3239 lines
108 KiB
JSON
3239 lines
108 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 2130,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.004694835680751174,
|
||
|
|
"grad_norm": 349.5201090470861,
|
||
|
|
"learning_rate": 2.1126760563380282e-08,
|
||
|
|
"logits/chosen": -4.224999904632568,
|
||
|
|
"logits/rejected": -4.053124904632568,
|
||
|
|
"logps/chosen": -546.2000122070312,
|
||
|
|
"logps/rejected": -632.0,
|
||
|
|
"loss": 2.2046,
|
||
|
|
"rewards/accuracies": 0.16718749701976776,
|
||
|
|
"rewards/chosen": -9.765625145519152e-05,
|
||
|
|
"rewards/margins": -0.09228515625,
|
||
|
|
"rewards/rejected": 0.09267578274011612,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.009389671361502348,
|
||
|
|
"grad_norm": 362.6757163538109,
|
||
|
|
"learning_rate": 4.460093896713615e-08,
|
||
|
|
"logits/chosen": -4.125,
|
||
|
|
"logits/rejected": -4.064062595367432,
|
||
|
|
"logps/chosen": -575.5999755859375,
|
||
|
|
"logps/rejected": -634.0,
|
||
|
|
"loss": 3.0842,
|
||
|
|
"rewards/accuracies": 0.24375000596046448,
|
||
|
|
"rewards/chosen": -0.13261719048023224,
|
||
|
|
"rewards/margins": -0.29277342557907104,
|
||
|
|
"rewards/rejected": 0.16044922173023224,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.014084507042253521,
|
||
|
|
"grad_norm": 292.98498991781304,
|
||
|
|
"learning_rate": 6.807511737089202e-08,
|
||
|
|
"logits/chosen": -4.129687309265137,
|
||
|
|
"logits/rejected": -3.9625000953674316,
|
||
|
|
"logps/chosen": -608.4000244140625,
|
||
|
|
"logps/rejected": -700.0,
|
||
|
|
"loss": 2.7602,
|
||
|
|
"rewards/accuracies": 0.25703126192092896,
|
||
|
|
"rewards/chosen": 0.4609375,
|
||
|
|
"rewards/margins": 0.20703125,
|
||
|
|
"rewards/rejected": 0.25341796875,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.018779342723004695,
|
||
|
|
"grad_norm": 335.938627794309,
|
||
|
|
"learning_rate": 9.154929577464789e-08,
|
||
|
|
"logits/chosen": -4.167187690734863,
|
||
|
|
"logits/rejected": -4.0390625,
|
||
|
|
"logps/chosen": -599.2000122070312,
|
||
|
|
"logps/rejected": -716.7999877929688,
|
||
|
|
"loss": 3.2419,
|
||
|
|
"rewards/accuracies": 0.25390625,
|
||
|
|
"rewards/chosen": -0.49101561307907104,
|
||
|
|
"rewards/margins": -0.22670897841453552,
|
||
|
|
"rewards/rejected": -0.2652343809604645,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.023474178403755867,
|
||
|
|
"grad_norm": 383.05083847560314,
|
||
|
|
"learning_rate": 1.1502347417840374e-07,
|
||
|
|
"logits/chosen": -4.123437404632568,
|
||
|
|
"logits/rejected": -4.067187309265137,
|
||
|
|
"logps/chosen": -576.0,
|
||
|
|
"logps/rejected": -640.7999877929688,
|
||
|
|
"loss": 3.2698,
|
||
|
|
"rewards/accuracies": 0.26484376192092896,
|
||
|
|
"rewards/chosen": -0.32695311307907104,
|
||
|
|
"rewards/margins": -0.0634765625,
|
||
|
|
"rewards/rejected": -0.263671875,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.028169014084507043,
|
||
|
|
"grad_norm": 349.1965000442969,
|
||
|
|
"learning_rate": 1.384976525821596e-07,
|
||
|
|
"logits/chosen": -4.199999809265137,
|
||
|
|
"logits/rejected": -4.032812595367432,
|
||
|
|
"logps/chosen": -590.7999877929688,
|
||
|
|
"logps/rejected": -697.5999755859375,
|
||
|
|
"loss": 2.7531,
|
||
|
|
"rewards/accuracies": 0.23593750596046448,
|
||
|
|
"rewards/chosen": -0.3087402284145355,
|
||
|
|
"rewards/margins": -0.02490234375,
|
||
|
|
"rewards/rejected": -0.283935546875,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03286384976525822,
|
||
|
|
"grad_norm": 433.25242575117915,
|
||
|
|
"learning_rate": 1.619718309859155e-07,
|
||
|
|
"logits/chosen": -4.092187404632568,
|
||
|
|
"logits/rejected": -4.006249904632568,
|
||
|
|
"logps/chosen": -566.2000122070312,
|
||
|
|
"logps/rejected": -628.0,
|
||
|
|
"loss": 2.7881,
|
||
|
|
"rewards/accuracies": 0.22578124701976776,
|
||
|
|
"rewards/chosen": 0.04750976711511612,
|
||
|
|
"rewards/margins": -0.10000000149011612,
|
||
|
|
"rewards/rejected": 0.1484375,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03755868544600939,
|
||
|
|
"grad_norm": 323.0287405163464,
|
||
|
|
"learning_rate": 1.8544600938967138e-07,
|
||
|
|
"logits/chosen": -4.199999809265137,
|
||
|
|
"logits/rejected": -4.112500190734863,
|
||
|
|
"logps/chosen": -578.0,
|
||
|
|
"logps/rejected": -670.0,
|
||
|
|
"loss": 3.2578,
|
||
|
|
"rewards/accuracies": 0.23984375596046448,
|
||
|
|
"rewards/chosen": -0.2548828125,
|
||
|
|
"rewards/margins": -0.34467774629592896,
|
||
|
|
"rewards/rejected": 0.08955077826976776,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04225352112676056,
|
||
|
|
"grad_norm": 606.9294632383828,
|
||
|
|
"learning_rate": 2.089201877934272e-07,
|
||
|
|
"logits/chosen": -4.221875190734863,
|
||
|
|
"logits/rejected": -4.129687309265137,
|
||
|
|
"logps/chosen": -552.2000122070312,
|
||
|
|
"logps/rejected": -619.0,
|
||
|
|
"loss": 2.8603,
|
||
|
|
"rewards/accuracies": 0.24765625596046448,
|
||
|
|
"rewards/chosen": -0.15292969346046448,
|
||
|
|
"rewards/margins": 0.014575195498764515,
|
||
|
|
"rewards/rejected": -0.16718749701976776,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.046948356807511735,
|
||
|
|
"grad_norm": 373.77763598016054,
|
||
|
|
"learning_rate": 2.323943661971831e-07,
|
||
|
|
"logits/chosen": -4.137499809265137,
|
||
|
|
"logits/rejected": -4.084374904632568,
|
||
|
|
"logps/chosen": -595.2000122070312,
|
||
|
|
"logps/rejected": -623.2000122070312,
|
||
|
|
"loss": 2.7484,
|
||
|
|
"rewards/accuracies": 0.25078123807907104,
|
||
|
|
"rewards/chosen": 0.10239257663488388,
|
||
|
|
"rewards/margins": 0.43437498807907104,
|
||
|
|
"rewards/rejected": -0.33222657442092896,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.051643192488262914,
|
||
|
|
"grad_norm": 379.39664695274985,
|
||
|
|
"learning_rate": 2.5586854460093895e-07,
|
||
|
|
"logits/chosen": -4.190625190734863,
|
||
|
|
"logits/rejected": -4.050000190734863,
|
||
|
|
"logps/chosen": -575.2000122070312,
|
||
|
|
"logps/rejected": -670.4000244140625,
|
||
|
|
"loss": 3.2888,
|
||
|
|
"rewards/accuracies": 0.24843749403953552,
|
||
|
|
"rewards/chosen": 0.02099609375,
|
||
|
|
"rewards/margins": 0.03076171875,
|
||
|
|
"rewards/rejected": -0.00927734375,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.056338028169014086,
|
||
|
|
"grad_norm": 332.1539670626537,
|
||
|
|
"learning_rate": 2.7934272300469483e-07,
|
||
|
|
"logits/chosen": -4.198437690734863,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -586.4000244140625,
|
||
|
|
"logps/rejected": -657.2000122070312,
|
||
|
|
"loss": 2.9951,
|
||
|
|
"rewards/accuracies": 0.2515625059604645,
|
||
|
|
"rewards/chosen": 0.09458007663488388,
|
||
|
|
"rewards/margins": -0.36796873807907104,
|
||
|
|
"rewards/rejected": 0.4627929627895355,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06103286384976526,
|
||
|
|
"grad_norm": 358.02210765226994,
|
||
|
|
"learning_rate": 3.0281690140845066e-07,
|
||
|
|
"logits/chosen": -4.1640625,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -593.0,
|
||
|
|
"logps/rejected": -676.0,
|
||
|
|
"loss": 2.2491,
|
||
|
|
"rewards/accuracies": 0.26171875,
|
||
|
|
"rewards/chosen": 0.8482910394668579,
|
||
|
|
"rewards/margins": 0.8780273199081421,
|
||
|
|
"rewards/rejected": -0.0302734375,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06572769953051644,
|
||
|
|
"grad_norm": 426.7438636061182,
|
||
|
|
"learning_rate": 3.2629107981220654e-07,
|
||
|
|
"logits/chosen": -4.128125190734863,
|
||
|
|
"logits/rejected": -4.034375190734863,
|
||
|
|
"logps/chosen": -581.4000244140625,
|
||
|
|
"logps/rejected": -691.5999755859375,
|
||
|
|
"loss": 2.896,
|
||
|
|
"rewards/accuracies": 0.24687500298023224,
|
||
|
|
"rewards/chosen": 0.449462890625,
|
||
|
|
"rewards/margins": -0.0478515625,
|
||
|
|
"rewards/rejected": 0.4970703125,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07042253521126761,
|
||
|
|
"grad_norm": 318.17414183804203,
|
||
|
|
"learning_rate": 3.497652582159624e-07,
|
||
|
|
"logits/chosen": -4.196875095367432,
|
||
|
|
"logits/rejected": -4.153124809265137,
|
||
|
|
"logps/chosen": -581.4000244140625,
|
||
|
|
"logps/rejected": -678.0,
|
||
|
|
"loss": 2.4536,
|
||
|
|
"rewards/accuracies": 0.28515625,
|
||
|
|
"rewards/chosen": 1.0592772960662842,
|
||
|
|
"rewards/margins": 0.843554675579071,
|
||
|
|
"rewards/rejected": 0.21484375,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07511737089201878,
|
||
|
|
"grad_norm": 402.2068723059072,
|
||
|
|
"learning_rate": 3.732394366197183e-07,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.053124904632568,
|
||
|
|
"logps/chosen": -600.4000244140625,
|
||
|
|
"logps/rejected": -680.2000122070312,
|
||
|
|
"loss": 2.5593,
|
||
|
|
"rewards/accuracies": 0.2679687440395355,
|
||
|
|
"rewards/chosen": 1.2078125476837158,
|
||
|
|
"rewards/margins": 0.7469238042831421,
|
||
|
|
"rewards/rejected": 0.4620117247104645,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07981220657276995,
|
||
|
|
"grad_norm": 329.38260711999607,
|
||
|
|
"learning_rate": 3.967136150234742e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.025000095367432,
|
||
|
|
"logps/chosen": -551.4000244140625,
|
||
|
|
"logps/rejected": -658.7999877929688,
|
||
|
|
"loss": 2.7635,
|
||
|
|
"rewards/accuracies": 0.2734375,
|
||
|
|
"rewards/chosen": 1.1056640148162842,
|
||
|
|
"rewards/margins": 0.679980456829071,
|
||
|
|
"rewards/rejected": 0.4246582090854645,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08450704225352113,
|
||
|
|
"grad_norm": 312.91437090204977,
|
||
|
|
"learning_rate": 4.2018779342723e-07,
|
||
|
|
"logits/chosen": -4.143750190734863,
|
||
|
|
"logits/rejected": -4.015625,
|
||
|
|
"logps/chosen": -538.5999755859375,
|
||
|
|
"logps/rejected": -649.5999755859375,
|
||
|
|
"loss": 2.0002,
|
||
|
|
"rewards/accuracies": 0.30156248807907104,
|
||
|
|
"rewards/chosen": 1.570703148841858,
|
||
|
|
"rewards/margins": 1.3400390148162842,
|
||
|
|
"rewards/rejected": 0.23125000298023224,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0892018779342723,
|
||
|
|
"grad_norm": 326.6984379520781,
|
||
|
|
"learning_rate": 4.436619718309859e-07,
|
||
|
|
"logits/chosen": -4.1484375,
|
||
|
|
"logits/rejected": -4.0859375,
|
||
|
|
"logps/chosen": -577.4000244140625,
|
||
|
|
"logps/rejected": -635.5999755859375,
|
||
|
|
"loss": 2.5892,
|
||
|
|
"rewards/accuracies": 0.31718748807907104,
|
||
|
|
"rewards/chosen": 1.8718750476837158,
|
||
|
|
"rewards/margins": 1.2676270008087158,
|
||
|
|
"rewards/rejected": 0.604296863079071,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09389671361502347,
|
||
|
|
"grad_norm": 268.53014556032116,
|
||
|
|
"learning_rate": 4.671361502347418e-07,
|
||
|
|
"logits/chosen": -4.123437404632568,
|
||
|
|
"logits/rejected": -4.025000095367432,
|
||
|
|
"logps/chosen": -561.2000122070312,
|
||
|
|
"logps/rejected": -644.7999877929688,
|
||
|
|
"loss": 2.4101,
|
||
|
|
"rewards/accuracies": 0.33515626192092896,
|
||
|
|
"rewards/chosen": 2.66015625,
|
||
|
|
"rewards/margins": 1.9445312023162842,
|
||
|
|
"rewards/rejected": 0.717578113079071,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09859154929577464,
|
||
|
|
"grad_norm": 283.1829513051546,
|
||
|
|
"learning_rate": 4.906103286384976e-07,
|
||
|
|
"logits/chosen": -4.173437595367432,
|
||
|
|
"logits/rejected": -4.040625095367432,
|
||
|
|
"logps/chosen": -571.0,
|
||
|
|
"logps/rejected": -677.2000122070312,
|
||
|
|
"loss": 2.9838,
|
||
|
|
"rewards/accuracies": 0.33515626192092896,
|
||
|
|
"rewards/chosen": 3.332812547683716,
|
||
|
|
"rewards/margins": 1.3986327648162842,
|
||
|
|
"rewards/rejected": 1.933203101158142,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10328638497652583,
|
||
|
|
"grad_norm": 304.4791346501985,
|
||
|
|
"learning_rate": 4.984350547730829e-07,
|
||
|
|
"logits/chosen": -4.25,
|
||
|
|
"logits/rejected": -4.099999904632568,
|
||
|
|
"logps/chosen": -584.2000122070312,
|
||
|
|
"logps/rejected": -656.0,
|
||
|
|
"loss": 2.7172,
|
||
|
|
"rewards/accuracies": 0.3515625,
|
||
|
|
"rewards/chosen": 3.3765625953674316,
|
||
|
|
"rewards/margins": 1.972070336341858,
|
||
|
|
"rewards/rejected": 1.4074218273162842,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.107981220657277,
|
||
|
|
"grad_norm": 321.7772351477239,
|
||
|
|
"learning_rate": 4.958268127282212e-07,
|
||
|
|
"logits/chosen": -4.199999809265137,
|
||
|
|
"logits/rejected": -4.087500095367432,
|
||
|
|
"logps/chosen": -573.5999755859375,
|
||
|
|
"logps/rejected": -684.7999877929688,
|
||
|
|
"loss": 2.6311,
|
||
|
|
"rewards/accuracies": 0.36796873807907104,
|
||
|
|
"rewards/chosen": 3.7828125953674316,
|
||
|
|
"rewards/margins": 2.497265577316284,
|
||
|
|
"rewards/rejected": 1.2843749523162842,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11267605633802817,
|
||
|
|
"grad_norm": 377.608437816389,
|
||
|
|
"learning_rate": 4.932185706833594e-07,
|
||
|
|
"logits/chosen": -4.131249904632568,
|
||
|
|
"logits/rejected": -4.078125,
|
||
|
|
"logps/chosen": -528.4000244140625,
|
||
|
|
"logps/rejected": -572.0,
|
||
|
|
"loss": 2.7644,
|
||
|
|
"rewards/accuracies": 0.36015623807907104,
|
||
|
|
"rewards/chosen": 3.546875,
|
||
|
|
"rewards/margins": 1.503320336341858,
|
||
|
|
"rewards/rejected": 2.043750047683716,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11737089201877934,
|
||
|
|
"grad_norm": 309.5342207630007,
|
||
|
|
"learning_rate": 4.906103286384976e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.107812404632568,
|
||
|
|
"logps/chosen": -574.4000244140625,
|
||
|
|
"logps/rejected": -689.4000244140625,
|
||
|
|
"loss": 3.0576,
|
||
|
|
"rewards/accuracies": 0.35078126192092896,
|
||
|
|
"rewards/chosen": 3.879687547683716,
|
||
|
|
"rewards/margins": 2.080127000808716,
|
||
|
|
"rewards/rejected": 1.8014647960662842,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12206572769953052,
|
||
|
|
"grad_norm": 408.1206998275862,
|
||
|
|
"learning_rate": 4.880020865936358e-07,
|
||
|
|
"logits/chosen": -4.137499809265137,
|
||
|
|
"logits/rejected": -4.067187309265137,
|
||
|
|
"logps/chosen": -540.7999877929688,
|
||
|
|
"logps/rejected": -636.0,
|
||
|
|
"loss": 2.6253,
|
||
|
|
"rewards/accuracies": 0.37890625,
|
||
|
|
"rewards/chosen": 4.121874809265137,
|
||
|
|
"rewards/margins": 1.921875,
|
||
|
|
"rewards/rejected": 2.203906297683716,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1267605633802817,
|
||
|
|
"grad_norm": 295.9696657695239,
|
||
|
|
"learning_rate": 4.853938445487741e-07,
|
||
|
|
"logits/chosen": -4.078125,
|
||
|
|
"logits/rejected": -4.003125190734863,
|
||
|
|
"logps/chosen": -573.2000122070312,
|
||
|
|
"logps/rejected": -664.0,
|
||
|
|
"loss": 2.9412,
|
||
|
|
"rewards/accuracies": 0.3921875059604645,
|
||
|
|
"rewards/chosen": 4.870312690734863,
|
||
|
|
