ds_coder_adamw_iter2 / trainer_state.json
yiran-wang3's picture
End of training
52c5948 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 47,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"debug/policy_chosen_logits": 31.700279235839844,
"debug/policy_chosen_logps": -434.26495361328125,
"debug/policy_rejected_logits": 33.99253845214844,
"debug/policy_rejected_logps": -441.9063720703125,
"debug/reference_chosen_logps": -434.26495361328125,
"debug/reference_rejected_logps": -441.9063720703125,
"epoch": 0.02127659574468085,
"grad_norm": 5.407328059506411,
"learning_rate": 1e-06,
"logits/chosen": 31.700279235839844,
"logits/rejected": 33.99253845214844,
"logps/chosen": -434.26495361328125,
"logps/rejected": -441.9063720703125,
"loss": 0.5,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"debug/policy_chosen_logits": 27.489120483398438,
"debug/policy_chosen_logps": -410.28472900390625,
"debug/policy_rejected_logits": 31.382970809936523,
"debug/policy_rejected_logps": -435.17218017578125,
"debug/reference_chosen_logps": -410.96771240234375,
"debug/reference_rejected_logps": -436.0491638183594,
"epoch": 0.0425531914893617,
"grad_norm": 4.951759612240564,
"learning_rate": 1e-06,
"logits/chosen": 27.489120483398438,
"logits/rejected": 31.382970809936523,
"logps/chosen": -410.28472900390625,
"logps/rejected": -435.17218017578125,
"loss": 0.5009,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.006829871796071529,
"rewards/margins": -0.001939887646585703,
"rewards/rejected": 0.00876975990831852,
"step": 2
},
{
"debug/policy_chosen_logits": 31.120014190673828,
"debug/policy_chosen_logps": -401.9127197265625,
"debug/policy_rejected_logits": 33.329689025878906,
"debug/policy_rejected_logps": -424.90576171875,
"debug/reference_chosen_logps": -402.24658203125,
"debug/reference_rejected_logps": -424.7574462890625,
"epoch": 0.06382978723404255,
"grad_norm": 5.411851250945231,
"learning_rate": 1e-06,
"logits/chosen": 31.120014190673828,
"logits/rejected": 33.329689025878906,
"logps/chosen": -401.9127197265625,
"logps/rejected": -424.90576171875,
"loss": 0.5007,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.003338394220918417,
"rewards/margins": 0.004821510519832373,
"rewards/rejected": -0.001483116764575243,
"step": 3
},
{
"debug/policy_chosen_logits": 30.066335678100586,
"debug/policy_chosen_logps": -403.6931457519531,
"debug/policy_rejected_logits": 33.15522384643555,
"debug/policy_rejected_logps": -436.77801513671875,
"debug/reference_chosen_logps": -403.52996826171875,
"debug/reference_rejected_logps": -436.29296875,
"epoch": 0.0851063829787234,
"grad_norm": 6.440647306952527,
"learning_rate": 1e-06,
"logits/chosen": 30.066335678100586,
"logits/rejected": 33.15522384643555,
"logps/chosen": -403.6931457519531,
"logps/rejected": -436.77801513671875,
"loss": 0.4993,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.0016318517737090588,
"rewards/margins": 0.003218421945348382,
"rewards/rejected": -0.00485027302056551,
"step": 4
},
{
"debug/policy_chosen_logits": 28.028703689575195,
"debug/policy_chosen_logps": -391.64715576171875,
"debug/policy_rejected_logits": 30.672496795654297,
"debug/policy_rejected_logps": -420.91143798828125,
"debug/reference_chosen_logps": -392.4825134277344,
"debug/reference_rejected_logps": -422.0054931640625,
"epoch": 0.10638297872340426,
"grad_norm": 5.433518269383661,
"learning_rate": 1e-06,
"logits/chosen": 28.028703689575195,
"logits/rejected": 30.672496795654297,
"logps/chosen": -391.64715576171875,
"logps/rejected": -420.91143798828125,
"loss": 0.4987,
"rewards/accuracies": 0.25,
"rewards/chosen": 0.008353347890079021,
"rewards/margins": -0.0025872797705233097,
"rewards/rejected": 0.010940628126263618,
"step": 5
},
{
"debug/policy_chosen_logits": 26.0015811920166,
"debug/policy_chosen_logps": -403.2935791015625,
"debug/policy_rejected_logits": 25.2414608001709,
"debug/policy_rejected_logps": -407.3106994628906,
"debug/reference_chosen_logps": -402.87139892578125,
"debug/reference_rejected_logps": -406.1798095703125,
"epoch": 0.1276595744680851,
"grad_norm": 5.048275333461177,
"learning_rate": 1e-06,
"logits/chosen": 26.0015811920166,
"logits/rejected": 25.2414608001709,
"logps/chosen": -403.2935791015625,
"logps/rejected": -407.3106994628906,
"loss": 0.496,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0042218780145049095,
