12260 lines
360 KiB
JSON
12260 lines
360 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 815,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.001226993865030675,
|
||
|
|
"grad_norm": 151.34863688436073,
|
||
|
|
"learning_rate": 4e-08,
|
||
|
|
"logits/chosen": -0.32421875,
|
||
|
|
"logits/rejected": -0.515625,
|
||
|
|
"logps/chosen": -474.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.6914,
|
||
|
|
"rewards/accuracies": 0.0,
|
||
|
|
"rewards/chosen": 0.0,
|
||
|
|
"rewards/margins": 0.0,
|
||
|
|
"rewards/rejected": 0.0,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.00245398773006135,
|
||
|
|
"grad_norm": 162.78084419893554,
|
||
|
|
"learning_rate": 8e-08,
|
||
|
|
"logits/chosen": -0.22265625,
|
||
|
|
"logits/rejected": -0.439453125,
|
||
|
|
"logps/chosen": -436.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.6914,
|
||
|
|
"rewards/accuracies": 0.0,
|
||
|
|
"rewards/chosen": 0.0,
|
||
|
|
"rewards/margins": 0.0,
|
||
|
|
"rewards/rejected": 0.0,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0036809815950920245,
|
||
|
|
"grad_norm": 106.82291187813345,
|
||
|
|
"learning_rate": 1.2e-07,
|
||
|
|
"logits/chosen": -0.12890625,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -272.0,
|
||
|
|
"loss": 0.6898,
|
||
|
|
"rewards/accuracies": 0.40625,
|
||
|
|
"rewards/chosen": 0.0250244140625,
|
||
|
|
"rewards/margins": 0.00946044921875,
|
||
|
|
"rewards/rejected": 0.01556396484375,
|
||
|
|
"step": 3
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0049079754601227,
|
||
|
|
"grad_norm": 134.12056184559836,
|
||
|
|
"learning_rate": 1.6e-07,
|
||
|
|
"logits/chosen": -0.166015625,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -242.0,
|
||
|
|
"loss": 0.7057,
|
||
|
|
"rewards/accuracies": 0.2578125,
|
||
|
|
"rewards/chosen": 0.00927734375,
|
||
|
|
"rewards/margins": -0.01495361328125,
|
||
|
|
"rewards/rejected": 0.0242919921875,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.006134969325153374,
|
||
|
|
"grad_norm": 303.3832578095922,
|
||
|
|
"learning_rate": 2e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -270.0,
|
||
|
|
"loss": 0.6884,
|
||
|
|
"rewards/accuracies": 0.34375,
|
||
|
|
"rewards/chosen": 0.03173828125,
|
||
|
|
"rewards/margins": 0.01458740234375,
|
||
|
|
"rewards/rejected": 0.01708984375,
|
||
|
|
"step": 5
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.007361963190184049,
|
||
|
|
"grad_norm": 323.77200794559565,
|
||
|
|
"learning_rate": 2.4e-07,
|
||
|
|
"logits/chosen": -0.255859375,
|
||
|
|
"logits/rejected": -0.345703125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.6888,
|
||
|
|
"rewards/accuracies": 0.4296875,
|
||
|
|
"rewards/chosen": 0.03076171875,
|
||
|
|
"rewards/margins": 0.0267333984375,
|
||
|
|
"rewards/rejected": 0.00408935546875,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.008588957055214725,
|
||
|
|
"grad_norm": 242.90815448785312,
|
||
|
|
"learning_rate": 2.8e-07,
|
||
|
|
"logits/chosen": -0.205078125,
|
||
|
|
"logits/rejected": -0.38671875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -266.0,
|
||
|
|
"loss": 0.6953,
|
||
|
|
"rewards/accuracies": 0.3828125,
|
||
|
|
"rewards/chosen": 0.0037841796875,
|
||
|
|
"rewards/margins": 0.00604248046875,
|
||
|
|
"rewards/rejected": -0.0022430419921875,
|
||
|
|
"step": 7
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0098159509202454,
|
||
|
|
"grad_norm": 127.34735766597795,
|
||
|
|
"learning_rate": 3.2e-07,
|
||
|
|
"logits/chosen": -0.1123046875,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.6919,
|
||
|
|
"rewards/accuracies": 0.3671875,
|
||
|
|
"rewards/chosen": 0.043701171875,
|
||
|
|
"rewards/margins": 0.01104736328125,
|
||
|
|
"rewards/rejected": 0.03271484375,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.011042944785276074,
|
||
|
|
"grad_norm": 115.54234285419575,
|
||
|
|
"learning_rate": 3.6e-07,
|
||
|
|
"logits/chosen": -0.1318359375,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.6835,
|
||
|
|
"rewards/accuracies": 0.3984375,
|
||
|
|
"rewards/chosen": 0.0537109375,
|
||
|
|
"rewards/margins": 0.0286865234375,
|
||
|
|
"rewards/rejected": 0.02490234375,
|
||
|
|
"step": 9
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.012269938650306749,
|
||
|
|
"grad_norm": 160.56175777535378,
|
||
|
|
"learning_rate": 4e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.3515625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -278.0,
|
||
|
|
"loss": 0.6455,
|
||
|
|
"rewards/accuracies": 0.5390625,
|
||
|
|
"rewards/chosen": 0.0810546875,
|
||
|
|
"rewards/margins": 0.10888671875,
|
||
|
|
"rewards/rejected": -0.02783203125,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.013496932515337423,
|
||
|
|
"grad_norm": 184.2665155417569,
|
||
|
|
"learning_rate": 4.3999999999999997e-07,
|
||
|
|
"logits/chosen": -0.38671875,
|
||
|
|
"logits/rejected": -0.52734375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.6427,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 0.0732421875,
|
||
|
|
"rewards/margins": 0.1201171875,
|
||
|
|
"rewards/rejected": -0.04638671875,
|
||
|
|
"step": 11
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.014723926380368098,
|
||
|
|
"grad_norm": 263.0691476498303,
|
||
|
|
"learning_rate": 4.8e-07,
|
||
|
|
"logits/chosen": -0.1396484375,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -262.0,
|
||
|
|
"loss": 0.6103,
|
||
|
|
"rewards/accuracies": 0.6328125,
|
||
|
|
"rewards/chosen": 0.1474609375,
|
||
|
|
"rewards/margins": 0.21875,
|
||
|
|
"rewards/rejected": -0.0712890625,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.015950920245398775,
|
||
|
|
"grad_norm": 102.73894172073577,
|
||
|
|
"learning_rate": 5.2e-07,
|
||
|
|
"logits/chosen": -0.2470703125,
|
||
|
|
"logits/rejected": -0.4140625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -250.0,
|
||
|
|
"loss": 0.5678,
|
||
|
|
"rewards/accuracies": 0.6484375,
|
||
|
|
"rewards/chosen": 0.220703125,
|
||
|
|
"rewards/margins": 0.357421875,
|
||
|
|
"rewards/rejected": -0.1357421875,
|
||
|
|
"step": 13
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01717791411042945,
|
||
|
|
"grad_norm": 91.63521794879738,
|
||
|
|
"learning_rate": 5.6e-07,
|
||
|
|
"logits/chosen": -0.2236328125,
|
||
|
|
"logits/rejected": -0.404296875,
|
||
|
|
"logps/chosen": -470.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.5302,
|
||
|
|
"rewards/accuracies": 0.71875,
|
||
|
|
"rewards/chosen": 0.396484375,
|
||
|
|
"rewards/margins": 0.52734375,
|
||
|
|
"rewards/rejected": -0.1298828125,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.018404907975460124,
|
||
|
|
"grad_norm": 64.07563855366159,
|
||
|
|
"learning_rate": 6e-07,
|
||
|
|
"logits/chosen": -0.1123046875,
|
||
|
|
"logits/rejected": -0.1826171875,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.5614,
|
||
|
|
"rewards/accuracies": 0.6171875,
|
||
|
|
"rewards/chosen": 0.310546875,
|
||
|
|
"rewards/margins": 0.435546875,
|
||
|
|
"rewards/rejected": -0.125,
|
||
|
|
"step": 15
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0196319018404908,
|
||
|
|
"grad_norm": 66.54828155920123,
|
||
|
|
"learning_rate": 6.4e-07,
|
||
|
|
"logits/chosen": -0.2578125,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.5093,
|
||
|
|
"rewards/accuracies": 0.71875,
|
||
|
|
"rewards/chosen": 0.412109375,
|
||
|
|
"rewards/margins": 0.62109375,
|
||
|
|
"rewards/rejected": -0.2080078125,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.020858895705521473,
|
||
|
|
"grad_norm": 69.61399588031946,
|
||
|
|
"learning_rate": 6.800000000000001e-07,
|
||
|
|
"logits/chosen": -0.1650390625,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -430.0,
|
||
|
|
"logps/rejected": -256.0,
|
||
|
|
"loss": 0.4793,
|
||
|
|
"rewards/accuracies": 0.7265625,
|
||
|
|
"rewards/chosen": 0.61328125,
|
||
|
|
"rewards/margins": 0.87890625,
|
||
|
|
"rewards/rejected": -0.267578125,
|
||
|
|
"step": 17
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.022085889570552148,
|
||
|
|
"grad_norm": 46.055018181732336,
|
||
|
|
"learning_rate": 7.2e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -278.0,
|
||
|
|
"loss": 0.4651,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.671875,
|
||
|
|
"rewards/margins": 1.09375,
|
||
|
|
"rewards/rejected": -0.41796875,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.023312883435582823,
|
||
|
|
"grad_norm": 77.14807287387903,
|
||
|
|
"learning_rate": 7.599999999999999e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.369140625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.5136,
|
||
|
|
"rewards/accuracies": 0.7109375,
|
||
|
|
"rewards/chosen": 0.56640625,
|
||
|
|
"rewards/margins": 1.1171875,
|
||
|
|
"rewards/rejected": -0.5546875,
|
||
|
|
"step": 19
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.024539877300613498,
|
||
|
|
"grad_norm": 52.96981054915384,
|
||
|
|
"learning_rate": 8e-07,
|
||
|
|
"logits/chosen": -0.279296875,
|
||
|
|
"logits/rejected": -0.44921875,
|
||
|
|
"logps/chosen": -460.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.4216,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.9609375,
|
||
|
|
"rewards/margins": 1.7890625,
|
||
|
|
"rewards/rejected": -0.83203125,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.025766871165644172,
|
||
|
|
"grad_norm": 49.55217029175373,
|
||
|
|
"learning_rate": 8.399999999999999e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -264.0,
|
||
|
|
"loss": 0.4123,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.7421875,
|
||
|
|
"rewards/margins": 1.390625,
|
||
|
|
"rewards/rejected": -0.6484375,
|
||
|
|
"step": 21
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.026993865030674847,
|
||
|
|
"grad_norm": 56.562310105836765,
|
||
|
|
"learning_rate": 8.799999999999999e-07,
|
||
|
|
"logits/chosen": -0.150390625,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.4564,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.6875,
|
||
|
|
"rewards/margins": 1.5,
|
||
|
|
"rewards/rejected": -0.81640625,
|
||
|
|
"step": 22
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02822085889570552,
|
||
|
|
"grad_norm": 96.05342097645857,
|
||
|
|
"learning_rate": 9.2e-07,
|
||
|
|
"logits/chosen": -0.271484375,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.4949,
|
||
|
|
"rewards/accuracies": 0.7421875,
|
||
|
|
"rewards/chosen": 0.96875,
|
||
|
|
"rewards/margins": 1.8359375,
|
||
|
|
"rewards/rejected": -0.86328125,
|
||
|
|
"step": 23
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.029447852760736196,
|
||
|
|
"grad_norm": 68.63010923117422,
|
||
|
|
"learning_rate": 9.6e-07,
|
||
|
|
"logits/chosen": -0.2021484375,
|
||
|
|
"logits/rejected": -0.388671875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -280.0,
|
||
|
|
"loss": 0.4852,
|
||
|
|
"rewards/accuracies": 0.7265625,
|
||
|
|
"rewards/chosen": 0.5703125,
|
||
|
|
"rewards/margins": 1.4375,
|
||
|
|
"rewards/rejected": -0.8671875,
|
||
|
|
"step": 24
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03067484662576687,
|
||
|
|
"grad_norm": 40.52591666206953,
|
||
|
|
"learning_rate": 1e-06,
|
||
|
|
"logits/chosen": -0.275390625,
|
||
|
|
"logits/rejected": -0.46484375,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.375,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 2.46875,
|
||
|
|
"rewards/rejected": -1.3984375,
|
||
|
|
"step": 25
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03190184049079755,
|
||
|
|
"grad_norm": 47.31107870817392,
|
||
|
|
"learning_rate": 9.99586776859504e-07,
|
||
|
|
"logits/chosen": -0.171875,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -280.0,
|
||
|
|
"loss": 0.4293,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.3828125,
|
||
|
|
"rewards/margins": 2.59375,
|
||
|
|
"rewards/rejected": -1.203125,
|
||
|
|
"step": 26
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.033128834355828224,
|
||
|
|
"grad_norm": 42.03740863478121,
|
||
|
|
"learning_rate": 9.991735537190081e-07,
|
||
|
|
"logits/chosen": -0.115234375,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -284.0,
|
||
|
|
"loss": 0.3939,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.5703125,
|
||
|
|
"rewards/margins": 3.0,
|
||
|
|
"rewards/rejected": -1.4375,
|
||
|
|
"step": 27
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0343558282208589,
|
||
|
|
"grad_norm": 38.22751255614571,
|
||
|
|
"learning_rate": 9.987603305785124e-07,
|
||
|
|
"logits/chosen": -0.0517578125,
|
||
|
|
"logits/rejected": -0.1640625,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -266.0,
|
||
|
|
"loss": 0.3875,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.6015625,
|
||
|
|
"rewards/margins": 2.578125,
|
||
|
|
"rewards/rejected": -0.98828125,
|
||
|
|
"step": 28
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03558282208588957,
|
||
|
|
"grad_norm": 40.69846954289912,
|
||
|
|
"learning_rate": 9.983471074380165e-07,
|
||
|
|
"logits/chosen": -0.287109375,
|
||
|
|
"logits/rejected": -0.439453125,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3733,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.703125,
|
||
|
|
"rewards/margins": 3.125,
|
||
|
|
"rewards/rejected": -1.4140625,
|
||
|
|
"step": 29
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03680981595092025,
|
||
|
|
"grad_norm": 46.55169582197367,
|
||
|
|
"learning_rate": 9.979338842975206e-07,
|
||
|
|
"logits/chosen": -0.15625,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -332.0,
|
||
|
|
"logps/rejected": -268.0,
|
||
|
|
"loss": 0.4096,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.25,
|
||
|
|
"rewards/margins": 2.484375,
|
||
|
|
"rewards/rejected": -1.234375,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03803680981595092,
|
||
|
|
"grad_norm": 37.971298083652066,
|
||
|
|
"learning_rate": 9.975206611570247e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -278.0,
|
||
|
|
"loss": 0.3889,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.828125,
|
||
|
|
"rewards/margins": 3.25,
|
||
|
|
"rewards/rejected": -1.421875,
|
||
|
|
"step": 31
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0392638036809816,
|
||
|
|
"grad_norm": 40.712562776553135,
|
||
|
|
"learning_rate": 9.971074380165288e-07,
|
||
|
|
"logits/chosen": -0.1884765625,
|
||
|
|
"logits/rejected": -0.37890625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.384,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.640625,
|
||
|
|
"rewards/margins": 3.125,
|
||
|
|
"rewards/rejected": -1.4765625,
|
||
|
|
"step": 32
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04049079754601227,
|
||
|
|
"grad_norm": 37.991087949351716,
|
||
|
|
"learning_rate": 9.96694214876033e-07,
|
||
|
|
"logits/chosen": -0.07958984375,
|
||
|
|
"logits/rejected": -0.2177734375,
|
||
|
|
"logps/chosen": -328.0,
|
||
|
|
"logps/rejected": -238.0,
|
||
|
|
"loss": 0.3728,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 2.640625,
|
||
|
|
"rewards/rejected": -1.484375,
|
||
|
|
"step": 33
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04171779141104295,
|
||
|
|
"grad_norm": 45.39723737748533,
|
||
|
|
"learning_rate": 9.962809917355372e-07,
|
||
|
|
"logits/chosen": -0.228515625,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.4037,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 3.34375,
|
||
|
|
"rewards/rejected": -1.8984375,
|
||
|
|
"step": 34
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04294478527607362,
|
||
|
|
"grad_norm": 46.41786525801273,
|
||
|
|
"learning_rate": 9.958677685950413e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -270.0,
|
||
|
|
"loss": 0.3792,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.046875,
|
||
|
|
"rewards/margins": 3.15625,
|
||
|
|
"rewards/rejected": -2.109375,
|
||
|
|
"step": 35
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.044171779141104296,
|
||
|
|
"grad_norm": 42.51091401927911,
|
||
|
|
"learning_rate": 9.954545454545454e-07,
|
||
|
|
"logits/chosen": -0.193359375,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.3676,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 3.734375,
|
||
|
|
"rewards/rejected": -2.25,
|
||
|
|
"step": 36
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04539877300613497,
|
||
|
|
"grad_norm": 40.33063661852525,
|
||
|
|
"learning_rate": 9.950413223140495e-07,
|
||
|
|
"logits/chosen": -0.15625,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.362,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.890625,
|
||
|
|
"rewards/margins": 3.8125,
|
||
|
|
"rewards/rejected": -1.921875,
|
||
|
|
"step": 37
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.046625766871165646,
|
||
|
|
"grad_norm": 35.01835466546152,
|
||
|
|
"learning_rate": 9.946280991735538e-07,
|
||
|
|
"logits/chosen": -0.0947265625,
|
||
|
|
"logits/rejected": -0.2314453125,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3457,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.140625,
|
||
|
|
"rewards/margins": 4.25,
|
||
|
|
"rewards/rejected": -2.09375,
|
||
|
|
"step": 38
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04785276073619632,
|
||
|
|
"grad_norm": 46.14448752240649,
|
||
|
|
"learning_rate": 9.942148760330577e-07,
|
||
|
|
"logits/chosen": -0.2099609375,
|
||
|
|
"logits/rejected": -0.380859375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3766,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.7734375,
|
||
|
|
"rewards/margins": 4.1875,
|
||
|
|
"rewards/rejected": -2.40625,
|
||
|
|
"step": 39
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.049079754601226995,
|
||
|
|
"grad_norm": 33.12245165013039,
|
||
|
|
"learning_rate": 9.93801652892562e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.2986,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.734375,
|
||
|
|
"rewards/margins": 4.1875,
|
||
|
|
"rewards/rejected": -2.46875,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05030674846625767,
|
||
|
|
"grad_norm": 35.950653523061156,
|
||
|
|
"learning_rate": 9.93388429752066e-07,
|
||
|
|
"logits/chosen": -0.1337890625,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -282.0,
|
||
|
|
"loss": 0.3625,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.65625,
|
||
|
|
"rewards/margins": 3.984375,
|
||
|
|
"rewards/rejected": -2.328125,
|
||
|
|
"step": 41
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.051533742331288344,
|
||
|
|
"grad_norm": 33.25822119251682,
|
||
|
|
"learning_rate": 9.929752066115702e-07,
|
||
|
|
"logits/chosen": -0.083984375,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.3156,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.5234375,
|
||
|
|
"rewards/margins": 3.90625,
|
||
|
|
"rewards/rejected": -2.390625,
|
||
|
|
"step": 42
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05276073619631902,
|
||
|
|
"grad_norm": 35.99668142644002,
|
||
|
|
"learning_rate": 9.925619834710743e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2946,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.8125,
|
||
|
|
"rewards/margins": 4.53125,
|
||
|
|
"rewards/rejected": -2.734375,
|
||
|
|
"step": 43
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.053987730061349694,
|
||
|
|
"grad_norm": 45.219634249357064,
|
||
|
|
"learning_rate": 9.921487603305784e-07,
|
||
|
|
"logits/chosen": -0.212890625,
|
||
|
|
"logits/rejected": -0.333984375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.4138,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.25,
|
||
|
|
"rewards/margins": 4.0625,
|
||
|
|
"rewards/rejected": -2.8125,
|
||
|
|
"step": 44
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05521472392638037,
|
||
|
|
"grad_norm": 37.8165937134258,
|
||
|
|
"learning_rate": 9.917355371900827e-07,
|
||
|
|
"logits/chosen": -0.27734375,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.3291,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 4.375,
|
||
|
|
"rewards/rejected": -2.875,
|
||
|
|
"step": 45
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05644171779141104,
|
||
|
|
"grad_norm": 40.067988147487945,
|
||
|
|
"learning_rate": 9.913223140495868e-07,
|
||
|
|
"logits/chosen": -0.1484375,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.302,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.109375,
|
||
|
|
"rewards/margins": 4.9375,
|
||
|
|
"rewards/rejected": -2.8125,
|
||
|
|
"step": 46
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05766871165644172,
|
||
|
|
"grad_norm": 62.331909497612145,
|
||
|
|
"learning_rate": 9.909090909090909e-07,
|
||
|
|
"logits/chosen": -0.19921875,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.4649,
|
||
|
|
"rewards/accuracies": 0.7109375,
|
||
|
|
"rewards/chosen": 1.3203125,
|
||
|
|
"rewards/margins": 3.828125,
|
||
|
|
"rewards/rejected": -2.5,
|
||
|
|
"step": 47
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05889570552147239,
|
||
|
|
"grad_norm": 35.63996127780391,
|
||
|
|
"learning_rate": 9.90495867768595e-07,
|
||
|
|
"logits/chosen": -0.181640625,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -284.0,
|
||
|
|
"loss": 0.3776,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.109375,
|
||
|
|
"rewards/margins": 3.5,
|
||
|
|
"rewards/rejected": -2.390625,
|
||
|
|
"step": 48
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06012269938650307,
|
||
|
|
"grad_norm": 30.545142903578636,
|
||
|
|
"learning_rate": 9.90082644628099e-07,
|
||
|
|
"logits/chosen": -0.1953125,
|
||
|
|
"logits/rejected": -0.41015625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.2534,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.0625,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -2.984375,
|
||
|
|
"step": 49
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06134969325153374,
|
||
|
|
"grad_norm": 46.15545512472155,
|
||
|
|
"learning_rate": 9.896694214876032e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3808,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.5234375,
|
||
|
|
"rewards/margins": 4.21875,
|
||
|
|
"rewards/rejected": -2.6875,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06257668711656442,
|
||
|
|
"grad_norm": 42.158829592416716,
|
||
|
|
"learning_rate": 9.892561983471075e-07,
|
||
|
|
"logits/chosen": -0.126953125,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -442.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.3026,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.9375,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -2.90625,
|
||
|
|
"step": 51
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0638036809815951,
|
||
|
|
"grad_norm": 36.72373668526655,
|
||
|
|
"learning_rate": 9.888429752066116e-07,
|
||
|
|
"logits/chosen": -0.1572265625,
|
||
|
|
"logits/rejected": -0.3359375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.32,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.5546875,
|
||
|
|
"rewards/margins": 4.53125,
|
||
|
|
"rewards/rejected": -2.96875,
|
||
|
|
"step": 52
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06503067484662577,
|
||
|
|
"grad_norm": 35.57136422165816,
|
||
|
|
"learning_rate": 9.884297520661157e-07,
|
||
|
|
"logits/chosen": -0.10302734375,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2962,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.828125,
|
||
|
|
"rewards/margins": 4.28125,
|
||
|
|
"rewards/rejected": -2.46875,
|
||
|
|
"step": 53
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06625766871165645,
|
||
|
|
"grad_norm": 36.12409955593161,
|
||
|
|
"learning_rate": 9.880165289256197e-07,
|
||
|
|
"logits/chosen": -0.000728607177734375,
|
||
|
|
"logits/rejected": -0.17578125,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -262.0,
|
||
|
|
"loss": 0.2715,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 4.15625,
|
||
|
|
"rewards/rejected": -2.859375,
|
||
|
|
"step": 54
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06748466257668712,
|
||
|
|
"grad_norm": 39.59210158362732,
|
||
|
|
"learning_rate": 9.87603305785124e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.283203125,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3207,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 4.4375,
|
||
|
|
"rewards/rejected": -2.75,
|
||
|
|
"step": 55
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0687116564417178,
|
||
|
|
"grad_norm": 39.25115329003044,
|
||
|
|
"learning_rate": 9.871900826446281e-07,
|
||
|
|
"logits/chosen": -0.091796875,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3727,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 4.34375,
|
||
|
|
"rewards/rejected": -2.4375,
|
||
|
|
"step": 56
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06993865030674846,
|
||
|
|
"grad_norm": 46.72232187427657,
|
||
|
|
"learning_rate": 9.867768595041322e-07,
|
||
|
|
"logits/chosen": -0.09326171875,
|
||
|
|
"logits/rejected": -0.169921875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.357,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.671875,
|
||
|
|
"rewards/margins": 4.28125,
|
||
|
|
"rewards/rejected": -2.59375,
|
||
|
|
"step": 57
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07116564417177915,
|
||
|
|
"grad_norm": 43.616877681053715,
|
||
|
|
"learning_rate": 9.863636363636363e-07,
|
||
|
|
"logits/chosen": -0.1025390625,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3416,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.828125,
|
||
|
|
"rewards/margins": 4.4375,
|
||
|
|
"rewards/rejected": -2.609375,
|
||
|
|
"step": 58
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07239263803680981,
|
||
|
|
"grad_norm": 57.340851139568066,
|
||
|
|
"learning_rate": 9.859504132231404e-07,
|
||
|
|
"logits/chosen": -0.028564453125,
|
||
|
|
"logits/rejected": -0.11181640625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.4351,
|
||
|
|
"rewards/accuracies": 0.765625,
|
||
|
|
"rewards/chosen": 1.359375,
|
||
|
|
"rewards/margins": 3.578125,
|
||
|
|
"rewards/rejected": -2.21875,
|
||
|
|
"step": 59
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0736196319018405,
|
||
|
|
"grad_norm": 40.41299854217699,
|
||
|
|
"learning_rate": 9.855371900826445e-07,
|
||
|
|
"logits/chosen": -0.1015625,
|
||
|
|
"logits/rejected": -0.2333984375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -286.0,
|
||
|
|
"loss": 0.3642,
|
||
|
|
"rewards/accuracies": 0.765625,
|
||
|
|
"rewards/chosen": 1.2421875,
|
||
|
|
"rewards/margins": 3.75,
|
||
|
|
"rewards/rejected": -2.5,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07484662576687116,
|
||
|
|
"grad_norm": 38.38684468795477,
|
||
|
|
"learning_rate": 9.851239669421486e-07,
|
||
|
|
"logits/chosen": -0.1025390625,
|
||
|
|
"logits/rejected": -0.2373046875,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3147,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.125,
|
||
|
|
"rewards/margins": 4.90625,
|
||
|
|
"rewards/rejected": -2.796875,
|
||
|
|
"step": 61
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07607361963190185,
|
||
|
|
"grad_norm": 42.11986809638378,
|
||
|
|
"learning_rate": 9.84710743801653e-07,
|
||
|
|
"logits/chosen": -0.0869140625,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3517,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 2.609375,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -2.796875,
|
||
|
|
"step": 62
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07730061349693251,
|
||
|
|
"grad_norm": 48.46764704043021,
|
||
|
|
"learning_rate": 9.84297520661157e-07,
|
||
|
|
"logits/chosen": -0.005523681640625,
|
||
|
|
"logits/rejected": -0.1689453125,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3154,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.7421875,
|
||
|
|
"rewards/margins": 4.21875,
|
||
|
|
"rewards/rejected": -2.5,
|
||
|
|
"step": 63
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0785276073619632,
|
||
|
|
"grad_norm": 44.88518598351617,
|
||
|
|
"learning_rate": 9.838842975206611e-07,
|
||
|
|
"logits/chosen": -0.15234375,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3722,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 2.25,
|
||
|
|
"rewards/margins": 4.8125,
|
||
|
|
"rewards/rejected": -2.546875,
|
||
|
|
"step": 64
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07975460122699386,
|
||
|
|
"grad_norm": 35.350496206715135,
|
||
|
|
"learning_rate": 9.834710743801652e-07,
|
||
|
|
"logits/chosen": -0.2041015625,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.2996,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.078125,
|
||
|
|
"rewards/margins": 4.59375,
|
||
|
|
"rewards/rejected": -2.515625,
|
||
|
|
"step": 65
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08098159509202454,
|
||
|
|
"grad_norm": 34.14184871107505,
|
||
|
|
"learning_rate": 9.830578512396693e-07,
|
||
|
|
"logits/chosen": -0.18359375,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -314.0,
|
||
|
|
"logps/rejected": -260.0,
|
||
|
|
"loss": 0.2818,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.328125,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -2.890625,
|
||
|
|
"step": 66
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08220858895705521,
|
||
|
|
"grad_norm": 42.64233149605358,
|
||
|
|
"learning_rate": 9.826446280991734e-07,
|
||
|
|
"logits/chosen": -0.0869140625,
|
||
|
|
"logits/rejected": -0.205078125,
|
||
|
|
"logps/chosen": -330.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.3617,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.421875,
|
||
|
|
"rewards/margins": 4.125,
|
||
|
|
"rewards/rejected": -2.703125,
|
||
|
|
"step": 67
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0834355828220859,
|
||
|
|
"grad_norm": 41.39506967156979,
|
||
|
|
"learning_rate": 9.822314049586777e-07,
|
||
|
|
"logits/chosen": -0.1708984375,
|
||
|
|
"logits/rejected": -0.2392578125,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3635,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.78125,
|
||
|
|
"rewards/margins": 4.28125,
|
||
|
|
"rewards/rejected": -2.5,
|
||
|
|
"step": 68
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08466257668711656,
|
||
|
|
"grad_norm": 32.22625490427397,
|
||
|
|
"learning_rate": 9.818181818181818e-07,
|
||
|
|
"logits/chosen": -0.2333984375,
|
||
|
|
"logits/rejected": -0.400390625,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -282.0,
|
||
|
|
"loss": 0.263,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.609375,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -2.703125,
|
||
|
|
"step": 69
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08588957055214724,
|
||
|
|
"grad_norm": 43.90551994332239,
|
||
|
|
"learning_rate": 9.814049586776859e-07,
|
||
|
|
"logits/chosen": -0.107421875,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.315,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.59375,
|
||
|
|
"rewards/margins": 4.96875,
|
||
|
|
"rewards/rejected": -2.390625,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08711656441717791,
|
||
|
|
"grad_norm": 39.99506870265264,
|
||
|
|
"learning_rate": 9.8099173553719e-07,
|
||
|
|
"logits/chosen": -0.06201171875,
|
||
|
|
"logits/rejected": -0.1396484375,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -278.0,
|
||
|
|
"loss": 0.3872,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 3.9375,
|
||
|
|
"rewards/rejected": -2.34375,
|
||
|
|
"step": 71
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08834355828220859,
|
||
|
|
"grad_norm": 41.349559255204774,
|
||
|
|
"learning_rate": 9.805785123966943e-07,
|
||
|
|
"logits/chosen": -0.2197265625,
|
||
|
|
"logits/rejected": -0.375,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.311,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 2.171875,
|
||
|
|
"rewards/margins": 4.5625,
|
||
|
|
"rewards/rejected": -2.375,
|
||
|
|
"step": 72
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08957055214723926,
|
||
|
|
"grad_norm": 45.978363733587855,
|
||
|
|
"learning_rate": 9.801652892561984e-07,
|
||
|
|
"logits/chosen": -0.1669921875,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.4089,
|
||
|
|
"rewards/accuracies": 0.7578125,
|
||
|
|
"rewards/chosen": 1.75,
|
||
|
|
"rewards/margins": 4.4375,
|
||
|
|
"rewards/rejected": -2.671875,
|
||
|
|
"step": 73
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09079754601226994,
|
||
|
|
"grad_norm": 38.839980321987454,
|
||
|
|
"learning_rate": 9.797520661157025e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2848,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 4.28125,
|
||
|
|
"rewards/rejected": -2.8125,
|
||
|
|
"step": 74
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09202453987730061,
|
||
|
|
"grad_norm": 36.468204205869355,
|
||
|
|
"learning_rate": 9.793388429752066e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2732,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.1875,
|
||
|
|
"rewards/margins": 5.0625,
|
||
|
|
"rewards/rejected": -2.890625,
|
||
|
|
"step": 75
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09325153374233129,
|
||
|
|
"grad_norm": 51.75818303728974,
|
||
|
|
"learning_rate": 9.789256198347107e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -288.0,
|
||
|
|
"loss": 0.3498,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -2.65625,
|
||
|
|
"step": 76
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09447852760736196,
|
||
|
|
"grad_norm": 35.91522576533242,
|
||
|
|
"learning_rate": 9.785123966942148e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.284,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.6328125,
|
||
|
|
"rewards/margins": 4.625,
|
||
|
|
"rewards/rejected": -2.984375,
|
||
|
|
"step": 77
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09570552147239264,
|
||
|
|
"grad_norm": 39.01255683404935,
|
||
|
|
"learning_rate": 9.78099173553719e-07,
|
||
|
|
"logits/chosen": -0.12353515625,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2939,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.0,
|
||
|
|
"rewards/margins": 5.0,
|
||
|
|
"rewards/rejected": -3.0,
|
||
|
|
"step": 78
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09693251533742331,
|
||
|
|
"grad_norm": 41.678187193821934,
|
||
|
|
"learning_rate": 9.776859504132232e-07,
|
||
|
|
"logits/chosen": -0.1083984375,
|
||
|
|
"logits/rejected": -0.248046875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -288.0,
|
||
|
|
"loss": 0.3366,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.53125,
|
||
|
|
"rewards/margins": 4.5,
|
||
|
|
"rewards/rejected": -2.984375,
|
||
|
|
"step": 79
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09815950920245399,
|
||
|
|
"grad_norm": 36.89274945940705,
|
||
|
|
"learning_rate": 9.772727272727273e-07,
|
||
|
|
"logits/chosen": -0.162109375,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.285,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.671875,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -3.265625,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09938650306748466,
|
||
|
|
"grad_norm": 36.34213008191509,
|
||
|
|
"learning_rate": 9.768595041322313e-07,
|
||
|
|
"logits/chosen": -0.2255859375,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3048,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 2.28125,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -3.28125,
|
||
|
|
"step": 81
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10061349693251534,
|
||
|
|
"grad_norm": 51.83134303297906,
|
||
|
|
"learning_rate": 9.764462809917354e-07,
|
||
|
|
"logits/chosen": -0.1396484375,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3743,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.5390625,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -3.328125,
|
||
|
|
"step": 82
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10184049079754601,
|
||
|
|
"grad_norm": 44.6306513701737,
|
||
|
|
"learning_rate": 9.760330578512395e-07,
|
||
|
|
"logits/chosen": -0.12255859375,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.326,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.6015625,
|
||
|
|
"rewards/margins": 4.96875,
|
||
|
|
"rewards/rejected": -3.359375,
|
||
|
|
"step": 83
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10306748466257669,
|
||
|
|
"grad_norm": 43.368403842908165,
|
||
|
|
"learning_rate": 9.756198347107436e-07,
|
||
|
|
"logits/chosen": -0.10009765625,
|
||
|
|
"logits/rejected": -0.337890625,
|
||
|
|
"logps/chosen": -444.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2915,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.140625,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -3.6875,
|
||
|
|
"step": 84
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10429447852760736,
|
||
|
|
"grad_norm": 41.83545101757536,
|
||
|
|
"learning_rate": 9.75206611570248e-07,
|
||
|
|
"logits/chosen": -0.1962890625,
|
||
|
|
"logits/rejected": -0.36328125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2993,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.5,
|
||
|
|
"rewards/margins": 5.09375,
|
||
|
|
"rewards/rejected": -3.59375,
|
||
|
|
"step": 85
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10552147239263804,
|
||
|
|
"grad_norm": 42.79642482122795,
|
||
|
|
"learning_rate": 9.74793388429752e-07,
|
||
|
|
"logits/chosen": -0.12890625,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3469,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -3.109375,
|
||
|
|
"step": 86
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1067484662576687,
|
||
|
|
"grad_norm": 31.200859065220833,
|
||
|
|
"learning_rate": 9.743801652892561e-07,
|
||
|
|
"logits/chosen": -0.1689453125,
|
||
|
|
"logits/rejected": -0.30859375,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -260.0,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.140625,
|
||
|
|
"rewards/margins": 4.6875,
|
||
|
|
"rewards/rejected": -3.5625,
|
||
|
|
"step": 87
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10797546012269939,
|
||
|
|
"grad_norm": 33.089204895794104,
|
||
|
|
"learning_rate": 9.739669421487602e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -276.0,
|
||
|
|
"loss": 0.2632,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 4.9375,
|
||
|
|
"rewards/rejected": -3.328125,
|
||
|
|
"step": 88
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10920245398773006,
|
||
|
|
"grad_norm": 41.44285417719672,
|
||
|
|
"learning_rate": 9.735537190082645e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.283203125,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3377,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.796875,
|
||
|
|
"rewards/margins": 4.96875,
|
||
|
|
"rewards/rejected": -3.171875,
|
||
|
|
"step": 89
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11042944785276074,
|
||
|
|
"grad_norm": 33.95249948655705,
|
||
|
|
"learning_rate": 9.731404958677686e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3025,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.390625,
|
||
|
|
"rewards/margins": 4.78125,
|
||
|
|
"rewards/rejected": -3.390625,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1116564417177914,
|
||
|
|
"grad_norm": 39.291583688315264,
|
||
|
|
"learning_rate": 9.727272727272727e-07,
|
||
|
|
"logits/chosen": -0.033935546875,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.2872,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.6953125,
|
||
|
|
"rewards/margins": 4.90625,
|
||
|
|
"rewards/rejected": -3.203125,
|
||
|
|
"step": 91
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11288343558282209,