"rewards/margins": 3.128613233566284,
|
||
|
|
"rewards/rejected": 1.746484398841858,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13145539906103287,
|
||
|
|
"grad_norm": 343.11140208524813,
|
||
|
|
"learning_rate": 4.827856025039123e-07,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.099999904632568,
|
||
|
|
"logps/chosen": -570.2000122070312,
|
||
|
|
"logps/rejected": -670.4000244140625,
|
||
|
|
"loss": 3.3178,
|
||
|
|
"rewards/accuracies": 0.42500001192092896,
|
||
|
|
"rewards/chosen": 5.498437404632568,
|
||
|
|
"rewards/margins": 3.4000000953674316,
|
||
|
|
"rewards/rejected": 2.1011719703674316,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13615023474178403,
|
||
|
|
"grad_norm": 304.3885220361387,
|
||
|
|
"learning_rate": 4.801773604590506e-07,
|
||
|
|
"logits/chosen": -4.135937690734863,
|
||
|
|
"logits/rejected": -4.057812690734863,
|
||
|
|
"logps/chosen": -612.7999877929688,
|
||
|
|
"logps/rejected": -667.5999755859375,
|
||
|
|
"loss": 3.2102,
|
||
|
|
"rewards/accuracies": 0.4359374940395355,
|
||
|
|
"rewards/chosen": 5.34375,
|
||
|
|
"rewards/margins": 2.6572265625,
|
||
|
|
"rewards/rejected": 2.6861329078674316,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14084507042253522,
|
||
|
|
"grad_norm": 284.5528315210802,
|
||
|
|
"learning_rate": 4.775691184141888e-07,
|
||
|
|
"logits/chosen": -4.146874904632568,
|
||
|
|
"logits/rejected": -4.084374904632568,
|
||
|
|
"logps/chosen": -590.4000244140625,
|
||
|
|
"logps/rejected": -663.5999755859375,
|
||
|
|
"loss": 3.0293,
|
||
|
|
"rewards/accuracies": 0.4351562559604645,
|
||
|
|
"rewards/chosen": 6.224999904632568,
|
||
|
|
"rewards/margins": 3.686718702316284,
|
||
|
|
"rewards/rejected": 2.537890672683716,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14553990610328638,
|
||
|
|
"grad_norm": 340.14707313956575,
|
||
|
|
"learning_rate": 4.749608763693271e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.079687595367432,
|
||
|
|
"logps/chosen": -565.2000122070312,
|
||
|
|
"logps/rejected": -650.0,
|
||
|
|
"loss": 3.0758,
|
||
|
|
"rewards/accuracies": 0.4437499940395355,
|
||
|
|
"rewards/chosen": 6.237500190734863,
|
||
|
|
"rewards/margins": 4.151562690734863,
|
||
|
|
"rewards/rejected": 2.0859375,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15023474178403756,
|
||
|
|
"grad_norm": 357.8195713232441,
|
||
|
|
"learning_rate": 4.7235263432446533e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.046875,
|
||
|
|
"logps/chosen": -565.7999877929688,
|
||
|
|
"logps/rejected": -641.2000122070312,
|
||
|
|
"loss": 3.2169,
|
||
|
|
"rewards/accuracies": 0.4468750059604645,
|
||
|
|
"rewards/chosen": 6.146874904632568,
|
||
|
|
"rewards/margins": 3.5914063453674316,
|
||
|
|
"rewards/rejected": 2.5484375953674316,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15492957746478872,
|
||
|
|
"grad_norm": 387.3224371650379,
|
||
|
|
"learning_rate": 4.6974439227960353e-07,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.010937690734863,
|
||
|
|
"logps/chosen": -584.2000122070312,
|
||
|
|
"logps/rejected": -670.4000244140625,
|
||
|
|
"loss": 3.0496,
|
||
|
|
"rewards/accuracies": 0.45781248807907104,
|
||
|
|
"rewards/chosen": 6.340624809265137,
|
||
|
|
"rewards/margins": 4.669140815734863,
|
||
|
|
"rewards/rejected": 1.6662108898162842,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1596244131455399,
|
||
|
|
"grad_norm": 378.7521561914789,
|
||
|
|
"learning_rate": 4.671361502347418e-07,
|
||
|
|
"logits/chosen": -4.203125,
|
||
|
|
"logits/rejected": -4.118750095367432,
|
||
|
|
"logps/chosen": -555.7999877929688,
|
||
|
|
"logps/rejected": -601.4000244140625,
|
||
|
|
"loss": 3.6385,
|
||
|
|
"rewards/accuracies": 0.453125,
|
||
|
|
"rewards/chosen": 6.518750190734863,
|
||
|
|
"rewards/margins": 3.078320264816284,
|
||
|
|
"rewards/rejected": 3.4437499046325684,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1643192488262911,
|
||
|
|
"grad_norm": 280.186323729001,
|
||
|
|
"learning_rate": 4.6452790818988004e-07,
|
||
|
|
"logits/chosen": -4.214062690734863,
|
||
|
|
"logits/rejected": -4.126562595367432,
|
||
|
|
"logps/chosen": -593.2000122070312,
|
||
|
|
"logps/rejected": -669.2000122070312,
|
||
|
|
"loss": 3.508,
|
||
|
|
"rewards/accuracies": 0.4312500059604645,
|
||
|
|
"rewards/chosen": 6.784375190734863,
|
||
|
|
"rewards/margins": 4.580859184265137,
|
||
|
|
"rewards/rejected": 2.203418016433716,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16901408450704225,
|
||
|
|
"grad_norm": 289.7400713232728,
|
||
|
|
"learning_rate": 4.6191966614501824e-07,
|
||
|
|
"logits/chosen": -4.173437595367432,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -560.5999755859375,
|
||
|
|
"logps/rejected": -643.7999877929688,
|
||
|
|
"loss": 3.0314,
|
||
|
|
"rewards/accuracies": 0.46406251192092896,
|
||
|
|
"rewards/chosen": 7.434374809265137,
|
||
|
|
"rewards/margins": 4.764843940734863,
|
||
|
|
"rewards/rejected": 2.671093702316284,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17370892018779344,
|
||
|
|
"grad_norm": 305.1857644992596,
|
||
|
|
"learning_rate": 4.593114241001565e-07,
|
||
|
|
"logits/chosen": -4.115624904632568,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -602.4000244140625,
|
||
|
|
"logps/rejected": -672.0,
|
||
|
|
"loss": 3.0562,
|
||
|
|
"rewards/accuracies": 0.4429687559604645,
|
||
|
|
"rewards/chosen": 7.678124904632568,
|
||
|
|
"rewards/margins": 5.004687309265137,
|
||
|
|
"rewards/rejected": 2.671093702316284,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1784037558685446,
|
||
|
|
"grad_norm": 281.3726377297068,
|
||
|
|
"learning_rate": 4.5670318205529474e-07,
|
||
|
|
"logits/chosen": -4.146874904632568,
|
||
|
|
"logits/rejected": -4.026562690734863,
|
||
|
|
"logps/chosen": -570.7999877929688,
|
||
|
|
"logps/rejected": -666.4000244140625,
|
||
|
|
"loss": 3.4014,
|
||
|
|
"rewards/accuracies": 0.44453126192092896,
|
||
|
|
"rewards/chosen": 6.846875190734863,
|
||
|
|
"rewards/margins": 3.7099609375,
|
||
|
|
"rewards/rejected": 3.1382813453674316,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18309859154929578,
|
||
|
|
"grad_norm": 318.3140273505768,
|
||
|
|
"learning_rate": 4.54094940010433e-07,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.053124904632568,
|
||
|
|
"logps/chosen": -581.2000122070312,
|
||
|
|
"logps/rejected": -679.5999755859375,
|
||
|
|
"loss": 3.165,
|
||
|
|
"rewards/accuracies": 0.47187501192092896,
|
||
|
|
"rewards/chosen": 7.240624904632568,
|
||
|
|
"rewards/margins": 4.939062595367432,
|
||
|
|
"rewards/rejected": 2.298828125,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18779342723004694,
|
||
|
|
"grad_norm": 266.65165381199415,
|
||
|
|
"learning_rate": 4.514866979655712e-07,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.048437595367432,
|
||
|
|
"logps/chosen": -572.0,
|
||
|
|
"logps/rejected": -645.5999755859375,
|
||
|
|
"loss": 2.8849,
|
||
|
|
"rewards/accuracies": 0.46171873807907104,
|
||
|
|
"rewards/chosen": 7.662499904632568,
|
||
|
|
"rewards/margins": 5.106249809265137,
|
||
|
|
"rewards/rejected": 2.54736328125,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19248826291079812,
|
||
|
|
"grad_norm": 267.9080238940424,
|
||
|
|
"learning_rate": 4.4887845592070945e-07,
|
||
|
|
"logits/chosen": -4.109375,
|
||
|
|
"logits/rejected": -3.995312452316284,
|
||
|
|
"logps/chosen": -559.5999755859375,
|
||
|
|
"logps/rejected": -688.4000244140625,
|
||
|
|
"loss": 3.2426,
|
||
|
|
"rewards/accuracies": 0.46406251192092896,
|
||
|
|
"rewards/chosen": 7.618750095367432,
|
||
|
|
"rewards/margins": 4.5625,
|
||
|
|
"rewards/rejected": 3.0562500953674316,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19718309859154928,
|
||
|
|
"grad_norm": 312.69487450228394,
|
||
|
|
"learning_rate": 4.462702138758477e-07,
|
||
|
|
"logits/chosen": -4.121874809265137,
|
||
|
|
"logits/rejected": -4.03125,
|
||
|
|
"logps/chosen": -597.5999755859375,
|
||
|
|
"logps/rejected": -653.2000122070312,
|
||
|
|
"loss": 3.7578,
|
||
|
|
"rewards/accuracies": 0.43437498807907104,
|
||
|
|
"rewards/chosen": 7.662499904632568,
|
||
|
|
"rewards/margins": 3.947070360183716,
|
||
|
|
"rewards/rejected": 3.7164063453674316,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20187793427230047,
|
||
|
|
"grad_norm": 323.8865547251275,
|
||
|
|
"learning_rate": 4.436619718309859e-07,
|
||
|
|
"logits/chosen": -4.103125095367432,
|
||
|
|
"logits/rejected": -3.9546875953674316,
|
||
|
|
"logps/chosen": -581.4000244140625,
|
||
|
|
"logps/rejected": -698.7999877929688,
|
||
|
|
"loss": 3.3093,
|
||
|
|
"rewards/accuracies": 0.4632812440395355,
|
||
|
|
"rewards/chosen": 7.925000190734863,
|
||
|
|
"rewards/margins": 5.646874904632568,
|
||
|
|
"rewards/rejected": 2.2835936546325684,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20657276995305165,
|
||
|
|
"grad_norm": 393.9159806692629,
|
||
|
|
"learning_rate": 4.4105372978612415e-07,
|
||
|
|
"logits/chosen": -4.212500095367432,
|
||
|
|
"logits/rejected": -4.090624809265137,
|
||
|
|
"logps/chosen": -597.2000122070312,
|
||
|
|
"logps/rejected": -683.2000122070312,
|
||
|
|
"loss": 3.6426,
|
||
|
|
"rewards/accuracies": 0.4749999940395355,
|
||
|
|
"rewards/chosen": 7.912499904632568,
|
||
|
|
"rewards/margins": 5.670312404632568,
|
||
|
|
"rewards/rejected": 2.243359327316284,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2112676056338028,
|
||
|
|
"grad_norm": 373.9652792696506,
|
||
|
|
"learning_rate": 4.384454877412624e-07,
|
||
|
|
"logits/chosen": -4.189062595367432,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -582.5999755859375,
|
||
|
|
"logps/rejected": -746.7999877929688,
|
||
|
|
"loss": 3.6727,
|
||
|
|
"rewards/accuracies": 0.4554687440395355,
|
||
|
|
"rewards/chosen": 7.740624904632568,
|
||
|
|
"rewards/margins": 4.469531059265137,
|
||
|
|
"rewards/rejected": 3.26953125,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.215962441314554,
|
||
|
|
"grad_norm": 330.13898710319694,
|
||
|
|
"learning_rate": 4.358372456964006e-07,
|
||
|
|
"logits/chosen": -4.1328125,
|
||
|
|
"logits/rejected": -4.0703125,
|
||
|
|
"logps/chosen": -574.4000244140625,
|
||
|
|
"logps/rejected": -666.7999877929688,
|
||
|
|
"loss": 3.2602,
|
||
|
|
"rewards/accuracies": 0.4765625,
|
||
|
|
"rewards/chosen": 8.378125190734863,
|
||
|
|
"rewards/margins": 5.387499809265137,
|
||
|
|
"rewards/rejected": 2.9869141578674316,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22065727699530516,
|
||
|
|
"grad_norm": 293.6213926149627,
|
||
|
|
"learning_rate": 4.3322900365153886e-07,
|
||
|
|
"logits/chosen": -4.25,
|
||
|
|
"logits/rejected": -4.096875190734863,
|
||
|
|
"logps/chosen": -612.0,
|
||
|
|
"logps/rejected": -669.5999755859375,
|
||
|
|
"loss": 3.3312,
|
||
|
|
"rewards/accuracies": 0.46953123807907104,
|
||
|
|
"rewards/chosen": 8.553125381469727,
|
||
|
|
"rewards/margins": 5.612500190734863,
|
||
|
|
"rewards/rejected": 2.9410157203674316,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22535211267605634,
|
||
|
|
"grad_norm": 295.2353414406269,
|
||
|
|
"learning_rate": 4.306207616066771e-07,
|
||
|
|
"logits/chosen": -4.209374904632568,
|
||
|
|
"logits/rejected": -4.096875190734863,
|
||
|
|
"logps/chosen": -562.2000122070312,
|
||
|
|
"logps/rejected": -645.2000122070312,
|
||
|
|
"loss": 3.1204,
|
||
|
|
"rewards/accuracies": 0.46562498807907104,
|
||
|
|
"rewards/chosen": 8.978124618530273,
|
||
|
|
"rewards/margins": 5.035937309265137,
|
||
|
|
"rewards/rejected": 3.94140625,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2300469483568075,
|
||
|
|
"grad_norm": 254.67708489014822,
|
||
|
|
"learning_rate": 4.280125195618153e-07,
|
||
|
|
"logits/chosen": -4.153124809265137,
|
||
|
|
"logits/rejected": -4.010937690734863,
|
||
|
|
"logps/chosen": -573.0,
|
||
|
|
"logps/rejected": -675.4000244140625,
|
||
|
|
"loss": 2.9846,
|
||
|
|
"rewards/accuracies": 0.48359376192092896,
|
||
|
|
"rewards/chosen": 9.565625190734863,
|
||
|
|
"rewards/margins": 6.709374904632568,
|
||
|
|
"rewards/rejected": 2.8531250953674316,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2347417840375587,
|
||
|
|
"grad_norm": 274.79861750540084,
|
||
|
|
"learning_rate": 4.2540427751695357e-07,
|
||
|
|
"logits/chosen": -4.146874904632568,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -580.0,
|
||
|
|
"logps/rejected": -646.4000244140625,
|
||
|
|
"loss": 3.157,
|
||
|
|
"rewards/accuracies": 0.47968751192092896,
|
||
|
|
"rewards/chosen": 8.712499618530273,
|
||
|
|
"rewards/margins": 5.418749809265137,
|
||
|
|
"rewards/rejected": 3.296875,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23943661971830985,
|
||
|
|
"grad_norm": 250.05393627156735,
|
||
|
|
"learning_rate": 4.227960354720918e-07,
|
||
|
|
"logits/chosen": -4.221875190734863,
|
||
|
|
"logits/rejected": -4.1171875,
|
||
|
|
"logps/chosen": -566.5999755859375,
|
||
|
|
"logps/rejected": -685.2000122070312,
|
||
|
|
"loss": 2.8452,
|
||
|
|
"rewards/accuracies": 0.4945312440395355,
|
||
|
|
"rewards/chosen": 9.324999809265137,
|
||
|
|
"rewards/margins": 7.099999904632568,
|
||
|
|
"rewards/rejected": 2.210742235183716,
|
||
|
|
"step": 510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24413145539906103,
|
||
|
|
"grad_norm": 289.17642513775206,
|
||
|
|
"learning_rate": 4.2018779342723e-07,
|
||
|
|
"logits/chosen": -4.150000095367432,
|
||
|
|
"logits/rejected": -4.079687595367432,
|
||
|
|
"logps/chosen": -582.4000244140625,
|
||
|
|
"logps/rejected": -637.2000122070312,
|
||
|
|
"loss": 3.1881,
|
||
|
|
"rewards/accuracies": 0.47734373807907104,
|
||
|
|
"rewards/chosen": 9.550000190734863,
|
||
|
|
"rewards/margins": 5.614062309265137,
|
||
|
|
"rewards/rejected": 3.934765577316284,
|
||
|
|
"step": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24882629107981222,
|
||
|
|
"grad_norm": 301.4772408122623,
|
||
|
|
"learning_rate": 4.1757955138236827e-07,
|
||
|
|
"logits/chosen": -4.21875,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -565.4000244140625,
|
||
|
|
"logps/rejected": -671.2000122070312,
|
||
|
|
"loss": 3.2645,
|
||
|
|
"rewards/accuracies": 0.500781238079071,
|
||
|
|
"rewards/chosen": 9.428125381469727,
|
||
|
|
"rewards/margins": 7.178124904632568,
|
||
|
|
"rewards/rejected": 2.2474608421325684,
|
||
|
|
"step": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2535211267605634,
|
||
|
|
"grad_norm": 291.7949545654684,
|
||
|
|
"learning_rate": 4.149713093375065e-07,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -4.078125,
|
||
|
|
"logps/chosen": -566.4000244140625,
|
||
|
|
"logps/rejected": -618.7999877929688,
|
||
|
|
"loss": 3.2226,
|
||
|
|
"rewards/accuracies": 0.4976562559604645,
|
||
|
|
"rewards/chosen": 9.790624618530273,
|
||
|
|
"rewards/margins": 5.895312309265137,
|
||
|
|
"rewards/rejected": 3.8968749046325684,