"rewards/margins": 0.007086906582117081,
"rewards/rejected": -0.01130878459662199,
"step": 6
},
{
"debug/policy_chosen_logits": 27.019393920898438,
"debug/policy_chosen_logps": -424.65643310546875,
"debug/policy_rejected_logits": 29.134994506835938,
"debug/policy_rejected_logps": -420.3319396972656,
"debug/reference_chosen_logps": -423.5784912109375,
"debug/reference_rejected_logps": -420.0720520019531,
"epoch": 0.14893617021276595,
"grad_norm": 6.092801689841109,
"learning_rate": 1e-06,
"logits/chosen": 27.019393920898438,
"logits/rejected": 29.134994506835938,
"logps/chosen": -424.65643310546875,
"logps/rejected": -420.3319396972656,
"loss": 0.499,
"rewards/accuracies": 0.125,
"rewards/chosen": -0.010779608972370625,
"rewards/margins": -0.008180923759937286,
"rewards/rejected": -0.0025986863765865564,
"step": 7
},
{
"debug/policy_chosen_logits": 25.41461944580078,
"debug/policy_chosen_logps": -420.9305419921875,
"debug/policy_rejected_logits": 25.522966384887695,
"debug/policy_rejected_logps": -427.20623779296875,
"debug/reference_chosen_logps": -420.0583190917969,
"debug/reference_rejected_logps": -426.08453369140625,
"epoch": 0.1702127659574468,
"grad_norm": 5.248588154856571,
"learning_rate": 1e-06,
"logits/chosen": 25.41461944580078,
"logits/rejected": 25.522966384887695,
"logps/chosen": -420.9305419921875,
"logps/rejected": -427.20623779296875,
"loss": 0.4975,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.00872222799807787,
"rewards/margins": 0.0024948506616055965,
"rewards/rejected": -0.011217079125344753,
"step": 8
},
{
"debug/policy_chosen_logits": 27.28373146057129,
"debug/policy_chosen_logps": -413.857177734375,
"debug/policy_rejected_logits": 29.01516342163086,
"debug/policy_rejected_logps": -429.70623779296875,
"debug/reference_chosen_logps": -413.364501953125,
"debug/reference_rejected_logps": -429.416259765625,
"epoch": 0.19148936170212766,
"grad_norm": 5.836663309727503,
"learning_rate": 1e-06,
"logits/chosen": 27.28373146057129,
"logits/rejected": 29.01516342163086,
"logps/chosen": -413.857177734375,
"logps/rejected": -429.70623779296875,
"loss": 0.4954,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.004926986526697874,
"rewards/margins": -0.002027016133069992,
"rewards/rejected": -0.002899970393627882,
"step": 9
},
{
"debug/policy_chosen_logits": 29.55730628967285,
"debug/policy_chosen_logps": -418.648193359375,
"debug/policy_rejected_logits": 30.004676818847656,
"debug/policy_rejected_logps": -430.01788330078125,
"debug/reference_chosen_logps": -418.4599609375,
"debug/reference_rejected_logps": -429.8154296875,
"epoch": 0.2127659574468085,
"grad_norm": 4.978725766808406,
"learning_rate": 1e-06,
"logits/chosen": 29.55730628967285,
"logits/rejected": 30.004676818847656,
"logps/chosen": -418.648193359375,
"logps/rejected": -430.01788330078125,
"loss": 0.4991,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0018823242280632257,
"rewards/margins": 0.00014217384159564972,
"rewards/rejected": -0.002024497603997588,
"step": 10
},
{
"debug/policy_chosen_logits": 33.0296516418457,
"debug/policy_chosen_logps": -434.47308349609375,
"debug/policy_rejected_logits": 31.160263061523438,
"debug/policy_rejected_logps": -406.6353759765625,
"debug/reference_chosen_logps": -433.95892333984375,
"debug/reference_rejected_logps": -405.2964782714844,
"epoch": 0.23404255319148937,
"grad_norm": 5.822177618834045,
"learning_rate": 1e-06,
"logits/chosen": 33.0296516418457,
"logits/rejected": 31.160263061523438,
"logps/chosen": -434.47308349609375,
"logps/rejected": -406.6353759765625,
"loss": 0.493,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.005142059177160263,
"rewards/margins": 0.008247108198702335,
"rewards/rejected": -0.013389168307185173,
"step": 11
},
{
"debug/policy_chosen_logits": 30.720827102661133,
"debug/policy_chosen_logps": -455.06597900390625,
"debug/policy_rejected_logits": 32.45933151245117,
"debug/policy_rejected_logps": -462.2677307128906,
"debug/reference_chosen_logps": -454.6126403808594,
"debug/reference_rejected_logps": -459.7181396484375,
"epoch": 0.2553191489361702,
"grad_norm": 5.239137130887116,
"learning_rate": 1e-06,
"logits/chosen": 30.720827102661133,
"logits/rejected": 32.45933151245117,
"logps/chosen": -455.06597900390625,
"logps/rejected": -462.2677307128906,