|
||
|
|
"grad_norm": 34.11914530579394,
|
||
|
|
"learning_rate": 9.723140495867768e-07,
|
||
|
|
"logits/chosen": -0.1240234375,
|
||
|
|
"logits/rejected": -0.2451171875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.2706,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.65625,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -3.640625,
|
||
|
|
"step": 92
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11411042944785275,
|
||
|
|
"grad_norm": 35.0129736595644,
|
||
|
|
"learning_rate": 9.71900826446281e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.369140625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.271,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 2.359375,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -3.359375,
|
||
|
|
"step": 93
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11533742331288344,
|
||
|
|
"grad_norm": 39.56522119452336,
|
||
|
|
"learning_rate": 9.71487603305785e-07,
|
||
|
|
"logits/chosen": -0.1748046875,
|
||
|
|
"logits/rejected": -0.400390625,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2779,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.53125,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -3.328125,
|
||
|
|
"step": 94
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1165644171779141,
|
||
|
|
"grad_norm": 35.12017401888975,
|
||
|
|
"learning_rate": 9.710743801652893e-07,
|
||
|
|
"logits/chosen": -0.25390625,
|
||
|
|
"logits/rejected": -0.40625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2912,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.0,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -3.609375,
|
||
|
|
"step": 95
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11779141104294479,
|
||
|
|
"grad_norm": 35.951495412713705,
|
||
|
|
"learning_rate": 9.706611570247934e-07,
|
||
|
|
"logits/chosen": -0.23828125,
|
||
|
|
"logits/rejected": -0.41015625,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2652,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.40625,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -3.640625,
|
||
|
|
"step": 96
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11901840490797547,
|
||
|
|
"grad_norm": 46.49379123938547,
|
||
|
|
"learning_rate": 9.702479338842975e-07,
|
||
|
|
"logits/chosen": -0.236328125,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3607,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 2.09375,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -3.21875,
|
||
|
|
"step": 97
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12024539877300613,
|
||
|
|
"grad_norm": 38.4803512021978,
|
||
|
|
"learning_rate": 9.698347107438016e-07,
|
||
|
|
"logits/chosen": -0.04638671875,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -286.0,
|
||
|
|
"loss": 0.2977,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.8671875,
|
||
|
|
"rewards/margins": 5.0625,
|
||
|
|
"rewards/rejected": -3.1875,
|
||
|
|
"step": 98
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12147239263803682,
|
||
|
|
"grad_norm": 39.97761004406295,
|
||
|
|
"learning_rate": 9.694214876033057e-07,
|
||
|
|
"logits/chosen": -0.201171875,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -247.0,
|
||
|
|
"loss": 0.268,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.34375,
|
||
|
|
"rewards/margins": 4.4375,
|
||
|
|
"rewards/rejected": -3.109375,
|
||
|
|
"step": 99
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12269938650306748,
|
||
|
|
"grad_norm": 37.95641413102004,
|
||
|
|
"learning_rate": 9.6900826446281e-07,
|
||
|
|
"logits/chosen": -0.2255859375,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3193,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.8046875,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -3.5,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12392638036809817,
|
||
|
|
"grad_norm": 46.76166817261012,
|
||
|
|
"learning_rate": 9.685950413223139e-07,
|
||
|
|
"logits/chosen": -0.224609375,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.365,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.953125,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -3.390625,
|
||
|
|
"step": 101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12515337423312883,
|
||
|
|
"grad_norm": 41.79765734110343,
|
||
|
|
"learning_rate": 9.681818181818182e-07,
|
||
|
|
"logits/chosen": -0.1962890625,
|
||
|
|
"logits/rejected": -0.314453125,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3324,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 5.0625,
|
||
|
|
"rewards/rejected": -3.03125,
|
||
|
|
"step": 102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1263803680981595,
|
||
|
|
"grad_norm": 36.794153558575765,
|
||
|
|
"learning_rate": 9.677685950413223e-07,
|
||
|
|
"logits/chosen": -0.18359375,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.2922,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.7734375,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -3.09375,
|
||
|
|
"step": 103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1276073619631902,
|
||
|
|
"grad_norm": 50.370859060453284,
|
||
|
|
"learning_rate": 9.673553719008264e-07,
|
||
|
|
"logits/chosen": -0.1064453125,
|
||
|
|
"logits/rejected": -0.2470703125,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3734,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.78125,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -3.078125,
|
||
|
|
"step": 104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12883435582822086,
|
||
|
|
"grad_norm": 34.529612759458495,
|
||
|
|
"learning_rate": 9.669421487603305e-07,
|
||
|
|
"logits/chosen": -0.19140625,
|
||
|
|
"logits/rejected": -0.39453125,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.2608,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.078125,
|
||
|
|
"rewards/margins": 5.625,
|
||
|
|
"rewards/rejected": -3.5625,
|
||
|
|
"step": 105
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13006134969325153,
|
||
|
|
"grad_norm": 39.013908957263666,
|
||
|
|
"learning_rate": 9.665289256198348e-07,
|
||
|
|
"logits/chosen": -0.169921875,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3065,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.9375,
|
||
|
|
"rewards/margins": 5.28125,
|
||
|
|
"rewards/rejected": -3.328125,
|
||
|
|
"step": 106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1312883435582822,
|
||
|
|
"grad_norm": 42.150982723963395,
|
||
|
|
"learning_rate": 9.661157024793389e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.2956,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -3.359375,
|
||
|
|
"step": 107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1325153374233129,
|
||
|
|
"grad_norm": 29.969554041041413,
|
||
|
|
"learning_rate": 9.65702479338843e-07,
|
||
|
|
"logits/chosen": -0.2080078125,
|
||
|
|
"logits/rejected": -0.421875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2406,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -3.953125,
|
||
|
|
"step": 108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13374233128834356,
|
||
|
|
"grad_norm": 36.72032157988445,
|
||
|
|
"learning_rate": 9.65289256198347e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.375,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.317,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.734375,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -3.546875,
|
||
|
|
"step": 109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13496932515337423,
|
||
|
|
"grad_norm": 40.64788730782337,
|
||
|
|
"learning_rate": 9.648760330578511e-07,
|
||
|
|
"logits/chosen": -0.1796875,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3312,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -3.59375,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1361963190184049,
|
||
|
|
"grad_norm": 34.71283913194937,
|
||
|
|
"learning_rate": 9.644628099173552e-07,
|
||
|
|
"logits/chosen": -0.11279296875,
|
||
|
|
"logits/rejected": -0.2734375,
|
||
|
|
"logps/chosen": -330.0,
|
||
|
|
"logps/rejected": -274.0,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -3.234375,
|
||
|
|
"step": 111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1374233128834356,
|
||
|
|
"grad_norm": 36.76218297956249,
|
||
|
|
"learning_rate": 9.640495867768595e-07,
|
||
|
|
"logits/chosen": -0.1875,
|
||
|
|
"logits/rejected": -0.306640625,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -284.0,
|
||
|
|
"loss": 0.2795,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.53125,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -3.6875,
|
||
|
|
"step": 112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13865030674846626,
|
||
|
|
"grad_norm": 39.011108940957456,
|
||
|
|
"learning_rate": 9.636363636363636e-07,
|
||
|
|
"logits/chosen": -0.1015625,
|
||
|
|
"logits/rejected": -0.193359375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3276,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.859375,
|
||
|
|
"rewards/margins": 5.25,
|
||
|
|
"rewards/rejected": -3.40625,
|
||
|
|
"step": 113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13987730061349693,
|
||
|
|
"grad_norm": 42.342001118592165,
|
||
|
|
"learning_rate": 9.632231404958677e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.296875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3342,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.359375,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -3.46875,
|
||
|
|
"step": 114
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1411042944785276,
|
||
|
|
"grad_norm": 48.02563322504298,
|
||
|
|
"learning_rate": 9.628099173553718e-07,
|
||
|
|
"logits/chosen": -0.14453125,
|
||
|
|
"logits/rejected": -0.26953125,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3999,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.98046875,
|
||
|
|
"rewards/margins": 4.375,
|
||
|
|
"rewards/rejected": -3.40625,
|
||
|
|
"step": 115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1423312883435583,
|
||
|
|
"grad_norm": 31.044871089885095,
|
||
|
|
"learning_rate": 9.62396694214876e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.431640625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.262,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.171875,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -3.53125,
|
||
|
|
"step": 116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14355828220858896,
|
||
|
|
"grad_norm": 37.17488322550573,
|
||
|
|
"learning_rate": 9.619834710743802e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -326.0,
|
||
|
|
"logps/rejected": -266.0,
|
||
|
|
"loss": 0.3079,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.2421875,
|
||
|
|
"rewards/margins": 5.15625,
|
||
|
|
"rewards/rejected": -3.90625,
|
||
|
|
"step": 117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14478527607361963,
|
||
|
|
"grad_norm": 38.57684603177769,
|
||
|
|
"learning_rate": 9.615702479338843e-07,
|
||
|
|
"logits/chosen": -0.166015625,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3182,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 2.109375,
|
||
|
|
"rewards/margins": 5.625,
|
||
|
|
"rewards/rejected": -3.53125,
|
||
|
|
"step": 118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1460122699386503,
|
||
|
|
"grad_norm": 39.31837627767617,
|
||
|
|
"learning_rate": 9.611570247933884e-07,
|
||
|
|
"logits/chosen": -0.236328125,
|
||
|
|
"logits/rejected": -0.333984375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.308,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 5.1875,
|
||
|
|
"rewards/rejected": -3.734375,
|
||
|
|
"step": 119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.147239263803681,
|
||
|
|
"grad_norm": 47.322798683163334,
|
||
|
|
"learning_rate": 9.607438016528925e-07,
|
||
|
|
"logits/chosen": -0.21875,
|
||
|
|
"logits/rejected": -0.40625,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3156,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.53125,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -3.65625,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14846625766871166,
|
||
|
|
"grad_norm": 35.88276223707849,
|
||
|
|
"learning_rate": 9.603305785123966e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 2.34375,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -3.5625,
|
||
|
|
"step": 121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14969325153374233,
|
||
|
|
"grad_norm": 43.922306327606435,
|
||
|
|
"learning_rate": 9.599173553719007e-07,
|
||
|
|
"logits/chosen": -0.15234375,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.3011,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.7265625,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -3.5,
|
||
|
|
"step": 122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.150920245398773,
|
||
|
|
"grad_norm": 44.742388015169865,
|
||
|
|
"learning_rate": 9.59504132231405e-07,
|
||
|
|
"logits/chosen": -0.04931640625,
|
||
|
|
"logits/rejected": -0.2236328125,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -282.0,
|
||
|
|
"loss": 0.3574,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -3.609375,
|
||
|
|
"step": 123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1521472392638037,
|
||
|
|
"grad_norm": 34.160343795291915,
|
||
|
|
"learning_rate": 9.59090909090909e-07,
|
||
|
|
"logits/chosen": -0.2197265625,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -286.0,
|
||
|
|
"loss": 0.2674,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.109375,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -3.953125,
|
||
|
|
"step": 124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15337423312883436,
|
||
|
|
"grad_norm": 39.518799854999045,
|
||
|
|
"learning_rate": 9.586776859504132e-07,
|
||
|
|
"logits/chosen": -0.28125,
|
||
|
|
"logits/rejected": -0.458984375,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2871,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.828125,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -4.0625,
|
||
|
|
"step": 125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15460122699386503,
|
||
|
|
"grad_norm": 56.25861619658859,
|
||
|
|
"learning_rate": 9.582644628099173e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.3359375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3637,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.0078125,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -3.859375,
|
||
|
|
"step": 126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1558282208588957,
|
||
|
|
"grad_norm": 42.50340403129208,
|
||
|
|
"learning_rate": 9.578512396694216e-07,
|
||
|
|
"logits/chosen": -0.08837890625,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3851,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -3.609375,
|
||
|
|
"step": 127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1570552147239264,
|
||
|
|
"grad_norm": 44.339394645108435,
|
||
|
|
"learning_rate": 9.574380165289255e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3155,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.90625,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -3.90625,
|
||
|
|
"step": 128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15828220858895706,
|
||
|
|
"grad_norm": 35.34227561050985,
|
||
|
|
"learning_rate": 9.570247933884298e-07,
|
||
|
|
"logits/chosen": -0.040283203125,
|
||
|
|
"logits/rejected": -0.169921875,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.3153,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.3203125,
|
||
|
|
"rewards/margins": 4.875,
|
||
|
|
"rewards/rejected": -3.546875,
|
||
|
|
"step": 129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15950920245398773,
|
||
|
|
"grad_norm": 38.43919164614137,
|
||
|
|
"learning_rate": 9.566115702479339e-07,
|
||
|
|
"logits/chosen": -0.1435546875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3349,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.3125,
|
||
|
|
"rewards/margins": 5.28125,
|
||
|
|
"rewards/rejected": -3.96875,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1607361963190184,
|
||
|
|
"grad_norm": 34.097476299359684,
|
||
|
|
"learning_rate": 9.56198347107438e-07,
|
||
|
|
"logits/chosen": -0.25390625,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.2132,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.90625,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1619631901840491,
|
||
|
|
"grad_norm": 42.765865552519735,
|
||
|
|
"learning_rate": 9.55785123966942e-07,
|
||
|
|
"logits/chosen": -0.212890625,
|
||
|
|
"logits/rejected": -0.40234375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3402,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.765625,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -3.984375,
|
||
|
|
"step": 132
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16319018404907976,
|
||
|
|
"grad_norm": 40.18673593506736,
|
||
|
|
"learning_rate": 9.553719008264462e-07,
|
||
|
|
"logits/chosen": -0.2470703125,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2822,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.484375,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16441717791411042,
|
||
|
|
"grad_norm": 54.50121988482124,
|
||
|
|
"learning_rate": 9.549586776859505e-07,
|
||
|
|
"logits/chosen": -0.1787109375,
|
||
|
|
"logits/rejected": -0.390625,
|
||
|
|
"logps/chosen": -452.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.3145,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.3828125,
|
||
|
|
"rewards/margins": 5.09375,
|
||
|
|
"rewards/rejected": -3.734375,
|
||
|
|
"step": 134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1656441717791411,
|
||
|
|
"grad_norm": 44.50905430648629,
|
||
|
|
"learning_rate": 9.545454545454546e-07,
|
||
|
|
"logits/chosen": -0.05517578125,
|
||
|
|
"logits/rejected": -0.1669921875,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3672,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.78125,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -4.0625,
|
||
|
|
"step": 135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1668711656441718,
|
||
|
|
"grad_norm": 37.03632873479481,
|
||
|
|
"learning_rate": 9.541322314049586e-07,
|
||
|
|
"logits/chosen": -0.28515625,
|
||
|
|
"logits/rejected": -0.36328125,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3149,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2890625,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 136
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16809815950920245,
|
||
|
|
"grad_norm": 46.53968885524061,
|
||
|
|
"learning_rate": 9.537190082644627e-07,
|
||
|
|
"logits/chosen": -0.0576171875,
|
||
|
|
"logits/rejected": -0.2119140625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3795,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 4.75,
|
||
|
|
"rewards/rejected": -3.6875,
|
||
|
|
"step": 137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16932515337423312,
|
||
|
|
"grad_norm": 39.2528692217884,
|
||
|
|
"learning_rate": 9.533057851239668e-07,
|
||
|
|
"logits/chosen": -0.1416015625,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3313,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 5.09375,
|
||
|
|
"rewards/rejected": -4.09375,
|
||
|
|
"step": 138
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1705521472392638,
|
||
|
|
"grad_norm": 41.85887635916101,
|
||
|
|
"learning_rate": 9.52892561983471e-07,
|
||
|
|
"logits/chosen": -0.1328125,
|
||
|
|
"logits/rejected": -0.30859375,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.3652,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.5859375,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 139
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17177914110429449,
|
||
|
|
"grad_norm": 50.211692574184056,
|
||
|
|
"learning_rate": 9.524793388429751e-07,
|
||
|
|
"logits/chosen": -0.10400390625,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.3576,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.4296875,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17300613496932515,
|
||
|
|
"grad_norm": 34.61921033025654,
|
||
|
|
"learning_rate": 9.520661157024793e-07,
|
||
|
|
"logits/chosen": -0.08349609375,
|
||
|
|
"logits/rejected": -0.2099609375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.267,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.4140625,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17423312883435582,
|
||
|
|
"grad_norm": 39.84749847418945,
|
||
|
|
"learning_rate": 9.516528925619834e-07,
|
||
|
|
"logits/chosen": -0.12158203125,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3431,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.46875,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1754601226993865,
|
||
|
|
"grad_norm": 41.19383820823238,
|
||
|
|
"learning_rate": 9.512396694214876e-07,
|
||
|
|
"logits/chosen": -0.14453125,
|
||
|
|
"logits/rejected": -0.365234375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2931,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.09375,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17668711656441718,
|
||
|
|
"grad_norm": 39.48774183772262,
|
||
|
|
"learning_rate": 9.508264462809917e-07,
|
||
|
|
"logits/chosen": -0.240234375,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2937,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.9375,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 144
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17791411042944785,
|
||
|
|
"grad_norm": 39.31332603384939,
|
||
|
|
"learning_rate": 9.504132231404959e-07,
|
||
|
|
"logits/chosen": -0.2294921875,
|
||
|
|
"logits/rejected": -0.380859375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2665,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.71875,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17914110429447852,
|
||
|
|
"grad_norm": 32.69853161252682,
|
||
|
|
"learning_rate": 9.499999999999999e-07,
|
||
|
|
"logits/chosen": -0.216796875,
|
||
|
|
"logits/rejected": -0.3359375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2369,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.9609375,
|
||
|
|
"rewards/margins": 5.28125,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18036809815950922,
|
||
|
|
"grad_norm": 33.92758965649418,
|
||
|
|
"learning_rate": 9.495867768595041e-07,
|
||
|
|
"logits/chosen": -0.216796875,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2687,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.859375,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18159509202453988,
|
||
|
|
"grad_norm": 49.11649217825285,
|
||
|
|
"learning_rate": 9.491735537190082e-07,
|
||
|
|
"logits/chosen": -0.23828125,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3872,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18282208588957055,
|
||
|
|
"grad_norm": 42.0523811991046,
|
||
|
|
"learning_rate": 9.487603305785124e-07,
|
||
|
|
"logits/chosen": -0.11474609375,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -372.0,
|
||
|
|
"loss": 0.2569,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.921875,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -3.9375,
|
||
|
|
"step": 149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18404907975460122,
|
||
|
|
"grad_norm": 37.91815584989451,
|
||
|
|
"learning_rate": 9.483471074380165e-07,
|
||
|
|
"logits/chosen": -0.1435546875,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3146,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.625,
|
||
|
|
"rewards/margins": 5.1875,
|
||
|
|
"rewards/rejected": -3.546875,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18527607361963191,
|
||
|
|
"grad_norm": 39.01655480069067,
|
||
|
|
"learning_rate": 9.479338842975206e-07,
|
||
|
|
"logits/chosen": -0.1572265625,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3162,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.6328125,
|
||
|
|
"rewards/margins": 5.09375,
|
||
|
|
"rewards/rejected": -3.46875,
|
||
|
|
"step": 151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18650306748466258,
|
||
|
|
"grad_norm": 36.295557767109855,
|
||
|
|
"learning_rate": 9.475206611570248e-07,
|
||
|
|
"logits/chosen": -0.2470703125,
|
||
|
|
"logits/rejected": -0.4921875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.284,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -4.40625,
|
||
|
|
"step": 152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18773006134969325,
|
||
|
|
"grad_norm": 35.43408456309788,
|
||
|
|
"learning_rate": 9.471074380165289e-07,
|
||
|
|
"logits/chosen": -0.216796875,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3077,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.671875,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -3.765625,
|
||
|
|
"step": 153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18895705521472392,
|
||
|
|
"grad_norm": 45.691475338279055,
|
||
|
|
"learning_rate": 9.466942148760331e-07,
|
||
|
|
"logits/chosen": -0.09033203125,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3274,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -3.359375,
|
||
|
|
"step": 154
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1901840490797546,
|
||
|
|
"grad_norm": 28.365158188278386,
|
||
|
|
"learning_rate": 9.462809917355371e-07,
|
||
|
|
"logits/chosen": -0.2001953125,
|
||
|
|
"logits/rejected": -0.44921875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.1938,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.625,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -4.0,
|
||
|
|
"step": 155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19141104294478528,
|
||
|
|
"grad_norm": 43.775694095180505,
|
||
|
|
"learning_rate": 9.458677685950413e-07,
|
||
|
|
"logits/chosen": -0.11181640625,
|
||
|
|
"logits/rejected": -0.22265625,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.315,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.4296875,
|
||
|
|
"rewards/margins": 4.96875,
|
||
|
|
"rewards/rejected": -3.546875,
|
||
|
|
"step": 156
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19263803680981595,
|
||
|
|
"grad_norm": 35.92254576769146,
|
||
|
|
"learning_rate": 9.454545454545454e-07,
|
||
|
|
"logits/chosen": -0.103515625,
|
||
|
|
"logits/rejected": -0.232421875,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.2914,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.546875,
|
||
|
|
"rewards/margins": 5.25,
|
||
|
|
"rewards/rejected": -3.703125,
|
||
|
|
"step": 157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19386503067484662,
|
||
|
|
"grad_norm": 32.68696317336219,
|
||
|
|
"learning_rate": 9.450413223140496e-07,
|
||
|
|
"logits/chosen": -0.3046875,
|
||
|
|
"logits/rejected": -0.490234375,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2534,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1950920245398773,
|
||
|
|
"grad_norm": 40.79749046301349,
|
||
|
|
"learning_rate": 9.446280991735537e-07,
|
||
|
|
"logits/chosen": -0.203125,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.5859375,
|
||
|
|
"rewards/margins": 5.1875,
|
||
|
|
"rewards/rejected": -3.609375,
|
||
|
|
"step": 159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19631901840490798,
|
||
|
|
"grad_norm": 28.669147296818444,
|
||
|
|
"learning_rate": 9.442148760330579e-07,
|
||
|
|
"logits/chosen": -0.0908203125,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.2185,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 1.1171875,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -4.28125,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19754601226993865,
|
||
|
|
"grad_norm": 35.436535721930035,
|
||
|
|
"learning_rate": 9.43801652892562e-07,
|
||
|
|
"logits/chosen": -0.267578125,
|
||
|
|
"logits/rejected": -0.47265625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.2404,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.09375,
|
||
|
|
"step": 161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19877300613496932,
|
||
|
|
"grad_norm": 43.95617745827823,
|
||
|
|
"learning_rate": 9.433884297520662e-07,
|
||
|
|
"logits/chosen": -0.18359375,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3681,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.4453125,
|
||
|
|
"rewards/margins": 5.1875,
|
||
|
|
"rewards/rejected": -3.734375,
|
||
|
|
"step": 162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"grad_norm": 41.651239703295666,
|
||
|
|
"learning_rate": 9.429752066115701e-07,
|
||
|
|
"logits/chosen": -0.125,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2972,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -3.890625,
|
||
|
|
"step": 163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20122699386503068,
|
||
|
|
"grad_norm": 30.82750182223144,
|
||
|
|
"learning_rate": 9.425619834710743e-07,
|
||
|
|
"logits/chosen": -0.2734375,
|
||
|
|
"logits/rejected": -0.5078125,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.15625,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20245398773006135,
|
||
|
|
"grad_norm": 44.48878395259188,
|
||
|
|
"learning_rate": 9.421487603305784e-07,
|
||
|
|
"logits/chosen": -0.0703125,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -442.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3043,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.71875,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -3.828125,
|
||
|
|
"step": 165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20368098159509201,
|
||
|
|
"grad_norm": 41.12003400664833,
|
||
|
|
"learning_rate": 9.417355371900826e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.369140625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3598,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 2.046875,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -3.640625,
|
||
|
|
"step": 166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2049079754601227,
|
||
|
|
"grad_norm": 40.52632483767134,
|
||
|
|
"learning_rate": 9.413223140495867e-07,
|
||
|
|
"logits/chosen": -0.1171875,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3636,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 2.3125,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -3.625,
|
||
|
|
"step": 167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20613496932515338,
|
||
|
|
"grad_norm": 38.13990823039671,
|
||
|
|
"learning_rate": 9.409090909090908e-07,
|
||
|
|
"logits/chosen": -0.0986328125,
|
||
|
|
"logits/rejected": -0.2099609375,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -278.0,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.1953125,
|
||
|
|
"rewards/margins": 5.46875,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 168
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20736196319018405,
|
||
|
|
"grad_norm": 37.3898766928124,
|
||
|
|
"learning_rate": 9.40495867768595e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.337890625,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2887,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 5.25,
|
||
|
|
"rewards/rejected": -3.96875,
|
||
|
|
"step": 169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2085889570552147,
|
||
|
|
"grad_norm": 42.62803720646298,
|
||
|
|
"learning_rate": 9.400826446280991e-07,
|
||
|
|
"logits/chosen": -0.16015625,
|
||
|
|
"logits/rejected": -0.40234375,
|
||
|
|
"logps/chosen": -446.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3073,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.1875,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -4.03125,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2098159509202454,
|
||
|
|
"grad_norm": 33.8264512159167,
|
||
|
|
"learning_rate": 9.396694214876033e-07,
|
||
|
|
"logits/chosen": -0.1494140625,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2986,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -4.3125,
|
||
|
|
"step": 171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21104294478527608,
|
||
|
|
"grad_norm": 35.98076154174604,
|
||
|
|
"learning_rate": 9.392561983471073e-07,
|
||
|
|
"logits/chosen": -0.1142578125,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -434.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.27,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21226993865030674,
|
||
|
|
"grad_norm": 27.651041582591024,
|
||
|
|
"learning_rate": 9.388429752066115e-07,
|
||
|
|
"logits/chosen": -0.1708984375,
|
||
|
|
"logits/rejected": -0.40625,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.1967,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.453125,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2134969325153374,
|
||
|
|
"grad_norm": 42.78349510928123,
|
||
|
|
"learning_rate": 9.384297520661156e-07,
|
||
|
|
"logits/chosen": -0.01483154296875,
|
||
|
|
"logits/rejected": -0.2265625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3777,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 174
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2147239263803681,
|
||
|
|
"grad_norm": 49.466708668886746,
|
||
|
|
"learning_rate": 9.380165289256198e-07,
|
||
|
|
"logits/chosen": -0.1611328125,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.361,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21595092024539878,
|
||
|
|
"grad_norm": 37.891097032789595,
|
||
|
|
"learning_rate": 9.376033057851239e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.37890625,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3265,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 2.125,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21717791411042944,
|
||
|
|
"grad_norm": 29.591204044267382,
|
||
|
|
"learning_rate": 9.371900826446281e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2365,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.578125,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 177
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2184049079754601,
|
||
|
|
"grad_norm": 39.29643939569424,
|
||
|
|
"learning_rate": 9.367768595041322e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -324.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3276,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.7421875,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2196319018404908,
|
||
|
|
"grad_norm": 32.63548187592793,
|
||
|
|
"learning_rate": 9.363636363636364e-07,
|
||
|
|
"logits/chosen": -0.1748046875,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2755,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.98046875,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22085889570552147,
|
||
|
|
"grad_norm": 41.35724310032657,
|
||
|
|
"learning_rate": 9.359504132231405e-07,
|
||
|
|
"logits/chosen": -0.171875,
|
||
|
|
"logits/rejected": -0.296875,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.3359,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.3125,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22208588957055214,
|
||
|
|
"grad_norm": 41.06602522381907,
|
||
|
|
"learning_rate": 9.355371900826446e-07,
|
||
|
|
"logits/chosen": -0.1630859375,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.3587,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -4.65625,
|
||
|
|
"step": 181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2233128834355828,
|
||
|
|
"grad_norm": 38.85418184494936,
|
||
|
|
"learning_rate": 9.351239669421487e-07,
|
||
|
|
"logits/chosen": -0.1953125,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3357,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.85546875,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2245398773006135,
|
||
|
|
"grad_norm": 42.84335154848466,
|
||
|
|
"learning_rate": 9.347107438016529e-07,
|
||
|
|
"logits/chosen": -0.07275390625,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -382.0,
|
||
|
|
"loss": 0.342,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.83984375,
|
||
|
|
"rewards/margins": 5.1875,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22576687116564417,
|
||
|
|
"grad_norm": 46.28637855648245,
|
||
|
|
"learning_rate": 9.34297520661157e-07,
|
||
|
|
"logits/chosen": -0.10693359375,
|
||
|
|
"logits/rejected": -0.224609375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3705,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.6640625,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -4.65625,
|
||
|
|
"step": 184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22699386503067484,
|
||
|
|
"grad_norm": 49.113931033261764,
|
||
|
|
"learning_rate": 9.338842975206612e-07,
|
||
|
|
"logits/chosen": -0.1484375,
|
||
|
|
"logits/rejected": -0.2177734375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3514,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.65625,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.3125,
|
||
|
|
"step": 185
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2282208588957055,
|
||
|
|
"grad_norm": 37.05498191674873,
|
||
|
|
"learning_rate": 9.334710743801653e-07,
|
||
|
|
"logits/chosen": -0.1630859375,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3089,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.3671875,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2294478527607362,
|
||
|
|
"grad_norm": 37.71647333523593,
|
||
|
|
"learning_rate": 9.330578512396694e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.2659,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.3671875,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23067484662576687,
|
||
|
|
"grad_norm": 33.868390403245435,
|
||
|
|
"learning_rate": 9.326446280991736e-07,
|
||
|
|
"logits/chosen": -0.056884765625,
|
||
|
|
"logits/rejected": -0.15234375,
|
||
|
|
"logps/chosen": -332.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.3049,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.296875,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.5625,
|
||
|
|
"step": 188
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23190184049079754,
|
||
|
|
"grad_norm": 32.6551143996549,
|
||
|
|
"learning_rate": 9.322314049586776e-07,
|
||
|
|
"logits/chosen": -0.1484375,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2185,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.8515625,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2331288343558282,
|
||
|
|
"grad_norm": 34.45171325611373,
|
||
|
|
"learning_rate": 9.318181818181817e-07,
|
||
|
|
"logits/chosen": -0.04296875,
|
||
|
|
"logits/rejected": -0.205078125,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.292,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 5.5,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2343558282208589,
|
||
|
|
"grad_norm": 50.99221887208661,
|
||
|
|
"learning_rate": 9.314049586776858e-07,
|
||
|
|
"logits/chosen": -0.1162109375,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3858,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.8125,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23558282208588957,
|
||
|
|
"grad_norm": 47.091436431958705,
|
||
|
|
"learning_rate": 9.3099173553719e-07,
|
||
|
|
"logits/chosen": -0.1298828125,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.3259,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 5.125,
|
||
|
|
"rewards/rejected": -3.859375,
|
||
|
|
"step": 192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23680981595092024,
|
||
|
|
"grad_norm": 36.36482495891595,
|
||
|
|
"learning_rate": 9.305785123966941e-07,