|
||
|
|
"step": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25821596244131456,
|
||
|
|
"grad_norm": 295.46300897462413,
|
||
|
|
"learning_rate": 4.123630672926447e-07,
|
||
|
|
"logits/chosen": -4.189062595367432,
|
||
|
|
"logits/rejected": -4.09375,
|
||
|
|
"logps/chosen": -555.4000244140625,
|
||
|
|
"logps/rejected": -664.5999755859375,
|
||
|
|
"loss": 3.3349,
|
||
|
|
"rewards/accuracies": 0.4828124940395355,
|
||
|
|
"rewards/chosen": 9.003125190734863,
|
||
|
|
"rewards/margins": 5.984375,
|
||
|
|
"rewards/rejected": 3.02392578125,
|
||
|
|
"step": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26291079812206575,
|
||
|
|
"grad_norm": 282.55243937448216,
|
||
|
|
"learning_rate": 4.09754825247783e-07,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.034375190734863,
|
||
|
|
"logps/chosen": -575.2000122070312,
|
||
|
|
"logps/rejected": -645.5999755859375,
|
||
|
|
"loss": 3.3186,
|
||
|
|
"rewards/accuracies": 0.4976562559604645,
|
||
|
|
"rewards/chosen": 9.703125,
|
||
|
|
"rewards/margins": 6.403124809265137,
|
||
|
|
"rewards/rejected": 3.303906202316284,
|
||
|
|
"step": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2676056338028169,
|
||
|
|
"grad_norm": 243.28520451102895,
|
||
|
|
"learning_rate": 4.0714658320292123e-07,
|
||
|
|
"logits/chosen": -4.217187404632568,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -574.0,
|
||
|
|
"logps/rejected": -659.5999755859375,
|
||
|
|
"loss": 3.5655,
|
||
|
|
"rewards/accuracies": 0.4820312559604645,
|
||
|
|
"rewards/chosen": 8.893750190734863,
|
||
|
|
"rewards/margins": 6.009375095367432,
|
||
|
|
"rewards/rejected": 2.874218702316284,
|
||
|
|
"step": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27230046948356806,
|
||
|
|
"grad_norm": 290.7842698495749,
|
||
|
|
"learning_rate": 4.045383411580595e-07,
|
||
|
|
"logits/chosen": -4.215624809265137,
|
||
|
|
"logits/rejected": -4.096875190734863,
|
||
|
|
"logps/chosen": -555.0,
|
||
|
|
"logps/rejected": -658.4000244140625,
|
||
|
|
"loss": 2.9852,
|
||
|
|
"rewards/accuracies": 0.4976562559604645,
|
||
|
|
"rewards/chosen": 10.081250190734863,
|
||
|
|
"rewards/margins": 7.290625095367432,
|
||
|
|
"rewards/rejected": 2.79833984375,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27699530516431925,
|
||
|
|
"grad_norm": 291.2604745601705,
|
||
|
|
"learning_rate": 4.019300991131977e-07,
|
||
|
|
"logits/chosen": -4.140625,
|
||
|
|
"logits/rejected": -4.015625,
|
||
|
|
"logps/chosen": -557.0,
|
||
|
|
"logps/rejected": -648.7999877929688,
|
||
|
|
"loss": 3.018,
|
||
|
|
"rewards/accuracies": 0.5023437738418579,
|
||
|
|
"rewards/chosen": 10.199999809265137,
|
||
|
|
"rewards/margins": 6.995312690734863,
|
||
|
|
"rewards/rejected": 3.2113280296325684,
|
||
|
|
"step": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28169014084507044,
|
||
|
|
"grad_norm": 227.7522974477189,
|
||
|
|
"learning_rate": 3.9932185706833594e-07,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -583.2000122070312,
|
||
|
|
"logps/rejected": -650.0,
|
||
|
|
"loss": 3.2984,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 10.106249809265137,
|
||
|
|
"rewards/margins": 6.421875,
|
||
|
|
"rewards/rejected": 3.671093702316284,
|
||
|
|
"step": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2863849765258216,
|
||
|
|
"grad_norm": 287.0075813999204,
|
||
|
|
"learning_rate": 3.967136150234742e-07,
|
||
|
|
"logits/chosen": -4.151562690734863,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -584.2000122070312,
|
||
|
|
"logps/rejected": -664.7999877929688,
|
||
|
|
"loss": 3.377,
|
||
|
|
"rewards/accuracies": 0.4828124940395355,
|
||
|
|
"rewards/chosen": 10.190625190734863,
|
||
|
|
"rewards/margins": 6.734375,
|
||
|
|
"rewards/rejected": 3.453906297683716,
|
||
|
|
"step": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29107981220657275,
|
||
|
|
"grad_norm": 280.2718266050094,
|
||
|
|
"learning_rate": 3.941053729786124e-07,
|
||
|
|
"logits/chosen": -4.154687404632568,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -587.4000244140625,
|
||
|
|
"logps/rejected": -661.5999755859375,
|
||
|
|
"loss": 3.1514,
|
||
|
|
"rewards/accuracies": 0.5179687738418579,
|
||
|
|
"rewards/chosen": 9.859375,
|
||
|
|
"rewards/margins": 7.278124809265137,
|
||
|
|
"rewards/rejected": 2.56640625,
|
||
|
|
"step": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29577464788732394,
|
||
|
|
"grad_norm": 324.67176703959694,
|
||
|
|
"learning_rate": 3.9149713093375064e-07,
|
||
|
|
"logits/chosen": -4.1015625,
|
||
|
|
"logits/rejected": -3.957812547683716,
|
||
|
|
"logps/chosen": -580.0,
|
||
|
|
"logps/rejected": -683.5999755859375,
|
||
|
|
"loss": 3.7381,
|
||
|
|
"rewards/accuracies": 0.504687488079071,
|
||
|
|
"rewards/chosen": 9.53125,
|
||
|
|
"rewards/margins": 6.6875,
|
||
|
|
"rewards/rejected": 2.8443360328674316,
|
||
|
|
"step": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3004694835680751,
|
||
|
|
"grad_norm": 316.9452197327162,
|
||
|
|
"learning_rate": 3.888888888888889e-07,
|
||
|
|
"logits/chosen": -4.185937404632568,
|
||
|
|
"logits/rejected": -4.026562690734863,
|
||
|
|
"logps/chosen": -568.5999755859375,
|
||
|
|
"logps/rejected": -696.0,
|
||
|
|
"loss": 3.5189,
|
||
|
|
"rewards/accuracies": 0.4765625,
|
||
|
|
"rewards/chosen": 9.512499809265137,
|
||
|
|
"rewards/margins": 6.237500190734863,
|
||
|
|
"rewards/rejected": 3.28076171875,
|
||
|
|
"step": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3051643192488263,
|
||
|
|
"grad_norm": 297.1688628970196,
|
||
|
|
"learning_rate": 3.862806468440271e-07,
|
||
|
|
"logits/chosen": -4.284375190734863,
|
||
|
|
"logits/rejected": -4.184374809265137,
|
||
|
|
"logps/chosen": -527.0,
|
||
|
|
"logps/rejected": -620.0,
|
||
|
|
"loss": 3.1752,
|
||
|
|
"rewards/accuracies": 0.4945312440395355,
|
||
|
|
"rewards/chosen": 9.568750381469727,
|
||
|
|
"rewards/margins": 5.458984375,
|
||
|
|
"rewards/rejected": 4.130053520202637,
|
||
|
|
"step": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30985915492957744,
|
||
|
|
"grad_norm": 251.23055282468778,
|
||
|
|
"learning_rate": 3.8367240479916535e-07,
|
||
|
|
"logits/chosen": -4.192187309265137,
|
||
|
|
"logits/rejected": -4.151562690734863,
|
||
|
|
"logps/chosen": -560.4000244140625,
|
||
|
|
"logps/rejected": -623.0,
|
||
|
|
"loss": 3.2523,
|
||
|
|
"rewards/accuracies": 0.48046875,
|
||
|
|
"rewards/chosen": 9.712499618530273,
|
||
|
|
"rewards/margins": 5.959374904632568,
|
||
|
|
"rewards/rejected": 3.7593750953674316,
|
||
|
|
"step": 660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3145539906103286,
|
||
|
|
"grad_norm": 322.7495965594595,
|
||
|
|
"learning_rate": 3.810641627543036e-07,
|
||
|
|
"logits/chosen": -4.196875095367432,
|
||
|
|
"logits/rejected": -4.09375,
|
||
|
|
"logps/chosen": -576.4000244140625,
|
||
|
|
"logps/rejected": -658.2000122070312,
|
||
|
|
"loss": 3.4558,
|
||
|
|
"rewards/accuracies": 0.49609375,
|
||
|
|
"rewards/chosen": 9.318750381469727,
|
||
|
|
"rewards/margins": 6.964062690734863,
|
||
|
|
"rewards/rejected": 2.3499999046325684,
|
||
|
|
"step": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3192488262910798,
|
||
|
|
"grad_norm": 304.2542609024998,
|
||
|
|
"learning_rate": 3.784559207094418e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -595.2000122070312,
|
||
|
|
"logps/rejected": -651.2000122070312,
|
||
|
|
"loss": 3.9372,
|
||
|
|
"rewards/accuracies": 0.484375,
|
||
|
|
"rewards/chosen": 9.699999809265137,
|
||
|
|
"rewards/margins": 5.493750095367432,
|
||
|
|
"rewards/rejected": 4.212500095367432,
|
||
|
|
"step": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.323943661971831,
|
||
|
|
"grad_norm": 242.7782318272521,
|
||
|
|
"learning_rate": 3.7584767866458005e-07,
|
||
|
|
"logits/chosen": -4.162499904632568,
|
||
|
|
"logits/rejected": -4.042187690734863,
|
||
|
|
"logps/chosen": -578.4000244140625,
|
||
|
|
"logps/rejected": -621.7999877929688,
|
||
|
|
"loss": 3.8324,
|
||
|
|
"rewards/accuracies": 0.4859375059604645,
|
||
|
|
"rewards/chosen": 8.943750381469727,
|
||
|
|
"rewards/margins": 6.134375095367432,
|
||
|
|
"rewards/rejected": 2.81640625,
|
||
|
|
"step": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3286384976525822,
|
||
|
|
"grad_norm": 387.5093883157667,
|
||
|
|
"learning_rate": 3.732394366197183e-07,
|
||
|
|
"logits/chosen": -4.212500095367432,
|
||
|
|
"logits/rejected": -4.109375,
|
||
|
|
"logps/chosen": -606.5999755859375,
|
||
|
|
"logps/rejected": -635.5999755859375,
|
||
|
|
"loss": 4.5264,
|
||
|
|
"rewards/accuracies": 0.500781238079071,
|
||
|
|
"rewards/chosen": 8.987500190734863,
|
||
|
|
"rewards/margins": 5.770312309265137,
|
||
|
|
"rewards/rejected": 3.219531297683716,
|
||
|
|
"step": 700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3333333333333333,
|
||
|
|
"grad_norm": 357.2057827892341,
|
||
|
|
"learning_rate": 3.706311945748565e-07,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -623.5999755859375,
|
||
|
|
"logps/rejected": -645.5999755859375,
|
||
|
|
"loss": 3.9486,
|
||
|
|
"rewards/accuracies": 0.49921876192092896,
|
||
|
|
"rewards/chosen": 9.634374618530273,
|
||
|
|
"rewards/margins": 6.546875,
|
||
|
|
"rewards/rejected": 3.098437547683716,
|
||
|
|
"step": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3380281690140845,
|
||
|
|
"grad_norm": 289.48952564262845,
|
||
|
|
"learning_rate": 3.6802295252999476e-07,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -3.9703125953674316,
|
||
|
|
"logps/chosen": -590.7999877929688,
|
||
|
|
"logps/rejected": -686.5999755859375,
|
||
|
|
"loss": 3.6266,
|
||
|
|
"rewards/accuracies": 0.49687498807907104,
|
||
|
|
"rewards/chosen": 9.737500190734863,
|
||
|
|
"rewards/margins": 7.662499904632568,
|
||
|
|
"rewards/rejected": 2.065869092941284,
|
||
|
|
"step": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3427230046948357,
|
||
|
|
"grad_norm": 322.44572436537794,
|
||
|
|
"learning_rate": 3.65414710485133e-07,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.09375,
|
||
|
|
"logps/chosen": -585.2000122070312,
|
||
|
|
"logps/rejected": -628.5999755859375,
|
||
|
|
"loss": 3.707,
|
||
|
|
"rewards/accuracies": 0.515625,
|
||
|
|
"rewards/chosen": 9.618749618530273,
|
||
|
|
"rewards/margins": 7.190625190734863,
|
||
|
|
"rewards/rejected": 2.4375,
|
||
|
|
"step": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3474178403755869,
|
||
|
|
"grad_norm": 327.4047879299755,
|
||
|
|
"learning_rate": 3.6280646844027127e-07,
|
||
|
|
"logits/chosen": -4.135937690734863,
|
||
|
|
"logits/rejected": -4.040625095367432,
|
||
|
|
"logps/chosen": -564.0,
|
||
|
|
"logps/rejected": -702.0,
|
||
|
|
"loss": 2.9845,
|
||
|
|
"rewards/accuracies": 0.5257812738418579,
|
||
|
|
"rewards/chosen": 9.962499618530273,
|
||
|
|
"rewards/margins": 8.574999809265137,
|
||
|
|
"rewards/rejected": 1.39453125,
|
||
|
|
"step": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.352112676056338,
|
||
|
|
"grad_norm": 268.40285966149014,
|
||
|
|
"learning_rate": 3.6019822639540947e-07,
|
||
|
|
"logits/chosen": -4.150000095367432,
|
||
|
|
"logits/rejected": -4.042187690734863,
|
||
|
|
"logps/chosen": -599.7999877929688,
|
||
|
|
"logps/rejected": -627.2000122070312,
|
||
|
|
"loss": 2.9559,
|
||
|
|
"rewards/accuracies": 0.4984374940395355,
|
||
|
|
"rewards/chosen": 10.800000190734863,
|
||
|
|
"rewards/margins": 7.465624809265137,
|
||
|
|
"rewards/rejected": 3.3382811546325684,
|
||
|
|
"step": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3568075117370892,
|
||
|
|
"grad_norm": 228.49576724117685,
|
||
|
|
"learning_rate": 3.575899843505477e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.071875095367432,
|
||
|
|
"logps/chosen": -571.5999755859375,
|
||
|
|
"logps/rejected": -611.5999755859375,
|
||
|
|
"loss": 3.3682,
|
||
|
|
"rewards/accuracies": 0.5328124761581421,
|
||
|
|
"rewards/chosen": 9.946874618530273,
|
||
|
|
"rewards/margins": 6.965234279632568,
|
||
|
|
"rewards/rejected": 2.9810547828674316,
|
||
|
|
"step": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3615023474178404,
|
||
|
|
"grad_norm": 339.19249872693797,
|
||
|
|
"learning_rate": 3.5498174230568597e-07,
|
||
|
|
"logits/chosen": -4.110937595367432,
|
||
|
|
"logits/rejected": -4.0234375,
|
||
|
|
"logps/chosen": -570.4000244140625,
|
||
|
|
"logps/rejected": -650.7999877929688,
|
||
|
|
"loss": 2.9888,
|
||
|
|
"rewards/accuracies": 0.51953125,
|
||
|
|
"rewards/chosen": 9.78125,
|
||
|
|
"rewards/margins": 8.318750381469727,
|
||
|
|
"rewards/rejected": 1.464257836341858,
|
||
|
|
"step": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36619718309859156,
|
||
|
|
"grad_norm": 307.4418590636236,
|
||
|
|
"learning_rate": 3.5237350026082417e-07,
|
||
|
|
"logits/chosen": -4.125,
|
||
|
|
"logits/rejected": -4.046875,
|
||
|
|
"logps/chosen": -578.4000244140625,
|
||
|
|
"logps/rejected": -609.2000122070312,
|
||
|
|
"loss": 3.5426,
|
||
|
|
"rewards/accuracies": 0.49140626192092896,
|
||
|
|
"rewards/chosen": 9.300000190734863,
|
||
|
|
"rewards/margins": 6.425000190734863,
|
||
|
|
"rewards/rejected": 2.8755860328674316,
|
||
|
|
"step": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37089201877934275,
|
||
|
|
"grad_norm": 304.8766546196448,
|
||
|
|
"learning_rate": 3.497652582159624e-07,
|
||
|
|
"logits/chosen": -4.140625,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -582.7999877929688,
|
||
|
|
"logps/rejected": -674.0,
|
||
|
|
"loss": 3.1521,
|
||
|
|
"rewards/accuracies": 0.503125011920929,
|
||
|
|
"rewards/chosen": 9.399999618530273,
|
||
|
|
"rewards/margins": 7.859375,
|
||
|
|
"rewards/rejected": 1.552148461341858,
|
||
|
|
"step": 790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3755868544600939,
|
||
|
|
"grad_norm": 253.18672796971052,
|
||
|
|
"learning_rate": 3.471570161711007e-07,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.042187690734863,
|
||
|
|
"logps/chosen": -609.7999877929688,
|
||
|
|
"logps/rejected": -685.7999877929688,
|
||
|
|
"loss": 3.4785,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 9.543749809265137,
|
||
|
|
"rewards/margins": 8.240625381469727,
|
||
|
|
"rewards/rejected": 1.3092772960662842,
|
||
|
|
"step": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38028169014084506,
|
||
|
|
"grad_norm": 267.7610739280651,
|
||
|
|
"learning_rate": 3.445487741262389e-07,
|
||
|
|
"logits/chosen": -4.184374809265137,
|
||
|
|
"logits/rejected": -4.103125095367432,
|
||
|
|
"logps/chosen": -561.5999755859375,
|
||
|
|
"logps/rejected": -647.5999755859375,
|
||
|
|
"loss": 3.9487,
|
||
|
|
"rewards/accuracies": 0.5140625238418579,
|
||
|
|
"rewards/chosen": 9.259374618530273,
|
||
|
|
"rewards/margins": 7.068749904632568,
|
||
|
|
"rewards/rejected": 2.196093797683716,
|
||
|
|