"loss": 0.4991,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.00453338585793972,
"rewards/margins": 0.020962638780474663,
"rewards/rejected": -0.025496024638414383,
"step": 12
},
{
"debug/policy_chosen_logits": 30.186174392700195,
"debug/policy_chosen_logps": -412.29742431640625,
"debug/policy_rejected_logits": 28.243711471557617,
"debug/policy_rejected_logps": -426.9504089355469,
"debug/reference_chosen_logps": -411.92120361328125,
"debug/reference_rejected_logps": -425.7698974609375,
"epoch": 0.2765957446808511,
"grad_norm": 5.6983956081800855,
"learning_rate": 1e-06,
"logits/chosen": 30.186174392700195,
"logits/rejected": 28.243711471557617,
"logps/chosen": -412.29742431640625,
"logps/rejected": -426.9504089355469,
"loss": 0.4955,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.003762359730899334,
"rewards/margins": 0.008042870089411736,
"rewards/rejected": -0.011805228888988495,
"step": 13
},
{
"debug/policy_chosen_logits": 29.8179931640625,
"debug/policy_chosen_logps": -402.04205322265625,
"debug/policy_rejected_logits": 27.887521743774414,
"debug/policy_rejected_logps": -406.5090637207031,
"debug/reference_chosen_logps": -402.81463623046875,
"debug/reference_rejected_logps": -406.35760498046875,
"epoch": 0.2978723404255319,
"grad_norm": 5.185829515819964,
"learning_rate": 1e-06,
"logits/chosen": 29.8179931640625,
"logits/rejected": 27.887521743774414,
"logps/chosen": -402.04205322265625,
"logps/rejected": -406.5090637207031,
"loss": 0.4892,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.007725906558334827,
"rewards/margins": 0.009240342304110527,
"rewards/rejected": -0.0015144352801144123,
"step": 14
},
{
"debug/policy_chosen_logits": 30.058448791503906,
"debug/policy_chosen_logps": -412.29827880859375,
"debug/policy_rejected_logits": 29.466854095458984,
"debug/policy_rejected_logps": -412.73504638671875,
"debug/reference_chosen_logps": -411.7734680175781,
"debug/reference_rejected_logps": -413.09912109375,
"epoch": 0.3191489361702128,
"grad_norm": 5.147098230721813,
"learning_rate": 1e-06,
"logits/chosen": 30.058448791503906,
"logits/rejected": 29.466854095458984,
"logps/chosen": -412.29827880859375,
"logps/rejected": -412.73504638671875,
"loss": 0.4893,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.005248222034424543,
"rewards/margins": -0.008889121934771538,
"rewards/rejected": 0.0036408999003469944,
"step": 15
},
{
"debug/policy_chosen_logits": 26.801280975341797,
"debug/policy_chosen_logps": -453.10833740234375,
"debug/policy_rejected_logits": 28.296146392822266,
"debug/policy_rejected_logps": -433.0950927734375,
"debug/reference_chosen_logps": -453.86102294921875,
"debug/reference_rejected_logps": -432.42510986328125,
"epoch": 0.3404255319148936,
"grad_norm": 5.3650227932794765,
"learning_rate": 1e-06,
"logits/chosen": 26.801280975341797,
"logits/rejected": 28.296146392822266,
"logps/chosen": -453.10833740234375,
"logps/rejected": -433.0950927734375,
"loss": 0.4907,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.007526512257754803,
"rewards/margins": 0.014225959777832031,
"rewards/rejected": -0.0066994475200772285,
"step": 16
},
{
"debug/policy_chosen_logits": 25.992467880249023,
"debug/policy_chosen_logps": -436.1377258300781,
"debug/policy_rejected_logits": 27.410860061645508,
"debug/policy_rejected_logps": -426.43035888671875,
"debug/reference_chosen_logps": -434.6832275390625,
"debug/reference_rejected_logps": -424.5072021484375,
"epoch": 0.3617021276595745,
"grad_norm": 5.048216336945854,
"learning_rate": 1e-06,
"logits/chosen": 25.992467880249023,
"logits/rejected": 27.410860061645508,
"logps/chosen": -436.1377258300781,
"logps/rejected": -426.43035888671875,
"loss": 0.494,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.014545059762895107,
"rewards/margins": 0.004686659201979637,
"rewards/rejected": -0.01923171989619732,
"step": 17
},
{
"debug/policy_chosen_logits": 27.924072265625,
"debug/policy_chosen_logps": -456.6978759765625,
"debug/policy_rejected_logits": 27.263843536376953,
"debug/policy_rejected_logps": -411.67791748046875,
"debug/reference_chosen_logps": -455.6437683105469,
"debug/reference_rejected_logps": -408.3628234863281,
"epoch": 0.3829787234042553,
"grad_norm": 4.959644897985259,
"learning_rate": 1e-06,
"logits/chosen": 27.924072265625,
"logits/rejected": 27.263843536376953,