|
||
|
|
"logits/chosen": -0.248046875,
|
||
|
|
"logits/rejected": -0.498046875,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2485,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.90625,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23803680981595093,
|
||
|
|
"grad_norm": 37.6633231814014,
|
||
|
|
"learning_rate": 9.301652892561983e-07,
|
||
|
|
"logits/chosen": -0.294921875,
|
||
|
|
"logits/rejected": -0.4140625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2939,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.1640625,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.65625,
|
||
|
|
"step": 194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2392638036809816,
|
||
|
|
"grad_norm": 34.69492122435969,
|
||
|
|
"learning_rate": 9.297520661157024e-07,
|
||
|
|
"logits/chosen": -0.10595703125,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2298,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.359375,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24049079754601227,
|
||
|
|
"grad_norm": 41.61809014057929,
|
||
|
|
"learning_rate": 9.293388429752066e-07,
|
||
|
|
"logits/chosen": -0.07373046875,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.3133,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.0625,
|
||
|
|
"step": 196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24171779141104294,
|
||
|
|
"grad_norm": 31.620797247972778,
|
||
|
|
"learning_rate": 9.289256198347107e-07,
|
||
|
|
"logits/chosen": -0.189453125,
|
||
|
|
"logits/rejected": -0.435546875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.282,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.859375,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24294478527607363,
|
||
|
|
"grad_norm": 39.95412220257755,
|
||
|
|
"learning_rate": 9.285123966942149e-07,
|
||
|
|
"logits/chosen": -0.11279296875,
|
||
|
|
"logits/rejected": -0.20703125,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -260.0,
|
||
|
|
"loss": 0.3762,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.296875,
|
||
|
|
"rewards/margins": 5.28125,
|
||
|
|
"rewards/rejected": -3.96875,
|
||
|
|
"step": 198
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2441717791411043,
|
||
|
|
"grad_norm": 27.11893031942543,
|
||
|
|
"learning_rate": 9.280991735537189e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.1959,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.46875,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24539877300613497,
|
||
|
|
"grad_norm": 42.12057014134938,
|
||
|
|
"learning_rate": 9.276859504132231e-07,
|
||
|
|
"logits/chosen": -0.11669921875,
|
||
|
|
"logits/rejected": -0.341796875,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -374.0,
|
||
|
|
"loss": 0.3142,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.234375,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24662576687116564,
|
||
|
|
"grad_norm": 32.34510842493663,
|
||
|
|
"learning_rate": 9.272727272727272e-07,
|
||
|
|
"logits/chosen": -0.125,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2316,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.109375,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24785276073619633,
|
||
|
|
"grad_norm": 34.14824944726285,
|
||
|
|
"learning_rate": 9.268595041322314e-07,
|
||
|
|
"logits/chosen": -0.10205078125,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2685,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.09375,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -4.125,
|
||
|
|
"step": 202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.249079754601227,
|
||
|
|
"grad_norm": 40.81109270790126,
|
||
|
|
"learning_rate": 9.264462809917355e-07,
|
||
|
|
"logits/chosen": -0.06787109375,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.4324,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.828125,
|
||
|
|
"rewards/margins": 5.375,
|
||
|
|
"rewards/rejected": -3.5625,
|
||
|
|
"step": 203
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25030674846625767,
|
||
|
|
"grad_norm": 39.08482579609809,
|
||
|
|
"learning_rate": 9.260330578512396e-07,
|
||
|
|
"logits/chosen": -0.05712890625,
|
||
|
|
"logits/rejected": -0.203125,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -276.0,
|
||
|
|
"loss": 0.3833,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25153374233128833,
|
||
|
|
"grad_norm": 36.559212995740964,
|
||
|
|
"learning_rate": 9.256198347107438e-07,
|
||
|
|
"logits/chosen": -0.00885009765625,
|
||
|
|
"logits/rejected": -0.1533203125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3464,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.625,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -3.953125,
|
||
|
|
"step": 205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.252760736196319,
|
||
|
|
"grad_norm": 26.131004803554553,
|
||
|
|
"learning_rate": 9.252066115702479e-07,
|
||
|
|
"logits/chosen": -0.1142578125,
|
||
|
|
"logits/rejected": -0.400390625,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2208,
|
||
|
|
"rewards/accuracies": 0.9296875,
|
||
|
|
"rewards/chosen": 1.875,
|
||
|
|
"rewards/margins": 7.09375,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25398773006134967,
|
||
|
|
"grad_norm": 42.01910176739678,
|
||
|
|
"learning_rate": 9.247933884297521e-07,
|
||
|
|
"logits/chosen": -0.240234375,
|
||
|
|
"logits/rejected": -0.447265625,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.3729,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.8125,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2552147239263804,
|
||
|
|
"grad_norm": 38.03191943662351,
|
||
|
|
"learning_rate": 9.243801652892561e-07,
|
||
|
|
"logits/chosen": -0.208984375,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3343,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25644171779141106,
|
||
|
|
"grad_norm": 36.80831687011056,
|
||
|
|
"learning_rate": 9.239669421487603e-07,
|
||
|
|
"logits/chosen": -0.2080078125,
|
||
|
|
"logits/rejected": -0.3359375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3148,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.1015625,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25766871165644173,
|
||
|
|
"grad_norm": 44.11101399224612,
|
||
|
|
"learning_rate": 9.235537190082644e-07,
|
||
|
|
"logits/chosen": -0.16015625,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -374.0,
|
||
|
|
"loss": 0.3628,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.63671875,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2588957055214724,
|
||
|
|
"grad_norm": 38.561232670471654,
|
||
|
|
"learning_rate": 9.231404958677686e-07,
|
||
|
|
"logits/chosen": -0.251953125,
|
||
|
|
"logits/rejected": -0.458984375,
|
||
|
|
"logps/chosen": -436.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3366,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26012269938650306,
|
||
|
|
"grad_norm": 36.86026213129111,
|
||
|
|
"learning_rate": 9.227272727272727e-07,
|
||
|
|
"logits/chosen": -0.181640625,
|
||
|
|
"logits/rejected": -0.34765625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2974,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26134969325153373,
|
||
|
|
"grad_norm": 33.76063146504063,
|
||
|
|
"learning_rate": 9.223140495867769e-07,
|
||
|
|
"logits/chosen": -0.12451171875,
|
||
|
|
"logits/rejected": -0.1962890625,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.2679,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.58984375,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -4.8125,
|
||
|
|
"step": 213
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2625766871165644,
|
||
|
|
"grad_norm": 44.147098970181965,
|
||
|
|
"learning_rate": 9.21900826446281e-07,
|
||
|
|
"logits/chosen": -0.10302734375,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3562,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.984375,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26380368098159507,
|
||
|
|
"grad_norm": 44.646862605268325,
|
||
|
|
"learning_rate": 9.214876033057852e-07,
|
||
|
|
"logits/chosen": -0.0791015625,
|
||
|
|
"logits/rejected": -0.1796875,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3235,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.62890625,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2650306748466258,
|
||
|
|
"grad_norm": 40.25809763769465,
|
||
|
|
"learning_rate": 9.210743801652892e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.3204,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.447265625,
|
||
|
|
"rewards/margins": 5.0625,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26625766871165646,
|
||
|
|
"grad_norm": 42.423324639307225,
|
||
|
|
"learning_rate": 9.206611570247933e-07,
|
||
|
|
"logits/chosen": -0.04248046875,
|
||
|
|
"logits/rejected": -0.1923828125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.4054,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.3359375,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2674846625766871,
|
||
|
|
"grad_norm": 46.69438445358285,
|
||
|
|
"learning_rate": 9.202479338842974e-07,
|
||
|
|
"logits/chosen": -0.0849609375,
|
||
|
|
"logits/rejected": -0.2021484375,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3932,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.0234375,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -4.28125,
|
||
|
|
"step": 218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2687116564417178,
|
||
|
|
"grad_norm": 35.2004707186147,
|
||
|
|
"learning_rate": 9.198347107438016e-07,
|
||
|
|
"logits/chosen": -0.181640625,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3107,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.109375,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -4.5625,
|
||
|
|
"step": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26993865030674846,
|
||
|
|
"grad_norm": 50.169633893127326,
|
||
|
|
"learning_rate": 9.194214876033057e-07,
|
||
|
|
"logits/chosen": -0.0322265625,
|
||
|
|
"logits/rejected": -0.1123046875,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.4295,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.68359375,
|
||
|
|
"rewards/margins": 4.65625,
|
||
|
|
"rewards/rejected": -3.96875,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27116564417177913,
|
||
|
|
"grad_norm": 42.093475734809076,
|
||
|
|
"learning_rate": 9.190082644628098e-07,
|
||
|
|
"logits/chosen": -0.1328125,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3922,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.953125,
|
||
|
|
"rewards/margins": 5.25,
|
||
|
|
"rewards/rejected": -4.3125,
|
||
|
|
"step": 221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2723926380368098,
|
||
|
|
"grad_norm": 36.0657090040612,
|
||
|
|
"learning_rate": 9.18595041322314e-07,
|
||
|
|
"logits/chosen": -0.306640625,
|
||
|
|
"logits/rejected": -0.451171875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2895,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.9375,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 222
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27361963190184047,
|
||
|
|
"grad_norm": 37.91090520092486,
|
||
|
|
"learning_rate": 9.181818181818181e-07,
|
||
|
|
"logits/chosen": -0.11328125,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 5.46875,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 223
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2748466257668712,
|
||
|
|
"grad_norm": 39.87455871283598,
|
||
|
|
"learning_rate": 9.177685950413223e-07,
|
||
|
|
"logits/chosen": -0.2216796875,
|
||
|
|
"logits/rejected": -0.408203125,
|
||
|
|
"logps/chosen": -446.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2872,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.890625,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27607361963190186,
|
||
|
|
"grad_norm": 35.385311266968216,
|
||
|
|
"learning_rate": 9.173553719008264e-07,
|
||
|
|
"logits/chosen": -0.09228515625,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2718,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.171875,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2773006134969325,
|
||
|
|
"grad_norm": 37.620291584299586,
|
||
|
|
"learning_rate": 9.169421487603305e-07,
|
||
|
|
"logits/chosen": -0.10888671875,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.353,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2785276073619632,
|
||
|
|
"grad_norm": 41.70527475343198,
|
||
|
|
"learning_rate": 9.165289256198346e-07,
|
||
|
|
"logits/chosen": -0.14453125,
|
||
|
|
"logits/rejected": -0.22265625,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.352,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.265625,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 227
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27975460122699386,
|
||
|
|
"grad_norm": 47.672730884256275,
|
||
|
|
"learning_rate": 9.161157024793388e-07,
|
||
|
|
"logits/chosen": -0.15234375,
|
||
|
|
"logits/rejected": -0.341796875,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3342,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.125,
|
||
|
|
"rewards/margins": 5.5,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2809815950920245,
|
||
|
|
"grad_norm": 36.5310600090198,
|
||
|
|
"learning_rate": 9.157024793388429e-07,
|
||
|
|
"logits/chosen": -0.05908203125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -3.78125,
|
||
|
|
"step": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2822085889570552,
|
||
|
|
"grad_norm": 34.01874266449764,
|
||
|
|
"learning_rate": 9.152892561983471e-07,
|
||
|
|
"logits/chosen": -0.06005859375,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.253,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.4140625,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28343558282208586,
|
||
|
|
"grad_norm": 43.76870205345571,
|
||
|
|
"learning_rate": 9.148760330578512e-07,
|
||
|
|
"logits/chosen": -0.1357421875,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.4191,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.46875,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -3.875,
|
||
|
|
"step": 231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2846625766871166,
|
||
|
|
"grad_norm": 36.17795323546657,
|
||
|
|
"learning_rate": 9.144628099173554e-07,
|
||
|
|
"logits/chosen": -0.2080078125,
|
||
|
|
"logits/rejected": -0.345703125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3191,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28588957055214725,
|
||
|
|
"grad_norm": 35.1635442877376,
|
||
|
|
"learning_rate": 9.140495867768595e-07,
|
||
|
|
"logits/chosen": -0.10205078125,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.8828125,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2871165644171779,
|
||
|
|
"grad_norm": 37.827856451736956,
|
||
|
|
"learning_rate": 9.136363636363637e-07,
|
||
|
|
"logits/chosen": -0.248046875,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3271,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.6953125,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -3.75,
|
||
|
|
"step": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2883435582822086,
|
||
|
|
"grad_norm": 36.88124628560585,
|
||
|
|
"learning_rate": 9.132231404958677e-07,
|
||
|
|
"logits/chosen": -0.26171875,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3114,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.125,
|
||
|
|
"step": 235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28957055214723926,
|
||
|
|
"grad_norm": 48.08754663310893,
|
||
|
|
"learning_rate": 9.128099173553719e-07,
|
||
|
|
"logits/chosen": -0.0145263671875,
|
||
|
|
"logits/rejected": -0.1962890625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.4279,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 2.234375,
|
||
|
|
"rewards/margins": 5.46875,
|
||
|
|
"rewards/rejected": -3.25,
|
||
|
|
"step": 236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2907975460122699,
|
||
|
|
"grad_norm": 33.733681477930965,
|
||
|
|
"learning_rate": 9.12396694214876e-07,
|
||
|
|
"logits/chosen": -0.1669921875,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2814,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.875,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.0625,
|
||
|
|
"step": 237
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2920245398773006,
|
||
|
|
"grad_norm": 41.709782097463915,
|
||
|
|
"learning_rate": 9.119834710743802e-07,
|
||
|
|
"logits/chosen": -0.181640625,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3809,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -3.765625,
|
||
|
|
"step": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29325153374233126,
|
||
|
|
"grad_norm": 41.613433419635136,
|
||
|
|
"learning_rate": 9.115702479338843e-07,
|
||
|
|
"logits/chosen": -0.1943359375,
|
||
|
|
"logits/rejected": -0.37890625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3411,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.3125,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -4.0,
|
||
|
|
"step": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.294478527607362,
|
||
|
|
"grad_norm": 33.76902045077368,
|
||
|
|
"learning_rate": 9.111570247933884e-07,
|
||
|
|
"logits/chosen": -0.212890625,
|
||
|
|
"logits/rejected": -0.36328125,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2174,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.125,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29570552147239265,
|
||
|
|
"grad_norm": 35.54809707331976,
|
||
|
|
"learning_rate": 9.107438016528926e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2793,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.25,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2969325153374233,
|
||
|
|
"grad_norm": 31.39332422823268,
|
||
|
|
"learning_rate": 9.103305785123967e-07,
|
||
|
|
"logits/chosen": -0.1611328125,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.238,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.298159509202454,
|
||
|
|
"grad_norm": 42.489213771046316,
|
||
|
|
"learning_rate": 9.099173553719007e-07,
|
||
|
|
"logits/chosen": -0.244140625,
|
||
|
|
"logits/rejected": -0.390625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.3768,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.46875,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -4.28125,
|
||
|
|
"step": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29938650306748466,
|
||
|
|
"grad_norm": 42.520272970122726,
|
||
|
|
"learning_rate": 9.095041322314048e-07,
|
||
|
|
"logits/chosen": -0.236328125,
|
||
|
|
"logits/rejected": -0.45703125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.4879,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.3515625,
|
||
|
|
"rewards/margins": 5.5,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3006134969325153,
|
||
|
|
"grad_norm": 36.80182572222518,
|
||
|
|
"learning_rate": 9.09090909090909e-07,
|
||
|
|
"logits/chosen": -0.2080078125,
|
||
|
|
"logits/rejected": -0.349609375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2633,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.301840490797546,
|
||
|
|
"grad_norm": 36.10259081942398,
|
||
|
|
"learning_rate": 9.086776859504131e-07,
|
||
|
|
"logits/chosen": -0.2119140625,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3021,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 2.296875,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3030674846625767,
|
||
|
|
"grad_norm": 37.745736662906765,
|
||
|
|
"learning_rate": 9.082644628099173e-07,
|
||
|
|
"logits/chosen": -0.20703125,
|
||
|
|
"logits/rejected": -0.390625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.317,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.8125,
|
||
|
|
"step": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3042944785276074,
|
||
|
|
"grad_norm": 35.08194100895569,
|
||
|
|
"learning_rate": 9.078512396694214e-07,
|
||
|
|
"logits/chosen": -0.125,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2352,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.6640625,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 248
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30552147239263805,
|
||
|
|
"grad_norm": 39.937675917368395,
|
||
|
|
"learning_rate": 9.074380165289256e-07,
|
||
|
|
"logits/chosen": -0.2470703125,
|
||
|
|
"logits/rejected": -0.443359375,
|
||
|
|
"logps/chosen": -474.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.294,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 2.625,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3067484662576687,
|
||
|
|
"grad_norm": 40.869769785212625,
|
||
|
|
"learning_rate": 9.070247933884297e-07,
|
||
|
|
"logits/chosen": -0.3359375,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3549,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3079754601226994,
|
||
|
|
"grad_norm": 40.52486042802848,
|
||
|
|
"learning_rate": 9.066115702479339e-07,
|
||
|
|
"logits/chosen": -0.02978515625,
|
||
|
|
"logits/rejected": -0.2001953125,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3659,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -3.828125,
|
||
|
|
"step": 251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30920245398773005,
|
||
|
|
"grad_norm": 39.2129534588987,
|
||
|
|
"learning_rate": 9.061983471074379e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3449,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.9375,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -3.96875,
|
||
|
|
"step": 252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3104294478527607,
|
||
|
|
"grad_norm": 32.24326164619934,
|
||
|
|
"learning_rate": 9.057851239669421e-07,
|
||
|
|
"logits/chosen": -0.19921875,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.2689,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3116564417177914,
|
||
|
|
"grad_norm": 42.0234467969428,
|
||
|
|
"learning_rate": 9.053719008264462e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3892,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.8203125,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3128834355828221,
|
||
|
|
"grad_norm": 39.369143269027276,
|
||
|
|
"learning_rate": 9.049586776859504e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.375,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.8671875,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -3.703125,
|
||
|
|
"step": 255
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3141104294478528,
|
||
|
|
"grad_norm": 33.202524065959715,
|
||
|
|
"learning_rate": 9.045454545454545e-07,
|
||
|
|
"logits/chosen": -0.1875,
|
||
|
|
"logits/rejected": -0.35546875,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2491,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.375,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -4.5625,
|
||
|
|
"step": 256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31533742331288345,
|
||
|
|
"grad_norm": 41.51571179035507,
|
||
|
|
"learning_rate": 9.041322314049586e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3173,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.265625,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -4.0,
|
||
|
|
"step": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3165644171779141,
|
||
|
|
"grad_norm": 35.738504133116166,
|
||
|
|
"learning_rate": 9.037190082644628e-07,
|
||
|
|
"logits/chosen": -0.1298828125,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2726,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.921875,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.40625,
|
||
|
|
"step": 258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3177914110429448,
|
||
|
|
"grad_norm": 34.59247141687721,
|
||
|
|
"learning_rate": 9.033057851239669e-07,
|
||
|
|
"logits/chosen": -0.19140625,
|
||
|
|
"logits/rejected": -0.3828125,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31901840490797545,
|
||
|
|
"grad_norm": 37.65821660368043,
|
||
|
|
"learning_rate": 9.028925619834711e-07,
|
||
|
|
"logits/chosen": -0.2197265625,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3336,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -4.28125,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3202453987730061,
|
||
|
|
"grad_norm": 40.17093324908733,
|
||
|
|
"learning_rate": 9.024793388429751e-07,
|
||
|
|
"logits/chosen": -0.150390625,
|
||
|
|
"logits/rejected": -0.30078125,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3215,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.6171875,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.40625,
|
||
|
|
"step": 261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3214723926380368,
|
||
|
|
"grad_norm": 41.77028106523845,
|
||
|
|
"learning_rate": 9.020661157024793e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3204,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.203125,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -4.0625,
|
||
|
|
"step": 262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3226993865030675,
|
||
|
|
"grad_norm": 45.33050032219676,
|
||
|
|
"learning_rate": 9.016528925619834e-07,
|
||
|
|
"logits/chosen": -0.1826171875,
|
||
|
|
"logits/rejected": -0.30859375,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3834,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.0859375,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -3.953125,
|
||
|
|
"step": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3239263803680982,
|
||
|
|
"grad_norm": 41.9279090830257,
|
||
|
|
"learning_rate": 9.012396694214876e-07,
|
||
|
|
"logits/chosen": -0.140625,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.34,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -3.921875,
|
||
|
|
"step": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32515337423312884,
|
||
|
|
"grad_norm": 47.31340400046535,
|
||
|
|
"learning_rate": 9.008264462809917e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -274.0,
|
||
|
|
"loss": 0.2994,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.375,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -4.4375,
|
||
|
|
"step": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3263803680981595,
|
||
|
|
"grad_norm": 51.49399504173386,
|
||
|
|
"learning_rate": 9.004132231404959e-07,
|
||
|
|
"logits/chosen": -0.1220703125,
|
||
|
|
"logits/rejected": -0.26953125,
|
||
|
|
"logps/chosen": -320.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.464,
|
||
|
|
"rewards/accuracies": 0.7421875,
|
||
|
|
"rewards/chosen": 0.423828125,
|
||
|
|
"rewards/margins": 4.53125,
|
||
|
|
"rewards/rejected": -4.125,
|
||
|
|
"step": 266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3276073619631902,
|
||
|
|
"grad_norm": 49.994468797582975,
|
||
|
|
"learning_rate": 9e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.361328125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3743,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.8046875,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -3.953125,
|
||
|
|
"step": 267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32883435582822085,
|
||
|
|
"grad_norm": 31.9275570965626,
|
||
|
|
"learning_rate": 8.995867768595042e-07,
|
||
|
|
"logits/chosen": -0.2890625,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2019,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.484375,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3300613496932515,
|
||
|
|
"grad_norm": 34.908633509533956,
|
||
|
|
"learning_rate": 8.991735537190083e-07,
|
||
|
|
"logits/chosen": -0.1728515625,
|
||
|
|
"logits/rejected": -0.478515625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2272,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.7265625,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -4.8125,
|
||
|
|
"step": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3312883435582822,
|
||
|
|
"grad_norm": 42.11747091260519,
|
||
|
|
"learning_rate": 8.987603305785123e-07,
|
||
|
|
"logits/chosen": -0.171875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -328.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3885,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.0390625,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -3.984375,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3325153374233129,
|
||
|
|
"grad_norm": 31.100938375184207,
|
||
|
|
"learning_rate": 8.983471074380164e-07,
|
||
|
|
"logits/chosen": -0.1865234375,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.262,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.078125,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3337423312883436,
|
||
|
|
"grad_norm": 36.40948876277434,
|
||
|
|
"learning_rate": 8.979338842975206e-07,
|
||
|
|
"logits/chosen": -0.0791015625,
|
||
|
|
"logits/rejected": -0.244140625,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.263,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.0859375,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 272
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33496932515337424,
|
||
|
|
"grad_norm": 43.97482452623596,
|
||
|
|
"learning_rate": 8.975206611570247e-07,
|
||
|
|
"logits/chosen": -0.025146484375,
|
||
|
|
"logits/rejected": -0.10498046875,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.4597,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.86328125,
|
||
|
|
"rewards/margins": 5.15625,
|
||
|
|
"rewards/rejected": -4.28125,
|
||
|
|
"step": 273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3361963190184049,
|
||
|
|
"grad_norm": 38.58556750213493,
|
||
|
|
"learning_rate": 8.971074380165289e-07,
|
||
|
|
"logits/chosen": -0.04345703125,
|
||
|
|
"logits/rejected": -0.189453125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3517,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3374233128834356,
|
||
|
|
"grad_norm": 36.32564369328471,
|
||
|
|
"learning_rate": 8.96694214876033e-07,
|
||
|
|
"logits/chosen": -0.1103515625,
|
||
|
|
"logits/rejected": -0.20703125,
|
||
|
|
"logps/chosen": -318.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3319,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -4.40625,
|
||
|
|
"step": 275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33865030674846625,
|
||
|
|
"grad_norm": 28.631593511584413,
|
||
|
|
"learning_rate": 8.962809917355371e-07,
|
||
|
|
"logits/chosen": -0.034912109375,
|
||
|
|
"logits/rejected": -0.244140625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.295,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3398773006134969,
|
||
|
|
"grad_norm": 30.563119880156428,
|
||
|
|
"learning_rate": 8.958677685950413e-07,
|
||
|
|
"logits/chosen": -0.2099609375,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2232,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.8828125,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3411042944785276,
|
||
|
|
"grad_norm": 43.29625763685435,
|
||
|
|
"learning_rate": 8.954545454545454e-07,
|
||
|
|
"logits/chosen": -0.125,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3369,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.3359375,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3423312883435583,
|
||
|
|
"grad_norm": 23.231526168499595,
|
||
|
|
"learning_rate": 8.950413223140495e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.1765,
|
||
|
|
"rewards/accuracies": 0.9375,
|
||
|
|
"rewards/chosen": 1.421875,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -5.34375,
|
||
|
|
"step": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34355828220858897,
|
||
|
|
"grad_norm": 42.428969483206515,
|
||
|
|
"learning_rate": 8.946280991735536e-07,
|
||
|
|
"logits/chosen": -0.16015625,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.3827,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34478527607361964,
|
||
|
|
"grad_norm": 37.34866118523232,
|
||
|
|
"learning_rate": 8.942148760330578e-07,
|
||
|
|
"logits/chosen": -0.09765625,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3435,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.1875,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3460122699386503,
|
||
|
|
"grad_norm": 33.25917218474408,
|
||
|
|
"learning_rate": 8.938016528925619e-07,
|
||
|
|
"logits/chosen": -0.26953125,
|
||
|
|
"logits/rejected": -0.421875,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2298,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.347239263803681,
|
||
|
|
"grad_norm": 34.34252527463677,
|
||
|
|
"learning_rate": 8.933884297520661e-07,
|
||
|
|
"logits/chosen": -0.294921875,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -288.0,
|
||
|
|
"loss": 0.2751,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.125,
|
||
|
|
"rewards/margins": 5.625,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34846625766871164,
|
||
|
|
"grad_norm": 38.92697687813909,
|
||
|
|
"learning_rate": 8.929752066115702e-07,
|
||
|
|
"logits/chosen": -0.1142578125,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.2851,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2890625,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3496932515337423,
|
||
|
|
"grad_norm": 30.695326350223162,
|
||
|
|
"learning_rate": 8.925619834710744e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.40234375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2414,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.350920245398773,
|
||
|
|
"grad_norm": 38.57430321053572,
|
||
|
|
"learning_rate": 8.921487603305785e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3505,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.375,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3521472392638037,
|
||
|
|
"grad_norm": 42.69834276963442,
|
||
|
|
"learning_rate": 8.917355371900827e-07,
|
||
|
|
"logits/chosen": -0.2265625,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3256,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.95703125,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.8125,
|
||
|
|
"step": 287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35337423312883437,
|
||
|
|
"grad_norm": 36.1832160066718,
|
||
|
|
"learning_rate": 8.913223140495867e-07,
|
||
|
|
"logits/chosen": -0.036376953125,
|
||
|
|
"logits/rejected": -0.1865234375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3261,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.95703125,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35460122699386504,
|
||
|
|
"grad_norm": 28.376305528750855,
|
||
|
|
"learning_rate": 8.909090909090909e-07,
|
||
|
|
"logits/chosen": -0.181640625,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.1858,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.6640625,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3558282208588957,
|
||
|
|
"grad_norm": 31.607934322430555,
|
||
|
|
"learning_rate": 8.90495867768595e-07,
|
||
|
|
"logits/chosen": -0.34375,
|
||
|
|
"logits/rejected": -0.48828125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.217,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.3984375,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3570552147239264,
|
||
|
|
"grad_norm": 38.81396858549568,
|
||
|
|
"learning_rate": 8.900826446280992e-07,
|
||
|
|
"logits/chosen": -0.28515625,
|
||
|
|
"logits/rejected": -0.412109375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3515,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35828220858895704,
|
||
|
|
"grad_norm": 37.738218399257526,
|
||
|
|
"learning_rate": 8.896694214876033e-07,
|
||
|
|
"logits/chosen": -0.2392578125,
|
||
|
|
"logits/rejected": -0.46484375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.2524,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.453125,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3595092024539877,
|
||
|
|
"grad_norm": 29.993709808838254,
|
||
|
|
"learning_rate": 8.892561983471074e-07,
|
||
|
|
"logits/chosen": -0.16015625,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2178,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 0.9375,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36073619631901843,
|
||
|
|
"grad_norm": 26.82462113028186,
|
||
|
|
"learning_rate": 8.888429752066116e-07,
|
||
|
|
"logits/chosen": -0.4140625,
|
||
|
|
"logits/rejected": -0.5625,
|
||
|
|
"logps/chosen": -462.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2557,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.6171875,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -5.59375,
|
||
|
|
"step": 294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3619631901840491,
|
||
|
|
"grad_norm": 31.872360219352924,
|
||
|
|
"learning_rate": 8.884297520661157e-07,
|
||
|
|
"logits/chosen": -0.12451171875,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2396,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36319018404907977,
|
||
|
|
"grad_norm": 40.252951579438076,
|
||
|
|
"learning_rate": 8.880165289256199e-07,
|
||
|
|
"logits/chosen": -0.0966796875,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3687,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36441717791411044,
|
||
|
|
"grad_norm": 31.43163856844169,
|
||
|
|
"learning_rate": 8.876033057851238e-07,
|
||
|
|
"logits/chosen": -0.2275390625,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2373,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.6328125,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3656441717791411,
|
||
|
|
"grad_norm": 40.16169035445508,
|
||
|
|
"learning_rate": 8.87190082644628e-07,
|
||
|
|
"logits/chosen": -0.10400390625,
|
||
|
|
"logits/rejected": -0.2041015625,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.322,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.03125,
|
||
|
|
"step": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36687116564417177,
|
||
|
|
"grad_norm": 44.752661502125484,
|
||
|
|
"learning_rate": 8.867768595041321e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3433,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.96484375,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36809815950920244,
|
||
|
|
"grad_norm": 40.61630087470012,
|
||
|
|
"learning_rate": 8.863636363636363e-07,
|
||
|
|
"logits/chosen": -0.25,
|
||
|
|
"logits/rejected": -0.40625,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3683,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3693251533742331,
|
||
|
|
"grad_norm": 34.184057413647636,
|
||
|
|
"learning_rate": 8.859504132231404e-07,
|
||
|
|
"logits/chosen": -0.193359375,
|
||
|
|
"logits/rejected": -0.38671875,
|
||
|
|
"logps/chosen": -448.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.3007,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37055214723926383,
|
||
|
|
"grad_norm": 28.52262909170181,
|
||
|
|
"learning_rate": 8.855371900826446e-07,
|
||
|
|
"logits/chosen": -0.23046875,
|
||
|
|
"logits/rejected": -0.4296875,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.2335,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.578125,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.4375,
|
||
|
|
"step": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3717791411042945,
|
||
|
|
"grad_norm": 45.58508428540745,
|
||
|
|
"learning_rate": 8.851239669421487e-07,
|
||
|
|
"logits/chosen": -0.2294921875,
|
||
|
|
"logits/rejected": -0.34765625,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3977,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.99609375,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37300613496932516,
