479 lines
15 KiB
JSON
479 lines
15 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.0,
|
|
"eval_steps": 500,
|
|
"global_step": 292,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.034327397554172925,
|
|
"grad_norm": 0.027326168841644558,
|
|
"learning_rate": 1.5e-07,
|
|
"logits/chosen": -1.796875,
|
|
"logits/rejected": -1.78125,
|
|
"logps/chosen": -0.279296875,
|
|
"logps/rejected": -0.287109375,
|
|
"loss": 0.6927052736282349,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": 2.3672189712524414,
|
|
"rewards/margins": 0.8975496292114258,
|
|
"rewards/rejected": 1.4696693420410156,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.06865479510834585,
|
|
"grad_norm": 0.09921485553278094,
|
|
"learning_rate": 3.166666666666666e-07,
|
|
"logits/chosen": -1.71875,
|
|
"logits/rejected": -1.6484375,
|
|
"logps/chosen": -0.28515625,
|
|
"logps/rejected": -0.302734375,
|
|
"loss": 0.6922475099563599,
|
|
"rewards/accuracies": 0.8968750238418579,
|
|
"rewards/chosen": 3.4436473846435547,
|
|
"rewards/margins": 1.8109512329101562,
|
|
"rewards/rejected": 1.6326961517333984,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.10298219266251878,
|
|
"grad_norm": 0.10420540005955876,
|
|
"learning_rate": 4.833333333333333e-07,
|
|
"logits/chosen": -1.8359375,
|
|
"logits/rejected": -1.7734375,
|
|
"logps/chosen": -0.275390625,
|
|
"logps/rejected": -0.291015625,
|
|
"loss": 0.6926367282867432,
|
|
"rewards/accuracies": 0.859375,
|
|
"rewards/chosen": 2.249516487121582,
|
|
"rewards/margins": 1.0263261795043945,
|
|
"rewards/rejected": 1.2231903076171875,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.1373095902166917,
|
|
"grad_norm": 0.03849402823923644,
|
|
"learning_rate": 4.985456442051682e-07,
|
|
"logits/chosen": -1.8515625,
|
|
"logits/rejected": -1.8203125,
|
|
"logps/chosen": -0.2890625,
|
|
"logps/rejected": -0.306640625,
|
|
"loss": 0.6926412582397461,
|
|
"rewards/accuracies": 0.8843749761581421,
|
|
"rewards/chosen": 2.4276199340820312,
|
|
"rewards/margins": 1.0192298889160156,
|
|
"rewards/rejected": 1.4083900451660156,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.1716369877708646,
|
|
"grad_norm": 0.0374673819672732,
|
|
"learning_rate": 4.935399618791793e-07,
|
|
"logits/chosen": -1.828125,
|
|
"logits/rejected": -1.7890625,
|
|
"logps/chosen": -0.279296875,
|
|
"logps/rejected": -0.296875,
|
|
"loss": 0.6924989223480225,
|
|
"rewards/accuracies": 0.8531249761581421,
|
|
"rewards/chosen": 2.6443328857421875,
|
|
"rewards/margins": 1.3047637939453125,
|
|
"rewards/rejected": 1.339569091796875,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.20596438532503755,
|
|
"grad_norm": 0.03365931058119091,
|
|
"learning_rate": 4.850368660819092e-07,
|
|
"logits/chosen": -1.921875,
|
|
"logits/rejected": -1.8125,
|
|
"logps/chosen": -0.283203125,
|
|
"logps/rejected": -0.30859375,
|
|
"loss": 0.6927420496940613,
|
|
"rewards/accuracies": 0.878125011920929,
|
|
"rewards/chosen": 1.7035331726074219,
|
|
"rewards/margins": 0.8122596740722656,
|
|
"rewards/rejected": 0.8912734985351562,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.24029178287921046,
|
|
"grad_norm": 0.17635150812873454,
|
|
"learning_rate": 4.731584675403184e-07,
|
|
"logits/chosen": -1.7265625,
|
|
"logits/rejected": -1.6484375,
|
|
"logps/chosen": -0.345703125,
|
|
"logps/rejected": -0.390625,
|
|
"loss": 0.6921142935752869,
|
|
"rewards/accuracies": 0.8999999761581421,
|
|
"rewards/chosen": 2.563936948776245,
|
|
"rewards/margins": 2.078953504562378,
|
|
"rewards/rejected": 0.4849834442138672,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.2746191804333834,
|
|
"grad_norm": 0.06545776403341849,