"logps/chosen": -456.6978759765625,
"logps/rejected": -411.67791748046875,
"loss": 0.4872,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.01054123044013977,
"rewards/margins": 0.022609787061810493,
"rewards/rejected": -0.033151015639305115,
"step": 18
},
{
"debug/policy_chosen_logits": 30.296974182128906,
"debug/policy_chosen_logps": -407.5791320800781,
"debug/policy_rejected_logits": 29.760583877563477,
"debug/policy_rejected_logps": -417.291748046875,
"debug/reference_chosen_logps": -410.10662841796875,
"debug/reference_rejected_logps": -418.6151123046875,
"epoch": 0.40425531914893614,
"grad_norm": 4.809418559441442,
"learning_rate": 1e-06,
"logits/chosen": 30.296974182128906,
"logits/rejected": 29.760583877563477,
"logps/chosen": -407.5791320800781,
"logps/rejected": -417.291748046875,
"loss": 0.4937,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.02527473494410515,
"rewards/margins": 0.012040939182043076,
"rewards/rejected": 0.013233794830739498,
"step": 19
},
{
"debug/policy_chosen_logits": 30.575592041015625,
"debug/policy_chosen_logps": -413.82574462890625,
"debug/policy_rejected_logits": 32.98490905761719,
"debug/policy_rejected_logps": -443.43548583984375,
"debug/reference_chosen_logps": -414.27642822265625,
"debug/reference_rejected_logps": -441.5928649902344,
"epoch": 0.425531914893617,
"grad_norm": 5.055368747694493,
"learning_rate": 1e-06,
"logits/chosen": 30.575592041015625,
"logits/rejected": 32.98490905761719,
"logps/chosen": -413.82574462890625,
"logps/rejected": -443.43548583984375,
"loss": 0.4771,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.0045069498009979725,
"rewards/margins": 0.02293361723423004,
"rewards/rejected": -0.018426666036248207,
"step": 20
},
{
"debug/policy_chosen_logits": 28.962617874145508,
"debug/policy_chosen_logps": -416.78582763671875,
"debug/policy_rejected_logits": 31.380332946777344,
"debug/policy_rejected_logps": -443.6494445800781,
"debug/reference_chosen_logps": -419.51043701171875,
"debug/reference_rejected_logps": -442.1171875,
"epoch": 0.44680851063829785,
"grad_norm": 5.2399641694392685,
"learning_rate": 1e-06,
"logits/chosen": 28.962617874145508,
"logits/rejected": 31.380332946777344,
"logps/chosen": -416.78582763671875,
"logps/rejected": -443.6494445800781,
"loss": 0.4861,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.02724616974592209,
"rewards/margins": 0.04256858676671982,
"rewards/rejected": -0.015322417952120304,
"step": 21
},
{
"debug/policy_chosen_logits": 33.89327621459961,
"debug/policy_chosen_logps": -439.533203125,
"debug/policy_rejected_logits": 32.8599853515625,
"debug/policy_rejected_logps": -468.5189208984375,
"debug/reference_chosen_logps": -437.9319763183594,
"debug/reference_rejected_logps": -460.019287109375,
"epoch": 0.46808510638297873,
"grad_norm": 5.600180944400873,
"learning_rate": 1e-06,
"logits/chosen": 33.89327621459961,
"logits/rejected": 32.8599853515625,
"logps/chosen": -439.533203125,
"logps/rejected": -468.5189208984375,
"loss": 0.4779,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.016012268140912056,
"rewards/margins": 0.06898414343595505,
"rewards/rejected": -0.08499641716480255,
"step": 22
},
{
"debug/policy_chosen_logits": 30.022546768188477,
"debug/policy_chosen_logps": -448.22540283203125,
"debug/policy_rejected_logits": 30.50183868408203,
"debug/policy_rejected_logps": -417.8683776855469,
"debug/reference_chosen_logps": -448.73858642578125,
"debug/reference_rejected_logps": -417.9635925292969,
"epoch": 0.48936170212765956,
"grad_norm": 5.053771531276715,
"learning_rate": 1e-06,
"logits/chosen": 30.022546768188477,
"logits/rejected": 30.50183868408203,
"logps/chosen": -448.22540283203125,
"logps/rejected": -417.8683776855469,
"loss": 0.4906,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0051317219622433186,
"rewards/margins": 0.004179535433650017,
"rewards/rejected": 0.0009521869942545891,
"step": 23
},
{
"debug/policy_chosen_logits": 31.068572998046875,
"debug/policy_chosen_logps": -408.3885192871094,
"debug/policy_rejected_logits": 30.79738426208496,
"debug/policy_rejected_logps": -432.73651123046875,
"debug/reference_chosen_logps": -406.73419189453125,
"debug/reference_rejected_logps": -432.0497131347656,
"epoch": 0.5106382978723404,
"grad_norm": 5.120783229464688,
"learning_rate": 1e-06,
"logits/chosen": 31.068572998046875,