"step": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38497652582159625,
|
||
|
|
"grad_norm": 278.98516681627376,
|
||
|
|
"learning_rate": 3.4194053208137713e-07,
|
||
|
|
"logits/chosen": -4.153124809265137,
|
||
|
|
"logits/rejected": -4.051562309265137,
|
||
|
|
"logps/chosen": -571.2000122070312,
|
||
|
|
"logps/rejected": -622.0,
|
||
|
|
"loss": 3.2654,
|
||
|
|
"rewards/accuracies": 0.5179687738418579,
|
||
|
|
"rewards/chosen": 9.9375,
|
||
|
|
"rewards/margins": 7.559374809265137,
|
||
|
|
"rewards/rejected": 2.372119188308716,
|
||
|
|
"step": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38967136150234744,
|
||
|
|
"grad_norm": 263.29366918290094,
|
||
|
|
"learning_rate": 3.393322900365154e-07,
|
||
|
|
"logits/chosen": -4.171875,
|
||
|
|
"logits/rejected": -4.009375095367432,
|
||
|
|
"logps/chosen": -540.5999755859375,
|
||
|
|
"logps/rejected": -624.0,
|
||
|
|
"loss": 3.855,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 8.768750190734863,
|
||
|
|
"rewards/margins": 6.830078125,
|
||
|
|
"rewards/rejected": 1.9441406726837158,
|
||
|
|
"step": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39436619718309857,
|
||
|
|
"grad_norm": 268.93242807228216,
|
||
|
|
"learning_rate": 3.367240479916536e-07,
|
||
|
|
"logits/chosen": -4.162499904632568,
|
||
|
|
"logits/rejected": -4.026562690734863,
|
||
|
|
"logps/chosen": -606.0,
|
||
|
|
"logps/rejected": -673.5999755859375,
|
||
|
|
"loss": 4.048,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 8.678125381469727,
|
||
|
|
"rewards/margins": 6.693749904632568,
|
||
|
|
"rewards/rejected": 1.9873046875,
|
||
|
|
"step": 840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39906103286384975,
|
||
|
|
"grad_norm": 312.88147844071045,
|
||
|
|
"learning_rate": 3.3411580594679184e-07,
|
||
|
|
"logits/chosen": -4.2109375,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -561.7999877929688,
|
||
|
|
"logps/rejected": -646.7999877929688,
|
||
|
|
"loss": 3.2428,
|
||
|
|
"rewards/accuracies": 0.5054687261581421,
|
||
|
|
"rewards/chosen": 9.328125,
|
||
|
|
"rewards/margins": 7.162499904632568,
|
||
|
|
"rewards/rejected": 2.1823487281799316,
|
||
|
|
"step": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40375586854460094,
|
||
|
|
"grad_norm": 260.7141136621247,
|
||
|
|
"learning_rate": 3.315075639019301e-07,
|
||
|
|
"logits/chosen": -4.173437595367432,
|
||
|
|
"logits/rejected": -4.095312595367432,
|
||
|
|
"logps/chosen": -555.5999755859375,
|
||
|
|
"logps/rejected": -642.0,
|
||
|
|
"loss": 2.9281,
|
||
|
|
"rewards/accuracies": 0.510937511920929,
|
||
|
|
"rewards/chosen": 9.393750190734863,
|
||
|
|
"rewards/margins": 8.515625,
|
||
|
|
"rewards/rejected": 0.88330078125,
|
||
|
|
"step": 860
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4084507042253521,
|
||
|
|
"grad_norm": 296.8891743971074,
|
||
|
|
"learning_rate": 3.288993218570683e-07,
|
||
|
|
"logits/chosen": -4.199999809265137,
|
||
|
|
"logits/rejected": -4.040625095367432,
|
||
|
|
"logps/chosen": -562.7999877929688,
|
||
|
|
"logps/rejected": -628.0,
|
||
|
|
"loss": 3.5136,
|
||
|
|
"rewards/accuracies": 0.510937511920929,
|
||
|
|
"rewards/chosen": 9.068750381469727,
|
||
|
|
"rewards/margins": 6.717187404632568,
|
||
|
|
"rewards/rejected": 2.3675780296325684,
|
||
|
|
"step": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4131455399061033,
|
||
|
|
"grad_norm": 265.0685724449779,
|
||
|
|
"learning_rate": 3.2629107981220654e-07,
|
||
|
|
"logits/chosen": -4.170312404632568,
|
||
|
|
"logits/rejected": -3.996875047683716,
|
||
|
|
"logps/chosen": -586.2000122070312,
|
||
|
|
"logps/rejected": -651.0,
|
||
|
|
"loss": 3.7241,
|
||
|
|
"rewards/accuracies": 0.48906248807907104,
|
||
|
|
"rewards/chosen": 8.490625381469727,
|
||
|
|
"rewards/margins": 5.989062309265137,
|
||
|
|
"rewards/rejected": 2.502978563308716,
|
||
|
|
"step": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41784037558685444,
|
||
|
|
"grad_norm": 290.79755612144265,
|
||
|
|
"learning_rate": 3.236828377673448e-07,
|
||
|
|
"logits/chosen": -4.107812404632568,
|
||
|
|
"logits/rejected": -4.0703125,
|
||
|
|
"logps/chosen": -609.2000122070312,
|
||
|
|
"logps/rejected": -644.0,
|
||
|
|
"loss": 4.4678,
|
||
|
|
"rewards/accuracies": 0.48124998807907104,
|
||
|
|
"rewards/chosen": 8.574999809265137,
|
||
|
|
"rewards/margins": 5.985937595367432,
|
||
|
|
"rewards/rejected": 2.5953125953674316,
|
||
|
|
"step": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4225352112676056,
|
||
|
|
"grad_norm": 261.2611410206578,
|
||
|
|
"learning_rate": 3.2107459572248305e-07,
|
||
|
|
"logits/chosen": -4.118750095367432,
|
||
|
|
"logits/rejected": -4.043749809265137,
|
||
|
|
"logps/chosen": -587.2000122070312,
|
||
|
|
"logps/rejected": -651.2000122070312,
|
||
|
|
"loss": 2.8678,
|
||
|
|
"rewards/accuracies": 0.5335937738418579,
|
||
|
|
"rewards/chosen": 9.399999618530273,
|
||
|
|
"rewards/margins": 8.534375190734863,
|
||
|
|
"rewards/rejected": 0.861328125,
|
||
|
|
"step": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4272300469483568,
|
||
|
|
"grad_norm": 362.8472585991743,
|
||
|
|
"learning_rate": 3.1846635367762125e-07,
|
||
|
|
"logits/chosen": -4.131249904632568,
|
||
|
|
"logits/rejected": -4.028124809265137,
|
||
|
|
"logps/chosen": -587.4000244140625,
|
||
|
|
"logps/rejected": -677.7999877929688,
|
||
|
|
"loss": 3.9947,
|
||
|
|
"rewards/accuracies": 0.4921875,
|
||
|
|
"rewards/chosen": 9.056249618530273,
|
||
|
|
"rewards/margins": 7.450781345367432,
|
||
|
|
"rewards/rejected": 1.5958983898162842,
|
||
|
|
"step": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.431924882629108,
|
||
|
|
"grad_norm": 428.9226775378459,
|
||
|
|
"learning_rate": 3.158581116327595e-07,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -586.0,
|
||
|
|
"logps/rejected": -707.5999755859375,
|
||
|
|
"loss": 3.3051,
|
||
|
|
"rewards/accuracies": 0.515625,
|
||
|
|
"rewards/chosen": 9.362500190734863,
|
||
|
|
"rewards/margins": 8.137499809265137,
|
||
|
|
"rewards/rejected": 1.2257812023162842,
|
||
|
|
"step": 920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43661971830985913,
|
||
|
|
"grad_norm": 363.0151726164607,
|
||
|
|
"learning_rate": 3.1324986958789775e-07,
|
||
|
|
"logits/chosen": -4.171875,
|
||
|
|
"logits/rejected": -4.067187309265137,
|
||
|
|
"logps/chosen": -547.4000244140625,
|
||
|
|
"logps/rejected": -624.4000244140625,
|
||
|
|
"loss": 3.1743,
|
||
|
|
"rewards/accuracies": 0.4898437559604645,
|
||
|
|
"rewards/chosen": 9.231249809265137,
|
||
|
|
"rewards/margins": 6.953125,
|
||
|
|
"rewards/rejected": 2.28125,
|
||
|
|
"step": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4413145539906103,
|
||
|
|
"grad_norm": 300.234697004928,
|
||
|
|
"learning_rate": 3.1064162754303595e-07,
|
||
|
|
"logits/chosen": -4.140625,
|
||
|
|
"logits/rejected": -3.9921875,
|
||
|
|
"logps/chosen": -585.0,
|
||
|
|
"logps/rejected": -686.0,
|
||
|
|
"loss": 3.6555,
|
||
|
|
"rewards/accuracies": 0.512499988079071,
|
||
|
|
"rewards/chosen": 8.868749618530273,
|
||
|
|
"rewards/margins": 8.421875,
|
||
|
|
"rewards/rejected": 0.443359375,
|
||
|
|
"step": 940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4460093896713615,
|
||
|
|
"grad_norm": 307.2370185290697,
|
||
|
|
"learning_rate": 3.080333854981742e-07,
|
||
|
|
"logits/chosen": -4.131249904632568,
|
||
|
|
"logits/rejected": -4.017187595367432,
|
||
|
|
"logps/chosen": -573.0,
|
||
|
|
"logps/rejected": -626.0,
|
||
|
|
"loss": 3.6289,
|
||
|
|
"rewards/accuracies": 0.520312488079071,
|
||
|
|
"rewards/chosen": 9.081250190734863,
|
||
|
|
"rewards/margins": 6.7421875,
|
||
|
|
"rewards/rejected": 2.350781202316284,
|
||
|
|
"step": 950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4507042253521127,
|
||
|
|
"grad_norm": 315.1420548177452,
|
||
|
|
"learning_rate": 3.0542514345331246e-07,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.032812595367432,
|
||
|
|
"logps/chosen": -582.5999755859375,
|
||
|
|
"logps/rejected": -710.0,
|
||
|
|
"loss": 3.708,
|
||
|
|
"rewards/accuracies": 0.4828124940395355,
|
||
|
|
"rewards/chosen": 8.524999618530273,
|
||
|
|
"rewards/margins": 7.0859375,
|
||
|
|
"rewards/rejected": 1.443359375,
|
||
|
|
"step": 960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45539906103286387,
|
||
|
|
"grad_norm": 292.1652064232398,
|
||
|
|
"learning_rate": 3.0281690140845066e-07,
|
||
|
|
"logits/chosen": -4.103125095367432,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -611.2000122070312,
|
||
|
|
"logps/rejected": -673.4000244140625,
|
||
|
|
"loss": 3.8467,
|
||
|
|
"rewards/accuracies": 0.48828125,
|
||
|
|
"rewards/chosen": 8.893750190734863,
|
||
|
|
"rewards/margins": 6.938281059265137,
|
||
|
|
"rewards/rejected": 1.955664038658142,
|
||
|
|
"step": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.460093896713615,
|
||
|
|
"grad_norm": 305.7034108659845,
|
||
|
|
"learning_rate": 3.002086593635889e-07,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.035937309265137,
|
||
|
|
"logps/chosen": -576.0,
|
||
|
|
"logps/rejected": -676.7999877929688,
|
||
|
|
"loss": 2.8016,
|
||
|
|
"rewards/accuracies": 0.5132812261581421,
|
||
|
|
"rewards/chosen": 9.1875,
|
||
|
|
"rewards/margins": 9.334375381469727,
|
||
|
|
"rewards/rejected": -0.13544921576976776,
|
||
|
|
"step": 980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4647887323943662,
|
||
|
|
"grad_norm": 276.83804710650116,
|
||
|
|
"learning_rate": 2.9760041731872716e-07,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -4.089062690734863,
|
||
|
|
"logps/chosen": -538.5999755859375,
|
||
|
|
"logps/rejected": -639.2000122070312,
|
||
|
|
"loss": 2.9199,
|
||
|
|
"rewards/accuracies": 0.535937488079071,
|
||
|
|
"rewards/chosen": 9.056249618530273,
|
||
|
|
"rewards/margins": 8.396875381469727,
|
||
|
|
"rewards/rejected": 0.6742187738418579,
|
||
|
|
"step": 990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4694835680751174,
|
||
|
|
"grad_norm": 994.2333919263641,
|
||
|
|
"learning_rate": 2.9499217527386536e-07,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.0859375,
|
||
|
|
"logps/chosen": -563.7999877929688,
|
||
|
|
"logps/rejected": -655.2000122070312,
|
||
|
|
"loss": 2.8529,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 9.096875190734863,
|
||
|
|
"rewards/margins": 8.024999618530273,
|
||
|
|
"rewards/rejected": 1.073144555091858,
|
||
|
|
"step": 1000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47417840375586856,
|
||
|
|
"grad_norm": 263.164923515219,
|
||
|
|
"learning_rate": 2.923839332290036e-07,
|
||
|
|
"logits/chosen": -4.123437404632568,
|
||
|
|
"logits/rejected": -4.012499809265137,
|
||
|
|
"logps/chosen": -560.0,
|
||
|
|
"logps/rejected": -667.2000122070312,
|
||
|
|
"loss": 3.148,
|
||
|
|
"rewards/accuracies": 0.49921876192092896,
|
||
|
|
"rewards/chosen": 8.240625381469727,
|
||
|
|
"rewards/margins": 7.603125095367432,
|
||
|
|
"rewards/rejected": 0.6279296875,
|
||
|
|
"step": 1010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4788732394366197,
|
||
|
|
"grad_norm": 318.0813412687802,
|
||
|
|
"learning_rate": 2.8977569118414187e-07,
|
||
|
|
"logits/chosen": -4.25,
|
||
|
|
"logits/rejected": -4.096875190734863,
|
||
|
|
"logps/chosen": -549.4000244140625,
|
||
|
|
"logps/rejected": -635.7999877929688,
|
||
|
|
"loss": 3.1516,
|
||
|
|
"rewards/accuracies": 0.526562511920929,
|
||
|
|
"rewards/chosen": 8.353124618530273,
|
||
|
|
"rewards/margins": 7.503125190734863,
|
||
|
|
"rewards/rejected": 0.833789050579071,
|
||
|
|
"step": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4835680751173709,
|
||
|
|
"grad_norm": 277.03569972426686,
|
||
|
|
"learning_rate": 2.8716744913928007e-07,
|
||
|
|
"logits/chosen": -4.214062690734863,
|
||
|
|
"logits/rejected": -4.106249809265137,
|
||
|
|
"logps/chosen": -572.7999877929688,
|
||
|
|
"logps/rejected": -632.2000122070312,
|
||
|
|
"loss": 2.8036,
|
||
|
|
"rewards/accuracies": 0.5296875238418579,
|
||
|
|
"rewards/chosen": 9.274999618530273,
|
||
|
|
"rewards/margins": 8.546875,
|
||
|
|
"rewards/rejected": 0.735546886920929,
|
||
|
|
"step": 1030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48826291079812206,
|
||
|
|
"grad_norm": 211.8844280559408,
|
||
|
|
"learning_rate": 2.845592070944183e-07,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -570.5999755859375,
|
||
|
|
"logps/rejected": -683.5999755859375,
|
||
|
|
"loss": 3.0776,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 9.168749809265137,
|
||
|
|
"rewards/margins": 7.628125190734863,
|
||
|
|
"rewards/rejected": 1.5320312976837158,
|
||
|
|
"step": 1040
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49295774647887325,
|
||
|
|
"grad_norm": 1084.9538807947201,
|
||
|
|
"learning_rate": 2.819509650495566e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.025000095367432,
|
||
|
|
"logps/chosen": -562.0,
|
||
|
|
"logps/rejected": -629.0,
|
||
|
|
"loss": 3.2087,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 8.762499809265137,
|
||
|
|
"rewards/margins": 7.685937404632568,
|
||
|
|
"rewards/rejected": 1.0810546875,
|
||
|
|
"step": 1050
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49765258215962443,
|
||
|
|
"grad_norm": 307.62081588002684,
|
||
|
|
"learning_rate": 2.7934272300469483e-07,
|
||
|
|
"logits/chosen": -4.134375095367432,
|
||
|
|
"logits/rejected": -4.035937309265137,
|
||
|
|
"logps/chosen": -598.0,
|
||
|
|
"logps/rejected": -641.0,
|
||
|
|
"loss": 3.1239,
|
||
|
|
"rewards/accuracies": 0.4976562559604645,
|
||
|
|
"rewards/chosen": 9.178125381469727,
|
||
|
|
"rewards/margins": 7.729687690734863,
|
||
|
|
"rewards/rejected": 1.441308617591858,
|
||
|
|
"step": 1060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5023474178403756,
|
||
|
|
"grad_norm": 291.91293524568334,
|
||
|
|
"learning_rate": 2.7673448095983303e-07,
|
||
|
|
"logits/chosen": -4.1484375,
|
||
|
|
"logits/rejected": -4.057812690734863,
|
||
|
|
"logps/chosen": -578.0,
|
||
|
|
"logps/rejected": -640.4000244140625,
|
||
|
|
"loss": 3.6736,
|
||
|
|
"rewards/accuracies": 0.4976562559604645,
|
||
|
|
"rewards/chosen": 8.696874618530273,
|
||
|
|
"rewards/margins": 7.473437309265137,
|
||
|
|
"rewards/rejected": 1.2189452648162842,
|
||
|
|
"step": 1070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5070422535211268,
|
||
|
|
"grad_norm": 1044.1991089251253,
|
||
|
|
"learning_rate": 2.741262389149713e-07,
|
||
|
|
"logits/chosen": -4.154687404632568,
|
||
|
|
"logits/rejected": -3.9671874046325684,
|
||
|
|
"logps/chosen": -578.0,
|
||
|
|
"logps/rejected": -662.0,
|
||
|
|
"loss": 3.184,
|
||
|
|
"rewards/accuracies": 0.535937488079071,
|
||
|
|
"rewards/chosen": 8.759374618530273,
|
||
|
|
"rewards/margins": 8.140625,
|
||
|
|
"rewards/rejected": 0.632031261920929,
|
||
|
|