|
|
"learning_rate": 4.5807534881817183e-07,
|
|
"logits/chosen": -1.90625,
|
|
"logits/rejected": -1.8359375,
|
|
"logps/chosen": -0.40234375,
|
|
"logps/rejected": -0.380859375,
|
|
"loss": 0.6927981376647949,
|
|
"rewards/accuracies": 0.8812500238418579,
|
|
"rewards/chosen": 1.0752201080322266,
|
|
"rewards/margins": 0.7200770378112793,
|
|
"rewards/rejected": 0.35514307022094727,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.3089465779875563,
|
|
"grad_norm": 0.09027203287016318,
|
|
"learning_rate": 4.400041146246136e-07,
|
|
"logits/chosen": -1.890625,
|
|
"logits/rejected": -1.8125,
|
|
"logps/chosen": -0.478515625,
|
|
"logps/rejected": -0.5546875,
|
|
"loss": 0.692075788974762,
|
|
"rewards/accuracies": 0.887499988079071,
|
|
"rewards/chosen": 1.1104850769042969,
|
|
"rewards/margins": 2.175220489501953,
|
|
"rewards/rejected": -1.0647354125976562,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.3432739755417292,
|
|
"grad_norm": 0.031074765621602243,
|
|
"learning_rate": 4.1920428121079e-07,
|
|
"logits/chosen": -1.9921875,
|
|
"logits/rejected": -1.9453125,
|
|
"logps/chosen": -0.337890625,
|
|
"logps/rejected": -0.63671875,
|
|
"loss": 0.6915220618247986,
|
|
"rewards/accuracies": 0.8656250238418579,
|
|
"rewards/chosen": 1.155731201171875,
|
|
"rewards/margins": 3.3451175689697266,
|
|
"rewards/rejected": -2.1893863677978516,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.3776013730959022,
|
|
"grad_norm": 0.16180777732568963,
|
|
"learning_rate": 3.959745495250818e-07,
|
|
"logits/chosen": -2.15625,
|
|
"logits/rejected": -2.09375,
|
|
"logps/chosen": -0.5625,
|
|
"logps/rejected": -0.84765625,
|
|
"loss": 0.6915071606636047,
|
|
"rewards/accuracies": 0.893750011920929,
|
|
"rewards/chosen": -0.6894073486328125,
|
|
"rewards/margins": 3.4209938049316406,
|
|
"rewards/rejected": -4.110401153564453,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.4119287706500751,
|
|
"grad_norm": 0.1511404746628019,
|
|
"learning_rate": 3.7064851564718353e-07,
|
|
"logits/chosen": -2.140625,
|
|
"logits/rejected": -2.0625,
|
|
"logps/chosen": -0.85546875,
|
|
"logps/rejected": -1.3515625,
|
|
"loss": 0.6901899576187134,
|
|
"rewards/accuracies": 0.859375,
|
|
"rewards/chosen": -2.8015074729919434,
|
|
"rewards/margins": 6.206118106842041,
|
|
"rewards/rejected": -9.007625579833984,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.446256168204248,
|
|
"grad_norm": 0.06160115608776609,
|
|
"learning_rate": 3.4358988010236103e-07,
|
|
"logits/chosen": -2.203125,
|
|
"logits/rejected": -2.15625,
|
|
"logps/chosen": -0.85546875,
|
|
"logps/rejected": -0.8359375,
|
|
"loss": 0.6927467584609985,
|
|
"rewards/accuracies": 0.8374999761581421,
|
|
"rewards/chosen": -3.158720016479492,
|
|
"rewards/margins": 1.0240888595581055,
|
|
"rewards/rejected": -4.182808876037598,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.4805835657584209,
|
|
"grad_norm": 0.10523155517981944,
|
|
"learning_rate": 3.1518722485366754e-07,
|
|
"logits/chosen": -2.203125,
|
|
"logits/rejected": -2.140625,
|
|
"logps/chosen": -0.8671875,
|
|
"logps/rejected": -1.21875,
|
|
"loss": 0.6911064386367798,
|
|
"rewards/accuracies": 0.84375,
|
|
"rewards/chosen": -3.4415740966796875,
|
|
"rewards/margins": 4.292812347412109,
|
|
"rewards/rejected": -7.734386444091797,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.5149109633125939,
|
|
"grad_norm": 0.035179212929538226,
|
|
"learning_rate": 2.8584843297836277e-07,
|
|
"logits/chosen": -2.25,
|
|
"logits/rejected": -2.203125,
|
|
"logps/chosen": -1.1328125,
|
|
"logps/rejected": -1.7578125,
|
|
"loss": 0.6899595856666565,
|
|
"rewards/accuracies": 0.828125,