"logits/rejected": 30.79738426208496,
"logps/chosen": -408.3885192871094,
"logps/rejected": -432.73651123046875,
"loss": 0.478,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.016543272882699966,
"rewards/margins": -0.009675255045294762,
"rewards/rejected": -0.006868018768727779,
"step": 24
},
{
"debug/policy_chosen_logits": 28.878725051879883,
"debug/policy_chosen_logps": -434.04144287109375,
"debug/policy_rejected_logits": 30.279621124267578,
"debug/policy_rejected_logps": -457.3016357421875,
"debug/reference_chosen_logps": -433.21746826171875,
"debug/reference_rejected_logps": -453.06280517578125,
"epoch": 0.5319148936170213,
"grad_norm": 5.297697739276052,
"learning_rate": 1e-06,
"logits/chosen": 28.878725051879883,
"logits/rejected": 30.279621124267578,
"logps/chosen": -434.04144287109375,
"logps/rejected": -457.3016357421875,
"loss": 0.485,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.00823978427797556,
"rewards/margins": 0.03414863348007202,
"rewards/rejected": -0.042388420552015305,
"step": 25
},
{
"debug/policy_chosen_logits": 30.609947204589844,
"debug/policy_chosen_logps": -402.73504638671875,
"debug/policy_rejected_logits": 29.12665367126465,
"debug/policy_rejected_logps": -411.1260986328125,
"debug/reference_chosen_logps": -404.96392822265625,
"debug/reference_rejected_logps": -411.49969482421875,
"epoch": 0.5531914893617021,
"grad_norm": 5.066192133102437,
"learning_rate": 1e-06,
"logits/chosen": 30.609947204589844,
"logits/rejected": 29.12665367126465,
"logps/chosen": -402.73504638671875,
"logps/rejected": -411.1260986328125,
"loss": 0.4781,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.022288817912340164,
"rewards/margins": 0.01855243556201458,
"rewards/rejected": 0.003736380487680435,
"step": 26
},
{
"debug/policy_chosen_logits": 26.225852966308594,
"debug/policy_chosen_logps": -434.0633544921875,
"debug/policy_rejected_logits": 27.547882080078125,
"debug/policy_rejected_logps": -460.3682861328125,
"debug/reference_chosen_logps": -434.0380859375,
"debug/reference_rejected_logps": -457.2252197265625,
"epoch": 0.574468085106383,
"grad_norm": 5.199359660864606,
"learning_rate": 1e-06,
"logits/chosen": 26.225852966308594,
"logits/rejected": 27.547882080078125,
"logps/chosen": -434.0633544921875,
"logps/rejected": -460.3682861328125,
"loss": 0.4775,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.0002528773620724678,
"rewards/margins": 0.031177710741758347,
"rewards/rejected": -0.03143058717250824,
"step": 27
},
{
"debug/policy_chosen_logits": 34.810482025146484,
"debug/policy_chosen_logps": -432.9176025390625,
"debug/policy_rejected_logits": 32.29673385620117,
"debug/policy_rejected_logps": -435.6657409667969,
"debug/reference_chosen_logps": -433.37603759765625,
"debug/reference_rejected_logps": -432.38958740234375,
"epoch": 0.5957446808510638,
"grad_norm": 5.456307945174751,
"learning_rate": 1e-06,
"logits/chosen": 34.810482025146484,
"logits/rejected": 32.29673385620117,
"logps/chosen": -432.9176025390625,
"logps/rejected": -435.6657409667969,
"loss": 0.4635,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.004584426060318947,
"rewards/margins": 0.03734596073627472,
"rewards/rejected": -0.03276153653860092,
"step": 28
},
{
"debug/policy_chosen_logits": 28.973360061645508,
"debug/policy_chosen_logps": -432.0859375,
"debug/policy_rejected_logits": 27.616941452026367,
"debug/policy_rejected_logps": -419.5810546875,
"debug/reference_chosen_logps": -432.6524658203125,
"debug/reference_rejected_logps": -413.3448181152344,
"epoch": 0.6170212765957447,
"grad_norm": 4.889651674840413,
"learning_rate": 1e-06,
"logits/chosen": 28.973360061645508,
"logits/rejected": 27.616941452026367,
"logps/chosen": -432.0859375,
"logps/rejected": -419.5810546875,
"loss": 0.4789,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.005664861761033535,
"rewards/margins": 0.06802742183208466,
"rewards/rejected": -0.062362559139728546,
"step": 29
},
{
"debug/policy_chosen_logits": 28.820457458496094,
"debug/policy_chosen_logps": -419.7233581542969,
"debug/policy_rejected_logits": 30.256000518798828,
"debug/policy_rejected_logps": -422.4107971191406,
"debug/reference_chosen_logps": -423.3680725097656,
"debug/reference_rejected_logps": -421.3091125488281,
"epoch": 0.6382978723404256,
"grad_norm": 5.137030977785722,
"learning_rate": 1e-06,