"step": 1080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5117370892018779,
|
||
|
|
"grad_norm": 268.621353868072,
|
||
|
|
"learning_rate": 2.7151799687010953e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.029687404632568,
|
||
|
|
"logps/chosen": -584.2000122070312,
|
||
|
|
"logps/rejected": -665.5999755859375,
|
||
|
|
"loss": 3.0666,
|
||
|
|
"rewards/accuracies": 0.530468761920929,
|
||
|
|
"rewards/chosen": 8.821874618530273,
|
||
|
|
"rewards/margins": 9.159375190734863,
|
||
|
|
"rewards/rejected": -0.3453125059604645,
|
||
|
|
"step": 1090
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5164319248826291,
|
||
|
|
"grad_norm": 275.9837003696594,
|
||
|
|
"learning_rate": 2.6890975482524773e-07,
|
||
|
|
"logits/chosen": -4.082812309265137,
|
||
|
|
"logits/rejected": -3.9781250953674316,
|
||
|
|
"logps/chosen": -603.0,
|
||
|
|
"logps/rejected": -655.2000122070312,
|
||
|
|
"loss": 3.5672,
|
||
|
|
"rewards/accuracies": 0.53125,
|
||
|
|
"rewards/chosen": 9.087499618530273,
|
||
|
|
"rewards/margins": 8.928125381469727,
|
||
|
|
"rewards/rejected": 0.14448241889476776,
|
||
|
|
"step": 1100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5211267605633803,
|
||
|
|
"grad_norm": 331.47009226887195,
|
||
|
|
"learning_rate": 2.66301512780386e-07,
|
||
|
|
"logits/chosen": -4.153124809265137,
|
||
|
|
"logits/rejected": -3.9859375953674316,
|
||
|
|
"logps/chosen": -552.2000122070312,
|
||
|
|
"logps/rejected": -645.5999755859375,
|
||
|
|
"loss": 3.3857,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 8.665624618530273,
|
||
|
|
"rewards/margins": 8.559374809265137,
|
||
|
|
"rewards/rejected": 0.111328125,
|
||
|
|
"step": 1110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5258215962441315,
|
||
|
|
"grad_norm": 317.89046309154855,
|
||
|
|
"learning_rate": 2.6369327073552424e-07,
|
||
|
|
"logits/chosen": -4.203125,
|
||
|
|
"logits/rejected": -4.082812309265137,
|
||
|
|
"logps/chosen": -560.5999755859375,
|
||
|
|
"logps/rejected": -618.7999877929688,
|
||
|
|
"loss": 2.9995,
|
||
|
|
"rewards/accuracies": 0.5171874761581421,
|
||
|
|
"rewards/chosen": 8.984375,
|
||
|
|
"rewards/margins": 7.115624904632568,
|
||
|
|
"rewards/rejected": 1.881250023841858,
|
||
|
|
"step": 1120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5305164319248826,
|
||
|
|
"grad_norm": 309.94253870846245,
|
||
|
|
"learning_rate": 2.6108502869066244e-07,
|
||
|
|
"logits/chosen": -4.162499904632568,
|
||
|
|
"logits/rejected": -4.0546875,
|
||
|
|
"logps/chosen": -593.5999755859375,
|
||
|
|
"logps/rejected": -688.7999877929688,
|
||
|
|
"loss": 3.2947,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 9.215624809265137,
|
||
|
|
"rewards/margins": 8.375,
|
||
|
|
"rewards/rejected": 0.826953113079071,
|
||
|
|
"step": 1130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5352112676056338,
|
||
|
|
"grad_norm": 366.55680036700505,
|
||
|
|
"learning_rate": 2.584767866458007e-07,
|
||
|
|
"logits/chosen": -4.217187404632568,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -568.4000244140625,
|
||
|
|
"logps/rejected": -646.7999877929688,
|
||
|
|
"loss": 3.5653,
|
||
|
|
"rewards/accuracies": 0.500781238079071,
|
||
|
|
"rewards/chosen": 9.065625190734863,
|
||
|
|
"rewards/margins": 7.668749809265137,
|
||
|
|
"rewards/rejected": 1.3982422351837158,
|
||
|
|
"step": 1140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.539906103286385,
|
||
|
|
"grad_norm": 272.5639470143928,
|
||
|
|
"learning_rate": 2.5586854460093895e-07,
|
||
|
|
"logits/chosen": -4.146874904632568,
|
||
|
|
"logits/rejected": -4.025000095367432,
|
||
|
|
"logps/chosen": -578.5999755859375,
|
||
|
|
"logps/rejected": -671.2000122070312,
|
||
|
|
"loss": 3.3055,
|
||
|
|
"rewards/accuracies": 0.534375011920929,
|
||
|
|
"rewards/chosen": 8.943750381469727,
|
||
|
|
"rewards/margins": 8.596875190734863,
|
||
|
|
"rewards/rejected": 0.3472656309604645,
|
||
|
|
"step": 1150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5446009389671361,
|
||
|
|
"grad_norm": 249.74677763574044,
|
||
|
|
"learning_rate": 2.5326030255607715e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.057812690734863,
|
||
|
|
"logps/chosen": -584.0,
|
||
|
|
"logps/rejected": -654.7999877929688,
|
||
|
|
"loss": 3.5447,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 9.278124809265137,
|
||
|
|
"rewards/margins": 8.045312881469727,
|
||
|
|
"rewards/rejected": 1.235937476158142,
|
||
|
|
"step": 1160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5492957746478874,
|
||
|
|
"grad_norm": 259.21404711583165,
|
||
|
|
"learning_rate": 2.506520605112154e-07,
|
||
|
|
"logits/chosen": -4.1640625,
|
||
|
|
"logits/rejected": -4.081250190734863,
|
||
|
|
"logps/chosen": -576.2000122070312,
|
||
|
|
"logps/rejected": -656.0,
|
||
|
|
"loss": 3.4939,
|
||
|
|
"rewards/accuracies": 0.4749999940395355,
|
||
|
|
"rewards/chosen": 8.59375,
|
||
|
|
"rewards/margins": 6.532812595367432,
|
||
|
|
"rewards/rejected": 2.0582032203674316,
|
||
|
|
"step": 1170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5539906103286385,
|
||
|
|
"grad_norm": 262.41014591871937,
|
||
|
|
"learning_rate": 2.4804381846635365e-07,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.109375,
|
||
|
|
"logps/chosen": -593.4000244140625,
|
||
|
|
"logps/rejected": -650.5999755859375,
|
||
|
|
"loss": 3.4565,
|
||
|
|
"rewards/accuracies": 0.514843761920929,
|
||
|
|
"rewards/chosen": 8.637499809265137,
|
||
|
|
"rewards/margins": 7.8359375,
|
||
|
|
"rewards/rejected": 0.8057616949081421,
|
||
|
|
"step": 1180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5586854460093896,
|
||
|
|
"grad_norm": 285.70918665861836,
|
||
|
|
"learning_rate": 2.454355764214919e-07,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.089062690734863,
|
||
|
|
"logps/chosen": -554.5999755859375,
|
||
|
|
"logps/rejected": -676.5999755859375,
|
||
|
|
"loss": 3.6881,
|
||
|
|
"rewards/accuracies": 0.5015624761581421,
|
||
|
|
"rewards/chosen": 8.412500381469727,
|
||
|
|
"rewards/margins": 7.448437690734863,
|
||
|
|
"rewards/rejected": 0.95751953125,
|
||
|
|
"step": 1190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5633802816901409,
|
||
|
|
"grad_norm": 2395.3317758278217,
|
||
|
|
"learning_rate": 2.4282733437663016e-07,
|
||
|
|
"logits/chosen": -4.190625190734863,
|
||
|
|
"logits/rejected": -4.140625,
|
||
|
|
"logps/chosen": -583.4000244140625,
|
||
|
|
"logps/rejected": -633.2000122070312,
|
||
|
|
"loss": 3.8918,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 8.962499618530273,
|
||
|
|
"rewards/margins": 6.510937690734863,
|
||
|
|
"rewards/rejected": 2.446093797683716,
|
||
|
|
"step": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.568075117370892,
|
||
|
|
"grad_norm": 291.79485364660576,
|
||
|
|
"learning_rate": 2.4021909233176836e-07,
|
||
|
|
"logits/chosen": -4.153124809265137,
|
||
|
|
"logits/rejected": -4.067187309265137,
|
||
|
|
"logps/chosen": -548.5999755859375,
|
||
|
|
"logps/rejected": -653.2000122070312,
|
||
|
|
"loss": 3.6139,
|
||
|
|
"rewards/accuracies": 0.503125011920929,
|
||
|
|
"rewards/chosen": 8.703125,
|
||
|
|
"rewards/margins": 6.978906154632568,
|
||
|
|
"rewards/rejected": 1.7251465320587158,
|
||
|
|
"step": 1210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5727699530516432,
|
||
|
|
"grad_norm": 293.112373801934,
|
||
|
|
"learning_rate": 2.376108502869066e-07,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.125,
|
||
|
|
"logps/chosen": -585.7999877929688,
|
||
|
|
"logps/rejected": -678.4000244140625,
|
||
|
|
"loss": 3.2837,
|
||
|
|
"rewards/accuracies": 0.514843761920929,
|
||
|
|
"rewards/chosen": 8.96875,
|
||
|
|
"rewards/margins": 7.993750095367432,
|
||
|
|
"rewards/rejected": 0.983593761920929,
|
||
|
|
"step": 1220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5774647887323944,
|
||
|
|
"grad_norm": 409.90103938295374,
|
||
|
|
"learning_rate": 2.3500260824204484e-07,
|
||
|
|
"logits/chosen": -4.129687309265137,
|
||
|
|
"logits/rejected": -4.009375095367432,
|
||
|
|
"logps/chosen": -613.2000122070312,
|
||
|
|
"logps/rejected": -688.0,
|
||
|
|
"loss": 3.2107,
|
||
|
|
"rewards/accuracies": 0.5078125,
|
||
|
|
"rewards/chosen": 8.831250190734863,
|
||
|
|
"rewards/margins": 9.434374809265137,
|
||
|
|
"rewards/rejected": -0.60546875,
|
||
|
|
"step": 1230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5821596244131455,
|
||
|
|
"grad_norm": 262.57777744316803,
|
||
|
|
"learning_rate": 2.323943661971831e-07,
|
||
|
|
"logits/chosen": -4.221875190734863,
|
||
|
|
"logits/rejected": -4.092187404632568,
|
||
|
|
"logps/chosen": -573.4000244140625,
|
||
|
|
"logps/rejected": -648.7999877929688,
|
||
|
|
"loss": 3.164,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 9.140625,
|
||
|
|
"rewards/margins": 8.3203125,
|
||
|
|
"rewards/rejected": 0.8245605230331421,
|
||
|
|
"step": 1240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5868544600938967,
|
||
|
|
"grad_norm": 267.13392657233817,
|
||
|
|
"learning_rate": 2.2978612415232132e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -605.5999755859375,
|
||
|
|
"logps/rejected": -678.0,
|
||
|
|
"loss": 3.06,
|
||
|
|
"rewards/accuracies": 0.5101562738418579,
|
||
|
|
"rewards/chosen": 9.199999809265137,
|
||
|
|
"rewards/margins": 8.590624809265137,
|
||
|
|
"rewards/rejected": 0.608593761920929,
|
||
|
|
"step": 1250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5915492957746479,
|
||
|
|
"grad_norm": 365.075647857907,
|
||
|
|
"learning_rate": 2.2717788210745957e-07,
|
||
|
|
"logits/chosen": -4.171875,
|
||
|
|
"logits/rejected": -4.065625190734863,
|
||
|
|
"logps/chosen": -568.2000122070312,
|
||
|
|
"logps/rejected": -661.5999755859375,
|
||
|
|
"loss": 3.8677,
|
||
|
|
"rewards/accuracies": 0.512499988079071,
|
||
|
|
"rewards/chosen": 8.6328125,
|
||
|
|
"rewards/margins": 8.237500190734863,
|
||
|
|
"rewards/rejected": 0.39375001192092896,
|
||
|
|
"step": 1260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.596244131455399,
|
||
|
|
"grad_norm": 258.1348728222688,
|
||
|
|
"learning_rate": 2.245696400625978e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.073437690734863,
|
||
|
|
"logps/chosen": -597.4000244140625,
|
||
|
|
"logps/rejected": -670.4000244140625,
|
||
|
|
"loss": 3.3881,
|
||
|
|
"rewards/accuracies": 0.49687498807907104,
|
||
|
|
"rewards/chosen": 8.578125,
|
||
|
|
"rewards/margins": 8.340624809265137,
|
||
|
|
"rewards/rejected": 0.23769530653953552,
|
||
|
|
"step": 1270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6009389671361502,
|
||
|
|
"grad_norm": 253.69792112851948,
|
||
|
|
"learning_rate": 2.2196139801773602e-07,
|
||
|
|
"logits/chosen": -4.153124809265137,
|
||
|
|
"logits/rejected": -4.0078125,
|
||
|
|
"logps/chosen": -562.2000122070312,
|
||
|
|
"logps/rejected": -628.7999877929688,
|
||
|
|
"loss": 3.3145,
|
||
|
|
"rewards/accuracies": 0.48359376192092896,
|
||
|
|
"rewards/chosen": 8.212499618530273,
|
||
|
|
"rewards/margins": 7.349999904632568,
|
||
|
|
"rewards/rejected": 0.857617199420929,
|
||
|
|
"step": 1280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6056338028169014,
|
||
|
|
"grad_norm": 335.2679465369597,
|
||
|
|
"learning_rate": 2.1935315597287428e-07,
|
||
|
|
"logits/chosen": -4.140625,
|
||
|
|
"logits/rejected": -4.045312404632568,
|
||
|
|
"logps/chosen": -582.0,
|
||
|
|
"logps/rejected": -628.7999877929688,
|
||
|
|
"loss": 3.2686,
|
||
|
|
"rewards/accuracies": 0.53125,
|
||
|
|
"rewards/chosen": 8.487500190734863,
|
||
|
|
"rewards/margins": 7.296875,
|
||
|
|
"rewards/rejected": 1.1824219226837158,
|
||
|
|
"step": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6103286384976526,
|
||
|
|
"grad_norm": 231.44130498480763,
|
||
|
|
"learning_rate": 2.167449139280125e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -588.2000122070312,
|
||
|
|
"logps/rejected": -672.0,
|
||
|
|
"loss": 2.9305,
|
||
|
|
"rewards/accuracies": 0.518750011920929,
|
||
|
|
"rewards/chosen": 9.725000381469727,
|
||
|
|
"rewards/margins": 9.03125,
|
||
|
|
"rewards/rejected": 0.7081054449081421,
|
||
|
|
"step": 1300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6150234741784038,
|
||
|
|
"grad_norm": 266.1692522688036,
|
||
|
|
"learning_rate": 2.1413667188315073e-07,
|
||
|
|
"logits/chosen": -4.237500190734863,
|
||
|
|
"logits/rejected": -4.192187309265137,
|
||
|
|
"logps/chosen": -556.7999877929688,
|
||
|
|
"logps/rejected": -617.2000122070312,
|
||
|
|
"loss": 3.648,
|
||
|
|
"rewards/accuracies": 0.5078125,
|
||
|
|
"rewards/chosen": 8.921875,
|
||
|
|
"rewards/margins": 8.4375,
|
||
|
|
"rewards/rejected": 0.472412109375,
|
||
|
|
"step": 1310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6197183098591549,
|
||
|
|
"grad_norm": 391.06496825559407,
|
||
|
|
"learning_rate": 2.1152842983828898e-07,
|
||
|
|
"logits/chosen": -4.126562595367432,
|
||
|
|
"logits/rejected": -4.03125,
|
||
|
|
"logps/chosen": -588.2000122070312,
|
||
|
|
"logps/rejected": -645.2000122070312,
|
||
|
|
"loss": 4.0506,
|
||
|
|
"rewards/accuracies": 0.5101562738418579,
|
||
|
|
"rewards/chosen": 8.337499618530273,
|
||
|
|
"rewards/margins": 7.263281345367432,
|
||
|
|
"rewards/rejected": 1.096093773841858,
|
||
|
|
"step": 1320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6244131455399061,
|
||
|
|
"grad_norm": 278.7704596502006,
|
||
|
|
"learning_rate": 2.089201877934272e-07,
|
||
|
|
"logits/chosen": -4.128125190734863,
|
||
|
|
"logits/rejected": -4.021874904632568,
|
||
|
|
"logps/chosen": -601.5999755859375,
|
||
|
|
"logps/rejected": -694.7999877929688,
|
||
|
|
"loss": 3.4997,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 8.03125,
|
||
|
|
"rewards/margins": 8.348437309265137,
|
||
|
|
"rewards/rejected": -0.3167968690395355,
|
||
|
|
"step": 1330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6291079812206573,
|
||
|
|
"grad_norm": 250.9766144500523,
|
||
|
|
"learning_rate": 2.0631194574856543e-07,
|
||
|
|
"logits/chosen": -4.224999904632568,
|
||
|
|
"logits/rejected": -4.146874904632568,
|
||
|
|
"logps/chosen": -557.0,
|
||
|
|
"logps/rejected": -599.7999877929688,
|
||
|
|
"loss": 3.2297,
|
||
|
|
"rewards/accuracies": 0.50390625,
|
||
|
|
"rewards/chosen": 8.934374809265137,
|
||
|
|
"rewards/margins": 6.598437309265137,
|
||
|
|
"rewards/rejected": 2.334765672683716,
|
||
|
|
"step": 1340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6338028169014085,
|
||
|
|
"grad_norm": 286.2932148393472,
|
||
|
|
"learning_rate": 2.0370370370370369e-07,
|
||
|
|
"logits/chosen": -4.114062309265137,
|
||
|
|
"logits/rejected": -4.025000095367432,
|
||
|
|
"logps/chosen": -581.5999755859375,
|
||
|
|
"logps/rejected": -696.0,
|
||
|
|
"loss": 3.2912,
|
||
|
|
"rewards/accuracies": 0.5289062261581421,
|
||
|
|
"rewards/chosen": 8.237500190734863,