|
||
|
|
"grad_norm": 29.98788371317914,
|
||
|
|
"learning_rate": 8.847107438016529e-07,
|
||
|
|
"logits/chosen": -0.11767578125,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2267,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.421875,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37423312883435583,
|
||
|
|
"grad_norm": 35.17479410559687,
|
||
|
|
"learning_rate": 8.84297520661157e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.34765625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3025,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.78125,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3754601226993865,
|
||
|
|
"grad_norm": 39.17340101734157,
|
||
|
|
"learning_rate": 8.838842975206611e-07,
|
||
|
|
"logits/chosen": -0.294921875,
|
||
|
|
"logits/rejected": -0.45703125,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2699,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.2109375,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37668711656441717,
|
||
|
|
"grad_norm": 37.77108585187058,
|
||
|
|
"learning_rate": 8.834710743801652e-07,
|
||
|
|
"logits/chosen": -0.15234375,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3981,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.21875,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37791411042944784,
|
||
|
|
"grad_norm": 37.13471968173906,
|
||
|
|
"learning_rate": 8.830578512396694e-07,
|
||
|
|
"logits/chosen": -0.109375,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3474,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.89453125,
|
||
|
|
"rewards/margins": 5.25,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3791411042944785,
|
||
|
|
"grad_norm": 32.80133096174931,
|
||
|
|
"learning_rate": 8.826446280991735e-07,
|
||
|
|
"logits/chosen": -0.283203125,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2405,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 1.6015625,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 309
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3803680981595092,
|
||
|
|
"grad_norm": 36.675395105919584,
|
||
|
|
"learning_rate": 8.822314049586776e-07,
|
||
|
|
"logits/chosen": -0.061767578125,
|
||
|
|
"logits/rejected": -0.158203125,
|
||
|
|
"logps/chosen": -312.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3773,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.15234375,
|
||
|
|
"rewards/margins": 5.03125,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3815950920245399,
|
||
|
|
"grad_norm": 33.25381441708521,
|
||
|
|
"learning_rate": 8.818181818181818e-07,
|
||
|
|
"logits/chosen": -0.2421875,
|
||
|
|
"logits/rejected": -0.423828125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2785,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.3984375,
|
||
|
|
"rewards/margins": 7.09375,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38282208588957056,
|
||
|
|
"grad_norm": 42.64167876910758,
|
||
|
|
"learning_rate": 8.814049586776859e-07,
|
||
|
|
"logits/chosen": -0.09765625,
|
||
|
|
"logits/rejected": -0.236328125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.4387,
|
||
|
|
"rewards/accuracies": 0.765625,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38404907975460123,
|
||
|
|
"grad_norm": 35.315697341783846,
|
||
|
|
"learning_rate": 8.809917355371901e-07,
|
||
|
|
"logits/chosen": -0.109375,
|
||
|
|
"logits/rejected": -0.388671875,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3074,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 2.046875,
|
||
|
|
"rewards/margins": 7.5625,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3852760736196319,
|
||
|
|
"grad_norm": 28.577677082306792,
|
||
|
|
"learning_rate": 8.805785123966942e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -452.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.1843,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 2.625,
|
||
|
|
"rewards/margins": 8.5,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38650306748466257,
|
||
|
|
"grad_norm": 34.67651692357887,
|
||
|
|
"learning_rate": 8.801652892561983e-07,
|
||
|
|
"logits/chosen": -0.03564453125,
|
||
|
|
"logits/rejected": -0.23046875,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -288.0,
|
||
|
|
"loss": 0.2637,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.89453125,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.25,
|
||
|
|
"step": 315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38773006134969323,
|
||
|
|
"grad_norm": 31.070595317274808,
|
||
|
|
"learning_rate": 8.797520661157024e-07,
|
||
|
|
"logits/chosen": -0.04541015625,
|
||
|
|
"logits/rejected": -0.19140625,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.2748,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.2109375,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3889570552147239,
|
||
|
|
"grad_norm": 39.24846705876142,
|
||
|
|
"learning_rate": 8.793388429752066e-07,
|
||
|
|
"logits/chosen": -0.17578125,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.2469,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.3828125,
|
||
|
|
"rewards/margins": 7.40625,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3901840490797546,
|
||
|
|
"grad_norm": 41.29363946363224,
|
||
|
|
"learning_rate": 8.789256198347107e-07,
|
||
|
|
"logits/chosen": -0.224609375,
|
||
|
|
"logits/rejected": -0.388671875,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3595,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.375,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3914110429447853,
|
||
|
|
"grad_norm": 21.922537381657932,
|
||
|
|
"learning_rate": 8.785123966942149e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.345703125,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.1571,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 7.9375,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39263803680981596,
|
||
|
|
"grad_norm": 37.85813864992483,
|
||
|
|
"learning_rate": 8.78099173553719e-07,
|
||
|
|
"logits/chosen": -0.1533203125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -318.0,
|
||
|
|
"logps/rejected": -276.0,
|
||
|
|
"loss": 0.3097,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.671875,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.5625,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39386503067484663,
|
||
|
|
"grad_norm": 44.42516577854287,
|
||
|
|
"learning_rate": 8.776859504132232e-07,
|
||
|
|
"logits/chosen": -0.275390625,
|
||
|
|
"logits/rejected": -0.4453125,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -388.0,
|
||
|
|
"loss": 0.3335,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.8671875,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3950920245398773,
|
||
|
|
"grad_norm": 29.9172278345542,
|
||
|
|
"learning_rate": 8.772727272727273e-07,
|
||
|
|
"logits/chosen": -0.1943359375,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2506,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.5234375,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39631901840490796,
|
||
|
|
"grad_norm": 38.156948353448634,
|
||
|
|
"learning_rate": 8.768595041322315e-07,
|
||
|
|
"logits/chosen": -0.220703125,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.4208,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39754601226993863,
|
||
|
|
"grad_norm": 32.767921456046125,
|
||
|
|
"learning_rate": 8.764462809917354e-07,
|
||
|
|
"logits/chosen": -0.265625,
|
||
|
|
"logits/rejected": -0.439453125,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3067,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.96875,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3987730061349693,
|
||
|
|
"grad_norm": 40.13577337526863,
|
||
|
|
"learning_rate": 8.760330578512396e-07,
|
||
|
|
"logits/chosen": -0.119140625,
|
||
|
|
"logits/rejected": -0.1962890625,
|
||
|
|
"logps/chosen": -328.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.4481,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.6796875,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"grad_norm": 31.397049512750392,
|
||
|
|
"learning_rate": 8.756198347107437e-07,
|
||
|
|
"logits/chosen": -0.296875,
|
||
|
|
"logits/rejected": -0.44921875,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2169,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 2.046875,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4012269938650307,
|
||
|
|
"grad_norm": 43.09847438786523,
|
||
|
|
"learning_rate": 8.752066115702479e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.36328125,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.4159,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.4296875,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.09375,
|
||
|
|
"step": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40245398773006136,
|
||
|
|
"grad_norm": 38.59522801863003,
|
||
|
|
"learning_rate": 8.74793388429752e-07,
|
||
|
|
"logits/chosen": -0.2177734375,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -276.0,
|
||
|
|
"loss": 0.3241,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.78125,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.403680981595092,
|
||
|
|
"grad_norm": 35.517969492099304,
|
||
|
|
"learning_rate": 8.743801652892561e-07,
|
||
|
|
"logits/chosen": -0.162109375,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -456.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.3229,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 2.0,
|
||
|
|
"rewards/margins": 7.625,
|
||
|
|
"rewards/rejected": -5.59375,
|
||
|
|
"step": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4049079754601227,
|
||
|
|
"grad_norm": 32.81462048360016,
|
||
|
|
"learning_rate": 8.739669421487603e-07,
|
||
|
|
"logits/chosen": -0.298828125,
|
||
|
|
"logits/rejected": -0.5078125,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2512,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.796875,
|
||
|
|
"rewards/margins": 7.25,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40613496932515336,
|
||
|
|
"grad_norm": 31.55365723251363,
|
||
|
|
"learning_rate": 8.735537190082644e-07,
|
||
|
|
"logits/chosen": -0.26171875,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.2528,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.7890625,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40736196319018403,
|
||
|
|
"grad_norm": 25.058536630302516,
|
||
|
|
"learning_rate": 8.731404958677685e-07,
|
||
|
|
"logits/chosen": -0.271484375,
|
||
|
|
"logits/rejected": -0.46484375,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2033,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.984375,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4085889570552147,
|
||
|
|
"grad_norm": 37.27877790438504,
|
||
|
|
"learning_rate": 8.727272727272726e-07,
|
||
|
|
"logits/chosen": -0.087890625,
|
||
|
|
"logits/rejected": -0.216796875,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.332,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.3515625,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4098159509202454,
|
||
|
|
"grad_norm": 32.893281250723994,
|
||
|
|
"learning_rate": 8.723140495867768e-07,
|
||
|
|
"logits/chosen": -0.2021484375,
|
||
|
|
"logits/rejected": -0.439453125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2882,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.5859375,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4110429447852761,
|
||
|
|
"grad_norm": 37.617815711991454,
|
||
|
|
"learning_rate": 8.719008264462809e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.380859375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2311,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.84375,
|
||
|
|
"step": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41226993865030676,
|
||
|
|
"grad_norm": 34.178158756977034,
|
||
|
|
"learning_rate": 8.714876033057851e-07,
|
||
|
|
"logits/chosen": -0.2734375,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3112,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.3359375,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4134969325153374,
|
||
|
|
"grad_norm": 43.74423728378528,
|
||
|
|
"learning_rate": 8.710743801652892e-07,
|
||
|
|
"logits/chosen": -0.0673828125,
|
||
|
|
"logits/rejected": -0.2373046875,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.4057,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 1.3046875,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4147239263803681,
|
||
|
|
"grad_norm": 34.34678654972963,
|
||
|
|
"learning_rate": 8.706611570247934e-07,
|
||
|
|
"logits/chosen": -0.11083984375,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3277,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.359375,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41595092024539876,
|
||
|
|
"grad_norm": 38.33964943892026,
|
||
|
|
"learning_rate": 8.702479338842975e-07,
|
||
|
|
"logits/chosen": -0.046630859375,
|
||
|
|
"logits/rejected": -0.18359375,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3247,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.1328125,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.875,
|
||
|
|
"step": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4171779141104294,
|
||
|
|
"grad_norm": 42.20123185254217,
|
||
|
|
"learning_rate": 8.698347107438017e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3163,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.9765625,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41840490797546015,
|
||
|
|
"grad_norm": 30.617090577606344,
|
||
|
|
"learning_rate": 8.694214876033057e-07,
|
||
|
|
"logits/chosen": -0.1904296875,
|
||
|
|
"logits/rejected": -0.412109375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3055,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.5390625,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4196319018404908,
|
||
|
|
"grad_norm": 36.98686744550473,
|
||
|
|
"learning_rate": 8.690082644628099e-07,
|
||
|
|
"logits/chosen": -0.14453125,
|
||
|
|
"logits/rejected": -0.248046875,
|
||
|
|
"logps/chosen": -330.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3403,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.9609375,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4208588957055215,
|
||
|
|
"grad_norm": 38.00637794749286,
|
||
|
|
"learning_rate": 8.68595041322314e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.2734375,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3859,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.89453125,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -4.4375,
|
||
|
|
"step": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42208588957055215,
|
||
|
|
"grad_norm": 39.71440557055468,
|
||
|
|
"learning_rate": 8.681818181818182e-07,
|
||
|
|
"logits/chosen": -0.17578125,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3756,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4233128834355828,
|
||
|
|
"grad_norm": 35.94521943416743,
|
||
|
|
"learning_rate": 8.677685950413223e-07,
|
||
|
|
"logits/chosen": -0.158203125,
|
||
|
|
"logits/rejected": -0.345703125,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3392,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4245398773006135,
|
||
|
|
"grad_norm": 38.26590405063427,
|
||
|
|
"learning_rate": 8.673553719008264e-07,
|
||
|
|
"logits/chosen": -0.1591796875,
|
||
|
|
"logits/rejected": -0.3515625,
|
||
|
|
"logps/chosen": -332.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.39,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.88671875,
|
||
|
|
"rewards/margins": 5.46875,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42576687116564416,
|
||
|
|
"grad_norm": 28.157756450395475,
|
||
|
|
"learning_rate": 8.669421487603306e-07,
|
||
|
|
"logits/chosen": -0.1796875,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.2541,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.1015625,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -4.65625,
|
||
|
|
"step": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4269938650306748,
|
||
|
|
"grad_norm": 30.247474673307313,
|
||
|
|
"learning_rate": 8.665289256198347e-07,
|
||
|
|
"logits/chosen": -0.2177734375,
|
||
|
|
"logits/rejected": -0.42578125,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2592,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.9140625,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42822085889570555,
|
||
|
|
"grad_norm": 40.56583556395029,
|
||
|
|
"learning_rate": 8.661157024793389e-07,
|
||
|
|
"logits/chosen": -0.205078125,
|
||
|
|
"logits/rejected": -0.3515625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3461,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.83203125,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4294478527607362,
|
||
|
|
"grad_norm": 37.32296255724777,
|
||
|
|
"learning_rate": 8.657024793388428e-07,
|
||
|
|
"logits/chosen": -0.0595703125,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.3201,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.078125,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -4.625,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4306748466257669,
|
||
|
|
"grad_norm": 38.61905817763815,
|
||
|
|
"learning_rate": 8.65289256198347e-07,
|
||
|
|
"logits/chosen": -0.2265625,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3683,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.55078125,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43190184049079755,
|
||
|
|
"grad_norm": 32.77930853827487,
|
||
|
|
"learning_rate": 8.648760330578511e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.337890625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3505,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.7734375,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.1875,
|
||
|
|
"step": 352
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4331288343558282,
|
||
|
|
"grad_norm": 28.429883368705962,
|
||
|
|
"learning_rate": 8.644628099173553e-07,
|
||
|
|
"logits/chosen": -0.1884765625,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.6875,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4343558282208589,
|
||
|
|
"grad_norm": 32.10596291611156,
|
||
|
|
"learning_rate": 8.640495867768594e-07,
|
||
|
|
"logits/chosen": -0.1083984375,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.301,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.09375,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.0,
|
||
|
|
"step": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43558282208588955,
|
||
|
|
"grad_norm": 35.70256329945714,
|
||
|
|
"learning_rate": 8.636363636363636e-07,
|
||
|
|
"logits/chosen": -0.26171875,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3752,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.3125,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4368098159509202,
|
||
|
|
"grad_norm": 33.540082224759566,
|
||
|
|
"learning_rate": 8.632231404958677e-07,
|
||
|
|
"logits/chosen": -0.05908203125,
|
||
|
|
"logits/rejected": -0.2041015625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.351,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.4296875,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.125,
|
||
|
|
"step": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43803680981595094,
|
||
|
|
"grad_norm": 32.54050040447977,
|
||
|
|
"learning_rate": 8.628099173553719e-07,
|
||
|
|
"logits/chosen": -0.318359375,
|
||
|
|
"logits/rejected": -0.470703125,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3107,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4392638036809816,
|
||
|
|
"grad_norm": 34.06624107320208,
|
||
|
|
"learning_rate": 8.62396694214876e-07,
|
||
|
|
"logits/chosen": -0.267578125,
|
||
|
|
"logits/rejected": -0.423828125,
|
||
|
|
"logps/chosen": -438.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2835,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 2.171875,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.40625,
|
||
|
|
"step": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4404907975460123,
|
||
|
|
"grad_norm": 36.767095986046016,
|
||
|
|
"learning_rate": 8.619834710743801e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3176,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 2.09375,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -4.03125,
|
||
|
|
"step": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44171779141104295,
|
||
|
|
"grad_norm": 34.37658092994157,
|
||
|
|
"learning_rate": 8.615702479338842e-07,
|
||
|
|
"logits/chosen": -0.1240234375,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.27,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.4296875,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.9375,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4429447852760736,
|
||
|
|
"grad_norm": 32.664077801869844,
|
||
|
|
"learning_rate": 8.611570247933884e-07,
|
||
|
|
"logits/chosen": -0.0732421875,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2781,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.953125,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4441717791411043,
|
||
|
|
"grad_norm": 31.9733261958367,
|
||
|
|
"learning_rate": 8.607438016528925e-07,
|
||
|
|
"logits/chosen": -0.20703125,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2931,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.78125,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44539877300613495,
|
||
|
|
"grad_norm": 29.251958446930967,
|
||
|
|
"learning_rate": 8.603305785123967e-07,
|
||
|
|
"logits/chosen": -0.1162109375,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2371,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.8671875,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4466257668711656,
|
||
|
|
"grad_norm": 37.57398787735362,
|
||
|
|
"learning_rate": 8.599173553719008e-07,
|
||
|
|
"logits/chosen": -0.042724609375,
|
||
|
|
"logits/rejected": -0.2265625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.3235,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.625,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44785276073619634,
|
||
|
|
"grad_norm": 29.870035318324586,
|
||
|
|
"learning_rate": 8.595041322314049e-07,
|
||
|
|
"logits/chosen": -0.10302734375,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.2644,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.3828125,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.449079754601227,
|
||
|
|
"grad_norm": 34.90800717205546,
|
||
|
|
"learning_rate": 8.590909090909091e-07,
|
||
|
|
"logits/chosen": -0.08056640625,
|
||
|
|
"logits/rejected": -0.18359375,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3873,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.7421875,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -3.8125,
|
||
|
|
"step": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4503067484662577,
|
||
|
|
"grad_norm": 29.571048038433712,
|
||
|
|
"learning_rate": 8.586776859504132e-07,
|
||
|
|
"logits/chosen": -0.2255859375,
|
||
|
|
"logits/rejected": -0.40625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.671875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45153374233128835,
|
||
|
|
"grad_norm": 42.056035274022705,
|
||
|
|
"learning_rate": 8.582644628099173e-07,
|
||
|
|
"logits/chosen": -0.034423828125,
|
||
|
|
"logits/rejected": -0.0859375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3686,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.2109375,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.452760736196319,
|
||
|
|
"grad_norm": 38.8917843400358,
|
||
|
|
"learning_rate": 8.578512396694214e-07,
|
||
|
|
"logits/chosen": -0.05712890625,
|
||
|
|
"logits/rejected": -0.1689453125,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.3427,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.62109375,
|
||
|
|
"rewards/margins": 5.28125,
|
||
|
|
"rewards/rejected": -4.65625,
|
||
|
|
"step": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4539877300613497,
|
||
|
|
"grad_norm": 35.30535390014327,
|
||
|
|
"learning_rate": 8.574380165289256e-07,
|
||
|
|
"logits/chosen": -0.2265625,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3331,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 2.03125,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45521472392638035,
|
||
|
|
"grad_norm": 31.840615729545142,
|
||
|
|
"learning_rate": 8.570247933884297e-07,
|
||
|
|
"logits/chosen": -0.2119140625,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -456.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2692,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.890625,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.03125,
|
||
|
|
"step": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.456441717791411,
|
||
|
|
"grad_norm": 29.955978159088186,
|
||
|
|
"learning_rate": 8.566115702479339e-07,
|
||
|
|
"logits/chosen": -0.12353515625,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2245,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.8125,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 372
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45766871165644174,
|
||
|
|
"grad_norm": 36.82441545636138,
|
||
|
|
"learning_rate": 8.56198347107438e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.37,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.828125,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4588957055214724,
|
||
|
|
"grad_norm": 29.01028759814529,
|
||
|
|
"learning_rate": 8.557851239669422e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2316,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 2.0625,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4601226993865031,
|
||
|
|
"grad_norm": 34.54866122563966,
|
||
|
|
"learning_rate": 8.553719008264463e-07,
|
||
|
|
"logits/chosen": -0.2314453125,
|
||
|
|
"logits/rejected": -0.314453125,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2607,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.7421875,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.21875,
|
||
|
|
"step": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46134969325153374,
|
||
|
|
"grad_norm": 33.10026205924125,
|
||
|
|
"learning_rate": 8.549586776859505e-07,
|
||
|
|
"logits/chosen": -0.1044921875,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2614,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.7421875,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.25,
|
||
|
|
"step": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4625766871165644,
|
||
|
|
"grad_norm": 35.75040093231158,
|
||
|
|
"learning_rate": 8.545454545454544e-07,
|
||
|
|
"logits/chosen": -0.1826171875,
|
||
|
|
"logits/rejected": -0.24609375,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3217,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4638036809815951,
|
||
|
|
"grad_norm": 37.937967119942094,
|
||
|
|
"learning_rate": 8.541322314049586e-07,
|
||
|
|
"logits/chosen": -0.2158203125,
|
||
|
|
"logits/rejected": -0.462890625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.332,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.5703125,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46503067484662575,
|
||
|
|
"grad_norm": 32.72768732817253,
|
||
|
|
"learning_rate": 8.537190082644627e-07,
|
||
|
|
"logits/chosen": -0.267578125,
|
||
|
|
"logits/rejected": -0.43359375,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2597,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.640625,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -5.59375,
|
||
|
|
"step": 379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4662576687116564,
|
||
|
|
"grad_norm": 33.93477462013243,
|
||
|
|
"learning_rate": 8.533057851239669e-07,
|
||
|
|
"logits/chosen": -0.177734375,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2747,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -5.34375,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46748466257668714,
|
||
|
|
"grad_norm": 33.523029253094535,
|
||
|
|
"learning_rate": 8.52892561983471e-07,
|
||
|
|
"logits/chosen": -0.263671875,
|
||
|
|
"logits/rejected": -0.439453125,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3434,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.875,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -6.125,
|
||
|
|
"step": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4687116564417178,
|
||
|
|
"grad_norm": 30.950990880904673,
|
||
|
|
"learning_rate": 8.524793388429751e-07,
|
||
|
|
"logits/chosen": -0.26953125,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.291,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.92578125,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4699386503067485,
|
||
|
|
"grad_norm": 37.335141994131185,
|
||
|
|
"learning_rate": 8.520661157024793e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -382.0,
|
||
|
|
"loss": 0.3653,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.85546875,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47116564417177914,
|
||
|
|
"grad_norm": 35.708837972051896,
|
||
|
|
"learning_rate": 8.516528925619834e-07,
|
||
|
|
"logits/chosen": -0.2734375,
|
||
|
|
"logits/rejected": -0.515625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2715,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.546875,
|
||
|
|
"rewards/margins": 7.5,
|
||
|
|
"rewards/rejected": -6.9375,
|
||
|
|
"step": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4723926380368098,
|
||
|
|
"grad_norm": 38.095621644948935,
|
||
|
|
"learning_rate": 8.512396694214876e-07,
|
||
|
|
"logits/chosen": -0.1787109375,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3249,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.279296875,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4736196319018405,
|
||
|
|
"grad_norm": 42.56085647887984,
|
||
|
|
"learning_rate": 8.508264462809916e-07,
|
||
|
|
"logits/chosen": -0.1337890625,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3519,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": -0.337890625,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47484662576687114,
|
||
|
|
"grad_norm": 63.07829836210694,
|
||
|
|
"learning_rate": 8.504132231404958e-07,
|
||
|
|
"logits/chosen": -0.1552734375,
|
||
|
|
"logits/rejected": -0.353515625,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3377,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.87890625,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47607361963190187,
|
||
|
|
"grad_norm": 44.508149553651,
|
||
|
|
"learning_rate": 8.499999999999999e-07,
|
||
|
|
"logits/chosen": -0.1728515625,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.4576,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.318359375,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47730061349693254,
|
||
|
|
"grad_norm": 43.15219698694189,
|
||
|
|
"learning_rate": 8.495867768595041e-07,
|
||
|
|
"logits/chosen": -0.2314453125,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.4252,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.765625,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -5.5625,
|
||
|
|
"step": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4785276073619632,
|
||
|
|
"grad_norm": 43.3947576259124,
|
||
|
|
"learning_rate": 8.491735537190082e-07,
|
||
|
|
"logits/chosen": -0.294921875,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.4287,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.10595703125,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47975460122699387,
|
||
|
|
"grad_norm": 42.71401974468104,
|
||
|
|
"learning_rate": 8.487603305785124e-07,
|
||
|
|
"logits/chosen": -0.2041015625,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.4278,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.69140625,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48098159509202454,
|
||
|
|
"grad_norm": 31.928442662001206,
|
||
|
|
"learning_rate": 8.483471074380165e-07,
|
||
|
|
"logits/chosen": -0.154296875,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2866,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.890625,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4822085889570552,
|
||
|
|
"grad_norm": 34.01837933004929,
|
||
|
|
"learning_rate": 8.479338842975207e-07,
|
||
|
|
"logits/chosen": -0.1181640625,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2862,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.796875,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4834355828220859,
|
||
|
|
"grad_norm": 34.069017331837635,
|
||
|
|
"learning_rate": 8.475206611570248e-07,
|
||
|
|
"logits/chosen": -0.1865234375,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.255,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.82421875,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48466257668711654,
|
||
|
|
"grad_norm": 36.10911763794192,
|
||
|
|
"learning_rate": 8.471074380165289e-07,
|
||
|
|
"logits/chosen": -0.189453125,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -438.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3367,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.85546875,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48588957055214727,
|
||
|
|
"grad_norm": 24.87975181789429,
|
||
|
|
"learning_rate": 8.46694214876033e-07,
|
||
|
|
"logits/chosen": -0.2578125,
|
||
|
|
"logits/rejected": -0.466796875,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2201,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 0.439453125,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48711656441717793,
|
||
|
|
"grad_norm": 37.31242821688829,
|
||
|
|
"learning_rate": 8.462809917355372e-07,
|
||
|
|
"logits/chosen": -0.232421875,
|
||
|
|
"logits/rejected": -0.404296875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3358,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.64453125,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4883435582822086,
|
||
|
|
"grad_norm": 38.435432269740204,
|
||
|
|
"learning_rate": 8.458677685950413e-07,
|
||
|
|
"logits/chosen": -0.220703125,
|
||
|
|
"logits/rejected": -0.3671875,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3625,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.0234375,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48957055214723927,
|
||
|
|
"grad_norm": 31.226281146954268,
|
||
|
|
"learning_rate": 8.454545454545454e-07,
|
||
|
|
"logits/chosen": -0.2451171875,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2934,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.90625,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49079754601226994,
|
||
|
|
"grad_norm": 45.39512753724527,
|
||
|
|
"learning_rate": 8.450413223140496e-07,
|
||
|
|
"logits/chosen": -0.1904296875,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.4427,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.5625,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4920245398773006,
|
||
|
|
"grad_norm": 30.28556425067376,
|
||
|
|
"learning_rate": 8.446280991735537e-07,
|
||
|
|
"logits/chosen": -0.314453125,
|
||
|
|
"logits/rejected": -0.4609375,
|
||
|
|
"logps/chosen": -446.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.2465,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.390625,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49325153374233127,
|
||
|
|
"grad_norm": 34.97541893034036,
|
||
|
|
"learning_rate": 8.442148760330579e-07,
|
||
|
|
"logits/chosen": -0.1787109375,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3267,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.7890625,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49447852760736194,
|
||
|
|
"grad_norm": 35.34876316760721,
|
||
|
|
"learning_rate": 8.43801652892562e-07,
|
||
|
|
"logits/chosen": -0.1728515625,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.31,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.71875,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49570552147239266,
|
||
|
|
"grad_norm": 30.378373289318976,
|
||
|
|
"learning_rate": 8.43388429752066e-07,
|
||
|
|
"logits/chosen": -0.11865234375,
|
||
|
|
"logits/rejected": -0.259765625,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2139,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 0.96875,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49693251533742333,
|
||
|
|
"grad_norm": 38.10870942161704,
|
||
|
|
"learning_rate": 8.429752066115701e-07,
|
||
|
|
"logits/chosen": -0.201171875,
|
||
|
|
"logits/rejected": -0.4296875,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3558,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.234375,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.498159509202454,
|
||
|
|
"grad_norm": 41.99213740680795,
|
||
|
|
"learning_rate": 8.425619834710743e-07,
|
||
|
|
"logits/chosen": -0.09521484375,
|
||
|
|
"logits/rejected": -0.1875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.44,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.90234375,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49938650306748467,
|
||
|
|
"grad_norm": 41.75199039490215,
|
||
|
|
"learning_rate": 8.421487603305784e-07,
|
||
|
|
"logits/chosen": -0.08544921875,
|
||
|
|
"logits/rejected": -0.2431640625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3833,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 2.015625,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5006134969325153,
|
||
|
|
"grad_norm": 46.46791360512976,
|
||
|
|
"learning_rate": 8.417355371900826e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.4566,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.55078125,
|
||
|
|
"rewards/margins": 5.3125,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.501840490797546,
|
||
|
|
"grad_norm": 44.673203673536904,
|
||
|
|
"learning_rate": 8.413223140495867e-07,
|
||
|
|
"logits/chosen": -0.11474609375,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3461,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.2890625,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 409
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5030674846625767,
|
||
|
|
"grad_norm": 38.196100384677806,
|
||
|
|
"learning_rate": 8.409090909090909e-07,
|
||
|
|
"logits/chosen": -0.1025390625,
|
||
|
|
"logits/rejected": -0.306640625,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3237,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.9296875,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5042944785276073,
|
||
|
|
"grad_norm": 36.65382709836008,
|
||
|
|
"learning_rate": 8.40495867768595e-07,
|
||
|
|
"logits/chosen": -0.09326171875,
|
||
|
|
"logits/rejected": -0.234375,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2836,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.6796875,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.505521472392638,
|
||
|
|
"grad_norm": 46.13136333228446,
|
||
|
|
"learning_rate": 8.400826446280992e-07,
|
||
|
|
"logits/chosen": -0.09765625,
|
||
|
|
"logits/rejected": -0.2353515625,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3856,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.95703125,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5067484662576687,
|
||
|
|
"grad_norm": 40.318779960108294,
|
||
|
|