"logits/chosen": 28.820457458496094,
"logits/rejected": 30.256000518798828,
"logps/chosen": -419.7233581542969,
"logps/rejected": -422.4107971191406,
"loss": 0.4823,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.03644702956080437,
"rewards/margins": 0.047463756054639816,
"rewards/rejected": -0.011016730219125748,
"step": 30
},
{
"debug/policy_chosen_logits": 29.283926010131836,
"debug/policy_chosen_logps": -392.496826171875,
"debug/policy_rejected_logits": 31.77328109741211,
"debug/policy_rejected_logps": -434.51806640625,
"debug/reference_chosen_logps": -395.81146240234375,
"debug/reference_rejected_logps": -434.8221435546875,
"epoch": 0.6595744680851063,
"grad_norm": 4.951189622094444,
"learning_rate": 1e-06,
"logits/chosen": 29.283926010131836,
"logits/rejected": 31.77328109741211,
"logps/chosen": -392.496826171875,
"logps/rejected": -434.51806640625,
"loss": 0.4638,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.033146705478429794,
"rewards/margins": 0.03010578267276287,
"rewards/rejected": 0.003040926530957222,
"step": 31
},
{
"debug/policy_chosen_logits": 29.353422164916992,
"debug/policy_chosen_logps": -414.32415771484375,
"debug/policy_rejected_logits": 30.822248458862305,
"debug/policy_rejected_logps": -430.6376037597656,
"debug/reference_chosen_logps": -415.54888916015625,
"debug/reference_rejected_logps": -431.1400146484375,
"epoch": 0.6808510638297872,
"grad_norm": 4.999575032095535,
"learning_rate": 1e-06,
"logits/chosen": 29.353422164916992,
"logits/rejected": 30.822248458862305,
"logps/chosen": -414.32415771484375,
"logps/rejected": -430.6376037597656,
"loss": 0.4834,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.012247240170836449,
"rewards/margins": 0.007223015185445547,
"rewards/rejected": 0.00502422172576189,
"step": 32
},
{
"debug/policy_chosen_logits": 27.81666374206543,
"debug/policy_chosen_logps": -437.5671081542969,
"debug/policy_rejected_logits": 29.937236785888672,
"debug/policy_rejected_logps": -429.474853515625,
"debug/reference_chosen_logps": -440.85504150390625,
"debug/reference_rejected_logps": -431.23309326171875,
"epoch": 0.7021276595744681,
"grad_norm": 5.116678809536897,
"learning_rate": 1e-06,
"logits/chosen": 27.81666374206543,
"logits/rejected": 29.937236785888672,
"logps/chosen": -437.5671081542969,
"logps/rejected": -429.474853515625,
"loss": 0.4783,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.03287952393293381,
"rewards/margins": 0.01529712788760662,
"rewards/rejected": 0.017582397907972336,
"step": 33
},
{
"debug/policy_chosen_logits": 32.981014251708984,
"debug/policy_chosen_logps": -450.77899169921875,
"debug/policy_rejected_logits": 29.245454788208008,
"debug/policy_rejected_logps": -421.23468017578125,
"debug/reference_chosen_logps": -454.22735595703125,
"debug/reference_rejected_logps": -425.0584716796875,
"epoch": 0.723404255319149,
"grad_norm": 5.485713658124459,
"learning_rate": 1e-06,
"logits/chosen": 32.981014251708984,
"logits/rejected": 29.245454788208008,
"logps/chosen": -450.77899169921875,
"logps/rejected": -421.23468017578125,
"loss": 0.4736,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.03448398783802986,
"rewards/margins": -0.003754120320081711,
"rewards/rejected": 0.038238104432821274,
"step": 34
},
{
"debug/policy_chosen_logits": 30.9548397064209,
"debug/policy_chosen_logps": -427.41632080078125,
"debug/policy_rejected_logits": 29.430871963500977,
"debug/policy_rejected_logps": -436.83050537109375,
"debug/reference_chosen_logps": -430.6944580078125,
"debug/reference_rejected_logps": -433.17041015625,
"epoch": 0.7446808510638298,
"grad_norm": 5.73163901812078,
"learning_rate": 1e-06,
"logits/chosen": 30.9548397064209,
"logits/rejected": 29.430871963500977,
"logps/chosen": -427.41632080078125,
"logps/rejected": -436.83050537109375,
"loss": 0.4459,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.03278125822544098,
"rewards/margins": 0.06938225030899048,
"rewards/rejected": -0.0366009883582592,
"step": 35
},
{
"debug/policy_chosen_logits": 28.33571434020996,
"debug/policy_chosen_logps": -400.690673828125,
"debug/policy_rejected_logits": 24.813756942749023,
"debug/policy_rejected_logps": -445.54791259765625,
"debug/reference_chosen_logps": -404.2062072753906,
"debug/reference_rejected_logps": -433.4603576660156,
"epoch": 0.7659574468085106,