|
|
"rewards/chosen": -5.875,
|
|
"rewards/margins": 6.893577575683594,
|
|
"rewards/rejected": -12.768577575683594,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.5492383608667668,
|
|
"grad_norm": 0.1450101366240538,
|
|
"learning_rate": 2.5599483116609544e-07,
|
|
"logits/chosen": -2.296875,
|
|
"logits/rejected": -2.25,
|
|
"logps/chosen": -0.87890625,
|
|
"logps/rejected": -1.5625,
|
|
"loss": 0.6895779371261597,
|
|
"rewards/accuracies": 0.831250011920929,
|
|
"rewards/chosen": -3.4518814086914062,
|
|
"rewards/margins": 7.646453857421875,
|
|
"rewards/rejected": -11.098335266113281,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.5835657584209397,
|
|
"grad_norm": 0.1420016679190622,
|
|
"learning_rate": 2.2605513915689874e-07,
|
|
"logits/chosen": -2.359375,
|
|
"logits/rejected": -2.328125,
|
|
"logps/chosen": -0.94921875,
|
|
"logps/rejected": -1.5078125,
|
|
"loss": 0.6901618242263794,
|
|
"rewards/accuracies": 0.8187500238418579,
|
|
"rewards/chosen": -3.8529891967773438,
|
|
"rewards/margins": 6.436831474304199,
|
|
"rewards/rejected": -10.28982162475586,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.6178931559751126,
|
|
"grad_norm": 0.20453016696700213,
|
|
"learning_rate": 1.9645931300952795e-07,
|
|
"logits/chosen": -2.359375,
|
|
"logits/rejected": -2.296875,
|
|
"logps/chosen": -1.0,
|
|
"logps/rejected": -1.4375,
|
|
"loss": 0.6908394694328308,
|
|
"rewards/accuracies": 0.8500000238418579,
|
|
"rewards/chosen": -4.656373023986816,
|
|
"rewards/margins": 5.2327375411987305,
|
|
"rewards/rejected": -9.889110565185547,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.6522205535292855,
|
|
"grad_norm": 0.21323000370650325,
|
|
"learning_rate": 1.6763237061535008e-07,
|
|
"logits/chosen": -2.34375,
|
|
"logits/rejected": -2.3125,
|
|
"logps/chosen": -0.73828125,
|
|
"logps/rejected": -1.2734375,
|
|
"loss": 0.6903473734855652,
|
|
"rewards/accuracies": 0.796875,
|
|
"rewards/chosen": -2.4973716735839844,
|
|
"rewards/margins": 5.884622573852539,
|
|
"rewards/rejected": -8.381994247436523,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.6865479510834585,
|
|
"grad_norm": 0.0442944969938989,
|
|
"learning_rate": 1.3998828812795932e-07,
|
|
"logits/chosen": -2.46875,
|
|
"logits/rejected": -2.375,
|
|
"logps/chosen": -0.82421875,
|
|
"logps/rejected": -1.1484375,
|
|
"loss": 0.6911455392837524,
|
|
"rewards/accuracies": 0.8687499761581421,
|
|
"rewards/chosen": -3.016143798828125,
|
|
"rewards/margins": 4.1826019287109375,
|
|
"rewards/rejected": -7.1987457275390625,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.7208753486376314,
|
|
"grad_norm": 0.0814027642592345,
|
|
"learning_rate": 1.1392405496029076e-07,
|
|
"logits/chosen": -2.5,
|
|
"logits/rejected": -2.453125,
|
|
"logps/chosen": -1.0078125,
|
|
"logps/rejected": -1.546875,
|
|
"loss": 0.6901921629905701,
|
|
"rewards/accuracies": 0.828125,
|
|
"rewards/chosen": -4.940196990966797,
|
|
"rewards/margins": 6.411693572998047,
|
|
"rewards/rejected": -11.351890563964844,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.7552027461918044,
|
|
"grad_norm": 0.10252141292193413,
|
|
"learning_rate": 8.981397272385738e-08,
|
|
"logits/chosen": -2.46875,
|
|
"logits/rejected": -2.34375,
|
|
"logps/chosen": -0.6328125,
|
|
"logps/rejected": -0.9765625,
|
|
"loss": 0.6912047266960144,
|
|
"rewards/accuracies": 0.862500011920929,
|
|
"rewards/chosen": -1.6905670166015625,
|
|
"rewards/margins": 4.028224945068359,
|
|
"rewards/rejected": -5.718791961669922,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.7895301437459773,
|
|
"grad_norm": 0.18710535460212382,
|
|
"learning_rate": 6.800427998154968e-08,