"learning_rate": 8.396694214876032e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.36328125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3462,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.4375,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5079754601226993,
|
||
|
|
"grad_norm": 34.87712228181845,
|
||
|
|
"learning_rate": 8.392561983471074e-07,
|
||
|
|
"logits/chosen": -0.1806640625,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2741,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.296875,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.50920245398773,
|
||
|
|
"grad_norm": 34.161638850062126,
|
||
|
|
"learning_rate": 8.388429752066115e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2949,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.046875,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5104294478527608,
|
||
|
|
"grad_norm": 28.62101759801655,
|
||
|
|
"learning_rate": 8.384297520661157e-07,
|
||
|
|
"logits/chosen": -0.10791015625,
|
||
|
|
"logits/rejected": -0.328125,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.1788,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 0.8828125,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5116564417177915,
|
||
|
|
"grad_norm": 32.11517958415963,
|
||
|
|
"learning_rate": 8.380165289256198e-07,
|
||
|
|
"logits/chosen": -0.30859375,
|
||
|
|
"logits/rejected": -0.482421875,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -372.0,
|
||
|
|
"loss": 0.302,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.9765625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5128834355828221,
|
||
|
|
"grad_norm": 36.91670370496033,
|
||
|
|
"learning_rate": 8.376033057851239e-07,
|
||
|
|
"logits/chosen": -0.224609375,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3162,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5141104294478528,
|
||
|
|
"grad_norm": 29.74220467370693,
|
||
|
|
"learning_rate": 8.371900826446281e-07,
|
||
|
|
"logits/chosen": -0.15625,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2755,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.328125,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 419
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5153374233128835,
|
||
|
|
"grad_norm": 35.53103540854566,
|
||
|
|
"learning_rate": 8.367768595041322e-07,
|
||
|
|
"logits/chosen": -0.08837890625,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.335,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.09375,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5165644171779141,
|
||
|
|
"grad_norm": 32.97503154682443,
|
||
|
|
"learning_rate": 8.363636363636363e-07,
|
||
|
|
"logits/chosen": -0.1298828125,
|
||
|
|
"logits/rejected": -0.333984375,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3545,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5177914110429448,
|
||
|
|
"grad_norm": 38.11996274683729,
|
||
|
|
"learning_rate": 8.359504132231404e-07,
|
||
|
|
"logits/chosen": -0.0546875,
|
||
|
|
"logits/rejected": -0.2158203125,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3392,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.71484375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5190184049079755,
|
||
|
|
"grad_norm": 39.06478111720593,
|
||
|
|
"learning_rate": 8.355371900826446e-07,
|
||
|
|
"logits/chosen": -0.173828125,
|
||
|
|
"logits/rejected": -0.333984375,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.3845,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.67578125,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5202453987730061,
|
||
|
|
"grad_norm": 35.69529736649316,
|
||
|
|
"learning_rate": 8.351239669421487e-07,
|
||
|
|
"logits/chosen": -0.09619140625,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3743,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.1953125,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5214723926380368,
|
||
|
|
"grad_norm": 41.60888435807761,
|
||
|
|
"learning_rate": 8.347107438016529e-07,
|
||
|
|
"logits/chosen": -0.1328125,
|
||
|
|
"logits/rejected": -0.2470703125,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.4486,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.40625,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5226993865030675,
|
||
|
|
"grad_norm": 42.36087992828029,
|
||
|
|
"learning_rate": 8.34297520661157e-07,
|
||
|
|
"logits/chosen": -0.1787109375,
|
||
|
|
"logits/rejected": -0.30078125,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3332,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5239263803680981,
|
||
|
|
"grad_norm": 77.1683428566809,
|
||
|
|
"learning_rate": 8.338842975206612e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3382,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.09375,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5251533742331288,
|
||
|
|
"grad_norm": 35.0971325671411,
|
||
|
|
"learning_rate": 8.334710743801653e-07,
|
||
|
|
"logits/chosen": -0.166015625,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3745,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.546875,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5263803680981595,
|
||
|
|
"grad_norm": 30.4960218875002,
|
||
|
|
"learning_rate": 8.330578512396695e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -372.0,
|
||
|
|
"loss": 0.2397,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.275390625,
|
||
|
|
"rewards/margins": 7.40625,
|
||
|
|
"rewards/rejected": -7.125,
|
||
|
|
"step": 429
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5276073619631901,
|
||
|
|
"grad_norm": 27.007644146694663,
|
||
|
|
"learning_rate": 8.326446280991735e-07,
|
||
|
|
"logits/chosen": -0.2353515625,
|
||
|
|
"logits/rejected": -0.404296875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2181,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.80859375,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5288343558282208,
|
||
|
|
"grad_norm": 46.10911626487466,
|
||
|
|
"learning_rate": 8.322314049586777e-07,
|
||
|
|
"logits/chosen": -0.11767578125,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.4699,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.337890625,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5300613496932516,
|
||
|
|
"grad_norm": 39.706928116253934,
|
||
|
|
"learning_rate": 8.318181818181817e-07,
|
||
|
|
"logits/chosen": -0.06591796875,
|
||
|
|
"logits/rejected": -0.1083984375,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3755,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.1171875,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -5.34375,
|
||
|
|
"step": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5312883435582823,
|
||
|
|
"grad_norm": 44.60896484929849,
|
||
|
|
"learning_rate": 8.314049586776859e-07,
|
||
|
|
"logits/chosen": -0.04736328125,
|
||
|
|
"logits/rejected": -0.2734375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3756,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.2353515625,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5325153374233129,
|
||
|
|
"grad_norm": 39.167901096522954,
|
||
|
|
"learning_rate": 8.3099173553719e-07,
|
||
|
|
"logits/chosen": -0.20703125,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.34765625,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5337423312883436,
|
||
|
|
"grad_norm": 41.42002661257632,
|
||
|
|
"learning_rate": 8.305785123966941e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.349609375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.349,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.93359375,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5349693251533743,
|
||
|
|
"grad_norm": 36.28330999120649,
|
||
|
|
"learning_rate": 8.301652892561983e-07,
|
||
|
|
"logits/chosen": -0.1494140625,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.231,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.5546875,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5361963190184049,
|
||
|
|
"grad_norm": 29.931742040479904,
|
||
|
|
"learning_rate": 8.297520661157024e-07,
|
||
|
|
"logits/chosen": -0.19140625,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.2471,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.9375,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5374233128834356,
|
||
|
|
"grad_norm": 40.51322990584841,
|
||
|
|
"learning_rate": 8.293388429752066e-07,
|
||
|
|
"logits/chosen": -0.10546875,
|
||
|
|
"logits/rejected": -0.2060546875,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.4088,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.58984375,
|
||
|
|
"rewards/margins": 5.5,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5386503067484663,
|
||
|
|
"grad_norm": 36.14747773969047,
|
||
|
|
"learning_rate": 8.289256198347106e-07,
|
||
|
|
"logits/chosen": -0.057373046875,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2942,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.59765625,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 439
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5398773006134969,
|
||
|
|
"grad_norm": 47.70699134185717,
|
||
|
|
"learning_rate": 8.285123966942148e-07,
|
||
|
|
"logits/chosen": -0.099609375,
|
||
|
|
"logits/rejected": -0.1982421875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.4004,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.169921875,
|
||
|
|
"rewards/margins": 5.34375,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5411042944785276,
|
||
|
|
"grad_norm": 37.12029582406457,
|
||
|
|
"learning_rate": 8.280991735537189e-07,
|
||
|
|
"logits/chosen": -0.03271484375,
|
||
|
|
"logits/rejected": -0.14453125,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.3617,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.4609375,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5423312883435583,
|
||
|
|
"grad_norm": 36.833773823132546,
|
||
|
|
"learning_rate": 8.276859504132231e-07,
|
||
|
|
"logits/chosen": -0.18359375,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3691,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.99609375,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5435582822085889,
|
||
|
|
"grad_norm": 38.764256019654724,
|
||
|
|
"learning_rate": 8.272727272727272e-07,
|
||
|
|
"logits/chosen": -0.134765625,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3263,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.265625,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5447852760736196,
|
||
|
|
"grad_norm": 36.57479438910318,
|
||
|
|
"learning_rate": 8.268595041322314e-07,
|
||
|
|
"logits/chosen": -0.1748046875,
|
||
|
|
"logits/rejected": -0.412109375,
|
||
|
|
"logps/chosen": -448.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.2696,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.9453125,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5460122699386503,
|
||
|
|
"grad_norm": 37.423432013581284,
|
||
|
|
"learning_rate": 8.264462809917355e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.4224,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.234375,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5472392638036809,
|
||
|
|
"grad_norm": 32.85951175366752,
|
||
|
|
"learning_rate": 8.260330578512397e-07,
|
||
|
|
"logits/chosen": -0.1552734375,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.261,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.34375,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5484662576687117,
|
||
|
|
"grad_norm": 30.593381836628694,
|
||
|
|
"learning_rate": 8.256198347107438e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.35546875,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2775,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.796875,
|
||
|
|
"rewards/margins": 7.5625,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5496932515337424,
|
||
|
|
"grad_norm": 37.96919240385222,
|
||
|
|
"learning_rate": 8.252066115702479e-07,
|
||
|
|
"logits/chosen": -0.076171875,
|
||
|
|
"logits/rejected": -0.2236328125,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3504,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.75390625,
|
||
|
|
"rewards/margins": 5.4375,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.550920245398773,
|
||
|
|
"grad_norm": 32.94147744887095,
|
||
|
|
"learning_rate": 8.24793388429752e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.8359375,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5521472392638037,
|
||
|
|
"grad_norm": 32.316931210250736,
|
||
|
|
"learning_rate": 8.243801652892562e-07,
|
||
|
|
"logits/chosen": -0.2236328125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.2695,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.1484375,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5533742331288344,
|
||
|
|
"grad_norm": 32.65287065671748,
|
||
|
|
"learning_rate": 8.239669421487603e-07,
|
||
|
|
"logits/chosen": -0.09033203125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -286.0,
|
||
|
|
"loss": 0.3194,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.0703125,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -4.9375,
|
||
|
|
"step": 451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.554601226993865,
|
||
|
|
"grad_norm": 34.48235820988141,
|
||
|
|
"learning_rate": 8.235537190082645e-07,
|
||
|
|
"logits/chosen": -0.1533203125,
|
||
|
|
"logits/rejected": -0.40234375,
|
||
|
|
"logps/chosen": -456.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.259,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.34375,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5558282208588957,
|
||
|
|
"grad_norm": 28.400538678800142,
|
||
|
|
"learning_rate": 8.231404958677686e-07,
|
||
|
|
"logits/chosen": -0.234375,
|
||
|
|
"logits/rejected": -0.453125,
|
||
|
|
"logps/chosen": -436.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.2402,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.515625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 453
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5570552147239264,
|
||
|
|
"grad_norm": 33.9480760163604,
|
||
|
|
"learning_rate": 8.227272727272727e-07,
|
||
|
|
"logits/chosen": -0.078125,
|
||
|
|
"logits/rejected": -0.154296875,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.285,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.69140625,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 454
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.558282208588957,
|
||
|
|
"grad_norm": 30.854904595369923,
|
||
|
|
"learning_rate": 8.223140495867769e-07,
|
||
|
|
"logits/chosen": -0.193359375,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2265,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.921875,
|
||
|
|
"rewards/margins": 7.09375,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5595092024539877,
|
||
|
|
"grad_norm": 39.11103336894796,
|
||
|
|
"learning_rate": 8.21900826446281e-07,
|
||
|
|
"logits/chosen": -0.083984375,
|
||
|
|
"logits/rejected": -0.1845703125,
|
||
|
|
"logps/chosen": -312.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.34,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5607361963190184,
|
||
|
|
"grad_norm": 34.29324527378237,
|
||
|
|
"learning_rate": 8.214876033057851e-07,
|
||
|
|
"logits/chosen": -0.1123046875,
|
||
|
|
"logits/rejected": -0.2470703125,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3053,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.0859375,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.561963190184049,
|
||
|
|
"grad_norm": 32.239574863581325,
|
||
|
|
"learning_rate": 8.210743801652891e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.30859375,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.2319,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -5.59375,
|
||
|
|
"step": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5631901840490797,
|
||
|
|
"grad_norm": 33.179929411101874,
|
||
|
|
"learning_rate": 8.206611570247933e-07,
|
||
|
|
"logits/chosen": -0.1318359375,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5644171779141104,
|
||
|
|
"grad_norm": 33.55136315479753,
|
||
|
|
"learning_rate": 8.202479338842974e-07,
|
||
|
|
"logits/chosen": -0.1337890625,
|
||
|
|
"logits/rejected": -0.23828125,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.286,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.5625,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5656441717791411,
|
||
|
|
"grad_norm": 36.94174797842278,
|
||
|
|
"learning_rate": 8.198347107438016e-07,
|
||
|
|
"logits/chosen": -0.32421875,
|
||
|
|
"logits/rejected": -0.447265625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.5703125,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5668711656441717,
|
||
|
|
"grad_norm": 39.294473057532144,
|
||
|
|
"learning_rate": 8.194214876033057e-07,
|
||
|
|
"logits/chosen": -0.0751953125,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3504,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.38671875,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5680981595092025,
|
||
|
|
"grad_norm": 36.351270220398256,
|
||
|
|
"learning_rate": 8.190082644628099e-07,
|
||
|
|
"logits/chosen": -0.1923828125,
|
||
|
|
"logits/rejected": -0.30078125,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2884,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.84375,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5693251533742332,
|
||
|
|
"grad_norm": 39.910546519558004,
|
||
|
|
"learning_rate": 8.18595041322314e-07,
|
||
|
|
"logits/chosen": -0.0634765625,
|
||
|
|
"logits/rejected": -0.240234375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.312,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.703125,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5705521472392638,
|
||
|
|
"grad_norm": 25.392205352978074,
|
||
|
|
"learning_rate": 8.181818181818182e-07,
|
||
|
|
"logits/chosen": -0.177734375,
|
||
|
|
"logits/rejected": -0.421875,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.1799,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 1.015625,
|
||
|
|
"rewards/margins": 7.5625,
|
||
|
|
"rewards/rejected": -6.53125,
|
||
|
|
"step": 465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5717791411042945,
|
||
|
|
"grad_norm": 38.85817198852564,
|
||
|
|
"learning_rate": 8.177685950413222e-07,
|
||
|
|
"logits/chosen": -0.126953125,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3154,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.44921875,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5730061349693252,
|
||
|
|
"grad_norm": 37.87083062435175,
|
||
|
|
"learning_rate": 8.173553719008264e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3243,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.125,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5742331288343558,
|
||
|
|
"grad_norm": 39.325632460733274,
|
||
|
|
"learning_rate": 8.169421487603305e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.85546875,
|
||
|
|
"rewards/margins": 7.09375,
|
||
|
|
"rewards/rejected": -6.25,
|
||
|
|
"step": 468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5754601226993865,
|
||
|
|
"grad_norm": 28.14882814674194,
|
||
|
|
"learning_rate": 8.165289256198347e-07,
|
||
|
|
"logits/chosen": -0.08740234375,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.2011,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5766871165644172,
|
||
|
|
"grad_norm": 33.58329571055999,
|
||
|
|
"learning_rate": 8.161157024793388e-07,
|
||
|
|
"logits/chosen": -0.1181640625,
|
||
|
|
"logits/rejected": -0.1943359375,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3149,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.1796875,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5779141104294478,
|
||
|
|
"grad_norm": 33.52640840566966,
|
||
|
|
"learning_rate": 8.157024793388429e-07,
|
||
|
|
"logits/chosen": 0.002197265625,
|
||
|
|
"logits/rejected": -0.1650390625,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3098,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.75,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5791411042944785,
|
||
|
|
"grad_norm": 37.764625171132664,
|
||
|
|
"learning_rate": 8.152892561983471e-07,
|
||
|
|
"logits/chosen": -0.046142578125,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3377,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.87109375,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5803680981595092,
|
||
|
|
"grad_norm": 33.16690966573419,
|
||
|
|
"learning_rate": 8.148760330578512e-07,
|
||
|
|
"logits/chosen": -0.21484375,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -384.0,
|
||
|
|
"loss": 0.3486,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.84375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5815950920245399,
|
||
|
|
"grad_norm": 37.71975865074261,
|
||
|
|
"learning_rate": 8.144628099173554e-07,
|
||
|
|
"logits/chosen": -0.171875,
|
||
|
|
"logits/rejected": -0.37890625,
|
||
|
|
"logps/chosen": -432.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.296,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.73046875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5828220858895705,
|
||
|
|
"grad_norm": 31.866582554566868,
|
||
|
|
"learning_rate": 8.140495867768594e-07,
|
||
|
|
"logits/chosen": -0.1611328125,
|
||
|
|
"logits/rejected": -0.375,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 7.78125,
|
||
|
|
"rewards/rejected": -6.1875,
|
||
|
|
"step": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5840490797546012,
|
||
|
|
"grad_norm": 32.60067460423666,
|
||
|
|
"learning_rate": 8.136363636363636e-07,
|
||
|
|
"logits/chosen": -0.09619140625,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2944,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": -0.1923828125,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5852760736196319,
|
||
|
|
"grad_norm": 34.300412867411,
|
||
|
|
"learning_rate": 8.132231404958677e-07,
|
||
|
|
"logits/chosen": -0.2109375,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2808,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.0234375,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5865030674846625,
|
||
|
|
"grad_norm": 33.50508466908412,
|
||
|
|
"learning_rate": 8.128099173553719e-07,
|
||
|
|
"logits/chosen": -0.1357421875,
|
||
|
|
"logits/rejected": -0.30859375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3184,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.87890625,
|
||
|
|
"rewards/margins": 7.3125,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5877300613496933,
|
||
|
|
"grad_norm": 32.74590499905971,
|
||
|
|
"learning_rate": 8.12396694214876e-07,
|
||
|
|
"logits/chosen": -0.1318359375,
|
||
|
|
"logits/rejected": -0.388671875,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": -0.04638671875,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -6.53125,
|
||
|
|
"step": 479
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.588957055214724,
|
||
|
|
"grad_norm": 31.25419724588422,
|
||
|
|
"learning_rate": 8.119834710743802e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.306640625,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2338,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.296875,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5901840490797546,
|
||
|
|
"grad_norm": 40.0079772160971,
|
||
|
|
"learning_rate": 8.115702479338843e-07,
|
||
|
|
"logits/chosen": -0.1748046875,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -382.0,
|
||
|
|
"loss": 0.4119,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": 0.7578125,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 481
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5914110429447853,
|
||
|
|
"grad_norm": 35.87779767449304,
|
||
|
|
"learning_rate": 8.111570247933885e-07,
|
||
|
|
"logits/chosen": -0.16796875,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2926,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.592638036809816,
|
||
|
|
"grad_norm": 34.477936142548806,
|
||
|
|
"learning_rate": 8.107438016528926e-07,
|
||
|
|
"logits/chosen": -0.0306396484375,
|
||
|
|
"logits/rejected": -0.1708984375,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3076,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.7578125,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5938650306748466,
|
||
|
|
"grad_norm": 35.06510512407324,
|
||
|
|
"learning_rate": 8.103305785123967e-07,
|
||
|
|
"logits/chosen": -0.1064453125,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3428,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.3984375,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -4.84375,
|
||
|
|
"step": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5950920245398773,
|
||
|
|
"grad_norm": 29.58257600279775,
|
||
|
|
"learning_rate": 8.099173553719008e-07,
|
||
|
|
"logits/chosen": -0.083984375,
|
||
|
|
"logits/rejected": -0.2294921875,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.276,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.90234375,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.596319018404908,
|
||
|
|
"grad_norm": 35.63248619960468,
|
||
|
|
"learning_rate": 8.09504132231405e-07,
|
||
|
|
"logits/chosen": -0.11181640625,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3049,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5975460122699386,
|
||
|
|
"grad_norm": 27.478788009307355,
|
||
|
|
"learning_rate": 8.09090909090909e-07,
|
||
|
|
"logits/chosen": -0.1806640625,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2195,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.46875,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5987730061349693,
|
||
|
|
"grad_norm": 35.65492860883581,
|
||
|
|
"learning_rate": 8.086776859504131e-07,
|
||
|
|
"logits/chosen": -0.177734375,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.25,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.5,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"grad_norm": 29.439269568854748,
|
||
|
|
"learning_rate": 8.082644628099173e-07,
|
||
|
|
"logits/chosen": -0.09619140625,
|
||
|
|
"logits/rejected": -0.341796875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.890625,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -5.5625,
|
||
|
|
"step": 489
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6012269938650306,
|
||
|
|
"grad_norm": 40.34135361403921,
|
||
|
|
"learning_rate": 8.078512396694214e-07,
|
||
|
|
"logits/chosen": -0.06103515625,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3637,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.3203125,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -5.25,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6024539877300613,
|
||
|
|
"grad_norm": 33.6425847997183,
|
||
|
|
"learning_rate": 8.074380165289256e-07,
|
||
|
|
"logits/chosen": -0.04443359375,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.234375,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.603680981595092,
|
||
|
|
"grad_norm": 37.77385653884553,
|
||
|
|
"learning_rate": 8.070247933884297e-07,
|
||
|
|
"logits/chosen": -0.130859375,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -378.0,
|
||
|
|
"loss": 0.306,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.3203125,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.09375,
|
||
|
|
"step": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6049079754601226,
|
||
|
|
"grad_norm": 43.24053528343554,
|
||
|
|
"learning_rate": 8.066115702479338e-07,
|
||
|
|
"logits/chosen": 0.0159912109375,
|
||
|
|
"logits/rejected": -0.09228515625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.4301,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.98828125,
|
||
|
|
"rewards/margins": 5.375,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6061349693251534,
|
||
|
|
"grad_norm": 32.3711711931091,
|
||
|
|
"learning_rate": 8.061983471074379e-07,
|
||
|
|
"logits/chosen": -0.076171875,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2809,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.8125,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6073619631901841,
|
||
|
|
"grad_norm": 35.66825520473123,
|
||
|
|
"learning_rate": 8.057851239669421e-07,
|
||
|
|
"logits/chosen": -0.0859375,
|
||
|
|
"logits/rejected": -0.2353515625,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3078,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.6953125,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6085889570552148,
|
||
|
|
"grad_norm": 38.66039815304769,
|
||
|
|
"learning_rate": 8.053719008264462e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.34765625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3333,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.796875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6098159509202454,
|
||
|
|
"grad_norm": 39.75597363433562,
|
||
|
|
"learning_rate": 8.049586776859504e-07,
|
||
|
|
"logits/chosen": -0.08251953125,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.4202,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.8203125,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.59375,
|
||
|
|
"step": 497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6110429447852761,
|
||
|
|
"grad_norm": 33.630068628465295,
|
||
|
|
"learning_rate": 8.045454545454545e-07,
|
||
|
|
"logits/chosen": -0.208984375,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3002,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.5703125,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6122699386503068,
|
||
|
|
"grad_norm": 35.64431198341231,
|
||
|
|
"learning_rate": 8.041322314049587e-07,
|
||
|
|
"logits/chosen": -0.07666015625,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.291,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 499
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6134969325153374,
|
||
|
|
"grad_norm": 33.74414745074971,
|
||
|
|
"learning_rate": 8.037190082644628e-07,
|
||
|
|
"logits/chosen": -0.09716796875,
|
||
|
|
"logits/rejected": -0.216796875,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2982,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.59375,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6147239263803681,
|
||
|
|
"grad_norm": 33.58342944370235,
|
||
|
|
"learning_rate": 8.033057851239669e-07,
|
||
|
|
"logits/chosen": -0.1533203125,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2988,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.7578125,
|
||
|
|
"rewards/margins": 7.53125,
|
||
|
|
"rewards/rejected": -5.78125,
|
||
|
|
"step": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6159509202453988,
|
||
|
|
"grad_norm": 32.39121968024134,
|
||
|
|
"learning_rate": 8.02892561983471e-07,
|
||
|
|
"logits/chosen": -0.12353515625,
|
||
|
|
"logits/rejected": -0.39453125,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2782,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.4765625,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -6.125,
|
||
|
|
"step": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6171779141104294,
|
||
|
|
"grad_norm": 33.84549068331807,
|
||
|
|
"learning_rate": 8.024793388429752e-07,
|
||
|
|
"logits/chosen": 0.055908203125,
|
||
|
|
"logits/rejected": -0.1455078125,
|
||
|
|
"logps/chosen": -326.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.96484375,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6184049079754601,
|
||
|
|
"grad_norm": 27.907537064187128,
|
||
|
|
"learning_rate": 8.020661157024793e-07,
|
||
|
|
"logits/chosen": -0.1826171875,
|
||
|
|
"logits/rejected": -0.35546875,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2727,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.6328125,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6196319018404908,
|
||
|
|
"grad_norm": 42.97391892879846,
|
||
|
|
"learning_rate": 8.016528925619835e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.416015625,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -396.0,
|
||
|
|
"loss": 0.3986,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.30078125,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -6.375,
|
||
|
|
"step": 505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6208588957055214,
|
||
|
|
"grad_norm": 38.43298718810276,
|
||
|
|
"learning_rate": 8.012396694214876e-07,
|
||
|
|
"logits/chosen": -0.045654296875,
|
||
|
|
"logits/rejected": -0.349609375,
|
||
|
|
"logps/chosen": -420.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.268,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.92578125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6220858895705521,
|
||
|
|
"grad_norm": 27.936273038832482,
|
||
|
|
"learning_rate": 8.008264462809917e-07,
|
||
|
|
"logits/chosen": -0.1533203125,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.1868,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 0.953125,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6233128834355828,
|
||
|
|
"grad_norm": 32.8519316556872,
|
||
|
|
"learning_rate": 8.004132231404959e-07,
|
||
|
|
"logits/chosen": 0.0201416015625,
|
||
|
|
"logits/rejected": -0.16015625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2689,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.6171875,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 508
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6245398773006134,
|
||
|
|
"grad_norm": 41.63795799495562,
|
||
|
|
"learning_rate": 8e-07,
|
||
|
|
"logits/chosen": -0.1298828125,
|
||
|
|
"logits/rejected": -0.369140625,
|
||
|
|
"logps/chosen": -466.0,
|
||
|
|
"logps/rejected": -386.0,
|
||
|
|
"loss": 0.4381,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.9140625,
|
||
|
|
"rewards/margins": 8.25,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6257668711656442,
|
||
|
|
"grad_norm": 28.89555459383984,
|
||
|
|
"learning_rate": 7.995867768595041e-07,
|
||
|
|
"logits/chosen": -0.2314453125,
|
||
|
|
"logits/rejected": -0.404296875,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2237,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.341796875,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.78125,
|
||
|
|
"step": 510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6269938650306749,
|
||
|
|
"grad_norm": 38.19688149477771,
|
||
|
|
"learning_rate": 7.991735537190082e-07,
|
||
|
|
"logits/chosen": -0.2373046875,
|
||
|
|
"logits/rejected": -0.373046875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3636,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.7421875,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6282208588957056,
|
||
|
|
"grad_norm": 41.54065794607386,
|
||
|
|
"learning_rate": 7.987603305785124e-07,
|
||
|
|
"logits/chosen": -0.051025390625,
|
||
|
|
"logits/rejected": -0.2001953125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -388.0,
|
||
|
|
"loss": 0.3803,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.265625,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6294478527607362,
|
||
|
|
"grad_norm": 35.913329772336745,
|
||
|
|
"learning_rate": 7.983471074380164e-07,
|
||
|
|
"logits/chosen": -0.052490234375,
|
||
|
|
"logits/rejected": -0.19140625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3437,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.95703125,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 513
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6306748466257669,
|
||
|
|
"grad_norm": 29.78232755098569,
|
||
|
|
"learning_rate": 7.979338842975206e-07,
|
||
|
|
"logits/chosen": -0.038818359375,
|
||
|
|
"logits/rejected": -0.197265625,
|
||
|
|
"logps/chosen": -332.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.2618,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": -0.2451171875,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6319018404907976,
|
||
|
|
"grad_norm": 36.15654336969748,
|
||
|
|
"learning_rate": 7.975206611570247e-07,
|
||
|
|
"logits/chosen": -0.056396484375,
|
||
|
|
"logits/rejected": -0.2109375,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3156,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": -0.28125,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 515
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6331288343558282,
|
||
|
|
"grad_norm": 31.44788161564659,
|
||
|
|
"learning_rate": 7.971074380165289e-07,
|
||
|
|
"logits/chosen": -0.10400390625,
|
||
|
|
"logits/rejected": -0.20703125,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": -0.064453125,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6343558282208589,
|
||
|
|
"grad_norm": 42.37255733997886,
|
||
|
|
"learning_rate": 7.96694214876033e-07,
|
||
|
|
"logits/chosen": -0.07470703125,
|
||
|
|
"logits/rejected": -0.2333984375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.4386,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.3359375,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6355828220858896,
|
||
|
|
"grad_norm": 33.14166173125557,
|
||
|
|
"learning_rate": 7.962809917355372e-07,
|
||
|
|
"logits/chosen": -0.1669921875,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2235,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.078125,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6368098159509202,
|
||
|
|
"grad_norm": 40.145641610013044,
|
||
|
|
"learning_rate": 7.958677685950412e-07,
|
||
|
|
"logits/chosen": -0.1552734375,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3229,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.4765625,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6380368098159509,
|
||
|
|
"grad_norm": 38.367748711689615,
|
||
|
|
"learning_rate": 7.954545454545454e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.419921875,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3632,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.8984375,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6392638036809816,
|
||
|
|
"grad_norm": 31.706634028120178,
|
||
|
|
"learning_rate": 7.950413223140495e-07,
|
||
|
|
"logits/chosen": 0.01434326171875,
|
||
|
|
"logits/rejected": -0.125,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2611,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.2890625,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6404907975460122,
|
||
|
|
"grad_norm": 36.111031138767444,
|
||
|
|
"learning_rate": 7.946280991735537e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3182,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.90625,
|
||
|
|
"rewards/margins": 7.5,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6417177914110429,
|
||
|
|
"grad_norm": 43.661807402482516,
|
||
|
|
"learning_rate": 7.942148760330578e-07,
|
||
|
|
"logits/chosen": -0.11181640625,
|
||
|
|
"logits/rejected": -0.259765625,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.4124,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.2392578125,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6429447852760736,
|
||
|
|
"grad_norm": 44.04386819066262,
|
||
|
|
"learning_rate": 7.938016528925619e-07,
|
||
|
|
"logits/chosen": -0.07958984375,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.3794,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.251953125,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6441717791411042,
|
||
|
|
"grad_norm": 48.408238800052914,
|
||
|
|
"learning_rate": 7.933884297520661e-07,
|
||
|
|
"logits/chosen": -0.0751953125,
|
||
|
|