"grad_norm": 5.207209058021427,
"learning_rate": 1e-06,
"logits/chosen": 28.33571434020996,
"logits/rejected": 24.813756942749023,
"logps/chosen": -400.690673828125,
"logps/rejected": -445.54791259765625,
"loss": 0.4546,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.03515518084168434,
"rewards/margins": 0.15603074431419373,
"rewards/rejected": -0.12087554484605789,
"step": 36
},
{
"debug/policy_chosen_logits": 29.32408332824707,
"debug/policy_chosen_logps": -427.9587707519531,
"debug/policy_rejected_logits": 27.91067123413086,
"debug/policy_rejected_logps": -405.4839172363281,
"debug/reference_chosen_logps": -427.9952392578125,
"debug/reference_rejected_logps": -407.0904846191406,
"epoch": 0.7872340425531915,
"grad_norm": 5.293688440117633,
"learning_rate": 1e-06,
"logits/chosen": 29.32408332824707,
"logits/rejected": 27.91067123413086,
"logps/chosen": -427.9587707519531,
"logps/rejected": -405.4839172363281,
"loss": 0.5106,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.00036445818841457367,
"rewards/margins": -0.01570144295692444,
"rewards/rejected": 0.01606590300798416,
"step": 37
},
{
"debug/policy_chosen_logits": 28.01889419555664,
"debug/policy_chosen_logps": -404.5608825683594,
"debug/policy_rejected_logits": 28.02815055847168,
"debug/policy_rejected_logps": -413.2740173339844,
"debug/reference_chosen_logps": -405.4373779296875,
"debug/reference_rejected_logps": -406.2366027832031,
"epoch": 0.8085106382978723,
"grad_norm": 4.841695423817661,
"learning_rate": 1e-06,
"logits/chosen": 28.01889419555664,
"logits/rejected": 28.02815055847168,
"logps/chosen": -404.5608825683594,
"logps/rejected": -413.2740173339844,
"loss": 0.4719,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.00876510702073574,
"rewards/margins": 0.07913925498723984,
"rewards/rejected": -0.07037414610385895,
"step": 38
},
{
"debug/policy_chosen_logits": 26.369901657104492,
"debug/policy_chosen_logps": -418.9932556152344,
"debug/policy_rejected_logits": 24.119754791259766,
"debug/policy_rejected_logps": -405.0431213378906,
"debug/reference_chosen_logps": -421.83203125,
"debug/reference_rejected_logps": -404.04913330078125,
"epoch": 0.8297872340425532,
"grad_norm": 5.464322991896784,
"learning_rate": 1e-06,
"logits/chosen": 26.369901657104492,
"logits/rejected": 24.119754791259766,
"logps/chosen": -418.9932556152344,
"logps/rejected": -405.0431213378906,
"loss": 0.4666,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.028387565165758133,
"rewards/margins": 0.03832760080695152,
"rewards/rejected": -0.009940031915903091,
"step": 39
},
{
"debug/policy_chosen_logits": 31.222116470336914,
"debug/policy_chosen_logps": -408.92315673828125,
"debug/policy_rejected_logits": 30.565526962280273,
"debug/policy_rejected_logps": -438.00384521484375,
"debug/reference_chosen_logps": -410.3876037597656,
"debug/reference_rejected_logps": -428.6034851074219,
"epoch": 0.851063829787234,
"grad_norm": 5.087286245449022,
"learning_rate": 1e-06,
"logits/chosen": 31.222116470336914,
"logits/rejected": 30.565526962280273,
"logps/chosen": -408.92315673828125,
"logps/rejected": -438.00384521484375,
"loss": 0.4691,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.014644507318735123,
"rewards/margins": 0.10864795744419098,
"rewards/rejected": -0.09400344640016556,
"step": 40
},
{
"debug/policy_chosen_logits": 29.859107971191406,
"debug/policy_chosen_logps": -403.02581787109375,
"debug/policy_rejected_logits": 23.891035079956055,
"debug/policy_rejected_logps": -395.781005859375,
"debug/reference_chosen_logps": -409.97967529296875,
"debug/reference_rejected_logps": -398.636474609375,
"epoch": 0.8723404255319149,
"grad_norm": 4.953135237737809,
"learning_rate": 1e-06,
"logits/chosen": 29.859107971191406,
"logits/rejected": 23.891035079956055,
"logps/chosen": -403.02581787109375,
"logps/rejected": -395.781005859375,
"loss": 0.458,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.06953833997249603,
"rewards/margins": 0.04098331928253174,
"rewards/rejected": 0.02855503186583519,
"step": 41
},
{
"debug/policy_chosen_logits": 31.02197265625,
"debug/policy_chosen_logps": -424.50299072265625,
"debug/policy_rejected_logits": 29.632427215576172,
"debug/policy_rejected_logps": -422.94561767578125,
"debug/reference_chosen_logps": -429.08648681640625,
"debug/reference_rejected_logps": -424.3019104003906,