|
||
|
|
"rewards/margins": 7.896874904632568,
|
||
|
|
"rewards/rejected": 0.33769530057907104,
|
||
|
|
"step": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6384976525821596,
|
||
|
|
"grad_norm": 254.91504224414876,
|
||
|
|
"learning_rate": 2.010954616588419e-07,
|
||
|
|
"logits/chosen": -4.196875095367432,
|
||
|
|
"logits/rejected": -4.03125,
|
||
|
|
"logps/chosen": -569.7999877929688,
|
||
|
|
"logps/rejected": -640.7999877929688,
|
||
|
|
"loss": 3.3697,
|
||
|
|
"rewards/accuracies": 0.4906249940395355,
|
||
|
|
"rewards/chosen": 8.440625190734863,
|
||
|
|
"rewards/margins": 7.165625095367432,
|
||
|
|
"rewards/rejected": 1.27001953125,
|
||
|
|
"step": 1360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6431924882629108,
|
||
|
|
"grad_norm": 414.74835013532675,
|
||
|
|
"learning_rate": 1.9848721961398017e-07,
|
||
|
|
"logits/chosen": -4.209374904632568,
|
||
|
|
"logits/rejected": -4.139062404632568,
|
||
|
|
"logps/chosen": -578.4000244140625,
|
||
|
|
"logps/rejected": -629.5999755859375,
|
||
|
|
"loss": 3.3279,
|
||
|
|
"rewards/accuracies": 0.5015624761581421,
|
||
|
|
"rewards/chosen": 9.503125190734863,
|
||
|
|
"rewards/margins": 8.378125190734863,
|
||
|
|
"rewards/rejected": 1.1222655773162842,
|
||
|
|
"step": 1370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.647887323943662,
|
||
|
|
"grad_norm": 283.32788354346286,
|
||
|
|
"learning_rate": 1.958789775691184e-07,
|
||
|
|
"logits/chosen": -4.165625095367432,
|
||
|
|
"logits/rejected": -4.026562690734863,
|
||
|
|
"logps/chosen": -563.5999755859375,
|
||
|
|
"logps/rejected": -648.4000244140625,
|
||
|
|
"loss": 3.8283,
|
||
|
|
"rewards/accuracies": 0.51953125,
|
||
|
|
"rewards/chosen": 8.606249809265137,
|
||
|
|
"rewards/margins": 8.185937881469727,
|
||
|
|
"rewards/rejected": 0.4126953184604645,
|
||
|
|
"step": 1380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6525821596244131,
|
||
|
|
"grad_norm": 231.57307044550026,
|
||
|
|
"learning_rate": 1.9327073552425662e-07,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -4.050000190734863,
|
||
|
|
"logps/chosen": -555.5999755859375,
|
||
|
|
"logps/rejected": -622.2000122070312,
|
||
|
|
"loss": 3.011,
|
||
|
|
"rewards/accuracies": 0.5093749761581421,
|
||
|
|
"rewards/chosen": 9.303125381469727,
|
||
|
|
"rewards/margins": 9.512499809265137,
|
||
|
|
"rewards/rejected": -0.21181640028953552,
|
||
|
|
"step": 1390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6572769953051644,
|
||
|
|
"grad_norm": 245.7010459859998,
|
||
|
|
"learning_rate": 1.906624934793949e-07,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.084374904632568,
|
||
|
|
"logps/chosen": -575.0,
|
||
|
|
"logps/rejected": -676.0,
|
||
|
|
"loss": 3.4527,
|
||
|
|
"rewards/accuracies": 0.538281261920929,
|
||
|
|
"rewards/chosen": 9.899999618530273,
|
||
|
|
"rewards/margins": 9.184374809265137,
|
||
|
|
"rewards/rejected": 0.704882800579071,
|
||
|
|
"step": 1400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6619718309859155,
|
||
|
|
"grad_norm": 292.70988132691616,
|
||
|
|
"learning_rate": 1.8805425143453312e-07,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -3.9703125953674316,
|
||
|
|
"logps/chosen": -553.5999755859375,
|
||
|
|
"logps/rejected": -707.5999755859375,
|
||
|
|
"loss": 3.2431,
|
||
|
|
"rewards/accuracies": 0.512499988079071,
|
||
|
|
"rewards/chosen": 8.978124618530273,
|
||
|
|
"rewards/margins": 7.946875095367432,
|
||
|
|
"rewards/rejected": 1.0339844226837158,
|
||
|
|
"step": 1410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6666666666666666,
|
||
|
|
"grad_norm": 289.38318966753934,
|
||
|
|
"learning_rate": 1.8544600938967138e-07,
|
||
|
|
"logits/chosen": -4.167187690734863,
|
||
|
|
"logits/rejected": -4.0625,
|
||
|
|
"logps/chosen": -571.7999877929688,
|
||
|
|
"logps/rejected": -632.2000122070312,
|
||
|
|
"loss": 3.3539,
|
||
|
|
"rewards/accuracies": 0.515625,
|
||
|
|
"rewards/chosen": 8.84375,
|
||
|
|
"rewards/margins": 7.934374809265137,
|
||
|
|
"rewards/rejected": 0.91796875,
|
||
|
|
"step": 1420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6713615023474179,
|
||
|
|
"grad_norm": 222.43533121990208,
|
||
|
|
"learning_rate": 1.828377673448096e-07,
|
||
|
|
"logits/chosen": -4.150000095367432,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -607.0,
|
||
|
|
"logps/rejected": -652.7999877929688,
|
||
|
|
"loss": 3.7237,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 8.765625,
|
||
|
|
"rewards/margins": 8.543749809265137,
|
||
|
|
"rewards/rejected": 0.21718749403953552,
|
||
|
|
"step": 1430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.676056338028169,
|
||
|
|
"grad_norm": 216.86086222140142,
|
||
|
|
"learning_rate": 1.8022952529994783e-07,
|
||
|
|
"logits/chosen": -4.184374809265137,
|
||
|
|
"logits/rejected": -4.057812690734863,
|
||
|
|
"logps/chosen": -583.7999877929688,
|
||
|
|
"logps/rejected": -644.4000244140625,
|
||
|
|
"loss": 2.99,
|
||
|
|
"rewards/accuracies": 0.5210937261581421,
|
||
|
|
"rewards/chosen": 8.912500381469727,
|
||
|
|
"rewards/margins": 9.4921875,
|
||
|
|
"rewards/rejected": -0.573046863079071,
|
||
|
|
"step": 1440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6807511737089202,
|
||
|
|
"grad_norm": 4657.860857885418,
|
||
|
|
"learning_rate": 1.7762128325508608e-07,
|
||
|
|
"logits/chosen": -4.199999809265137,
|
||
|
|
"logits/rejected": -4.095312595367432,
|
||
|
|
"logps/chosen": -592.7999877929688,
|
||
|
|
"logps/rejected": -685.2000122070312,
|
||
|
|
"loss": 3.8062,
|
||
|
|
"rewards/accuracies": 0.5101562738418579,
|
||
|
|
"rewards/chosen": 9.371874809265137,
|
||
|
|
"rewards/margins": 8.243749618530273,
|
||
|
|
"rewards/rejected": 1.1334960460662842,
|
||
|
|
"step": 1450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6854460093896714,
|
||
|
|
"grad_norm": 304.75248514169795,
|
||
|
|
"learning_rate": 1.750130412102243e-07,
|
||
|
|
"logits/chosen": -4.087500095367432,
|
||
|
|
"logits/rejected": -4.010937690734863,
|
||
|
|
"logps/chosen": -622.4000244140625,
|
||
|
|
"logps/rejected": -668.4000244140625,
|
||
|
|
"loss": 3.8537,
|
||
|
|
"rewards/accuracies": 0.510937511920929,
|
||
|
|
"rewards/chosen": 9.203125,
|
||
|
|
"rewards/margins": 8.846875190734863,
|
||
|
|
"rewards/rejected": 0.35283201932907104,
|
||
|
|
"step": 1460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6901408450704225,
|
||
|
|
"grad_norm": 250.6636699927517,
|
||
|
|
"learning_rate": 1.7240479916536254e-07,
|
||
|
|
"logits/chosen": -4.084374904632568,
|
||
|
|
"logits/rejected": -4.014062404632568,
|
||
|
|
"logps/chosen": -561.4000244140625,
|
||
|
|
"logps/rejected": -684.0,
|
||
|
|
"loss": 3.4143,
|
||
|
|
"rewards/accuracies": 0.4937500059604645,
|
||
|
|
"rewards/chosen": 8.928125381469727,
|
||
|
|
"rewards/margins": 8.459375381469727,
|
||
|
|
"rewards/rejected": 0.4691406190395355,
|
||
|
|
"step": 1470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6948356807511737,
|
||
|
|
"grad_norm": 248.50954893198326,
|
||
|
|
"learning_rate": 1.697965571205008e-07,
|
||
|
|
"logits/chosen": -4.212500095367432,
|
||
|
|
"logits/rejected": -4.114062309265137,
|
||
|
|
"logps/chosen": -569.5999755859375,
|
||
|
|
"logps/rejected": -625.0,
|
||
|
|
"loss": 3.9249,
|
||
|
|
"rewards/accuracies": 0.5335937738418579,
|
||
|
|
"rewards/chosen": 8.978124618530273,
|
||
|
|
"rewards/margins": 7.28125,
|
||
|
|
"rewards/rejected": 1.696679711341858,
|
||
|
|
"step": 1480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6995305164319249,
|
||
|
|
"grad_norm": 259.50616073958525,
|
||
|
|
"learning_rate": 1.6718831507563902e-07,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.120312690734863,
|
||
|
|
"logps/chosen": -576.4000244140625,
|
||
|
|
"logps/rejected": -615.4000244140625,
|
||
|
|
"loss": 3.0162,
|
||
|
|
"rewards/accuracies": 0.5,
|
||
|
|
"rewards/chosen": 9.290624618530273,
|
||
|
|
"rewards/margins": 6.721875190734863,
|
||
|
|
"rewards/rejected": 2.567578077316284,
|
||
|
|
"step": 1490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.704225352112676,
|
||
|
|
"grad_norm": 237.4618737278568,
|
||
|
|
"learning_rate": 1.6458007303077727e-07,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -4.071875095367432,
|
||
|
|
"logps/chosen": -564.0,
|
||
|
|
"logps/rejected": -628.4000244140625,
|
||
|
|
"loss": 3.7618,
|
||
|
|
"rewards/accuracies": 0.5140625238418579,
|
||
|
|
"rewards/chosen": 8.762499809265137,
|
||
|
|
"rewards/margins": 7.009375095367432,
|
||
|
|
"rewards/rejected": 1.7527344226837158,
|
||
|
|
"step": 1500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7089201877934272,
|
||
|
|
"grad_norm": 304.6021709201427,
|
||
|
|
"learning_rate": 1.619718309859155e-07,
|
||
|
|
"logits/chosen": -4.176562309265137,
|
||
|
|
"logits/rejected": -4.03125,
|
||
|
|
"logps/chosen": -583.4000244140625,
|
||
|
|
"logps/rejected": -673.5999755859375,
|
||
|
|
"loss": 3.2058,
|
||
|
|
"rewards/accuracies": 0.5375000238418579,
|
||
|
|
"rewards/chosen": 9.234375,
|
||
|
|
"rewards/margins": 8.543749809265137,
|
||
|
|
"rewards/rejected": 0.6929687261581421,
|
||
|
|
"step": 1510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7136150234741784,
|
||
|
|
"grad_norm": 288.23056226017525,
|
||
|
|
"learning_rate": 1.5936358894105372e-07,
|
||
|
|
"logits/chosen": -4.196875095367432,
|
||
|
|
"logits/rejected": -4.078125,
|
||
|
|
"logps/chosen": -584.5999755859375,
|
||
|
|
"logps/rejected": -664.0,
|
||
|
|
"loss": 3.2153,
|
||
|
|
"rewards/accuracies": 0.5210937261581421,
|
||
|
|
"rewards/chosen": 9.600000381469727,
|
||
|
|
"rewards/margins": 8.106249809265137,
|
||
|
|
"rewards/rejected": 1.49609375,
|
||
|
|
"step": 1520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7183098591549296,
|
||
|
|
"grad_norm": 263.20042200814987,
|
||
|
|
"learning_rate": 1.5675534689619197e-07,
|
||
|
|
"logits/chosen": -4.184374809265137,
|
||
|
|
"logits/rejected": -4.112500190734863,
|
||
|
|
"logps/chosen": -583.5999755859375,
|
||
|
|
"logps/rejected": -611.5999755859375,
|
||
|
|
"loss": 3.699,
|
||
|
|
"rewards/accuracies": 0.51171875,
|
||
|
|
"rewards/chosen": 8.365625381469727,
|
||
|
|
"rewards/margins": 6.4921875,
|
||
|
|
"rewards/rejected": 1.866796851158142,
|
||
|
|
"step": 1530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7230046948356808,
|
||
|
|
"grad_norm": 256.34527394439255,
|
||
|
|
"learning_rate": 1.541471048513302e-07,
|
||
|
|
"logits/chosen": -4.1484375,
|
||
|
|
"logits/rejected": -4.0625,
|
||
|
|
"logps/chosen": -584.0,
|
||
|
|
"logps/rejected": -654.4000244140625,
|
||
|
|
"loss": 3.3438,
|
||
|
|
"rewards/accuracies": 0.5015624761581421,
|
||
|
|
"rewards/chosen": 9.071874618530273,
|
||
|
|
"rewards/margins": 8.068750381469727,
|
||
|
|
"rewards/rejected": 1.000878930091858,
|
||
|
|
"step": 1540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7276995305164319,
|
||
|
|
"grad_norm": 247.38632157274768,
|
||
|
|
"learning_rate": 1.5153886280646843e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.109375,
|
||
|
|
"logps/chosen": -541.0,
|
||
|
|
"logps/rejected": -602.5999755859375,
|
||
|
|
"loss": 2.9387,
|
||
|
|
"rewards/accuracies": 0.5453125238418579,
|
||
|
|
"rewards/chosen": 9.828125,
|
||
|
|
"rewards/margins": 8.837499618530273,
|
||
|
|
"rewards/rejected": 0.986523449420929,
|
||
|
|
"step": 1550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7323943661971831,
|
||
|
|
"grad_norm": 249.28526659037962,
|
||
|
|
"learning_rate": 1.4893062076160668e-07,
|
||
|
|
"logits/chosen": -4.184374809265137,
|
||
|
|
"logits/rejected": -4.131249904632568,
|
||
|
|
"logps/chosen": -572.7999877929688,
|
||
|
|
"logps/rejected": -662.7999877929688,
|
||
|
|
"loss": 2.9065,
|
||
|
|
"rewards/accuracies": 0.547656238079071,
|
||
|
|
"rewards/chosen": 10.112500190734863,
|
||
|
|
"rewards/margins": 9.84375,
|
||
|
|
"rewards/rejected": 0.25664061307907104,
|
||
|
|
"step": 1560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7370892018779343,
|
||
|
|
"grad_norm": 220.7522120460472,
|
||
|
|
"learning_rate": 1.463223787167449e-07,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.081250190734863,
|
||
|
|
"logps/chosen": -585.2000122070312,
|
||
|
|
"logps/rejected": -642.0,
|
||
|
|
"loss": 3.5233,
|
||
|
|
"rewards/accuracies": 0.510937511920929,
|
||
|
|
"rewards/chosen": 8.915624618530273,
|
||
|
|
"rewards/margins": 6.90625,
|
||
|
|
"rewards/rejected": 1.998046875,
|
||
|
|
"step": 1570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7417840375586855,
|
||
|
|
"grad_norm": 282.38518483309315,
|
||
|
|
"learning_rate": 1.4371413667188313e-07,
|
||
|
|
"logits/chosen": -4.215624809265137,
|
||
|
|
"logits/rejected": -4.0703125,
|
||
|
|
"logps/chosen": -568.4000244140625,
|
||
|
|
"logps/rejected": -680.4000244140625,
|
||
|
|
"loss": 3.2953,
|
||
|
|
"rewards/accuracies": 0.5414062738418579,
|
||
|
|
"rewards/chosen": 9.340624809265137,
|
||
|
|
"rewards/margins": 9.096875190734863,
|
||
|
|
"rewards/rejected": 0.23681640625,
|
||
|
|
"step": 1580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7464788732394366,
|
||
|
|
"grad_norm": 266.0114406524382,
|
||
|
|
"learning_rate": 1.4110589462702139e-07,
|
||
|
|
"logits/chosen": -4.237500190734863,
|
||
|
|
"logits/rejected": -4.134375095367432,
|
||
|
|
"logps/chosen": -567.5999755859375,
|
||
|
|
"logps/rejected": -626.2000122070312,
|
||
|
|
"loss": 3.4592,
|
||
|
|
"rewards/accuracies": 0.530468761920929,
|
||
|
|
"rewards/chosen": 9.65625,
|
||
|
|
"rewards/margins": 7.831250190734863,
|
||
|
|
"rewards/rejected": 1.8234374523162842,
|
||
|
|
"step": 1590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7511737089201878,
|
||
|
|
"grad_norm": 293.3811924171183,
|
||
|
|
"learning_rate": 1.384976525821596e-07,
|
||
|
|
"logits/chosen": -4.259375095367432,
|
||
|
|
"logits/rejected": -4.076562404632568,
|
||
|
|
"logps/chosen": -585.7999877929688,
|
||
|
|
"logps/rejected": -671.5999755859375,
|
||
|
|
"loss": 3.4814,
|
||
|
|
"rewards/accuracies": 0.5171874761581421,
|
||
|
|
"rewards/chosen": 9.199999809265137,
|
||
|
|
"rewards/margins": 9.306249618530273,
|
||
|
|
"rewards/rejected": -0.10209961235523224,
|
||
|
|
"step": 1600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.755868544600939,
|
||
|
|
"grad_norm": 299.3190882040113,
|
||
|
|
"learning_rate": 1.3588941053729787e-07,
|
||
|
|
"logits/chosen": -4.157812595367432,
|
||
|
|
"logits/rejected": -4.082812309265137,
|
||
|
|
"logps/chosen": -586.5999755859375,
|
||
|
|
"logps/rejected": -672.2000122070312,
|
||
|
|
"loss": 3.2615,
|
||
|
|
"rewards/accuracies": 0.5140625238418579,
|
||
|
|
"rewards/chosen": 9.368749618530273,
|
||
|
|
"rewards/margins": 9.240625381469727,
|
||
|
|
"rewards/rejected": 0.13740234076976776,
|
||
|
|
"step": 1610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7605633802816901,