"logits/rejected": -0.1337890625,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.4399,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.0625,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.645398773006135,
|
||
|
|
"grad_norm": 33.71138762956367,
|
||
|
|
"learning_rate": 7.929752066115702e-07,
|
||
|
|
"logits/chosen": -0.09814453125,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2684,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.734375,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6466257668711657,
|
||
|
|
"grad_norm": 35.72880037638548,
|
||
|
|
"learning_rate": 7.925619834710744e-07,
|
||
|
|
"logits/chosen": -0.031005859375,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3128,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.58203125,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6478527607361964,
|
||
|
|
"grad_norm": 30.922550042147858,
|
||
|
|
"learning_rate": 7.921487603305784e-07,
|
||
|
|
"logits/chosen": -0.1650390625,
|
||
|
|
"logits/rejected": -0.35546875,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3056,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.404296875,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 528
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.649079754601227,
|
||
|
|
"grad_norm": 33.98515479129212,
|
||
|
|
"learning_rate": 7.917355371900826e-07,
|
||
|
|
"logits/chosen": -0.10888671875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3684,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.609375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6503067484662577,
|
||
|
|
"grad_norm": 37.186485518201295,
|
||
|
|
"learning_rate": 7.913223140495867e-07,
|
||
|
|
"logits/chosen": -0.0703125,
|
||
|
|
"logits/rejected": -0.201171875,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3376,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.84375,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6515337423312884,
|
||
|
|
"grad_norm": 27.176074854888466,
|
||
|
|
"learning_rate": 7.909090909090909e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2583,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 0.79296875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.652760736196319,
|
||
|
|
"grad_norm": 31.28326806837223,
|
||
|
|
"learning_rate": 7.90495867768595e-07,
|
||
|
|
"logits/chosen": -0.0263671875,
|
||
|
|
"logits/rejected": -0.16796875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2852,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 0.75,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6539877300613497,
|
||
|
|
"grad_norm": 36.08815904584555,
|
||
|
|
"learning_rate": 7.900826446280992e-07,
|
||
|
|
"logits/chosen": -0.1904296875,
|
||
|
|
"logits/rejected": -0.3984375,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2781,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.53125,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6552147239263804,
|
||
|
|
"grad_norm": 31.873251516918746,
|
||
|
|
"learning_rate": 7.896694214876033e-07,
|
||
|
|
"logits/chosen": -0.051513671875,
|
||
|
|
"logits/rejected": -0.173828125,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2876,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.796875,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.656441717791411,
|
||
|
|
"grad_norm": 36.530095888975836,
|
||
|
|
"learning_rate": 7.892561983471075e-07,
|
||
|
|
"logits/chosen": -0.16015625,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.312,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.06103515625,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6576687116564417,
|
||
|
|
"grad_norm": 32.74817222483817,
|
||
|
|
"learning_rate": 7.888429752066116e-07,
|
||
|
|
"logits/chosen": -0.10986328125,
|
||
|
|
"logits/rejected": -0.2138671875,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2461,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6588957055214724,
|
||
|
|
"grad_norm": 44.707496162945326,
|
||
|
|
"learning_rate": 7.884297520661157e-07,
|
||
|
|
"logits/chosen": -0.087890625,
|
||
|
|
"logits/rejected": -0.2470703125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.4648,
|
||
|
|
"rewards/accuracies": 0.7890625,
|
||
|
|
"rewards/chosen": -0.0277099609375,
|
||
|
|
"rewards/margins": 5.125,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.660122699386503,
|
||
|
|
"grad_norm": 39.672945580644054,
|
||
|
|
"learning_rate": 7.880165289256198e-07,
|
||
|
|
"logits/chosen": -0.06689453125,
|
||
|
|
"logits/rejected": -0.2421875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.387,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.3125,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -5.25,
|
||
|
|
"step": 538
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6613496932515337,
|
||
|
|
"grad_norm": 35.46415326892615,
|
||
|
|
"learning_rate": 7.87603305785124e-07,
|
||
|
|
"logits/chosen": -0.06494140625,
|
||
|
|
"logits/rejected": -0.201171875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3394,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.21875,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6625766871165644,
|
||
|
|
"grad_norm": 37.23193708193184,
|
||
|
|
"learning_rate": 7.87190082644628e-07,
|
||
|
|
"logits/chosen": -0.04248046875,
|
||
|
|
"logits/rejected": -0.2138671875,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3496,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6638036809815951,
|
||
|
|
"grad_norm": 35.550266538453776,
|
||
|
|
"learning_rate": 7.867768595041322e-07,
|
||
|
|
"logits/chosen": -0.059814453125,
|
||
|
|
"logits/rejected": -0.1552734375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.313,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.55078125,
|
||
|
|
"rewards/margins": 5.75,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6650306748466258,
|
||
|
|
"grad_norm": 32.113839687383255,
|
||
|
|
"learning_rate": 7.863636363636363e-07,
|
||
|
|
"logits/chosen": -0.1015625,
|
||
|
|
"logits/rejected": -0.1982421875,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -4.9375,
|
||
|
|
"step": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6662576687116565,
|
||
|
|
"grad_norm": 35.51517611628603,
|
||
|
|
"learning_rate": 7.859504132231404e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -446.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.293,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.9765625,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6674846625766871,
|
||
|
|
"grad_norm": 44.19375136908175,
|
||
|
|
"learning_rate": 7.855371900826446e-07,
|
||
|
|
"logits/chosen": -0.11962890625,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.4179,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.7421875,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6687116564417178,
|
||
|
|
"grad_norm": 40.698950248901056,
|
||
|
|
"learning_rate": 7.851239669421487e-07,
|
||
|
|
"logits/chosen": -0.1708984375,
|
||
|
|
"logits/rejected": -0.341796875,
|
||
|
|
"logps/chosen": -454.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.3983,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.3984375,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -4.75,
|
||
|
|
"step": 545
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6699386503067485,
|
||
|
|
"grad_norm": 37.546979812882604,
|
||
|
|
"learning_rate": 7.847107438016528e-07,
|
||
|
|
"logits/chosen": -0.0537109375,
|
||
|
|
"logits/rejected": -0.1572265625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3606,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6711656441717792,
|
||
|
|
"grad_norm": 31.44946831622725,
|
||
|
|
"learning_rate": 7.842975206611569e-07,
|
||
|
|
"logits/chosen": -0.11279296875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2976,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.40625,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6723926380368098,
|
||
|
|
"grad_norm": 40.34732101502891,
|
||
|
|
"learning_rate": 7.838842975206611e-07,
|
||
|
|
"logits/chosen": -0.169921875,
|
||
|
|
"logits/rejected": -0.384765625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2922,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -4.71875,
|
||
|
|
"step": 548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6736196319018405,
|
||
|
|
"grad_norm": 36.61410881664837,
|
||
|
|
"learning_rate": 7.834710743801652e-07,
|
||
|
|
"logits/chosen": -0.0703125,
|
||
|
|
"logits/rejected": -0.283203125,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3224,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.0078125,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6748466257668712,
|
||
|
|
"grad_norm": 36.09415599886978,
|
||
|
|
"learning_rate": 7.830578512396694e-07,
|
||
|
|
"logits/chosen": -0.1416015625,
|
||
|
|
"logits/rejected": -0.2119140625,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3663,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.71484375,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.84375,
|
||
|
|
"step": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6760736196319018,
|
||
|
|
"grad_norm": 32.33592929698809,
|
||
|
|
"learning_rate": 7.826446280991735e-07,
|
||
|
|
"logits/chosen": -0.050048828125,
|
||
|
|
"logits/rejected": -0.171875,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.307,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.6640625,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 551
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6773006134969325,
|
||
|
|
"grad_norm": 34.998070605055936,
|
||
|
|
"learning_rate": 7.822314049586777e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3119,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.140625,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.09375,
|
||
|
|
"step": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6785276073619632,
|
||
|
|
"grad_norm": 36.85987881722463,
|
||
|
|
"learning_rate": 7.818181818181818e-07,
|
||
|
|
"logits/chosen": -0.08935546875,
|
||
|
|
"logits/rejected": -0.2353515625,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3871,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.267578125,
|
||
|
|
"rewards/margins": 5.46875,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6797546012269938,
|
||
|
|
"grad_norm": 33.055846271688495,
|
||
|
|
"learning_rate": 7.81404958677686e-07,
|
||
|
|
"logits/chosen": -0.03515625,
|
||
|
|
"logits/rejected": -0.2392578125,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2646,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.3671875,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 554
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6809815950920245,
|
||
|
|
"grad_norm": 35.60901600819588,
|
||
|
|
"learning_rate": 7.8099173553719e-07,
|
||
|
|
"logits/chosen": 0.07568359375,
|
||
|
|
"logits/rejected": -0.047119140625,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3521,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.90234375,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6822085889570552,
|
||
|
|
"grad_norm": 25.690739155417347,
|
||
|
|
"learning_rate": 7.805785123966942e-07,
|
||
|
|
"logits/chosen": -0.0576171875,
|
||
|
|
"logits/rejected": -0.2734375,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2129,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.7109375,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6834355828220859,
|
||
|
|
"grad_norm": 30.514742939928254,
|
||
|
|
"learning_rate": 7.801652892561983e-07,
|
||
|
|
"logits/chosen": -0.0654296875,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.2947,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6846625766871166,
|
||
|
|
"grad_norm": 29.65623720712839,
|
||
|
|
"learning_rate": 7.797520661157025e-07,
|
||
|
|
"logits/chosen": -0.0272216796875,
|
||
|
|
"logits/rejected": -0.1884765625,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2704,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 0.443359375,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -6.25,
|
||
|
|
"step": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6858895705521473,
|
||
|
|
"grad_norm": 37.74863410035385,
|
||
|
|
"learning_rate": 7.793388429752066e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3346,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.609375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6871165644171779,
|
||
|
|
"grad_norm": 22.447578429969834,
|
||
|
|
"learning_rate": 7.789256198347107e-07,
|
||
|
|
"logits/chosen": -0.26171875,
|
||
|
|
"logits/rejected": -0.400390625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2058,
|
||
|
|
"rewards/accuracies": 0.9375,
|
||
|
|
"rewards/chosen": 1.1015625,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6883435582822086,
|
||
|
|
"grad_norm": 28.553930750757928,
|
||
|
|
"learning_rate": 7.785123966942149e-07,
|
||
|
|
"logits/chosen": -0.11083984375,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -350.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.2121,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.0234375,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6895705521472393,
|
||
|
|
"grad_norm": 28.88112185443711,
|
||
|
|
"learning_rate": 7.78099173553719e-07,
|
||
|
|
"logits/chosen": 0.0203857421875,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2434,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 0.458984375,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69079754601227,
|
||
|
|
"grad_norm": 30.44618082475287,
|
||
|
|
"learning_rate": 7.776859504132232e-07,
|
||
|
|
"logits/chosen": -0.158203125,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.3087,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.671875,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6920245398773006,
|
||
|
|
"grad_norm": 23.53292610789856,
|
||
|
|
"learning_rate": 7.772727272727272e-07,
|
||
|
|
"logits/chosen": -0.1552734375,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2313,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 0.84375,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -5.34375,
|
||
|
|
"step": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6932515337423313,
|
||
|
|
"grad_norm": 32.54292501432025,
|
||
|
|
"learning_rate": 7.768595041322314e-07,
|
||
|
|
"logits/chosen": -0.099609375,
|
||
|
|
"logits/rejected": -0.294921875,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3924,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.7265625,
|
||
|
|
"rewards/margins": 5.6875,
|
||
|
|
"rewards/rejected": -4.96875,
|
||
|
|
"step": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.694478527607362,
|
||
|
|
"grad_norm": 32.727602830013666,
|
||
|
|
"learning_rate": 7.764462809917354e-07,
|
||
|
|
"logits/chosen": -0.11279296875,
|
||
|
|
"logits/rejected": -0.2119140625,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2518,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.625,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6957055214723926,
|
||
|
|
"grad_norm": 23.022890741215004,
|
||
|
|
"learning_rate": 7.760330578512397e-07,
|
||
|
|
"logits/chosen": -0.004852294921875,
|
||
|
|
"logits/rejected": -0.1357421875,
|
||
|
|
"logps/chosen": -314.0,
|
||
|
|
"logps/rejected": -274.0,
|
||
|
|
"loss": 0.193,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 0.58984375,
|
||
|
|
"rewards/margins": 5.625,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6969325153374233,
|
||
|
|
"grad_norm": 36.23377796534644,
|
||
|
|
"learning_rate": 7.756198347107437e-07,
|
||
|
|
"logits/chosen": -0.1435546875,
|
||
|
|
"logits/rejected": -0.2041015625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.326,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.296875,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.698159509202454,
|
||
|
|
"grad_norm": 29.465085248933423,
|
||
|
|
"learning_rate": 7.752066115702479e-07,
|
||
|
|
"logits/chosen": -0.060791015625,
|
||
|
|
"logits/rejected": -0.2060546875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -304.0,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.95703125,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 569
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6993865030674846,
|
||
|
|
"grad_norm": 24.94400322793083,
|
||
|
|
"learning_rate": 7.74793388429752e-07,
|
||
|
|
"logits/chosen": -0.2314453125,
|
||
|
|
"logits/rejected": -0.359375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.1676,
|
||
|
|
"rewards/accuracies": 0.9296875,
|
||
|
|
"rewards/chosen": 1.8046875,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -5.25,
|
||
|
|
"step": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7006134969325153,
|
||
|
|
"grad_norm": 31.9046154994296,
|
||
|
|
"learning_rate": 7.743801652892562e-07,
|
||
|
|
"logits/chosen": -0.0927734375,
|
||
|
|
"logits/rejected": -0.21875,
|
||
|
|
"logps/chosen": -324.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.2539,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.7578125,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.701840490797546,
|
||
|
|
"grad_norm": 31.18991534975819,
|
||
|
|
"learning_rate": 7.739669421487603e-07,
|
||
|
|
"logits/chosen": -0.0732421875,
|
||
|
|
"logits/rejected": -0.2236328125,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2406,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 1.96875,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -4.84375,
|
||
|
|
"step": 572
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7030674846625767,
|
||
|
|
"grad_norm": 40.389538917168906,
|
||
|
|
"learning_rate": 7.735537190082644e-07,
|
||
|
|
"logits/chosen": -0.0277099609375,
|
||
|
|
"logits/rejected": -0.1640625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.4632,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.65234375,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7042944785276074,
|
||
|
|
"grad_norm": 39.70707746238728,
|
||
|
|
"learning_rate": 7.731404958677685e-07,
|
||
|
|
"logits/chosen": -0.1611328125,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -322.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3206,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.0078125,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7055214723926381,
|
||
|
|
"grad_norm": 31.61799879254746,
|
||
|
|
"learning_rate": 7.727272727272727e-07,
|
||
|
|
"logits/chosen": -0.20703125,
|
||
|
|
"logits/rejected": -0.37109375,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.2326,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.8515625,
|
||
|
|
"rewards/margins": 7.53125,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7067484662576687,
|
||
|
|
"grad_norm": 24.65806387193105,
|
||
|
|
"learning_rate": 7.723140495867768e-07,
|
||
|
|
"logits/chosen": -0.06591796875,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.1948,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 1.7578125,
|
||
|
|
"rewards/margins": 7.65625,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7079754601226994,
|
||
|
|
"grad_norm": 36.42980548819235,
|
||
|
|
"learning_rate": 7.719008264462809e-07,
|
||
|
|
"logits/chosen": -0.00396728515625,
|
||
|
|
"logits/rejected": -0.1328125,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3269,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7092024539877301,
|
||
|
|
"grad_norm": 33.09293756143091,
|
||
|
|
"learning_rate": 7.714876033057851e-07,
|
||
|
|
"logits/chosen": -0.15625,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3248,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 7.65625,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7104294478527607,
|
||
|
|
"grad_norm": 41.891361592921854,
|
||
|
|
"learning_rate": 7.710743801652892e-07,
|
||
|
|
"logits/chosen": -0.134765625,
|
||
|
|
"logits/rejected": -0.2255859375,
|
||
|
|
"logps/chosen": -430.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.4104,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.2392578125,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -5.34375,
|
||
|
|
"step": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7116564417177914,
|
||
|
|
"grad_norm": 34.636238626793094,
|
||
|
|
"learning_rate": 7.706611570247934e-07,
|
||
|
|
"logits/chosen": -0.033447265625,
|
||
|
|
"logits/rejected": -0.2041015625,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.2367,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.6015625,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7128834355828221,
|
||
|
|
"grad_norm": 28.382108013456723,
|
||
|
|
"learning_rate": 7.702479338842975e-07,
|
||
|
|
"logits/chosen": -0.150390625,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2203,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.0859375,
|
||
|
|
"rewards/margins": 7.40625,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7141104294478527,
|
||
|
|
"grad_norm": 37.59655611342421,
|
||
|
|
"learning_rate": 7.698347107438016e-07,
|
||
|
|
"logits/chosen": 0.06982421875,
|
||
|
|
"logits/rejected": -0.1484375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.309,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.6796875,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7153374233128834,
|
||
|
|
"grad_norm": 36.220508133309536,
|
||
|
|
"learning_rate": 7.694214876033057e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.27734375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3079,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.1015625,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7165644171779141,
|
||
|
|
"grad_norm": 40.69950074705535,
|
||
|
|
"learning_rate": 7.690082644628099e-07,
|
||
|
|
"logits/chosen": -0.027099609375,
|
||
|
|
"logits/rejected": -0.09521484375,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -378.0,
|
||
|
|
"loss": 0.4144,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.09521484375,
|
||
|
|
"rewards/margins": 5.21875,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7177914110429447,
|
||
|
|
"grad_norm": 43.10567674720949,
|
||
|
|
"learning_rate": 7.68595041322314e-07,
|
||
|
|
"logits/chosen": 0.06591796875,
|
||
|
|
"logits/rejected": -0.0537109375,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.4515,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.86328125,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7190184049079754,
|
||
|
|
"grad_norm": 36.145469481313405,
|
||
|
|
"learning_rate": 7.681818181818182e-07,
|
||
|
|
"logits/chosen": -0.0673828125,
|
||
|
|
"logits/rejected": -0.17578125,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3494,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.470703125,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7202453987730061,
|
||
|
|
"grad_norm": 32.7614093132039,
|
||
|
|
"learning_rate": 7.677685950413223e-07,
|
||
|
|
"logits/chosen": -0.0966796875,
|
||
|
|
"logits/rejected": -0.20703125,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.333,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.53125,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 587
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7214723926380369,
|
||
|
|
"grad_norm": 25.043925429714594,
|
||
|
|
"learning_rate": 7.673553719008265e-07,
|
||
|
|
"logits/chosen": -0.11376953125,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2467,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.5234375,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7226993865030675,
|
||
|
|
"grad_norm": 34.35378043572271,
|
||
|
|
"learning_rate": 7.669421487603306e-07,
|
||
|
|
"logits/chosen": -0.1865234375,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3603,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.05224609375,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7239263803680982,
|
||
|
|
"grad_norm": 28.467941642279985,
|
||
|
|
"learning_rate": 7.665289256198347e-07,
|
||
|
|
"logits/chosen": -0.08154296875,
|
||
|
|
"logits/rejected": -0.197265625,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2867,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.59375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7251533742331289,
|
||
|
|
"grad_norm": 37.68214390316246,
|
||
|
|
"learning_rate": 7.661157024793388e-07,
|
||
|
|
"logits/chosen": -0.248046875,
|
||
|
|
"logits/rejected": -0.44140625,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2616,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.2109375,
|
||
|
|
"rewards/margins": 7.5625,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7263803680981595,
|
||
|
|
"grad_norm": 29.27377189590371,
|
||
|
|
"learning_rate": 7.65702479338843e-07,
|
||
|
|
"logits/chosen": -0.10693359375,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2369,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 7.78125,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7276073619631902,
|
||
|
|
"grad_norm": 40.332707371918964,
|
||
|
|
"learning_rate": 7.65289256198347e-07,
|
||
|
|
"logits/chosen": -0.0478515625,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -450.0,
|
||
|
|
"logps/rejected": -382.0,
|
||
|
|
"loss": 0.3628,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.109375,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7288343558282209,
|
||
|
|
"grad_norm": 33.289607036623224,
|
||
|
|
"learning_rate": 7.648760330578513e-07,
|
||
|
|
"logits/chosen": -0.11376953125,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3128,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.03125,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -6.90625,
|
||
|
|
"step": 594
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7300613496932515,
|
||
|
|
"grad_norm": 34.24935390172643,
|
||
|
|
"learning_rate": 7.644628099173553e-07,
|
||
|
|
"logits/chosen": -0.126953125,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -452.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.2221,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.80859375,
|
||
|
|
"rewards/margins": 7.625,
|
||
|
|
"rewards/rejected": -6.8125,
|
||
|
|
"step": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7312883435582822,
|
||
|
|
"grad_norm": 34.78409025027798,
|
||
|
|
"learning_rate": 7.640495867768594e-07,
|
||
|
|
"logits/chosen": -0.09521484375,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.3039,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.419921875,
|
||
|
|
"rewards/margins": 6.84375,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7325153374233129,
|
||
|
|
"grad_norm": 35.11934595828285,
|
||
|
|
"learning_rate": 7.636363636363636e-07,
|
||
|
|
"logits/chosen": 0.0028533935546875,
|
||
|
|
"logits/rejected": -0.171875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.3534,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.474609375,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7337423312883435,
|
||
|
|
"grad_norm": 28.132698718957986,
|
||
|
|
"learning_rate": 7.632231404958677e-07,
|
||
|
|
"logits/chosen": -0.09765625,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -446.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2167,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.34375,
|
||
|
|
"rewards/margins": 7.6875,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7349693251533742,
|
||
|
|
"grad_norm": 30.286056210482066,
|
||
|
|
"learning_rate": 7.628099173553718e-07,
|
||
|
|
"logits/chosen": -0.23046875,
|
||
|
|
"logits/rejected": -0.376953125,
|
||
|
|
"logps/chosen": -456.0,
|
||
|
|
"logps/rejected": -378.0,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.875,
|
||
|
|
"rewards/margins": 7.6875,
|
||
|
|
"rewards/rejected": -6.8125,
|
||
|
|
"step": 599
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7361963190184049,
|
||
|
|
"grad_norm": 36.213324901908116,
|
||
|
|
"learning_rate": 7.623966942148759e-07,
|
||
|
|
"logits/chosen": -0.2431640625,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.4124,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.2890625,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7374233128834355,
|
||
|
|
"grad_norm": 32.06779553662014,
|
||
|
|
"learning_rate": 7.619834710743801e-07,
|
||
|
|
"logits/chosen": -0.1142578125,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3017,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": -0.79296875,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -7.03125,
|
||
|
|
"step": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7386503067484662,
|
||
|
|
"grad_norm": 26.190653496247283,
|
||
|
|
"learning_rate": 7.615702479338842e-07,
|
||
|
|
"logits/chosen": 0.0299072265625,
|
||
|
|
"logits/rejected": -0.12890625,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2026,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 0.333984375,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -7.125,
|
||
|
|
"step": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7398773006134969,
|
||
|
|
"grad_norm": 27.143434858530537,
|
||
|
|
"learning_rate": 7.611570247933884e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.26953125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2978,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.578125,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7411042944785277,
|
||
|
|
"grad_norm": 34.631139500617245,
|
||
|
|
"learning_rate": 7.607438016528925e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.349609375,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.2591,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": -0.0869140625,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -7.125,
|
||
|
|
"step": 604
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7423312883435583,
|
||
|
|
"grad_norm": 31.74919177883954,
|
||
|
|
"learning_rate": 7.603305785123967e-07,
|
||
|
|
"logits/chosen": -0.1337890625,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.261,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": -0.0184326171875,
|
||
|
|
"rewards/margins": 7.375,
|
||
|
|
"rewards/rejected": -7.375,
|
||
|
|
"step": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.743558282208589,
|
||
|
|
"grad_norm": 36.161235720593524,
|
||
|
|
"learning_rate": 7.599173553719008e-07,
|
||
|
|
"logits/chosen": -0.2109375,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3577,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.109375,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7447852760736197,
|
||
|
|
"grad_norm": 32.105367044031574,
|
||
|
|
"learning_rate": 7.59504132231405e-07,
|
||
|
|
"logits/chosen": -0.09814453125,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.2672,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": -0.6796875,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -7.84375,
|
||
|
|
"step": 607
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7460122699386503,
|
||
|
|
"grad_norm": 33.485252705726495,
|
||
|
|
"learning_rate": 7.59090909090909e-07,
|
||
|
|
"logits/chosen": 0.01544189453125,
|
||
|
|
"logits/rejected": -0.2265625,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3304,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": -0.94921875,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -6.90625,
|
||
|
|
"step": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.747239263803681,
|
||
|
|
"grad_norm": 36.86942720098226,
|
||
|
|
"learning_rate": 7.586776859504132e-07,
|
||
|
|
"logits/chosen": -0.11376953125,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3437,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": -0.57421875,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -7.53125,
|
||
|
|
"step": 609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7484662576687117,
|
||
|
|
"grad_norm": 30.686568178530944,
|
||
|
|
"learning_rate": 7.582644628099173e-07,
|
||
|
|
"logits/chosen": -0.173828125,
|
||
|
|
"logits/rejected": -0.390625,
|
||
|
|
"logps/chosen": -454.0,
|
||
|
|
"logps/rejected": -388.0,
|
||
|
|
"loss": 0.2626,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 0.049072265625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -7.6875,
|
||
|
|
"step": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7496932515337423,
|
||
|
|
"grad_norm": 49.02806362281854,
|
||
|
|
"learning_rate": 7.578512396694215e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -442.0,
|
||
|
|
"logps/rejected": -376.0,
|
||
|
|
"loss": 0.4296,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": -1.2109375,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -7.40625,
|
||
|
|
"step": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.750920245398773,
|
||
|
|
"grad_norm": 37.48523436202966,
|
||
|
|
"learning_rate": 7.574380165289256e-07,
|
||
|
|
"logits/chosen": -0.1357421875,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.4003,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": -0.1787109375,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7521472392638037,
|
||
|
|
"grad_norm": 36.236160655246614,
|
||
|
|
"learning_rate": 7.570247933884297e-07,
|
||
|
|
"logits/chosen": -0.126953125,
|
||
|
|
"logits/rejected": -0.185546875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3752,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": -0.68359375,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -7.125,
|
||
|
|
"step": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7533742331288343,
|
||
|
|
"grad_norm": 36.75268374092951,
|
||
|
|
"learning_rate": 7.566115702479339e-07,
|
||
|
|
"logits/chosen": -0.1845703125,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3119,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": -0.1337890625,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -7.0,
|
||
|
|
"step": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.754601226993865,
|
||
|
|
"grad_norm": 33.698627786550716,
|
||
|
|
"learning_rate": 7.56198347107438e-07,
|
||
|
|
"logits/chosen": -0.1640625,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3104,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.37109375,
|
||
|
|
"rewards/margins": 7.3125,
|
||
|
|
"rewards/rejected": -6.9375,
|
||
|
|
"step": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7558282208588957,
|
||
|
|
"grad_norm": 30.559780985185174,
|
||
|
|
"learning_rate": 7.557851239669422e-07,
|
||
|
|
"logits/chosen": -0.109375,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -326.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.2703,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": -0.1494140625,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 616
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7570552147239263,
|
||
|
|
"grad_norm": 39.386718246924126,
|
||
|
|
"learning_rate": 7.553719008264462e-07,
|
||
|
|
"logits/chosen": -0.08544921875,
|
||
|
|
"logits/rejected": -0.1640625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3872,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.1982421875,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.758282208588957,
|
||
|
|
"grad_norm": 46.324705589027026,
|
||
|
|
"learning_rate": 7.549586776859504e-07,
|
||
|
|
"logits/chosen": -0.011474609375,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.4326,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.25,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7595092024539877,
|
||
|
|
"grad_norm": 30.627967348839967,
|
||
|
|
"learning_rate": 7.545454545454545e-07,
|
||
|
|
"logits/chosen": -0.037109375,
|
||
|
|
"logits/rejected": -0.240234375,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2452,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 0.9609375,
|
||
|
|
"rewards/margins": 7.3125,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7607361963190185,
|
||
|
|
"grad_norm": 44.10697642876468,
|
||
|
|
"learning_rate": 7.541322314049587e-07,
|
||
|
|
"logits/chosen": -0.041259765625,
|
||
|
|
"logits/rejected": -0.2421875,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3218,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.392578125,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7619631901840491,
|
||
|
|
"grad_norm": 32.581029083272384,
|
||
|
|
"learning_rate": 7.537190082644627e-07,
|
||
|
|
"logits/chosen": -0.162109375,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -478.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.2759,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.125,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7631901840490798,
|
||
|
|
"grad_norm": 31.98384527719372,
|
||
|
|
"learning_rate": 7.53305785123967e-07,
|
||
|
|
"logits/chosen": -0.046875,
|
||
|
|
"logits/rejected": -0.2138671875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2609,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.66796875,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7644171779141105,
|
||
|
|
"grad_norm": 36.238195387056656,
|
||
|
|
"learning_rate": 7.52892561983471e-07,
|
||
|
|
"logits/chosen": -0.076171875,
|
||
|
|
"logits/rejected": -0.251953125,
|
||
|
|
"logps/chosen": -452.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.3195,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.484375,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 623
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7656441717791411,
|
||
|
|
"grad_norm": 30.409808531447826,
|
||
|
|
"learning_rate": 7.524793388429752e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2453,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.2578125,
|
||
|
|
"rewards/margins": 7.875,
|
||
|
|
"rewards/rejected": -6.625,
|
||
|
|
"step": 624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7668711656441718,
|
||
|
|
"grad_norm": 30.338663314610898,
|
||
|
|
"learning_rate": 7.520661157024793e-07,
|
||
|
|
"logits/chosen": -0.10546875,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2278,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.1171875,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7680981595092025,
|
||
|
|
"grad_norm": 37.82309263616244,
|
||
|
|
"learning_rate": 7.516528925619834e-07,
|
||
|
|
"logits/chosen": -0.2294921875,
|
||
|
|
"logits/rejected": -0.369140625,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3596,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.5703125,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -4.9375,
|
||
|
|
"step": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7693251533742331,
|
||
|
|
"grad_norm": 32.50530579632077,
|
||
|
|
"learning_rate": 7.512396694214875e-07,
|
||
|
|
"logits/chosen": -0.0218505859375,
|
||
|
|
"logits/rejected": -0.10498046875,
|
||
|
|
"logps/chosen": -310.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3278,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": -0.1591796875,
|
||
|
|
"rewards/margins": 5.125,
|
||
|
|
"rewards/rejected": -5.28125,
|
||
|
|
"step": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7705521472392638,
|
||
|
|
"grad_norm": 37.587412820262834,
|
||
|
|
"learning_rate": 7.508264462809917e-07,
|
||
|
|
"logits/chosen": -0.047607421875,
|
||
|
|
"logits/rejected": -0.19140625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -296.0,
|
||
|
|
"loss": 0.3091,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7717791411042945,
|
||
|
|
"grad_norm": 21.71383824575051,
|
||
|
|
"learning_rate": 7.504132231404958e-07,
|
||
|
|
"logits/chosen": -0.07958984375,
|
||
|
|
"logits/rejected": -0.201171875,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -284.0,
|
||
|