"epoch": 0.8936170212765957,
"grad_norm": 5.514556663745466,
"learning_rate": 1e-06,
"logits/chosen": 31.02197265625,
"logits/rejected": 29.632427215576172,
"logps/chosen": -424.50299072265625,
"logps/rejected": -422.94561767578125,
"loss": 0.4735,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.04583461582660675,
"rewards/margins": 0.032271310687065125,
"rewards/rejected": 0.01356330793350935,
"step": 42
},
{
"debug/policy_chosen_logits": 32.29981231689453,
"debug/policy_chosen_logps": -437.9972229003906,
"debug/policy_rejected_logits": 30.15468978881836,
"debug/policy_rejected_logps": -440.30535888671875,
"debug/reference_chosen_logps": -436.08892822265625,
"debug/reference_rejected_logps": -434.7149353027344,
"epoch": 0.9148936170212766,
"grad_norm": 4.808288177536615,
"learning_rate": 1e-06,
"logits/chosen": 32.29981231689453,
"logits/rejected": 30.15468978881836,
"logps/chosen": -437.9972229003906,
"logps/rejected": -440.30535888671875,
"loss": 0.4598,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.01908310130238533,
"rewards/margins": 0.03682101517915726,
"rewards/rejected": -0.055904120206832886,
"step": 43
},
{
"debug/policy_chosen_logits": 26.440486907958984,
"debug/policy_chosen_logps": -391.61572265625,
"debug/policy_rejected_logits": 29.678592681884766,
"debug/policy_rejected_logps": -430.9735412597656,
"debug/reference_chosen_logps": -396.64862060546875,
"debug/reference_rejected_logps": -434.947998046875,
"epoch": 0.9361702127659575,
"grad_norm": 5.150941817603919,
"learning_rate": 1e-06,
"logits/chosen": 26.440486907958984,
"logits/rejected": 29.678592681884766,
"logps/chosen": -391.61572265625,
"logps/rejected": -430.9735412597656,
"loss": 0.4568,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.050329361110925674,
"rewards/margins": 0.010584792122244835,
"rewards/rejected": 0.03974456712603569,
"step": 44
},
{
"debug/policy_chosen_logits": 29.451526641845703,
"debug/policy_chosen_logps": -425.8896484375,
"debug/policy_rejected_logits": 32.46401596069336,
"debug/policy_rejected_logps": -428.1052551269531,
"debug/reference_chosen_logps": -424.65936279296875,
"debug/reference_rejected_logps": -427.16961669921875,
"epoch": 0.9574468085106383,
"grad_norm": 5.012447008649623,
"learning_rate": 1e-06,
"logits/chosen": 29.451526641845703,
"logits/rejected": 32.46401596069336,
"logps/chosen": -425.8896484375,
"logps/rejected": -428.1052551269531,
"loss": 0.4777,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.012303046882152557,
"rewards/margins": -0.0029468159191310406,
"rewards/rejected": -0.009356231428682804,
"step": 45
},
{
"debug/policy_chosen_logits": 31.168346405029297,
"debug/policy_chosen_logps": -426.1267395019531,
"debug/policy_rejected_logits": 29.51166534423828,
"debug/policy_rejected_logps": -444.91766357421875,
"debug/reference_chosen_logps": -429.6617736816406,
"debug/reference_rejected_logps": -439.4256591796875,
"epoch": 0.9787234042553191,
"grad_norm": 5.030625016447312,
"learning_rate": 1e-06,
"logits/chosen": 31.168346405029297,
"logits/rejected": 29.51166534423828,
"logps/chosen": -426.1267395019531,
"logps/rejected": -444.91766357421875,
"loss": 0.4647,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.03535018861293793,
"rewards/margins": 0.09027023613452911,
"rewards/rejected": -0.05492004379630089,
"step": 46
},
{
"debug/policy_chosen_logits": 28.43193244934082,
"debug/policy_chosen_logps": -419.4750671386719,
"debug/policy_rejected_logits": 27.273754119873047,
"debug/policy_rejected_logps": -438.4751892089844,
"debug/reference_chosen_logps": -421.2059326171875,
"debug/reference_rejected_logps": -437.2878723144531,
"epoch": 1.0,
"grad_norm": 5.486914446149956,
"learning_rate": 1e-06,
"logits/chosen": 28.43193244934082,
"logits/rejected": 27.273754119873047,
"logps/chosen": -419.4750671386719,
"logps/rejected": -438.4751892089844,
"loss": 0.4596,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.01730876788496971,
"rewards/margins": 0.02918224036693573,
"rewards/rejected": -0.011873474344611168,
"step": 47
},
{
"epoch": 1.0,
"step": 47,
"total_flos": 0.0,
"train_loss": 0.48215872493196044,
"train_runtime": 474.4995,
"train_samples_per_second": 6.327,
"train_steps_per_second": 0.099
}
],
"logging_steps": 1,
"max_steps": 47,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}