|
||
|
|
"grad_norm": 292.3874233512557,
|
||
|
|
"learning_rate": 1.332811684924361e-07,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -594.7999877929688,
|
||
|
|
"logps/rejected": -650.0,
|
||
|
|
"loss": 3.265,
|
||
|
|
"rewards/accuracies": 0.484375,
|
||
|
|
"rewards/chosen": 8.831250190734863,
|
||
|
|
"rewards/margins": 7.465624809265137,
|
||
|
|
"rewards/rejected": 1.368749976158142,
|
||
|
|
"step": 1620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7652582159624414,
|
||
|
|
"grad_norm": 257.6034386681777,
|
||
|
|
"learning_rate": 1.3067292644757432e-07,
|
||
|
|
"logits/chosen": -4.090624809265137,
|
||
|
|
"logits/rejected": -4.003125190734863,
|
||
|
|
"logps/chosen": -597.2000122070312,
|
||
|
|
"logps/rejected": -692.7999877929688,
|
||
|
|
"loss": 3.3718,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 8.643750190734863,
|
||
|
|
"rewards/margins": 8.181249618530273,
|
||
|
|
"rewards/rejected": 0.46630859375,
|
||
|
|
"step": 1630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7699530516431925,
|
||
|
|
"grad_norm": 351.04743008203644,
|
||
|
|
"learning_rate": 1.2806468440271257e-07,
|
||
|
|
"logits/chosen": -4.171875,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -560.2000122070312,
|
||
|
|
"logps/rejected": -632.5999755859375,
|
||
|
|
"loss": 2.8387,
|
||
|
|
"rewards/accuracies": 0.52734375,
|
||
|
|
"rewards/chosen": 9.603124618530273,
|
||
|
|
"rewards/margins": 8.171875,
|
||
|
|
"rewards/rejected": 1.4375,
|
||
|
|
"step": 1640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7746478873239436,
|
||
|
|
"grad_norm": 297.5853232790701,
|
||
|
|
"learning_rate": 1.254564423578508e-07,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -567.4000244140625,
|
||
|
|
"logps/rejected": -644.7999877929688,
|
||
|
|
"loss": 3.0734,
|
||
|
|
"rewards/accuracies": 0.5171874761581421,
|
||
|
|
"rewards/chosen": 9.106249809265137,
|
||
|
|
"rewards/margins": 8.315625190734863,
|
||
|
|
"rewards/rejected": 0.797656238079071,
|
||
|
|
"step": 1650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7793427230046949,
|
||
|
|
"grad_norm": 258.1808056870632,
|
||
|
|
"learning_rate": 1.2284820031298902e-07,
|
||
|
|
"logits/chosen": -4.165625095367432,
|
||
|
|
"logits/rejected": -4.040625095367432,
|
||
|
|
"logps/chosen": -606.4000244140625,
|
||
|
|
"logps/rejected": -671.4000244140625,
|
||
|
|
"loss": 3.1496,
|
||
|
|
"rewards/accuracies": 0.5218750238418579,
|
||
|
|
"rewards/chosen": 9.106249809265137,
|
||
|
|
"rewards/margins": 9.184374809265137,
|
||
|
|
"rewards/rejected": -0.07441405951976776,
|
||
|
|
"step": 1660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.784037558685446,
|
||
|
|
"grad_norm": 315.1161588076266,
|
||
|
|
"learning_rate": 1.2023995826812728e-07,
|
||
|
|
"logits/chosen": -4.170312404632568,
|
||
|
|
"logits/rejected": -4.03125,
|
||
|
|
"logps/chosen": -569.5999755859375,
|
||
|
|
"logps/rejected": -659.2000122070312,
|
||
|
|
"loss": 3.6683,
|
||
|
|
"rewards/accuracies": 0.5132812261581421,
|
||
|
|
"rewards/chosen": 8.653124809265137,
|
||
|
|
"rewards/margins": 8.026562690734863,
|
||
|
|
"rewards/rejected": 0.627685546875,
|
||
|
|
"step": 1670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7887323943661971,
|
||
|
|
"grad_norm": 339.45568377970034,
|
||
|
|
"learning_rate": 1.176317162232655e-07,
|
||
|
|
"logits/chosen": -4.175000190734863,
|
||
|
|
"logits/rejected": -4.090624809265137,
|
||
|
|
"logps/chosen": -565.0,
|
||
|
|
"logps/rejected": -670.2000122070312,
|
||
|
|
"loss": 3.4659,
|
||
|
|
"rewards/accuracies": 0.5218750238418579,
|
||
|
|
"rewards/chosen": 9.118749618530273,
|
||
|
|
"rewards/margins": 8.050000190734863,
|
||
|
|
"rewards/rejected": 1.072265625,
|
||
|
|
"step": 1680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7934272300469484,
|
||
|
|
"grad_norm": 309.46428057578476,
|
||
|
|
"learning_rate": 1.1502347417840374e-07,
|
||
|
|
"logits/chosen": -4.143750190734863,
|
||
|
|
"logits/rejected": -4.042187690734863,
|
||
|
|
"logps/chosen": -572.5999755859375,
|
||
|
|
"logps/rejected": -650.7999877929688,
|
||
|
|
"loss": 3.3681,
|
||
|
|
"rewards/accuracies": 0.526562511920929,
|
||
|
|
"rewards/chosen": 9.600000381469727,
|
||
|
|
"rewards/margins": 8.931249618530273,
|
||
|
|
"rewards/rejected": 0.663281261920929,
|
||
|
|
"step": 1690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7981220657276995,
|
||
|
|
"grad_norm": 276.7849077382129,
|
||
|
|
"learning_rate": 1.1241523213354198e-07,
|
||
|
|
"logits/chosen": -4.203125,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -574.4000244140625,
|
||
|
|
"logps/rejected": -672.7999877929688,
|
||
|
|
"loss": 3.2875,
|
||
|
|
"rewards/accuracies": 0.539843738079071,
|
||
|
|
"rewards/chosen": 9.787500381469727,
|
||
|
|
"rewards/margins": 9.084375381469727,
|
||
|
|
"rewards/rejected": 0.699414074420929,
|
||
|
|
"step": 1700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8028169014084507,
|
||
|
|
"grad_norm": 286.84592205572187,
|
||
|
|
"learning_rate": 1.0980699008868022e-07,
|
||
|
|
"logits/chosen": -4.162499904632568,
|
||
|
|
"logits/rejected": -4.0703125,
|
||
|
|
"logps/chosen": -586.4000244140625,
|
||
|
|
"logps/rejected": -637.2000122070312,
|
||
|
|
"loss": 3.9705,
|
||
|
|
"rewards/accuracies": 0.5210937261581421,
|
||
|
|
"rewards/chosen": 9.065625190734863,
|
||
|
|
"rewards/margins": 7.614062309265137,
|
||
|
|
"rewards/rejected": 1.4560546875,
|
||
|
|
"step": 1710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8075117370892019,
|
||
|
|
"grad_norm": 304.19812578369937,
|
||
|
|
"learning_rate": 1.0719874804381846e-07,
|
||
|
|
"logits/chosen": -4.154687404632568,
|
||
|
|
"logits/rejected": -4.0859375,
|
||
|
|
"logps/chosen": -568.4000244140625,
|
||
|
|
"logps/rejected": -603.4000244140625,
|
||
|
|
"loss": 2.9644,
|
||
|
|
"rewards/accuracies": 0.5023437738418579,
|
||
|
|
"rewards/chosen": 9.259374618530273,
|
||
|
|
"rewards/margins": 6.884375095367432,
|
||
|
|
"rewards/rejected": 2.369335889816284,
|
||
|
|
"step": 1720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.812206572769953,
|
||
|
|
"grad_norm": 277.7578193166386,
|
||
|
|
"learning_rate": 1.045905059989567e-07,
|
||
|
|
"logits/chosen": -4.209374904632568,
|
||
|
|
"logits/rejected": -4.151562690734863,
|
||
|
|
"logps/chosen": -542.7999877929688,
|
||
|
|
"logps/rejected": -594.5999755859375,
|
||
|
|
"loss": 3.0375,
|
||
|
|
"rewards/accuracies": 0.5140625238418579,
|
||
|
|
"rewards/chosen": 9.524999618530273,
|
||
|
|
"rewards/margins": 7.068749904632568,
|
||
|
|
"rewards/rejected": 2.4613280296325684,
|
||
|
|
"step": 1730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8169014084507042,
|
||
|
|
"grad_norm": 300.59459600097534,
|
||
|
|
"learning_rate": 1.0198226395409494e-07,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.029687404632568,
|
||
|
|
"logps/chosen": -551.4000244140625,
|
||
|
|
"logps/rejected": -619.4000244140625,
|
||
|
|
"loss": 3.3825,
|
||
|
|
"rewards/accuracies": 0.5210937261581421,
|
||
|
|
"rewards/chosen": 8.734375,
|
||
|
|
"rewards/margins": 7.996874809265137,
|
||
|
|
"rewards/rejected": 0.744140625,
|
||
|
|
"step": 1740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8215962441314554,
|
||
|
|
"grad_norm": 242.278341686424,
|
||
|
|
"learning_rate": 9.937402190923318e-08,
|
||
|
|
"logits/chosen": -4.128125190734863,
|
||
|
|
"logits/rejected": -4.021874904632568,
|
||
|
|
"logps/chosen": -541.2000122070312,
|
||
|
|
"logps/rejected": -658.7999877929688,
|
||
|
|
"loss": 3.2527,
|
||
|
|
"rewards/accuracies": 0.49296873807907104,
|
||
|
|
"rewards/chosen": 8.490625381469727,
|
||
|
|
"rewards/margins": 8.984375,
|
||
|
|
"rewards/rejected": -0.4996093809604645,
|
||
|
|
"step": 1750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8262910798122066,
|
||
|
|
"grad_norm": 264.74847983452094,
|
||
|
|
"learning_rate": 9.676577986437141e-08,
|
||
|
|
"logits/chosen": -4.1171875,
|
||
|
|
"logits/rejected": -3.981250047683716,
|
||
|
|
"logps/chosen": -601.5999755859375,
|
||
|
|
"logps/rejected": -652.4000244140625,
|
||
|
|
"loss": 3.3588,
|
||
|
|
"rewards/accuracies": 0.5140625238418579,
|
||
|
|
"rewards/chosen": 8.646875381469727,
|
||
|
|
"rewards/margins": 8.09375,
|
||
|
|
"rewards/rejected": 0.553759753704071,
|
||
|
|
"step": 1760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8309859154929577,
|
||
|
|
"grad_norm": 252.43600024697975,
|
||
|
|
"learning_rate": 9.415753781950965e-08,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -4.0859375,
|
||
|
|
"logps/chosen": -554.0,
|
||
|
|
"logps/rejected": -640.0,
|
||
|
|
"loss": 2.8763,
|
||
|
|
"rewards/accuracies": 0.546093761920929,
|
||
|
|
"rewards/chosen": 9.568750381469727,
|
||
|
|
"rewards/margins": 9.568750381469727,
|
||
|
|
"rewards/rejected": 0.013867187313735485,
|
||
|
|
"step": 1770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8356807511737089,
|
||
|
|
"grad_norm": 253.5525987973917,
|
||
|
|
"learning_rate": 9.154929577464789e-08,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.043749809265137,
|
||
|
|
"logps/chosen": -572.7999877929688,
|
||
|
|
"logps/rejected": -644.0,
|
||
|
|
"loss": 3.2664,
|
||
|
|
"rewards/accuracies": 0.514843761920929,
|
||
|
|
"rewards/chosen": 8.684374809265137,
|
||
|
|
"rewards/margins": 8.540624618530273,
|
||
|
|
"rewards/rejected": 0.15009765326976776,
|
||
|
|
"step": 1780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8403755868544601,
|
||
|
|
"grad_norm": 250.62939288036958,
|
||
|
|
"learning_rate": 8.894105372978613e-08,
|
||
|
|
"logits/chosen": -4.140625,
|
||
|
|
"logits/rejected": -4.053124904632568,
|
||
|
|
"logps/chosen": -590.7999877929688,
|
||
|
|
"logps/rejected": -653.0,
|
||
|
|
"loss": 3.257,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 9.178125381469727,
|
||
|
|
"rewards/margins": 9.496874809265137,
|
||
|
|
"rewards/rejected": -0.32539063692092896,
|
||
|
|
"step": 1790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8450704225352113,
|
||
|
|
"grad_norm": 276.0547109194086,
|
||
|
|
"learning_rate": 8.633281168492435e-08,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -4.125,
|
||
|
|
"logps/chosen": -569.2000122070312,
|
||
|
|
"logps/rejected": -633.4000244140625,
|
||
|
|
"loss": 3.2604,
|
||
|
|
"rewards/accuracies": 0.546875,
|
||
|
|
"rewards/chosen": 8.996874809265137,
|
||
|
|
"rewards/margins": 8.640625,
|
||
|
|
"rewards/rejected": 0.34589844942092896,
|
||
|
|
"step": 1800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8497652582159625,
|
||
|
|
"grad_norm": 315.86906591823214,
|
||
|
|
"learning_rate": 8.372456964006259e-08,
|
||
|
|
"logits/chosen": -4.120312690734863,
|
||
|
|
"logits/rejected": -4.009375095367432,
|
||
|
|
"logps/chosen": -564.7999877929688,
|
||
|
|
"logps/rejected": -642.0,
|
||
|
|
"loss": 3.4094,
|
||
|
|
"rewards/accuracies": 0.538281261920929,
|
||
|
|
"rewards/chosen": 9.053125381469727,
|
||
|
|
"rewards/margins": 8.796875,
|
||
|
|
"rewards/rejected": 0.25703126192092896,
|
||
|
|
"step": 1810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8544600938967136,
|
||
|
|
"grad_norm": 299.3337185435239,
|
||
|
|
"learning_rate": 8.111632759520083e-08,
|
||
|
|
"logits/chosen": -4.123437404632568,
|
||
|
|
"logits/rejected": -4.045312404632568,
|
||
|
|
"logps/chosen": -613.7999877929688,
|
||
|
|
"logps/rejected": -664.4000244140625,
|
||
|
|
"loss": 3.5758,
|
||
|
|
"rewards/accuracies": 0.5289062261581421,
|
||
|
|
"rewards/chosen": 8.653124809265137,
|
||
|
|
"rewards/margins": 8.795312881469727,
|
||
|
|
"rewards/rejected": -0.14658203721046448,
|
||
|
|
"step": 1820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8591549295774648,
|
||
|
|
"grad_norm": 305.90788242817615,
|
||
|
|
"learning_rate": 7.850808555033907e-08,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -4.009375095367432,
|
||
|
|
"logps/chosen": -573.2000122070312,
|
||
|
|
"logps/rejected": -666.7999877929688,
|
||
|
|
"loss": 3.1492,
|
||
|
|
"rewards/accuracies": 0.526562511920929,
|
||
|
|
"rewards/chosen": 9.293749809265137,
|
||
|
|
"rewards/margins": 9.759374618530273,
|
||
|
|
"rewards/rejected": -0.47089844942092896,
|
||
|
|
"step": 1830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.863849765258216,
|
||
|
|
"grad_norm": 283.60957647674786,
|
||
|
|
"learning_rate": 7.58998435054773e-08,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.012499809265137,
|
||
|
|
"logps/chosen": -592.0,
|
||
|
|
"logps/rejected": -664.4000244140625,
|
||
|
|
"loss": 3.4984,
|
||
|
|
"rewards/accuracies": 0.5015624761581421,
|
||
|
|
"rewards/chosen": 9.246874809265137,
|
||
|
|
"rewards/margins": 7.703125,
|
||
|
|
"rewards/rejected": 1.5480468273162842,
|
||
|
|
"step": 1840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8685446009389671,
|
||
|
|
"grad_norm": 264.6730900596976,
|
||
|
|
"learning_rate": 7.329160146061554e-08,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.190625190734863,
|
||
|
|
"logps/chosen": -561.5999755859375,
|
||
|
|
"logps/rejected": -650.0,
|
||
|
|
"loss": 2.9436,
|
||
|
|
"rewards/accuracies": 0.522656261920929,
|
||
|
|
"rewards/chosen": 9.556249618530273,
|
||
|
|
"rewards/margins": 8.175000190734863,
|
||
|
|
"rewards/rejected": 1.376953125,
|
||
|
|
"step": 1850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8732394366197183,
|
||
|
|
"grad_norm": 315.0386051657591,
|
||
|
|
"learning_rate": 7.068335941575378e-08,
|
||
|
|
"logits/chosen": -4.125,
|
||
|
|
"logits/rejected": -4.079687595367432,
|
||
|
|
"logps/chosen": -563.2000122070312,
|
||
|
|
"logps/rejected": -622.2000122070312,
|
||
|
|
"loss": 3.8326,
|
||
|
|
"rewards/accuracies": 0.5132812261581421,
|
||
|
|
"rewards/chosen": 8.432812690734863,
|
||
|
|
"rewards/margins": 8.271875381469727,
|
||
|
|
"rewards/rejected": 0.1669921875,
|
||
|
|
"step": 1860
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8779342723004695,
|
||
|
|
"grad_norm": 347.82291763141683,
|
||
|
|
"learning_rate": 6.807511737089202e-08,
|
||
|
|
"logits/chosen": -4.240624904632568,
|
||
|
|
"logits/rejected": -4.107812404632568,
|
||
|
|
"logps/chosen": -559.5999755859375,
|
||
|
|
"logps/rejected": -644.7999877929688,
|
||
|
|
"loss": 3.2637,
|
||
|
|
"rewards/accuracies": 0.508593738079071,
|
||
|
|
"rewards/chosen": 9.253125190734863,
|
||
|
|
"rewards/margins": 8.503125190734863,
|
||
|
|
"rewards/rejected": 0.744335949420929,
|
||
|
|
"step": 1870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8826291079812206,
|
||
|
|
"grad_norm": 370.90610454742574,
|
||
|
|
"learning_rate": 6.546687532603024e-08,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.057812690734863,
|
||
|
|
"logps/chosen": -566.5999755859375,
|
||
|
|
"logps/rejected": -654.4000244140625,
|
||
|
|
"loss": 3.2342,