|
"loss": 0.1735,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 0.8515625,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7730061349693251,
|
||
|
|
"grad_norm": 30.12378099002339,
|
||
|
|
"learning_rate": 7.5e-07,
|
||
|
|
"logits/chosen": -0.016845703125,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2675,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.734375,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.09375,
|
||
|
|
"step": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7742331288343558,
|
||
|
|
"grad_norm": 33.98419118837188,
|
||
|
|
"learning_rate": 7.495867768595041e-07,
|
||
|
|
"logits/chosen": -0.2158203125,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3587,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.53125,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -4.84375,
|
||
|
|
"step": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7754601226993865,
|
||
|
|
"grad_norm": 35.183542839414216,
|
||
|
|
"learning_rate": 7.491735537190082e-07,
|
||
|
|
"logits/chosen": -0.1240234375,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3281,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.625,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 632
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7766871165644171,
|
||
|
|
"grad_norm": 40.51431962905515,
|
||
|
|
"learning_rate": 7.487603305785124e-07,
|
||
|
|
"logits/chosen": -0.04248046875,
|
||
|
|
"logits/rejected": -0.1640625,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3706,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.796875,
|
||
|
|
"rewards/margins": 5.40625,
|
||
|
|
"rewards/rejected": -4.59375,
|
||
|
|
"step": 633
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7779141104294478,
|
||
|
|
"grad_norm": 30.98589205588177,
|
||
|
|
"learning_rate": 7.483471074380165e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.3515625,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2213,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 2.875,
|
||
|
|
"rewards/margins": 7.90625,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7791411042944786,
|
||
|
|
"grad_norm": 26.611336041491974,
|
||
|
|
"learning_rate": 7.479338842975206e-07,
|
||
|
|
"logits/chosen": -0.1455078125,
|
||
|
|
"logits/rejected": -0.208984375,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -294.0,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.078125,
|
||
|
|
"rewards/margins": 5.78125,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7803680981595092,
|
||
|
|
"grad_norm": 38.13714899235208,
|
||
|
|
"learning_rate": 7.475206611570247e-07,
|
||
|
|
"logits/chosen": -0.154296875,
|
||
|
|
"logits/rejected": -0.267578125,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.3472,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.90625,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -4.53125,
|
||
|
|
"step": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7815950920245399,
|
||
|
|
"grad_norm": 43.92799656490318,
|
||
|
|
"learning_rate": 7.471074380165289e-07,
|
||
|
|
"logits/chosen": 0.01190185546875,
|
||
|
|
"logits/rejected": -0.16796875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -298.0,
|
||
|
|
"loss": 0.4102,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.03125,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.8125,
|
||
|
|
"step": 637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7828220858895706,
|
||
|
|
"grad_norm": 31.881872275499614,
|
||
|
|
"learning_rate": 7.46694214876033e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.337890625,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2583,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.8671875,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -4.6875,
|
||
|
|
"step": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7840490797546013,
|
||
|
|
"grad_norm": 38.49434546180358,
|
||
|
|
"learning_rate": 7.462809917355372e-07,
|
||
|
|
"logits/chosen": -0.01019287109375,
|
||
|
|
"logits/rejected": -0.1845703125,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3851,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 1.78125,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.125,
|
||
|
|
"step": 639
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7852760736196319,
|
||
|
|
"grad_norm": 36.173889678321956,
|
||
|
|
"learning_rate": 7.458677685950413e-07,
|
||
|
|
"logits/chosen": -0.1728515625,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -440.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.3122,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -4.46875,
|
||
|
|
"step": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7865030674846626,
|
||
|
|
"grad_norm": 42.67341870666886,
|
||
|
|
"learning_rate": 7.454545454545455e-07,
|
||
|
|
"logits/chosen": -0.2080078125,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.4884,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.7421875,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -3.84375,
|
||
|
|
"step": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7877300613496933,
|
||
|
|
"grad_norm": 27.896406400263544,
|
||
|
|
"learning_rate": 7.450413223140496e-07,
|
||
|
|
"logits/chosen": -0.10888671875,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.268,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.5859375,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -4.34375,
|
||
|
|
"step": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7889570552147239,
|
||
|
|
"grad_norm": 36.50632780938656,
|
||
|
|
"learning_rate": 7.446280991735538e-07,
|
||
|
|
"logits/chosen": -0.251953125,
|
||
|
|
"logits/rejected": -0.41796875,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -370.0,
|
||
|
|
"loss": 0.2863,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.7578125,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -4.375,
|
||
|
|
"step": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7901840490797546,
|
||
|
|
"grad_norm": 36.90254530403193,
|
||
|
|
"learning_rate": 7.442148760330578e-07,
|
||
|
|
"logits/chosen": -0.1611328125,
|
||
|
|
"logits/rejected": -0.3046875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3362,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.6953125,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.0,
|
||
|
|
"step": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7914110429447853,
|
||
|
|
"grad_norm": 31.726545254851587,
|
||
|
|
"learning_rate": 7.43801652892562e-07,
|
||
|
|
"logits/chosen": -0.1162109375,
|
||
|
|
"logits/rejected": -0.31640625,
|
||
|
|
"logps/chosen": -328.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.84375,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7926380368098159,
|
||
|
|
"grad_norm": 32.26303661920976,
|
||
|
|
"learning_rate": 7.433884297520661e-07,
|
||
|
|
"logits/chosen": -0.051025390625,
|
||
|
|
"logits/rejected": -0.2294921875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2876,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -4.9375,
|
||
|
|
"step": 646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7938650306748466,
|
||
|
|
"grad_norm": 34.65899753363137,
|
||
|
|
"learning_rate": 7.429752066115703e-07,
|
||
|
|
"logits/chosen": -0.1201171875,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.3098,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.984375,
|
||
|
|
"rewards/margins": 7.375,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7950920245398773,
|
||
|
|
"grad_norm": 39.232231660651884,
|
||
|
|
"learning_rate": 7.425619834710743e-07,
|
||
|
|
"logits/chosen": -0.17578125,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -324.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.374,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.8359375,
|
||
|
|
"rewards/margins": 5.90625,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7963190184049079,
|
||
|
|
"grad_norm": 41.02647010732188,
|
||
|
|
"learning_rate": 7.421487603305784e-07,
|
||
|
|
"logits/chosen": -0.033447265625,
|
||
|
|
"logits/rejected": -0.12890625,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.5179,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 1.0078125,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -4.5625,
|
||
|
|
"step": 649
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7975460122699386,
|
||
|
|
"grad_norm": 43.926958102848964,
|
||
|
|
"learning_rate": 7.417355371900826e-07,
|
||
|
|
"logits/chosen": -0.05810546875,
|
||
|
|
"logits/rejected": -0.2197265625,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.4029,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7987730061349694,
|
||
|
|
"grad_norm": 37.4846322870058,
|
||
|
|
"learning_rate": 7.413223140495867e-07,
|
||
|
|
"logits/chosen": -0.203125,
|
||
|
|
"logits/rejected": -0.2734375,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3213,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.6484375,
|
||
|
|
"rewards/margins": 7.625,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 651
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"grad_norm": 39.82102192858506,
|
||
|
|
"learning_rate": 7.409090909090909e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3244,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.52734375,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8012269938650307,
|
||
|
|
"grad_norm": 39.721500208654376,
|
||
|
|
"learning_rate": 7.404958677685949e-07,
|
||
|
|
"logits/chosen": -0.0712890625,
|
||
|
|
"logits/rejected": -0.248046875,
|
||
|
|
"logps/chosen": -348.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2804,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.96484375,
|
||
|
|
"rewards/margins": 7.125,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 653
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8024539877300614,
|
||
|
|
"grad_norm": 29.27062388899099,
|
||
|
|
"learning_rate": 7.400826446280991e-07,
|
||
|
|
"logits/chosen": -0.05126953125,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2265,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.296875,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.803680981595092,
|
||
|
|
"grad_norm": 35.588927933308156,
|
||
|
|
"learning_rate": 7.396694214876032e-07,
|
||
|
|
"logits/chosen": -0.10986328125,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.953125,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8049079754601227,
|
||
|
|
"grad_norm": 38.2147503401273,
|
||
|
|
"learning_rate": 7.392561983471074e-07,
|
||
|
|
"logits/chosen": -0.0458984375,
|
||
|
|
"logits/rejected": -0.197265625,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3879,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.67578125,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 656
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8061349693251534,
|
||
|
|
"grad_norm": 31.082086756562887,
|
||
|
|
"learning_rate": 7.388429752066115e-07,
|
||
|
|
"logits/chosen": -0.1943359375,
|
||
|
|
"logits/rejected": -0.34375,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3235,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.8671875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.807361963190184,
|
||
|
|
"grad_norm": 34.397342278852506,
|
||
|
|
"learning_rate": 7.384297520661157e-07,
|
||
|
|
"logits/chosen": -0.1953125,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2744,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.72265625,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8085889570552147,
|
||
|
|
"grad_norm": 39.01877683461593,
|
||
|
|
"learning_rate": 7.380165289256198e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3594,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": -0.004241943359375,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8098159509202454,
|
||
|
|
"grad_norm": 29.796080293736573,
|
||
|
|
"learning_rate": 7.37603305785124e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2772,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.811042944785276,
|
||
|
|
"grad_norm": 43.970542180811556,
|
||
|
|
"learning_rate": 7.371900826446281e-07,
|
||
|
|
"logits/chosen": -0.06884765625,
|
||
|
|
"logits/rejected": -0.177734375,
|
||
|
|
"logps/chosen": -360.0,
|
||
|
|
"logps/rejected": -300.0,
|
||
|
|
"loss": 0.4211,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.1943359375,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8122699386503067,
|
||
|
|
"grad_norm": 27.972472797325533,
|
||
|
|
"learning_rate": 7.367768595041322e-07,
|
||
|
|
"logits/chosen": -0.07373046875,
|
||
|
|
"logits/rejected": -0.2060546875,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2248,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.328125,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8134969325153374,
|
||
|
|
"grad_norm": 31.48669247587522,
|
||
|
|
"learning_rate": 7.363636363636363e-07,
|
||
|
|
"logits/chosen": -0.0186767578125,
|
||
|
|
"logits/rejected": -0.2119140625,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3033,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.3125,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 663
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8147239263803681,
|
||
|
|
"grad_norm": 26.36831097405094,
|
||
|
|
"learning_rate": 7.359504132231405e-07,
|
||
|
|
"logits/chosen": -0.31640625,
|
||
|
|
"logits/rejected": -0.455078125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 7.84375,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8159509202453987,
|
||
|
|
"grad_norm": 41.87820109300682,
|
||
|
|
"learning_rate": 7.355371900826446e-07,
|
||
|
|
"logits/chosen": 0.000423431396484375,
|
||
|
|
"logits/rejected": -0.166015625,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -292.0,
|
||
|
|
"loss": 0.342,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.52734375,
|
||
|
|
"rewards/margins": 6.21875,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8171779141104294,
|
||
|
|
"grad_norm": 38.91570112915555,
|
||
|
|
"learning_rate": 7.351239669421487e-07,
|
||
|
|
"logits/chosen": -0.039794921875,
|
||
|
|
"logits/rejected": -0.216796875,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3886,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.265625,
|
||
|
|
"rewards/margins": 6.3125,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8184049079754602,
|
||
|
|
"grad_norm": 30.797773261094097,
|
||
|
|
"learning_rate": 7.347107438016529e-07,
|
||
|
|
"logits/chosen": -0.1181640625,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2568,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.1875,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8196319018404908,
|
||
|
|
"grad_norm": 30.324274734242014,
|
||
|
|
"learning_rate": 7.34297520661157e-07,
|
||
|
|
"logits/chosen": -0.1064453125,
|
||
|
|
"logits/rejected": -0.240234375,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2997,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -5.03125,
|
||
|
|
"step": 668
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8208588957055215,
|
||
|
|
"grad_norm": 39.407727222455065,
|
||
|
|
"learning_rate": 7.338842975206612e-07,
|
||
|
|
"logits/chosen": -0.166015625,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3349,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.0703125,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.0625,
|
||
|
|
"step": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8220858895705522,
|
||
|
|
"grad_norm": 29.030268011802306,
|
||
|
|
"learning_rate": 7.334710743801653e-07,
|
||
|
|
"logits/chosen": 0.038818359375,
|
||
|
|
"logits/rejected": -0.10595703125,
|
||
|
|
"logps/chosen": -320.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.2506,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 1.2890625,
|
||
|
|
"rewards/margins": 5.84375,
|
||
|
|
"rewards/rejected": -4.5625,
|
||
|
|
"step": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8233128834355828,
|
||
|
|
"grad_norm": 32.39302896464466,
|
||
|
|
"learning_rate": 7.330578512396694e-07,
|
||
|
|
"logits/chosen": -0.08056640625,
|
||
|
|
"logits/rejected": -0.2412109375,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -320.0,
|
||
|
|
"loss": 0.2618,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 2.046875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8245398773006135,
|
||
|
|
"grad_norm": 32.730627166939506,
|
||
|
|
"learning_rate": 7.326446280991735e-07,
|
||
|
|
"logits/chosen": -0.06396484375,
|
||
|
|
"logits/rejected": -0.330078125,
|
||
|
|
"logps/chosen": -444.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2935,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8257668711656442,
|
||
|
|
"grad_norm": 35.46284206569561,
|
||
|
|
"learning_rate": 7.322314049586777e-07,
|
||
|
|
"logits/chosen": 0.00579833984375,
|
||
|
|
"logits/rejected": -0.2255859375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2854,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.359375,
|
||
|
|
"rewards/margins": 5.53125,
|
||
|
|
"rewards/rejected": -4.15625,
|
||
|
|
"step": 673
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8269938650306748,
|
||
|
|
"grad_norm": 35.48501784668173,
|
||
|
|
"learning_rate": 7.318181818181818e-07,
|
||
|
|
"logits/chosen": 0.037841796875,
|
||
|
|
"logits/rejected": -0.12255859375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3332,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.390625,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -4.5,
|
||
|
|
"step": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8282208588957055,
|
||
|
|
"grad_norm": 34.07602348490195,
|
||
|
|
"learning_rate": 7.31404958677686e-07,
|
||
|
|
"logits/chosen": -0.037353515625,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3553,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.6796875,
|
||
|
|
"rewards/margins": 6.59375,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8294478527607362,
|
||
|
|
"grad_norm": 34.87811781040948,
|
||
|
|
"learning_rate": 7.3099173553719e-07,
|
||
|
|
"logits/chosen": -0.1416015625,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.2859,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.8359375,
|
||
|
|
"rewards/margins": 6.03125,
|
||
|
|
"rewards/rejected": -5.1875,
|
||
|
|
"step": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8306748466257668,
|
||
|
|
"grad_norm": 32.93168351329577,
|
||
|
|
"learning_rate": 7.305785123966942e-07,
|
||
|
|
"logits/chosen": -0.0242919921875,
|
||
|
|
"logits/rejected": -0.1953125,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -302.0,
|
||
|
|
"loss": 0.3128,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.546875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8319018404907975,
|
||
|
|
"grad_norm": 34.12760581259703,
|
||
|
|
"learning_rate": 7.301652892561983e-07,
|
||
|
|
"logits/chosen": -0.052001953125,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3125,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.85546875,
|
||
|
|
"rewards/margins": 6.0,
|
||
|
|
"rewards/rejected": -5.125,
|
||
|
|
"step": 678
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8331288343558282,
|
||
|
|
"grad_norm": 36.89463698994786,
|
||
|
|
"learning_rate": 7.297520661157024e-07,
|
||
|
|
"logits/chosen": 0.037841796875,
|
||
|
|
"logits/rejected": -0.0712890625,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.3702,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.859375,
|
||
|
|
"rewards/margins": 5.65625,
|
||
|
|
"rewards/rejected": -4.78125,
|
||
|
|
"step": 679
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8343558282208589,
|
||
|
|
"grad_norm": 29.256142841767886,
|
||
|
|
"learning_rate": 7.293388429752065e-07,
|
||
|
|
"logits/chosen": -0.09814453125,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -454.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2672,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.4140625,
|
||
|
|
"rewards/margins": 6.9375,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8355828220858895,
|
||
|
|
"grad_norm": 26.757154748897623,
|
||
|
|
"learning_rate": 7.289256198347107e-07,
|
||
|
|
"logits/chosen": -0.1396484375,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -344.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.94921875,
|
||
|
|
"rewards/margins": 7.21875,
|
||
|
|
"rewards/rejected": -6.25,
|
||
|
|
"step": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8368098159509203,
|
||
|
|
"grad_norm": 34.48662870784652,
|
||
|
|
"learning_rate": 7.285123966942148e-07,
|
||
|
|
"logits/chosen": -0.1298828125,
|
||
|
|
"logits/rejected": -0.2490234375,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.36,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.671875,
|
||
|
|
"rewards/margins": 6.40625,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.838036809815951,
|
||
|
|
"grad_norm": 31.6242482051113,
|
||
|
|
"learning_rate": 7.28099173553719e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.2923,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.890625,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.09375,
|
||
|
|
"step": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8392638036809816,
|
||
|
|
"grad_norm": 33.064102799803514,
|
||
|
|
"learning_rate": 7.276859504132231e-07,
|
||
|
|
"logits/chosen": 0.047119140625,
|
||
|
|
"logits/rejected": -0.0693359375,
|
||
|
|
"logps/chosen": -318.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.3284,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": -0.388671875,
|
||
|
|
"rewards/margins": 5.5,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8404907975460123,
|
||
|
|
"grad_norm": 32.22229237207361,
|
||
|
|
"learning_rate": 7.272727272727272e-07,
|
||
|
|
"logits/chosen": -0.08544921875,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.298,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.498046875,
|
||
|
|
"rewards/margins": 7.53125,
|
||
|
|
"rewards/rejected": -7.03125,
|
||
|
|
"step": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.841717791411043,
|
||
|
|
"grad_norm": 27.326122854839305,
|
||
|
|
"learning_rate": 7.268595041322314e-07,
|
||
|
|
"logits/chosen": -0.09814453125,
|
||
|
|
"logits/rejected": -0.23046875,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.266,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.40625,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -6.5,
|
||
|
|
"step": 686
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8429447852760736,
|
||
|
|
"grad_norm": 34.77177398420566,
|
||
|
|
"learning_rate": 7.264462809917355e-07,
|
||
|
|
"logits/chosen": -0.08642578125,
|
||
|
|
"logits/rejected": -0.234375,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2982,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.47265625,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8441717791411043,
|
||
|
|
"grad_norm": 32.90103137234285,
|
||
|
|
"learning_rate": 7.260330578512396e-07,
|
||
|
|
"logits/chosen": -0.130859375,
|
||
|
|
"logits/rejected": -0.291015625,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.286,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": -0.53515625,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -6.90625,
|
||
|
|
"step": 688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.845398773006135,
|
||
|
|
"grad_norm": 35.4677538273116,
|
||
|
|
"learning_rate": 7.256198347107437e-07,
|
||
|
|
"logits/chosen": -0.095703125,
|
||
|
|
"logits/rejected": -0.1884765625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3169,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.408203125,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8466257668711656,
|
||
|
|
"grad_norm": 47.82010746411854,
|
||
|
|
"learning_rate": 7.252066115702479e-07,
|
||
|
|
"logits/chosen": 0.03857421875,
|
||
|
|
"logits/rejected": -0.16015625,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.5312,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.5390625,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -6.375,
|
||
|
|
"step": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8478527607361963,
|
||
|
|
"grad_norm": 28.75628225752802,
|
||
|
|
"learning_rate": 7.24793388429752e-07,
|
||
|
|
"logits/chosen": -0.27734375,
|
||
|
|
"logits/rejected": -0.392578125,
|
||
|
|
"logps/chosen": -364.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.2626,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.40625,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -6.875,
|
||
|
|
"step": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.849079754601227,
|
||
|
|
"grad_norm": 42.35779725913325,
|
||
|
|
"learning_rate": 7.243801652892562e-07,
|
||
|
|
"logits/chosen": -0.1044921875,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2985,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.87890625,
|
||
|
|
"rewards/margins": 7.375,
|
||
|
|
"rewards/rejected": -6.5,
|
||
|
|
"step": 692
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8503067484662576,
|
||
|
|
"grad_norm": 33.77138589351699,
|
||
|
|
"learning_rate": 7.239669421487603e-07,
|
||
|
|
"logits/chosen": -0.12109375,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -450.0,
|
||
|
|
"logps/rejected": -376.0,
|
||
|
|
"loss": 0.333,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.265625,
|
||
|
|
"rewards/margins": 8.0625,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8515337423312883,
|
||
|
|
"grad_norm": 45.20119760328906,
|
||
|
|
"learning_rate": 7.235537190082645e-07,
|
||
|
|
"logits/chosen": -0.10498046875,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -342.0,
|
||
|
|
"loss": 0.4492,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": -0.2080078125,
|
||
|
|
"rewards/margins": 5.59375,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.852760736196319,
|
||
|
|
"grad_norm": 45.718560051692215,
|
||
|
|
"learning_rate": 7.231404958677686e-07,
|
||
|
|
"logits/chosen": -0.07666015625,
|
||
|
|
"logits/rejected": -0.16796875,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.4966,
|
||
|
|
"rewards/accuracies": 0.765625,
|
||
|
|
"rewards/chosen": -0.06884765625,
|
||
|
|
"rewards/margins": 4.84375,
|
||
|
|
"rewards/rejected": -4.90625,
|
||
|
|
"step": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8539877300613496,
|
||
|
|
"grad_norm": 35.81247616707383,
|
||
|
|
"learning_rate": 7.227272727272728e-07,
|
||
|
|
"logits/chosen": -0.039306640625,
|
||
|
|
"logits/rejected": -0.216796875,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.309,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.84375,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8552147239263803,
|
||
|
|
"grad_norm": 43.825897731212095,
|
||
|
|
"learning_rate": 7.223140495867768e-07,
|
||
|
|
"logits/chosen": -0.1953125,
|
||
|
|
"logits/rejected": -0.23828125,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -372.0,
|
||
|
|
"loss": 0.3419,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.46875,
|
||
|
|
"rewards/margins": 7.40625,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8564417177914111,
|
||
|
|
"grad_norm": 35.78736283192075,
|
||
|
|
"learning_rate": 7.21900826446281e-07,
|
||
|
|
"logits/chosen": -0.040771484375,
|
||
|
|
"logits/rejected": -0.171875,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3256,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.0,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 698
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8576687116564418,
|
||
|
|
"grad_norm": 36.453198015080005,
|
||
|
|
"learning_rate": 7.214876033057851e-07,
|
||
|
|
"logits/chosen": -0.10693359375,
|
||
|
|
"logits/rejected": -0.1982421875,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2859,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.357421875,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8588957055214724,
|
||
|
|
"grad_norm": 37.8768222815939,
|
||
|
|
"learning_rate": 7.210743801652893e-07,
|
||
|
|
"logits/chosen": -0.00787353515625,
|
||
|
|
"logits/rejected": -0.1123046875,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3363,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.7890625,
|
||
|
|
"rewards/margins": 5.9375,
|
||
|
|
"rewards/rejected": -5.15625,
|
||
|
|
"step": 700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8601226993865031,
|
||
|
|
"grad_norm": 39.25723778290927,
|
||
|
|
"learning_rate": 7.206611570247934e-07,
|
||
|
|
"logits/chosen": 0.0167236328125,
|
||
|
|
"logits/rejected": -0.09521484375,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.3918,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.5078125,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 701
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8613496932515338,
|
||
|
|
"grad_norm": 29.45196545763872,
|
||
|
|
"learning_rate": 7.202479338842974e-07,
|
||
|
|
"logits/chosen": -0.1337890625,
|
||
|
|
"logits/rejected": -0.427734375,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.2043,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.6328125,
|
||
|
|
"rewards/margins": 7.8125,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8625766871165644,
|
||
|
|
"grad_norm": 30.738694822628716,
|
||
|
|
"learning_rate": 7.198347107438016e-07,
|
||
|
|
"logits/chosen": -0.004669189453125,
|
||
|
|
"logits/rejected": -0.1435546875,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2985,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.9765625,
|
||
|
|
"rewards/margins": 6.375,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8638036809815951,
|
||
|
|
"grad_norm": 32.62903012160973,
|
||
|
|
"learning_rate": 7.194214876033057e-07,
|
||
|
|
"logits/chosen": -0.08154296875,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -326.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.3484,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.8125,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 704
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8650306748466258,
|
||
|
|
"grad_norm": 37.268433179228865,
|
||
|
|
"learning_rate": 7.190082644628099e-07,
|
||
|
|
"logits/chosen": -0.1982421875,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -364.0,
|
||
|
|
"loss": 0.2941,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.640625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8662576687116564,
|
||
|
|
"grad_norm": 35.93317835979168,
|
||
|
|
"learning_rate": 7.185950413223139e-07,
|
||
|
|
"logits/chosen": -0.130859375,
|
||
|
|
"logits/rejected": -0.333984375,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -306.0,
|
||
|
|
"loss": 0.3444,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.890625,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8674846625766871,
|
||
|
|
"grad_norm": 32.83261430682924,
|
||
|
|
"learning_rate": 7.181818181818181e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.3125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.27,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.640625,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8687116564417178,
|
||
|
|
"grad_norm": 31.02174274459476,
|
||
|
|
"learning_rate": 7.177685950413222e-07,
|
||
|
|
"logits/chosen": -0.125,
|
||
|
|
"logits/rejected": -0.345703125,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.2265625,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8699386503067484,
|
||
|
|
"grad_norm": 24.67697083946156,
|
||
|
|
"learning_rate": 7.173553719008264e-07,
|
||
|
|
"logits/chosen": -0.025146484375,
|
||
|
|
"logits/rejected": -0.1865234375,
|
||
|
|
"logps/chosen": -440.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2054,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.4453125,
|
||
|
|
"rewards/margins": 8.25,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8711656441717791,
|
||
|
|
"grad_norm": 31.37694718110565,
|
||
|
|
"learning_rate": 7.169421487603305e-07,
|
||
|
|
"logits/chosen": 0.0296630859375,
|
||
|
|
"logits/rejected": -0.1044921875,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.6484375,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.6875,
|
||
|
|
"step": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8723926380368098,
|
||
|
|
"grad_norm": 27.905237479227235,
|
||
|
|
"learning_rate": 7.165289256198347e-07,
|
||
|
|
"logits/chosen": 0.003997802734375,
|
||
|
|
"logits/rejected": -0.1884765625,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2746,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.015625,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8736196319018404,
|
||
|
|
"grad_norm": 38.712158048146684,
|
||
|
|
"learning_rate": 7.161157024793388e-07,
|
||
|
|
"logits/chosen": 0.005615234375,
|
||
|
|
"logits/rejected": -0.14453125,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3845,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.6171875,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -5.40625,
|
||
|
|
"step": 712
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8748466257668711,
|
||
|
|
"grad_norm": 28.765796451166402,
|
||
|
|
"learning_rate": 7.15702479338843e-07,
|
||
|
|
"logits/chosen": -0.08203125,
|
||
|
|
"logits/rejected": -0.28125,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2628,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.8046875,
|
||
|
|
"rewards/margins": 7.125,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8760736196319019,
|
||
|
|
"grad_norm": 33.59381716607132,
|
||
|
|
"learning_rate": 7.152892561983471e-07,
|
||
|
|
"logits/chosen": 0.03515625,
|
||
|
|
"logits/rejected": -0.1826171875,
|
||
|
|
"logps/chosen": -370.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.3023,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 0.91796875,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -5.75,
|
||
|
|
"step": 714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8773006134969326,
|
||
|
|
"grad_norm": 28.144956963469994,
|
||
|
|
"learning_rate": 7.148760330578512e-07,
|
||
|
|
"logits/chosen": -0.185546875,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2619,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.0703125,
|
||
|
|
"rewards/margins": 7.59375,
|
||
|
|
"rewards/rejected": -6.53125,
|
||
|
|
"step": 715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8785276073619632,
|
||
|
|
"grad_norm": 31.37035640952665,
|
||
|
|
"learning_rate": 7.144628099173553e-07,
|
||
|
|
"logits/chosen": -0.14453125,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -322.0,
|
||
|
|
"loss": 0.3094,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.1796875,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.46875,
|
||
|
|
"step": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8797546012269939,
|
||
|
|
"grad_norm": 31.006168306577173,
|
||
|
|
"learning_rate": 7.140495867768595e-07,
|
||
|
|
"logits/chosen": -0.0498046875,
|
||
|
|
"logits/rejected": -0.208984375,
|
||
|
|
"logps/chosen": -346.0,
|
||
|
|
"logps/rejected": -282.0,
|
||
|
|
"loss": 0.2815,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.64453125,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8809815950920246,
|
||
|
|
"grad_norm": 30.73153346857215,
|
||
|
|
"learning_rate": 7.136363636363636e-07,
|
||
|
|
"logits/chosen": -0.1474609375,
|
||
|
|
"logits/rejected": -0.287109375,
|
||
|
|
"logps/chosen": -400.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2255,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.7265625,
|
||
|
|
"rewards/margins": 7.6875,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 718
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8822085889570552,
|
||
|
|
"grad_norm": 38.41588372897521,
|
||
|
|
"learning_rate": 7.132231404958677e-07,
|
||
|
|
"logits/chosen": -0.1865234375,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.3139,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 2.046875,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 719
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8834355828220859,
|
||
|
|
"grad_norm": 26.827242556670615,
|
||
|
|
"learning_rate": 7.128099173553719e-07,
|
||
|
|
"logits/chosen": -0.1796875,
|
||
|
|
"logits/rejected": -0.38671875,
|
||
|
|
"logps/chosen": -410.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2223,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 2.21875,
|
||
|
|
"rewards/margins": 7.65625,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8846625766871166,
|
||
|
|
"grad_norm": 25.00213343436406,
|
||
|
|
"learning_rate": 7.12396694214876e-07,
|
||
|
|
"logits/chosen": -0.1708984375,
|
||
|
|
"logits/rejected": -0.4296875,
|
||
|
|
"logps/chosen": -412.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.2234,
|
||
|
|
"rewards/accuracies": 0.9375,
|
||
|
|
"rewards/chosen": 1.7890625,
|
||
|
|
"rewards/margins": 8.375,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8858895705521472,
|
||
|
|
"grad_norm": 28.769798448354592,
|
||
|
|
"learning_rate": 7.119834710743802e-07,
|
||
|
|
"logits/chosen": -0.150390625,
|
||
|
|
"logits/rejected": -0.244140625,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.2256,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.9609375,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8871165644171779,
|
||
|
|
"grad_norm": 33.86575550751347,
|
||
|
|
"learning_rate": 7.115702479338843e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.255859375,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3119,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.328125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 723
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8883435582822086,
|
||
|
|
"grad_norm": 31.590261922698286,
|
||
|
|
"learning_rate": 7.111570247933884e-07,
|
||
|
|
"logits/chosen": -0.1357421875,
|
||
|
|
"logits/rejected": -0.35546875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.2706,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.75,
|
||
|
|
"rewards/margins": 7.125,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8895705521472392,
|
||
|
|
"grad_norm": 33.780056784071235,
|
||
|
|
"learning_rate": 7.107438016528925e-07,
|
||
|
|
"logits/chosen": -0.002960205078125,
|
||
|
|
"logits/rejected": -0.1767578125,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3451,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.515625,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8907975460122699,
|
||
|
|
"grad_norm": 30.012891436757428,
|
||
|
|
"learning_rate": 7.103305785123967e-07,
|
||
|
|
"logits/chosen": -0.1376953125,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -290.0,
|
||
|
|
"loss": 0.2249,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.0625,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8920245398773006,
|
||
|
|
"grad_norm": 35.1770485043286,
|
||
|
|
"learning_rate": 7.099173553719008e-07,
|
||
|
|
"logits/chosen": -0.1123046875,
|
||
|
|
"logits/rejected": -0.2578125,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.5234375,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8932515337423312,
|
||
|
|
"grad_norm": 36.51114365671209,
|
||
|
|
"learning_rate": 7.09504132231405e-07,
|
||
|
|
"logits/chosen": -0.12890625,
|
||
|
|
"logits/rejected": -0.271484375,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.3237,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.96875,
|
||
|
|
"rewards/margins": 7.84375,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.894478527607362,
|
||
|
|
"grad_norm": 28.430454437088613,
|