|
|
"logits/chosen": -2.5,
|
|
"logits/rejected": -2.390625,
|
|
"logps/chosen": -1.3359375,
|
|
"logps/rejected": -2.71875,
|
|
"loss": 0.6866771578788757,
|
|
"rewards/accuracies": 0.8187500238418579,
|
|
"rewards/chosen": -7.6492438316345215,
|
|
"rewards/margins": 14.053179740905762,
|
|
"rewards/rejected": -21.702423095703125,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.8238575413001502,
|
|
"grad_norm": 0.07263049786204931,
|
|
"learning_rate": 4.8808180006509356e-08,
|
|
"logits/chosen": -2.40625,
|
|
"logits/rejected": -2.34375,
|
|
"logps/chosen": -1.1015625,
|
|
"logps/rejected": -1.5625,
|
|
"loss": 0.6904851198196411,
|
|
"rewards/accuracies": 0.7906249761581421,
|
|
"rewards/chosen": -5.462257385253906,
|
|
"rewards/margins": 5.7797698974609375,
|
|
"rewards/rejected": -11.242027282714844,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.8581849388543231,
|
|
"grad_norm": 0.13311409487376344,
|
|
"learning_rate": 3.250134295213053e-08,
|
|
"logits/chosen": -2.453125,
|
|
"logits/rejected": -2.40625,
|
|
"logps/chosen": -0.859375,
|
|
"logps/rejected": -1.078125,
|
|
"loss": 0.6915887594223022,
|
|
"rewards/accuracies": 0.8374999761581421,
|
|
"rewards/chosen": -3.7892303466796875,
|
|
"rewards/margins": 3.248276472091675,
|
|
"rewards/rejected": -7.037507057189941,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.892512336408496,
|
|
"grad_norm": 0.18186649279991612,
|
|
"learning_rate": 1.9317947025340232e-08,
|
|
"logits/chosen": -2.390625,
|
|
"logits/rejected": -2.296875,
|
|
"logps/chosen": -1.734375,
|
|
"logps/rejected": -1.828125,
|
|
"loss": 0.6919444799423218,
|
|
"rewards/accuracies": 0.8218749761581421,
|
|
"rewards/chosen": -10.762340545654297,
|
|
"rewards/margins": 3.087606430053711,
|
|
"rewards/rejected": -13.849946975708008,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.9268397339626689,
|
|
"grad_norm": 0.08411697914723029,
|
|
"learning_rate": 9.447315514833353e-09,
|
|
"logits/chosen": -2.453125,
|
|
"logits/rejected": -2.390625,
|
|
"logps/chosen": -0.98828125,
|
|
"logps/rejected": -1.203125,
|
|
"loss": 0.6915868520736694,
|
|
"rewards/accuracies": 0.809374988079071,
|
|
"rewards/chosen": -4.930023193359375,
|
|
"rewards/margins": 3.2534637451171875,
|
|
"rewards/rejected": -8.183486938476562,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.9611671315168419,
|
|
"grad_norm": 0.0866651631371244,
|
|
"learning_rate": 3.0311979690147426e-09,
|
|
"logits/chosen": -2.421875,
|
|
"logits/rejected": -2.3125,
|
|
"logps/chosen": -1.1328125,
|
|
"logps/rejected": -1.5859375,
|
|
"loss": 0.6905649900436401,
|
|
"rewards/accuracies": 0.840624988079071,
|
|
"rewards/chosen": -5.979114532470703,
|
|
"rewards/margins": 5.5533447265625,
|
|
"rewards/rejected": -11.532459259033203,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.9954945290710148,
|
|
"grad_norm": 0.17593981054033778,
|
|
"learning_rate": 1.6173456793908135e-10,
|
|
"logits/chosen": -2.46875,
|
|
"logits/rejected": -2.4375,
|
|
"logps/chosen": -1.09375,
|
|
"logps/rejected": -1.5078125,
|
|
"loss": 0.690861165523529,
|
|
"rewards/accuracies": 0.8500000238418579,
|
|
"rewards/chosen": -5.743495941162109,
|
|
"rewards/margins": 5.161052703857422,
|
|
"rewards/rejected": -10.904548645019531,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"step": 292,
|
|
"total_flos": 0.0,
|
|
"train_loss": 0.6896600347675689,
|
|
"train_runtime": 29629.3533,
|
|
"train_samples_per_second": 0.629,
|
|
"train_steps_per_second": 0.01
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 292,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 1000,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|