|
||
|
|
"rewards/accuracies": 0.4906249940395355,
|
||
|
|
"rewards/chosen": 8.434374809265137,
|
||
|
|
"rewards/margins": 7.290625095367432,
|
||
|
|
"rewards/rejected": 1.1453125476837158,
|
||
|
|
"step": 1880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8873239436619719,
|
||
|
|
"grad_norm": 240.62246133414592,
|
||
|
|
"learning_rate": 6.285863328116848e-08,
|
||
|
|
"logits/chosen": -4.1875,
|
||
|
|
"logits/rejected": -4.074999809265137,
|
||
|
|
"logps/chosen": -521.0,
|
||
|
|
"logps/rejected": -667.5999755859375,
|
||
|
|
"loss": 3.2679,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 8.834375381469727,
|
||
|
|
"rewards/margins": 8.953125,
|
||
|
|
"rewards/rejected": -0.13417968153953552,
|
||
|
|
"step": 1890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.892018779342723,
|
||
|
|
"grad_norm": 269.3744968930516,
|
||
|
|
"learning_rate": 6.025039123630672e-08,
|
||
|
|
"logits/chosen": -4.15625,
|
||
|
|
"logits/rejected": -4.073437690734863,
|
||
|
|
"logps/chosen": -584.4000244140625,
|
||
|
|
"logps/rejected": -622.4000244140625,
|
||
|
|
"loss": 3.6584,
|
||
|
|
"rewards/accuracies": 0.516406238079071,
|
||
|
|
"rewards/chosen": 9.231249809265137,
|
||
|
|
"rewards/margins": 8.609375,
|
||
|
|
"rewards/rejected": 0.62744140625,
|
||
|
|
"step": 1900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8967136150234741,
|
||
|
|
"grad_norm": 286.7976408001139,
|
||
|
|
"learning_rate": 5.764214919144496e-08,
|
||
|
|
"logits/chosen": -4.184374809265137,
|
||
|
|
"logits/rejected": -4.106249809265137,
|
||
|
|
"logps/chosen": -588.7999877929688,
|
||
|
|
"logps/rejected": -616.5999755859375,
|
||
|
|
"loss": 3.2276,
|
||
|
|
"rewards/accuracies": 0.49531251192092896,
|
||
|
|
"rewards/chosen": 8.837499618530273,
|
||
|
|
"rewards/margins": 6.567187309265137,
|
||
|
|
"rewards/rejected": 2.2730469703674316,
|
||
|
|
"step": 1910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9014084507042254,
|
||
|
|
"grad_norm": 350.40438971050554,
|
||
|
|
"learning_rate": 5.50339071465832e-08,
|
||
|
|
"logits/chosen": -4.173437595367432,
|
||
|
|
"logits/rejected": -4.115624904632568,
|
||
|
|
"logps/chosen": -580.2000122070312,
|
||
|
|
"logps/rejected": -604.0,
|
||
|
|
"loss": 3.4644,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 8.759374618530273,
|
||
|
|
"rewards/margins": 7.09375,
|
||
|
|
"rewards/rejected": 1.672216773033142,
|
||
|
|
"step": 1920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9061032863849765,
|
||
|
|
"grad_norm": 324.283448378128,
|
||
|
|
"learning_rate": 5.2425665101721436e-08,
|
||
|
|
"logits/chosen": -4.135937690734863,
|
||
|
|
"logits/rejected": -3.984375,
|
||
|
|
"logps/chosen": -590.5999755859375,
|
||
|
|
"logps/rejected": -693.2000122070312,
|
||
|
|
"loss": 3.1065,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 8.921875,
|
||
|
|
"rewards/margins": 8.193750381469727,
|
||
|
|
"rewards/rejected": 0.7367187738418579,
|
||
|
|
"step": 1930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9107981220657277,
|
||
|
|
"grad_norm": 273.11521435297766,
|
||
|
|
"learning_rate": 4.9817423056859675e-08,
|
||
|
|
"logits/chosen": -4.193749904632568,
|
||
|
|
"logits/rejected": -4.067187309265137,
|
||
|
|
"logps/chosen": -586.0,
|
||
|
|
"logps/rejected": -665.2000122070312,
|
||
|
|
"loss": 3.2885,
|
||
|
|
"rewards/accuracies": 0.514843761920929,
|
||
|
|
"rewards/chosen": 9.646875381469727,
|
||
|
|
"rewards/margins": 9.609375,
|
||
|
|
"rewards/rejected": 0.03242187574505806,
|
||
|
|
"step": 1940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9154929577464789,
|
||
|
|
"grad_norm": 272.6089128679087,
|
||
|
|
"learning_rate": 4.720918101199791e-08,
|
||
|
|
"logits/chosen": -4.089062690734863,
|
||
|
|
"logits/rejected": -3.942187547683716,
|
||
|
|
"logps/chosen": -591.5999755859375,
|
||
|
|
"logps/rejected": -718.4000244140625,
|
||
|
|
"loss": 3.6842,
|
||
|
|
"rewards/accuracies": 0.49531251192092896,
|
||
|
|
"rewards/chosen": 8.921875,
|
||
|
|
"rewards/margins": 9.946874618530273,
|
||
|
|
"rewards/rejected": -1.0234375,
|
||
|
|
"step": 1950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92018779342723,
|
||
|
|
"grad_norm": 243.55218528138548,
|
||
|
|
"learning_rate": 4.460093896713615e-08,
|
||
|
|
"logits/chosen": -4.217187404632568,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -571.5999755859375,
|
||
|
|
"logps/rejected": -671.5999755859375,
|
||
|
|
"loss": 3.64,
|
||
|
|
"rewards/accuracies": 0.5101562738418579,
|
||
|
|
"rewards/chosen": 9.643750190734863,
|
||
|
|
"rewards/margins": 8.115625381469727,
|
||
|
|
"rewards/rejected": 1.5261719226837158,
|
||
|
|
"step": 1960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9248826291079812,
|
||
|
|
"grad_norm": 262.26311139456277,
|
||
|
|
"learning_rate": 4.199269692227438e-08,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.112500190734863,
|
||
|
|
"logps/chosen": -562.5999755859375,
|
||
|
|
"logps/rejected": -610.7999877929688,
|
||
|
|
"loss": 3.6316,
|
||
|
|
"rewards/accuracies": 0.4898437559604645,
|
||
|
|
"rewards/chosen": 8.578125,
|
||
|
|
"rewards/margins": 7.546875,
|
||
|
|
"rewards/rejected": 1.0242187976837158,
|
||
|
|
"step": 1970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9295774647887324,
|
||
|
|
"grad_norm": 301.82634449968526,
|
||
|
|
"learning_rate": 3.938445487741262e-08,
|
||
|
|
"logits/chosen": -4.112500190734863,
|
||
|
|
"logits/rejected": -3.9546875953674316,
|
||
|
|
"logps/chosen": -588.7999877929688,
|
||
|
|
"logps/rejected": -642.7999877929688,
|
||
|
|
"loss": 3.1398,
|
||
|
|
"rewards/accuracies": 0.500781238079071,
|
||
|
|
"rewards/chosen": 8.509374618530273,
|
||
|
|
"rewards/margins": 7.971875190734863,
|
||
|
|
"rewards/rejected": 0.547656238079071,
|
||
|
|
"step": 1980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9342723004694836,
|
||
|
|
"grad_norm": 256.70883534671344,
|
||
|
|
"learning_rate": 3.677621283255086e-08,
|
||
|
|
"logits/chosen": -4.206250190734863,
|
||
|
|
"logits/rejected": -4.120312690734863,
|
||
|
|
"logps/chosen": -596.0,
|
||
|
|
"logps/rejected": -644.0,
|
||
|
|
"loss": 3.2795,
|
||
|
|
"rewards/accuracies": 0.5328124761581421,
|
||
|
|
"rewards/chosen": 9.756250381469727,
|
||
|
|
"rewards/margins": 8.909375190734863,
|
||
|
|
"rewards/rejected": 0.83203125,
|
||
|
|
"step": 1990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9389671361502347,
|
||
|
|
"grad_norm": 282.63915129606966,
|
||
|
|
"learning_rate": 3.41679707876891e-08,
|
||
|
|
"logits/chosen": -4.154687404632568,
|
||
|
|
"logits/rejected": -4.0234375,
|
||
|
|
"logps/chosen": -578.2000122070312,
|
||
|
|
"logps/rejected": -702.4000244140625,
|
||
|
|
"loss": 3.399,
|
||
|
|
"rewards/accuracies": 0.52734375,
|
||
|
|
"rewards/chosen": 8.784375190734863,
|
||
|
|
"rewards/margins": 9.846875190734863,
|
||
|
|
"rewards/rejected": -1.062890648841858,
|
||
|
|
"step": 2000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9436619718309859,
|
||
|
|
"grad_norm": 281.2956650232833,
|
||
|
|
"learning_rate": 3.155972874282733e-08,
|
||
|
|
"logits/chosen": -4.215624809265137,
|
||
|
|
"logits/rejected": -4.079687595367432,
|
||
|
|
"logps/chosen": -555.5999755859375,
|
||
|
|
"logps/rejected": -639.5999755859375,
|
||
|
|
"loss": 3.2535,
|
||
|
|
"rewards/accuracies": 0.535937488079071,
|
||
|
|
"rewards/chosen": 9.4375,
|
||
|
|
"rewards/margins": 8.964062690734863,
|
||
|
|
"rewards/rejected": 0.468017578125,
|
||
|
|
"step": 2010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9483568075117371,
|
||
|
|
"grad_norm": 281.4878426700789,
|
||
|
|
"learning_rate": 2.8951486697965573e-08,
|
||
|
|
"logits/chosen": -4.139062404632568,
|
||
|
|
"logits/rejected": -4.048437595367432,
|
||
|
|
"logps/chosen": -589.4000244140625,
|
||
|
|
"logps/rejected": -659.5999755859375,
|
||
|
|
"loss": 3.4492,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 9.112500190734863,
|
||
|
|
"rewards/margins": 8.853124618530273,
|
||
|
|
"rewards/rejected": 0.255859375,
|
||
|
|
"step": 2020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9530516431924883,
|
||
|
|
"grad_norm": 296.87658522208113,
|
||
|
|
"learning_rate": 2.634324465310381e-08,
|
||
|
|
"logits/chosen": -4.185937404632568,
|
||
|
|
"logits/rejected": -4.042187690734863,
|
||
|
|
"logps/chosen": -564.2000122070312,
|
||
|
|
"logps/rejected": -644.7999877929688,
|
||
|
|
"loss": 3.4496,
|
||
|
|
"rewards/accuracies": 0.5093749761581421,
|
||
|
|
"rewards/chosen": 8.821874618530273,
|
||
|
|
"rewards/margins": 8.653124809265137,
|
||
|
|
"rewards/rejected": 0.16904297471046448,
|
||
|
|
"step": 2030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9577464788732394,
|
||
|
|
"grad_norm": 252.64030204060066,
|
||
|
|
"learning_rate": 2.3735002608242045e-08,
|
||
|
|
"logits/chosen": -4.168749809265137,
|
||
|
|
"logits/rejected": -4.056250095367432,
|
||
|
|
"logps/chosen": -584.2000122070312,
|
||
|
|
"logps/rejected": -617.2000122070312,
|
||
|
|
"loss": 2.8264,
|
||
|
|
"rewards/accuracies": 0.53515625,
|
||
|
|
"rewards/chosen": 9.75,
|
||
|
|
"rewards/margins": 8.509374618530273,
|
||
|
|
"rewards/rejected": 1.2374999523162842,
|
||
|
|
"step": 2040
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9624413145539906,
|
||
|
|
"grad_norm": 274.4937616307003,
|
||
|
|
"learning_rate": 2.1126760563380282e-08,
|
||
|
|
"logits/chosen": -4.181250095367432,
|
||
|
|
"logits/rejected": -4.096875190734863,
|
||
|
|
"logps/chosen": -569.7999877929688,
|
||
|
|
"logps/rejected": -599.2000122070312,
|
||
|
|
"loss": 3.5493,
|
||
|
|
"rewards/accuracies": 0.510937511920929,
|
||
|
|
"rewards/chosen": 8.875,
|
||
|
|
"rewards/margins": 6.400000095367432,
|
||
|
|
"rewards/rejected": 2.4742188453674316,
|
||
|
|
"step": 2050
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9671361502347418,
|
||
|
|
"grad_norm": 239.72102282540757,
|
||
|
|
"learning_rate": 1.8518518518518518e-08,
|
||
|
|
"logits/chosen": -4.162499904632568,
|
||
|
|
"logits/rejected": -4.0234375,
|
||
|
|
"logps/chosen": -595.2000122070312,
|
||
|
|
"logps/rejected": -648.7999877929688,
|
||
|
|
"loss": 3.7638,
|
||
|
|
"rewards/accuracies": 0.5062500238418579,
|
||
|
|
"rewards/chosen": 8.953125,
|
||
|
|
"rewards/margins": 7.721875190734863,
|
||
|
|
"rewards/rejected": 1.220703125,
|
||
|
|
"step": 2060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.971830985915493,
|
||
|
|
"grad_norm": 266.3637050395917,
|
||
|
|
"learning_rate": 1.5910276473656755e-08,
|
||
|
|
"logits/chosen": -4.143750190734863,
|
||
|
|
"logits/rejected": -3.9937500953674316,
|
||
|
|
"logps/chosen": -570.2000122070312,
|
||
|
|
"logps/rejected": -643.5999755859375,
|
||
|
|
"loss": 2.7844,
|
||
|
|
"rewards/accuracies": 0.520312488079071,
|
||
|
|
"rewards/chosen": 9.15625,
|
||
|
|
"rewards/margins": 9.082812309265137,
|
||
|
|
"rewards/rejected": 0.07871093600988388,
|
||
|
|
"step": 2070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9765258215962441,
|
||
|
|
"grad_norm": 244.26802244450215,
|
||
|
|
"learning_rate": 1.3302034428794991e-08,
|
||
|
|
"logits/chosen": -4.215624809265137,
|
||
|
|
"logits/rejected": -4.153124809265137,
|
||
|
|
"logps/chosen": -549.5999755859375,
|
||
|
|
"logps/rejected": -610.4000244140625,
|
||
|
|
"loss": 3.382,
|
||
|
|
"rewards/accuracies": 0.5179687738418579,
|
||
|
|
"rewards/chosen": 8.71875,
|
||
|
|
"rewards/margins": 8.443750381469727,
|
||
|
|
"rewards/rejected": 0.2803710997104645,
|
||
|
|
"step": 2080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9812206572769953,
|
||
|
|
"grad_norm": 273.08298980869785,
|
||
|
|
"learning_rate": 1.0693792383933229e-08,
|
||
|
|
"logits/chosen": -4.212500095367432,
|
||
|
|
"logits/rejected": -4.15625,
|
||
|
|
"logps/chosen": -582.0,
|
||
|
|
"logps/rejected": -658.4000244140625,
|
||
|
|
"loss": 3.4493,
|
||
|
|
"rewards/accuracies": 0.5249999761581421,
|
||
|
|
"rewards/chosen": 9.546875,
|
||
|
|
"rewards/margins": 8.981249809265137,
|
||
|
|
"rewards/rejected": 0.56201171875,
|
||
|
|
"step": 2090
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9859154929577465,
|
||
|
|
"grad_norm": 247.97793524019707,
|
||
|
|
"learning_rate": 8.085550339071465e-09,
|
||
|
|
"logits/chosen": -4.159375190734863,
|
||
|
|
"logits/rejected": -4.059374809265137,
|
||
|
|
"logps/chosen": -597.0,
|
||
|
|
"logps/rejected": -688.5999755859375,
|
||
|
|
"loss": 3.6896,
|
||
|
|
"rewards/accuracies": 0.52734375,
|
||
|
|
"rewards/chosen": 8.565625190734863,
|
||
|
|
"rewards/margins": 8.681249618530273,
|
||
|
|
"rewards/rejected": -0.12148437649011612,
|
||
|
|
"step": 2100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9906103286384976,
|
||
|
|
"grad_norm": 355.78181353320775,
|
||
|
|
"learning_rate": 5.4773082942097025e-09,
|
||
|
|
"logits/chosen": -4.178124904632568,
|
||
|
|
"logits/rejected": -4.104687690734863,
|
||
|
|
"logps/chosen": -565.4000244140625,
|
||
|
|
"logps/rejected": -622.2000122070312,
|
||
|
|
"loss": 3.1718,
|
||
|
|
"rewards/accuracies": 0.5234375,
|
||
|
|
"rewards/chosen": 9.175000190734863,
|
||
|
|
"rewards/margins": 8.899999618530273,
|
||
|
|
"rewards/rejected": 0.26640623807907104,
|
||
|
|
"step": 2110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9953051643192489,
|
||
|
|
"grad_norm": 302.36578049371593,
|
||
|
|
"learning_rate": 2.8690662493479393e-09,
|
||
|
|
"logits/chosen": -4.231249809265137,
|
||
|
|
"logits/rejected": -4.035937309265137,
|
||
|
|
"logps/chosen": -568.0,
|
||
|
|
"logps/rejected": -635.2000122070312,
|
||
|
|
"loss": 3.9098,
|
||
|
|
"rewards/accuracies": 0.528124988079071,
|
||
|
|
"rewards/chosen": 9.043749809265137,
|
||
|
|
"rewards/margins": 7.417578220367432,
|
||
|
|
"rewards/rejected": 1.6320312023162842,
|
||
|
|
"step": 2120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 246.1305241577918,
|
||
|
|
"learning_rate": 2.608242044861763e-10,
|
||
|
|
"logits/chosen": -4.165625095367432,
|
||
|
|
"logits/rejected": -4.1171875,
|
||
|
|
"logps/chosen": -566.0,
|
||
|
|
"logps/rejected": -658.0,
|
||
|
|
"loss": 3.576,
|
||
|
|
"rewards/accuracies": 0.5049871206283569,
|
||
|
|
"rewards/chosen": 9.1875,
|
||
|
|
"rewards/margins": 9.4375,
|
||
|
|
"rewards/rejected": -0.25664061307907104,
|
||
|
|
"step": 2130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"step": 2130,
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_loss": 3.2928951639524646,
|
||
|
|
"train_runtime": 18574.6633,
|
||
|
|
"train_samples_per_second": 14.675,
|
||
|
|
"train_steps_per_second": 0.115
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 10,
|
||
|
|
"max_steps": 2130,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"save_steps": 200,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_batch_size": 32,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|