||
|
|
"learning_rate": 7.09090909090909e-07,
|
||
|
|
"logits/chosen": -0.1435546875,
|
||
|
|
"logits/rejected": -0.283203125,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.221,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.609375,
|
||
|
|
"rewards/margins": 7.9375,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 729
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8957055214723927,
|
||
|
|
"grad_norm": 29.81908334892623,
|
||
|
|
"learning_rate": 7.086776859504132e-07,
|
||
|
|
"logits/chosen": -0.21484375,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.2212,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 2.140625,
|
||
|
|
"rewards/margins": 8.0625,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8969325153374234,
|
||
|
|
"grad_norm": 32.79564398674248,
|
||
|
|
"learning_rate": 7.082644628099173e-07,
|
||
|
|
"logits/chosen": -0.0034332275390625,
|
||
|
|
"logits/rejected": -0.125,
|
||
|
|
"logps/chosen": -430.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.2765,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.328125,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -6.125,
|
||
|
|
"step": 731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.898159509202454,
|
||
|
|
"grad_norm": 38.63951488407778,
|
||
|
|
"learning_rate": 7.078512396694215e-07,
|
||
|
|
"logits/chosen": -0.1650390625,
|
||
|
|
"logits/rejected": -0.32421875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3878,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 1.3125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -5.625,
|
||
|
|
"step": 732
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8993865030674847,
|
||
|
|
"grad_norm": 29.56794348682216,
|
||
|
|
"learning_rate": 7.074380165289255e-07,
|
||
|
|
"logits/chosen": -0.05712890625,
|
||
|
|
"logits/rejected": -0.2080078125,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.236,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.015625,
|
||
|
|
"rewards/margins": 7.125,
|
||
|
|
"rewards/rejected": -6.09375,
|
||
|
|
"step": 733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9006134969325154,
|
||
|
|
"grad_norm": 34.52211164354205,
|
||
|
|
"learning_rate": 7.070247933884297e-07,
|
||
|
|
"logits/chosen": -0.042236328125,
|
||
|
|
"logits/rejected": -0.2236328125,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.295,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.901840490797546,
|
||
|
|
"grad_norm": 30.150478068869884,
|
||
|
|
"learning_rate": 7.066115702479338e-07,
|
||
|
|
"logits/chosen": -0.146484375,
|
||
|
|
"logits/rejected": -0.375,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -374.0,
|
||
|
|
"loss": 0.1939,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.1875,
|
||
|
|
"rewards/margins": 8.25,
|
||
|
|
"rewards/rejected": -7.09375,
|
||
|
|
"step": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9030674846625767,
|
||
|
|
"grad_norm": 36.53381753968603,
|
||
|
|
"learning_rate": 7.06198347107438e-07,
|
||
|
|
"logits/chosen": 0.026123046875,
|
||
|
|
"logits/rejected": -0.09912109375,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3826,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.1220703125,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9042944785276074,
|
||
|
|
"grad_norm": 45.25449310101674,
|
||
|
|
"learning_rate": 7.057851239669421e-07,
|
||
|
|
"logits/chosen": -0.07470703125,
|
||
|
|
"logits/rejected": -0.1708984375,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.4576,
|
||
|
|
"rewards/accuracies": 0.7734375,
|
||
|
|
"rewards/chosen": 0.203125,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -6.34375,
|
||
|
|
"step": 737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.905521472392638,
|
||
|
|
"grad_norm": 29.955058399708445,
|
||
|
|
"learning_rate": 7.053719008264462e-07,
|
||
|
|
"logits/chosen": -0.142578125,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -442.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2561,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.3515625,
|
||
|
|
"rewards/margins": 7.71875,
|
||
|
|
"rewards/rejected": -6.375,
|
||
|
|
"step": 738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9067484662576687,
|
||
|
|
"grad_norm": 33.57627676756485,
|
||
|
|
"learning_rate": 7.049586776859504e-07,
|
||
|
|
"logits/chosen": -0.06103515625,
|
||
|
|
"logits/rejected": -0.1943359375,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3007,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.40234375,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9079754601226994,
|
||
|
|
"grad_norm": 33.78022485190333,
|
||
|
|
"learning_rate": 7.045454545454545e-07,
|
||
|
|
"logits/chosen": -0.13671875,
|
||
|
|
"logits/rejected": -0.29296875,
|
||
|
|
"logps/chosen": -340.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.3257,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.4609375,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.90920245398773,
|
||
|
|
"grad_norm": 33.1650640297202,
|
||
|
|
"learning_rate": 7.041322314049587e-07,
|
||
|
|
"logits/chosen": -0.12255859375,
|
||
|
|
"logits/rejected": -0.279296875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -318.0,
|
||
|
|
"loss": 0.3045,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 1.2734375,
|
||
|
|
"rewards/margins": 8.125,
|
||
|
|
"rewards/rejected": -6.84375,
|
||
|
|
"step": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9104294478527607,
|
||
|
|
"grad_norm": 30.90480025389736,
|
||
|
|
"learning_rate": 7.037190082644627e-07,
|
||
|
|
"logits/chosen": -0.1318359375,
|
||
|
|
"logits/rejected": -0.310546875,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2353,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.21875,
|
||
|
|
"rewards/margins": 8.0,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9116564417177914,
|
||
|
|
"grad_norm": 33.61329805299957,
|
||
|
|
"learning_rate": 7.033057851239669e-07,
|
||
|
|
"logits/chosen": -0.09716796875,
|
||
|
|
"logits/rejected": -0.1572265625,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.3054,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.8046875,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": -5.875,
|
||
|
|
"step": 743
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.912883435582822,
|
||
|
|
"grad_norm": 25.590518259699632,
|
||
|
|
"learning_rate": 7.02892561983471e-07,
|
||
|
|
"logits/chosen": -0.07666015625,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2006,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 1.4609375,
|
||
|
|
"rewards/margins": 8.0625,
|
||
|
|
"rewards/rejected": -6.59375,
|
||
|
|
"step": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9141104294478528,
|
||
|
|
"grad_norm": 34.43136654833863,
|
||
|
|
"learning_rate": 7.024793388429752e-07,
|
||
|
|
"logits/chosen": -0.0284423828125,
|
||
|
|
"logits/rejected": -0.125,
|
||
|
|
"logps/chosen": -342.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3174,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.53125,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9153374233128835,
|
||
|
|
"grad_norm": 40.51516468048696,
|
||
|
|
"learning_rate": 7.020661157024793e-07,
|
||
|
|
"logits/chosen": -0.0322265625,
|
||
|
|
"logits/rejected": -0.1982421875,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.3759,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.921875,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9165644171779141,
|
||
|
|
"grad_norm": 31.597769899241445,
|
||
|
|
"learning_rate": 7.016528925619835e-07,
|
||
|
|
"logits/chosen": -0.2138671875,
|
||
|
|
"logits/rejected": -0.326171875,
|
||
|
|
"logps/chosen": -338.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.2728,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.453125,
|
||
|
|
"rewards/margins": 6.34375,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9177914110429448,
|
||
|
|
"grad_norm": 39.04762321513073,
|
||
|
|
"learning_rate": 7.012396694214876e-07,
|
||
|
|
"logits/chosen": 0.00433349609375,
|
||
|
|
"logits/rejected": -0.10888671875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.416,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.01544189453125,
|
||
|
|
"rewards/margins": 5.8125,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 748
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9190184049079755,
|
||
|
|
"grad_norm": 38.417255638109644,
|
||
|
|
"learning_rate": 7.008264462809918e-07,
|
||
|
|
"logits/chosen": -0.138671875,
|
||
|
|
"logits/rejected": -0.33984375,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3824,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.9375,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -6.21875,
|
||
|
|
"step": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9202453987730062,
|
||
|
|
"grad_norm": 41.17828471732116,
|
||
|
|
"learning_rate": 7.004132231404959e-07,
|
||
|
|
"logits/chosen": -0.1044921875,
|
||
|
|
"logits/rejected": -0.2392578125,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.381,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.5859375,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -6.1875,
|
||
|
|
"step": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9214723926380368,
|
||
|
|
"grad_norm": 40.041781400868416,
|
||
|
|
"learning_rate": 7e-07,
|
||
|
|
"logits/chosen": -0.11279296875,
|
||
|
|
"logits/rejected": -0.283203125,
|
||
|
|
"logps/chosen": -424.0,
|
||
|
|
"logps/rejected": -356.0,
|
||
|
|
"loss": 0.3298,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.97265625,
|
||
|
|
"rewards/margins": 7.75,
|
||
|
|
"rewards/rejected": -6.75,
|
||
|
|
"step": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9226993865030675,
|
||
|
|
"grad_norm": 31.287358924979277,
|
||
|
|
"learning_rate": 6.995867768595041e-07,
|
||
|
|
"logits/chosen": -0.1240234375,
|
||
|
|
"logits/rejected": -0.380859375,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.2493,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.21875,
|
||
|
|
"rewards/margins": 8.0,
|
||
|
|
"rewards/rejected": -6.8125,
|
||
|
|
"step": 752
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9239263803680982,
|
||
|
|
"grad_norm": 35.190849602144475,
|
||
|
|
"learning_rate": 6.991735537190083e-07,
|
||
|
|
"logits/chosen": 0.01312255859375,
|
||
|
|
"logits/rejected": -0.12451171875,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3014,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 0.6015625,
|
||
|
|
"rewards/margins": 6.78125,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9251533742331288,
|
||
|
|
"grad_norm": 37.30128361063688,
|
||
|
|
"learning_rate": 6.987603305785124e-07,
|
||
|
|
"logits/chosen": -0.05908203125,
|
||
|
|
"logits/rejected": -0.2265625,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3875,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": 0.470703125,
|
||
|
|
"rewards/margins": 6.625,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9263803680981595,
|
||
|
|
"grad_norm": 31.219996632117976,
|
||
|
|
"learning_rate": 6.983471074380165e-07,
|
||
|
|
"logits/chosen": -0.10546875,
|
||
|
|
"logits/rejected": -0.2275390625,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2524,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.6328125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9276073619631902,
|
||
|
|
"grad_norm": 30.607420425359958,
|
||
|
|
"learning_rate": 6.979338842975207e-07,
|
||
|
|
"logits/chosen": -0.01458740234375,
|
||
|
|
"logits/rejected": -0.10546875,
|
||
|
|
"logps/chosen": -334.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2667,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 0.08984375,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 756
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9288343558282208,
|
||
|
|
"grad_norm": 25.673659823784536,
|
||
|
|
"learning_rate": 6.975206611570247e-07,
|
||
|
|
"logits/chosen": 0.0308837890625,
|
||
|
|
"logits/rejected": -0.1904296875,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2237,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": 1.4921875,
|
||
|
|
"rewards/margins": 7.5,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9300613496932515,
|
||
|
|
"grad_norm": 38.08378193724403,
|
||
|
|
"learning_rate": 6.971074380165289e-07,
|
||
|
|
"logits/chosen": -0.034423828125,
|
||
|
|
"logits/rejected": -0.203125,
|
||
|
|
"logps/chosen": -368.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.357,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.283203125,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9312883435582822,
|
||
|
|
"grad_norm": 33.845509877999,
|
||
|
|
"learning_rate": 6.96694214876033e-07,
|
||
|
|
"logits/chosen": -0.00537109375,
|
||
|
|
"logits/rejected": -0.2333984375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.2702,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.03125,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -5.78125,
|
||
|
|
"step": 759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9325153374233128,
|
||
|
|
"grad_norm": 28.0088409746571,
|
||
|
|
"learning_rate": 6.962809917355371e-07,
|
||
|
|
"logits/chosen": -0.193359375,
|
||
|
|
"logits/rejected": -0.453125,
|
||
|
|
"logps/chosen": -434.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2775,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 2.015625,
|
||
|
|
"rewards/margins": 8.3125,
|
||
|
|
"rewards/rejected": -6.3125,
|
||
|
|
"step": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9337423312883436,
|
||
|
|
"grad_norm": 34.61271932757203,
|
||
|
|
"learning_rate": 6.958677685950412e-07,
|
||
|
|
"logits/chosen": 0.0277099609375,
|
||
|
|
"logits/rejected": -0.10546875,
|
||
|
|
"logps/chosen": -374.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2985,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.2421875,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -5.90625,
|
||
|
|
"step": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9349693251533743,
|
||
|
|
"grad_norm": 36.436482750324764,
|
||
|
|
"learning_rate": 6.954545454545454e-07,
|
||
|
|
"logits/chosen": 0.04052734375,
|
||
|
|
"logits/rejected": -0.142578125,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.2972,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.98828125,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -6.28125,
|
||
|
|
"step": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9361963190184049,
|
||
|
|
"grad_norm": 31.885728071569915,
|
||
|
|
"learning_rate": 6.950413223140495e-07,
|
||
|
|
"logits/chosen": -0.0198974609375,
|
||
|
|
"logits/rejected": -0.11181640625,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.306,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 1.3828125,
|
||
|
|
"rewards/margins": 7.375,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9374233128834356,
|
||
|
|
"grad_norm": 32.14943706006715,
|
||
|
|
"learning_rate": 6.946280991735537e-07,
|
||
|
|
"logits/chosen": -0.2373046875,
|
||
|
|
"logits/rejected": -0.396484375,
|
||
|
|
"logps/chosen": -322.0,
|
||
|
|
"logps/rejected": -308.0,
|
||
|
|
"loss": 0.27,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.1904296875,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -6.625,
|
||
|
|
"step": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9386503067484663,
|
||
|
|
"grad_norm": 40.62569491353796,
|
||
|
|
"learning_rate": 6.942148760330578e-07,
|
||
|
|
"logits/chosen": 0.000492095947265625,
|
||
|
|
"logits/rejected": -0.12890625,
|
||
|
|
"logps/chosen": -396.0,
|
||
|
|
"logps/rejected": -310.0,
|
||
|
|
"loss": 0.3557,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": 0.6484375,
|
||
|
|
"rewards/margins": 6.09375,
|
||
|
|
"rewards/rejected": -5.4375,
|
||
|
|
"step": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.939877300613497,
|
||
|
|
"grad_norm": 31.76476132663633,
|
||
|
|
"learning_rate": 6.93801652892562e-07,
|
||
|
|
"logits/chosen": -0.0654296875,
|
||
|
|
"logits/rejected": -0.263671875,
|
||
|
|
"logps/chosen": -436.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3003,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.3359375,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -6.0,
|
||
|
|
"step": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9411042944785276,
|
||
|
|
"grad_norm": 35.07353126279611,
|
||
|
|
"learning_rate": 6.933884297520661e-07,
|
||
|
|
"logits/chosen": -0.0791015625,
|
||
|
|
"logits/rejected": -0.2265625,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3239,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.8984375,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9423312883435583,
|
||
|
|
"grad_norm": 34.24541590745614,
|
||
|
|
"learning_rate": 6.929752066115702e-07,
|
||
|
|
"logits/chosen": -0.10009765625,
|
||
|
|
"logits/rejected": -0.275390625,
|
||
|
|
"logps/chosen": -402.0,
|
||
|
|
"logps/rejected": -376.0,
|
||
|
|
"loss": 0.3416,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.3359375,
|
||
|
|
"rewards/margins": 7.4375,
|
||
|
|
"rewards/rejected": -6.09375,
|
||
|
|
"step": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.943558282208589,
|
||
|
|
"grad_norm": 29.496588883092915,
|
||
|
|
"learning_rate": 6.925619834710743e-07,
|
||
|
|
"logits/chosen": -0.11083984375,
|
||
|
|
"logits/rejected": -0.25390625,
|
||
|
|
"logps/chosen": -372.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.264,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.58984375,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -6.75,
|
||
|
|
"step": 769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9447852760736196,
|
||
|
|
"grad_norm": 37.888017377085696,
|
||
|
|
"learning_rate": 6.921487603305785e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.375,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.3036,
|
||
|
|
"rewards/accuracies": 0.8515625,
|
||
|
|
"rewards/chosen": 1.6953125,
|
||
|
|
"rewards/margins": 7.34375,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9460122699386503,
|
||
|
|
"grad_norm": 41.74591781088978,
|
||
|
|
"learning_rate": 6.917355371900826e-07,
|
||
|
|
"logits/chosen": -0.1181640625,
|
||
|
|
"logits/rejected": -0.1416015625,
|
||
|
|
"logps/chosen": -422.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.4706,
|
||
|
|
"rewards/accuracies": 0.765625,
|
||
|
|
"rewards/chosen": 0.1103515625,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.947239263803681,
|
||
|
|
"grad_norm": 31.969953781435493,
|
||
|
|
"learning_rate": 6.913223140495868e-07,
|
||
|
|
"logits/chosen": 0.0159912109375,
|
||
|
|
"logits/rejected": -0.2041015625,
|
||
|
|
"logps/chosen": -416.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3091,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 1.21875,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -5.8125,
|
||
|
|
"step": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9484662576687116,
|
||
|
|
"grad_norm": 33.70149897966481,
|
||
|
|
"learning_rate": 6.909090909090909e-07,
|
||
|
|
"logits/chosen": 0.021240234375,
|
||
|
|
"logits/rejected": -0.1689453125,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3331,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.1796875,
|
||
|
|
"rewards/margins": 6.125,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9496932515337423,
|
||
|
|
"grad_norm": 36.75815167674088,
|
||
|
|
"learning_rate": 6.90495867768595e-07,
|
||
|
|
"logits/chosen": 0.002044677734375,
|
||
|
|
"logits/rejected": -0.1396484375,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.3729,
|
||
|
|
"rewards/accuracies": 0.8046875,
|
||
|
|
"rewards/chosen": -0.287109375,
|
||
|
|
"rewards/margins": 4.96875,
|
||
|
|
"rewards/rejected": -5.25,
|
||
|
|
"step": 774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.950920245398773,
|
||
|
|
"grad_norm": 36.27807417125134,
|
||
|
|
"learning_rate": 6.900826446280992e-07,
|
||
|
|
"logits/chosen": 0.087890625,
|
||
|
|
"logits/rejected": -0.09814453125,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3326,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.76171875,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -5.21875,
|
||
|
|
"step": 775
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9521472392638037,
|
||
|
|
"grad_norm": 31.589189156684085,
|
||
|
|
"learning_rate": 6.896694214876033e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.298828125,
|
||
|
|
"logps/chosen": -404.0,
|
||
|
|
"logps/rejected": -354.0,
|
||
|
|
"loss": 0.3345,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.0859375,
|
||
|
|
"rewards/margins": 6.5625,
|
||
|
|
"rewards/rejected": -5.5,
|
||
|
|
"step": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9533742331288344,
|
||
|
|
"grad_norm": 34.426563462025776,
|
||
|
|
"learning_rate": 6.892561983471074e-07,
|
||
|
|
"logits/chosen": -0.09033203125,
|
||
|
|
"logits/rejected": -0.2275390625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3353,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.435546875,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -6.09375,
|
||
|
|
"step": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9546012269938651,
|
||
|
|
"grad_norm": 32.93001270220387,
|
||
|
|
"learning_rate": 6.888429752066115e-07,
|
||
|
|
"logits/chosen": -0.054443359375,
|
||
|
|
"logits/rejected": -0.26171875,
|
||
|
|
"logps/chosen": -390.0,
|
||
|
|
"logps/rejected": -316.0,
|
||
|
|
"loss": 0.2456,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.357421875,
|
||
|
|
"rewards/margins": 5.875,
|
||
|
|
"rewards/rejected": -5.53125,
|
||
|
|
"step": 778
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9558282208588957,
|
||
|
|
"grad_norm": 40.1711499351323,
|
||
|
|
"learning_rate": 6.884297520661157e-07,
|
||
|
|
"logits/chosen": -0.11328125,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -434.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.3138,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 0.93359375,
|
||
|
|
"rewards/margins": 7.15625,
|
||
|
|
"rewards/rejected": -6.25,
|
||
|
|
"step": 779
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9570552147239264,
|
||
|
|
"grad_norm": 35.21476896811508,
|
||
|
|
"learning_rate": 6.880165289256198e-07,
|
||
|
|
"logits/chosen": -0.00063323974609375,
|
||
|
|
"logits/rejected": -0.173828125,
|
||
|
|
"logps/chosen": -376.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3715,
|
||
|
|
"rewards/accuracies": 0.8203125,
|
||
|
|
"rewards/chosen": 0.06103515625,
|
||
|
|
"rewards/margins": 5.71875,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9582822085889571,
|
||
|
|
"grad_norm": 25.471392662636358,
|
||
|
|
"learning_rate": 6.87603305785124e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.357421875,
|
||
|
|
"logps/chosen": -418.0,
|
||
|
|
"logps/rejected": -346.0,
|
||
|
|
"loss": 0.2024,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 1.6015625,
|
||
|
|
"rewards/margins": 8.375,
|
||
|
|
"rewards/rejected": -6.78125,
|
||
|
|
"step": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9595092024539877,
|
||
|
|
"grad_norm": 27.786411223699663,
|
||
|
|
"learning_rate": 6.87190082644628e-07,
|
||
|
|
"logits/chosen": -0.08642578125,
|
||
|
|
"logits/rejected": -0.1904296875,
|
||
|
|
"logps/chosen": -362.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2199,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.74609375,
|
||
|
|
"rewards/margins": 6.8125,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9607361963190184,
|
||
|
|
"grad_norm": 35.92383214242273,
|
||
|
|
"learning_rate": 6.867768595041323e-07,
|
||
|
|
"logits/chosen": -0.10595703125,
|
||
|
|
"logits/rejected": -0.25,
|
||
|
|
"logps/chosen": -414.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.2667,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 1.109375,
|
||
|
|
"rewards/margins": 7.6875,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9619631901840491,
|
||
|
|
"grad_norm": 38.77277265712556,
|
||
|
|
"learning_rate": 6.863636363636363e-07,
|
||
|
|
"logits/chosen": -0.04931640625,
|
||
|
|
"logits/rejected": -0.0966796875,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.3682,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": -0.0244140625,
|
||
|
|
"rewards/margins": 6.15625,
|
||
|
|
"rewards/rejected": -6.1875,
|
||
|
|
"step": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9631901840490797,
|
||
|
|
"grad_norm": 31.90266851164007,
|
||
|
|
"learning_rate": 6.859504132231405e-07,
|
||
|
|
"logits/chosen": -0.0179443359375,
|
||
|
|
"logits/rejected": -0.185546875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -314.0,
|
||
|
|
"loss": 0.2512,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.5625,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9644171779141104,
|
||
|
|
"grad_norm": 27.113863566340694,
|
||
|
|
"learning_rate": 6.855371900826445e-07,
|
||
|
|
"logits/chosen": -0.0164794921875,
|
||
|
|
"logits/rejected": -0.224609375,
|
||
|
|
"logps/chosen": -366.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.2269,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 0.2314453125,
|
||
|
|
"rewards/margins": 7.1875,
|
||
|
|
"rewards/rejected": -6.96875,
|
||
|
|
"step": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9656441717791411,
|
||
|
|
"grad_norm": 38.48315867389771,
|
||
|
|
"learning_rate": 6.851239669421487e-07,
|
||
|
|
"logits/chosen": -0.10400390625,
|
||
|
|
"logits/rejected": -0.234375,
|
||
|
|
"logps/chosen": -408.0,
|
||
|
|
"logps/rejected": -340.0,
|
||
|
|
"loss": 0.3133,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.318359375,
|
||
|
|
"rewards/margins": 7.46875,
|
||
|
|
"rewards/rejected": -7.15625,
|
||
|
|
"step": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9668711656441717,
|
||
|
|
"grad_norm": 25.75799559419808,
|
||
|
|
"learning_rate": 6.847107438016528e-07,
|
||
|
|
"logits/chosen": -0.1259765625,
|
||
|
|
"logits/rejected": -0.302734375,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -360.0,
|
||
|
|
"loss": 0.1881,
|
||
|
|
"rewards/accuracies": 0.9140625,
|
||
|
|
"rewards/chosen": 0.859375,
|
||
|
|
"rewards/margins": 7.90625,
|
||
|
|
"rewards/rejected": -7.03125,
|
||
|
|
"step": 788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9680981595092024,
|
||
|
|
"grad_norm": 32.90135828255642,
|
||
|
|
"learning_rate": 6.84297520661157e-07,
|
||
|
|
"logits/chosen": -0.0203857421875,
|
||
|
|
"logits/rejected": -0.2060546875,
|
||
|
|
"logps/chosen": -378.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.3296,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.009033203125,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9693251533742331,
|
||
|
|
"grad_norm": 29.130488464702395,
|
||
|
|
"learning_rate": 6.838842975206611e-07,
|
||
|
|
"logits/chosen": -0.1533203125,
|
||
|
|
"logits/rejected": -0.3671875,
|
||
|
|
"logps/chosen": -384.0,
|
||
|
|
"logps/rejected": -326.0,
|
||
|
|
"loss": 0.1817,
|
||
|
|
"rewards/accuracies": 0.921875,
|
||
|
|
"rewards/chosen": 0.1689453125,
|
||
|
|
"rewards/margins": 7.03125,
|
||
|
|
"rewards/rejected": -6.875,
|
||
|
|
"step": 790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9705521472392638,
|
||
|
|
"grad_norm": 36.24126812710581,
|
||
|
|
"learning_rate": 6.834710743801652e-07,
|
||
|
|
"logits/chosen": -0.1396484375,
|
||
|
|
"logits/rejected": -0.34765625,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -338.0,
|
||
|
|
"loss": 0.3177,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.703125,
|
||
|
|
"rewards/margins": 7.90625,
|
||
|
|
"rewards/rejected": -7.21875,
|
||
|
|
"step": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9717791411042945,
|
||
|
|
"grad_norm": 34.75459181078778,
|
||
|
|
"learning_rate": 6.830578512396694e-07,
|
||
|
|
"logits/chosen": -0.12158203125,
|
||
|
|
"logits/rejected": -0.265625,
|
||
|
|
"logps/chosen": -434.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.3603,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.8359375,
|
||
|
|
"rewards/margins": 6.5,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9730061349693252,
|
||
|
|
"grad_norm": 36.36771763644911,
|
||
|
|
"learning_rate": 6.826446280991735e-07,
|
||
|
|
"logits/chosen": -0.099609375,
|
||
|
|
"logits/rejected": -0.234375,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -352.0,
|
||
|
|
"loss": 0.3444,
|
||
|
|
"rewards/accuracies": 0.875,
|
||
|
|
"rewards/chosen": 0.3984375,
|
||
|
|
"rewards/margins": 6.4375,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9742331288343559,
|
||
|
|
"grad_norm": 37.02795141772375,
|
||
|
|
"learning_rate": 6.822314049586777e-07,
|
||
|
|
"logits/chosen": -0.046875,
|
||
|
|
"logits/rejected": -0.1806640625,
|
||
|
|
"logps/chosen": -354.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.4307,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": -0.1904296875,
|
||
|
|
"rewards/margins": 6.1875,
|
||
|
|
"rewards/rejected": -6.375,
|
||
|
|
"step": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9754601226993865,
|
||
|
|
"grad_norm": 26.039807081549554,
|
||
|
|
"learning_rate": 6.818181818181817e-07,
|
||
|
|
"logits/chosen": -0.25390625,
|
||
|
|
"logits/rejected": -0.41796875,
|
||
|
|
"logps/chosen": -388.0,
|
||
|
|
"logps/rejected": -362.0,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 8.75,
|
||
|
|
"rewards/rejected": -7.59375,
|
||
|
|
"step": 795
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9766871165644172,
|
||
|
|
"grad_norm": 40.332433215846514,
|
||
|
|
"learning_rate": 6.814049586776859e-07,
|
||
|
|
"logits/chosen": -0.1416015625,
|
||
|
|
"logits/rejected": -0.349609375,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.3141,
|
||
|
|
"rewards/accuracies": 0.890625,
|
||
|
|
"rewards/chosen": -0.421875,
|
||
|
|
"rewards/margins": 6.46875,
|
||
|
|
"rewards/rejected": -6.875,
|
||
|
|
"step": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9779141104294479,
|
||
|
|
"grad_norm": 26.54735552093693,
|
||
|
|
"learning_rate": 6.8099173553719e-07,
|
||
|
|
"logits/chosen": -0.06298828125,
|
||
|
|
"logits/rejected": -0.2431640625,
|
||
|
|
"logps/chosen": -398.0,
|
||
|
|
"logps/rejected": -312.0,
|
||
|
|
"loss": 0.2381,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.52734375,
|
||
|
|
"rewards/margins": 6.53125,
|
||
|
|
"rewards/rejected": -6.03125,
|
||
|
|
"step": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9791411042944785,
|
||
|
|
"grad_norm": 35.069027268070755,
|
||
|
|
"learning_rate": 6.805785123966942e-07,
|
||
|
|
"logits/chosen": -0.10498046875,
|
||
|
|
"logits/rejected": -0.2890625,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.3053,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 0.53125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.4375,
|
||
|
|
"step": 798
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9803680981595092,
|
||
|
|
"grad_norm": 37.27286246589222,
|
||
|
|
"learning_rate": 6.801652892561983e-07,
|
||
|
|
"logits/chosen": -0.1513671875,
|
||
|
|
"logits/rejected": -0.314453125,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -348.0,
|
||
|
|
"loss": 0.3271,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.90234375,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.0625,
|
||
|
|
"step": 799
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9815950920245399,
|
||
|
|
"grad_norm": 36.01628957518016,
|
||
|
|
"learning_rate": 6.797520661157025e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.3203125,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -332.0,
|
||
|
|
"loss": 0.2853,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.48828125,
|
||
|
|
"rewards/margins": 6.65625,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9828220858895705,
|
||
|
|
"grad_norm": 39.65535558250296,
|
||
|
|
"learning_rate": 6.793388429752066e-07,
|
||
|
|
"logits/chosen": -0.1279296875,
|
||
|
|
"logits/rejected": -0.28515625,
|
||
|
|
"logps/chosen": -430.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.4193,
|
||
|
|
"rewards/accuracies": 0.8125,
|
||
|
|
"rewards/chosen": 0.6328125,
|
||
|
|
"rewards/margins": 5.96875,
|
||
|
|
"rewards/rejected": -5.3125,
|
||
|
|
"step": 801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9840490797546012,
|
||
|
|
"grad_norm": 41.69927659280073,
|
||
|
|
"learning_rate": 6.789256198347108e-07,
|
||
|
|
"logits/chosen": -0.177734375,
|
||
|
|
"logits/rejected": -0.322265625,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.426,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 0.400390625,
|
||
|
|
"rewards/margins": 6.25,
|
||
|
|
"rewards/rejected": -5.84375,
|
||
|
|
"step": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9852760736196319,
|
||
|
|
"grad_norm": 31.096547435657403,
|
||
|
|
"learning_rate": 6.785123966942149e-07,
|
||
|
|
"logits/chosen": -0.244140625,
|
||
|
|
"logits/rejected": -0.404296875,
|
||
|
|
"logps/chosen": -392.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.3024,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.1171875,
|
||
|
|
"rewards/margins": 6.875,
|
||
|
|
"rewards/rejected": -5.78125,
|
||
|
|
"step": 803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9865030674846625,
|
||
|
|
"grad_norm": 32.05019392351643,
|
||
|
|
"learning_rate": 6.78099173553719e-07,
|
||
|
|
"logits/chosen": -0.06884765625,
|
||
|
|
"logits/rejected": -0.1826171875,
|
||
|
|
"logps/chosen": -352.0,
|
||
|
|
"logps/rejected": -330.0,
|
||
|
|
"loss": 0.3229,
|
||
|
|
"rewards/accuracies": 0.859375,
|
||
|
|
"rewards/chosen": 0.26953125,
|
||
|
|
"rewards/margins": 6.0625,
|
||
|
|
"rewards/rejected": -5.78125,
|
||
|
|
"step": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9877300613496932,
|
||
|
|
"grad_norm": 38.743266159886026,
|
||
|
|
"learning_rate": 6.776859504132231e-07,
|
||
|
|
"logits/chosen": -0.1767578125,
|
||
|
|
"logits/rejected": -0.318359375,
|
||
|
|
"logps/chosen": -426.0,
|
||
|
|
"logps/rejected": -380.0,
|
||
|
|
"loss": 0.439,
|
||
|
|
"rewards/accuracies": 0.828125,
|
||
|
|
"rewards/chosen": 1.1328125,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -5.9375,
|
||
|
|
"step": 805
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9889570552147239,
|
||
|
|
"grad_norm": 39.2535896844492,
|
||
|
|
"learning_rate": 6.772727272727273e-07,
|
||
|
|
"logits/chosen": 0.04345703125,
|
||
|
|
"logits/rejected": -0.08740234375,
|
||
|
|
"logps/chosen": -406.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.3836,
|
||
|
|
"rewards/accuracies": 0.78125,
|
||
|
|
"rewards/chosen": 0.8984375,
|
||
|
|
"rewards/margins": 6.28125,
|
||
|
|
"rewards/rejected": -5.375,
|
||
|
|
"step": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9901840490797545,
|
||
|
|
"grad_norm": 34.46845540454637,
|
||
|
|
"learning_rate": 6.768595041322314e-07,
|
||
|
|
"logits/chosen": -0.048583984375,
|
||
|
|
"logits/rejected": -0.2080078125,
|
||
|
|
"logps/chosen": -394.0,
|
||
|
|
"logps/rejected": -334.0,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"rewards/accuracies": 0.8828125,
|
||
|
|
"rewards/chosen": 1.15625,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": -6.125,
|
||
|
|
"step": 807
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9914110429447853,
|
||
|
|
"grad_norm": 30.33905056557921,
|
||
|
|
"learning_rate": 6.764462809917355e-07,
|
||
|
|
"logits/chosen": -0.189453125,
|
||
|
|
"logits/rejected": -0.20703125,
|
||
|
|
"logps/chosen": -356.0,
|
||
|
|
"logps/rejected": -336.0,
|
||
|
|
"loss": 0.3216,
|
||
|
|
"rewards/accuracies": 0.8671875,
|
||
|
|
"rewards/chosen": 0.255859375,
|
||
|
|
"rewards/margins": 6.71875,
|
||
|
|
"rewards/rejected": -6.46875,
|
||
|
|
"step": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.992638036809816,
|
||
|
|
"grad_norm": 33.590212816681365,
|
||
|
|
"learning_rate": 6.760330578512397e-07,
|
||
|
|
"logits/chosen": -0.109375,
|
||
|
|
"logits/rejected": -0.3359375,
|
||
|
|
"logps/chosen": -428.0,
|
||
|
|
"logps/rejected": -344.0,
|
||
|
|
"loss": 0.2833,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.6875,
|
||
|
|
"rewards/margins": 7.78125,
|
||
|
|
"rewards/rejected": -6.09375,
|
||
|
|
"step": 809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9938650306748467,
|
||
|
|
"grad_norm": 33.059294007862654,
|
||
|
|
"learning_rate": 6.756198347107437e-07,
|
||
|
|
"logits/chosen": -0.2109375,
|
||
|
|
"logits/rejected": -0.390625,
|
||
|
|
"logps/chosen": -436.0,
|
||
|
|
"logps/rejected": -368.0,
|
||
|
|
"loss": 0.239,
|
||
|
|
"rewards/accuracies": 0.8359375,
|
||
|
|
"rewards/chosen": 1.28125,
|
||
|
|
"rewards/margins": 7.25,
|
||
|
|
"rewards/rejected": -5.96875,
|
||
|
|
"step": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9950920245398773,
|
||
|
|
"grad_norm": 30.59632262177732,
|
||
|
|
"learning_rate": 6.75206611570248e-07,
|
||
|
|
"logits/chosen": -0.1884765625,
|
||
|
|
"logits/rejected": -0.33203125,
|
||
|
|
"logps/chosen": -386.0,
|
||
|
|
"logps/rejected": -366.0,
|
||
|
|
"loss": 0.2783,
|
||
|
|
"rewards/accuracies": 0.84375,
|
||
|
|
"rewards/chosen": 1.078125,
|
||
|
|
"rewards/margins": 6.75,
|
||
|
|
"rewards/rejected": -5.65625,
|
||
|
|
"step": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.996319018404908,
|
||
|
|
"grad_norm": 38.69919040079445,
|
||
|
|
"learning_rate": 6.74793388429752e-07,
|
||
|
|
"logits/chosen": -0.166015625,
|
||
|
|
"logits/rejected": -0.23828125,
|
||
|
|
"logps/chosen": -358.0,
|
||
|
|
"logps/rejected": -328.0,
|
||
|
|
"loss": 0.2784,
|
||
|
|
"rewards/accuracies": 0.90625,
|
||
|
|
"rewards/chosen": 0.4609375,
|
||
|
|
"rewards/margins": 7.0,
|
||
|
|
"rewards/rejected": -6.5625,
|
||
|
|
"step": 812
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9975460122699387,
|
||
|
|
"grad_norm": 27.207499335475898,
|
||
|
|
"learning_rate": 6.743801652892561e-07,
|
||
|
|
"logits/chosen": -0.08935546875,
|
||
|
|
"logits/rejected": -0.2275390625,
|
||
|
|
"logps/chosen": -336.0,
|
||
|
|
"logps/rejected": -324.0,
|
||
|
|
"loss": 0.1876,
|
||
|
|
"rewards/accuracies": 0.8984375,
|
||
|
|
"rewards/chosen": 0.88671875,
|
||
|
|
"rewards/margins": 7.0625,
|
||
|
|
"rewards/rejected": -6.15625,
|
||
|
|
"step": 813
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9987730061349693,
|
||
|
|
"grad_norm": 35.79902409901904,
|
||
|
|
"learning_rate": 6.739669421487602e-07,
|
||
|
|
"logits/chosen": -0.0634765625,
|
||
|
|
"logits/rejected": -0.142578125,
|
||
|
|
"logps/chosen": -382.0,
|
||
|
|
"logps/rejected": -358.0,
|
||
|
|
"loss": 0.3407,
|
||
|
|
"rewards/accuracies": 0.796875,
|
||
|
|
"rewards/chosen": -0.166015625,
|
||
|
|
"rewards/margins": 5.5625,
|
||
|
|
"rewards/rejected": -5.71875,
|
||
|
|
"step": 814
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 42.42708489033231,
|
||
|
|
"learning_rate": 6.735537190082644e-07,
|
||
|
|
"logits/chosen": 0.006988525390625,
|
||
|
|
"logits/rejected": -0.18359375,
|
||
|
|
"logps/chosen": -380.0,
|
||
|
|
"logps/rejected": -350.0,
|
||
|
|
"loss": 0.2248,
|
||
|
|
"rewards/accuracies": 0.8823530077934265,
|
||
|
|
"rewards/chosen": 0.5703125,
|
||
|
|
"rewards/margins": 6.96875,
|
||
|
|
"rewards/rejected": -6.40625,
|
||
|
|
"step": 815
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 1,
|
||
|
|
"max_steps": 2445,
|
||
|
|
"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": false
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_batch_size": 4,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|