits/chosen": -1.7627315521240234,
|
|
"logits/rejected": -1.7557799816131592,
|
|
"logps/chosen": -555.1419677734375,
|
|
"logps/rejected": -536.8969116210938,
|
|
"loss": 0.5864,
|
|
"rewards/accuracies": 0.714062511920929,
|
|
"rewards/chosen": -0.9729493856430054,
|
|
"rewards/margins": 0.33462923765182495,
|
|
"rewards/rejected": -1.3075785636901855,
|
|
"step": 2730
|
|
},
|
|
{
|
|
"epoch": 2.953453099194445,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 4.849137931034483e-08,
|
|
"logits/chosen": -1.7903178930282593,
|
|
"logits/rejected": -1.7958800792694092,
|
|
"logps/chosen": -540.8969116210938,
|
|
"logps/rejected": -532.142578125,
|
|
"loss": 0.5838,
|
|
"rewards/accuracies": 0.723437488079071,
|
|
"rewards/chosen": -0.9786058664321899,
|
|
"rewards/margins": 0.3315025866031647,
|
|
"rewards/rejected": -1.3101084232330322,
|
|
"step": 2740
|
|
},
|
|
{
|
|
"epoch": 2.9642387677373687,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 3.771551724137931e-08,
|
|
"logits/chosen": -1.7704594135284424,
|
|
"logits/rejected": -1.7803184986114502,
|
|
"logps/chosen": -546.58349609375,
|
|
"logps/rejected": -545.2351684570312,
|
|
"loss": 0.5639,
|
|
"rewards/accuracies": 0.71875,
|
|
"rewards/chosen": -1.017333984375,
|
|
"rewards/margins": 0.38938578963279724,
|
|
"rewards/rejected": -1.4067199230194092,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 2.9642387677373687,
|
|
"eval_logits/chosen": -1.8192270994186401,
|
|
"eval_logits/rejected": -1.8834381103515625,
|
|
"eval_logps/chosen": -571.4424438476562,
|
|
"eval_logps/rejected": -520.3687744140625,
|
|
"eval_loss": 0.6384233236312866,
|
|
"eval_rewards/accuracies": 0.6442307829856873,
|
|
"eval_rewards/chosen": -1.1608668565750122,
|
|
"eval_rewards/margins": 0.20704683661460876,
|
|
"eval_rewards/rejected": -1.3679137229919434,
|
|
"eval_runtime": 12.8408,
|
|
"eval_samples_per_second": 7.788,
|
|
"eval_steps_per_second": 1.012,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 2.9750244362802927,
|
|
"grad_norm": 20.625,
|
|
"learning_rate": 2.6939655172413792e-08,
|
|
"logits/chosen": -1.8296130895614624,
|
|
"logits/rejected": -1.8367893695831299,
|
|
"logps/chosen": -521.9843139648438,
|
|
"logps/rejected": -498.4405822753906,
|
|
"loss": 0.597,
|
|
"rewards/accuracies": 0.7171875238418579,
|
|
"rewards/chosen": -1.0047184228897095,
|
|
"rewards/margins": 0.30621933937072754,
|
|
"rewards/rejected": -1.3109376430511475,
|
|
"step": 2760
|
|
},
|
|
{
|
|
"epoch": 2.985810104823216,
|
|
"grad_norm": 21.375,
|
|
"learning_rate": 1.6163793103448278e-08,
|
|
"logits/chosen": -1.8125426769256592,
|
|
"logits/rejected": -1.837774634361267,
|
|
"logps/chosen": -573.8707275390625,
|
|
"logps/rejected": -565.1752319335938,
|
|
"loss": 0.5891,
|
|
"rewards/accuracies": 0.7093750238418579,
|
|
"rewards/chosen": -1.0296361446380615,
|
|
"rewards/margins": 0.3325849175453186,
|
|
"rewards/rejected": -1.3622210025787354,
|
|
"step": 2770
|
|
},
|
|
{
|
|
"epoch": 2.9965957733661397,
|
|
"grad_norm": 20.5,
|
|
"learning_rate": 5.387931034482759e-09,
|
|
"logits/chosen": -1.765448808670044,
|
|
"logits/rejected": -1.7794584035873413,
|
|
"logps/chosen": -562.4325561523438,
|
|
"logps/rejected": -534.7550659179688,
|
|
"loss": 0.5723,
|
|
"rewards/accuracies": 0.739062488079071,
|
|
"rewards/chosen": -0.9770288467407227,
|
|
"rewards/margins": 0.3728591799736023,
|
|
"rewards/rejected": -1.3498878479003906,
|
|
"step": 2780
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 2784,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 3,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|