jackf857/llama-3-8b-base-slic-hf-ultrafeedback-4xh200
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: jackf857/llama-3-8b-base-slic-hf-ultrafeedback-4xh200
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-04-23 22:59:07 +08:00
commit c83c41b2f0
19 changed files with 3507 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

895
trainer_state.json Normal file
View File

@@ -0,0 +1,895 @@
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9989528795811519,
"eval_steps": 200,
"global_step": 477,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0020942408376963353,
"grad_norm": 6127.23876953125,
"learning_rate": 0.0,
"logits/chosen": -0.6042112112045288,
"logits/rejected": -0.6177991628646851,
"logps/chosen": -275.48590087890625,
"logps/rejected": -223.16470336914062,
"loss": 1522.3201,
"rewards/accuracies": 0.46875,
"rewards/chosen": -275.48590087890625,
"rewards/margins": -52.32117462158203,
"rewards/rejected": -223.16470336914062,
"slic/ce_loss": 275.48590087890625,
"slic/rank_loss": 105.09413146972656,
"step": 1
},
{
"epoch": 0.020942408376963352,
"grad_norm": 5911.2392578125,
"learning_rate": 9.375e-08,
"logits/chosen": -0.6444513201713562,
"logits/rejected": -0.6520731449127197,
"logps/chosen": -290.2828674316406,
"logps/rejected": -264.8289489746094,
"loss": 1545.4015,
"rewards/accuracies": 0.4618055522441864,
"rewards/chosen": -290.2828674316406,
"rewards/margins": -25.45393943786621,
"rewards/rejected": -264.8289489746094,
"slic/ce_loss": 290.2828674316406,
"slic/rank_loss": 96.06748962402344,
"step": 10
},
{
"epoch": 0.041884816753926704,
"grad_norm": 6397.529296875,
"learning_rate": 1.9791666666666664e-07,
"logits/chosen": -0.6179991364479065,
"logits/rejected": -0.6330140829086304,
"logps/chosen": -286.71014404296875,
"logps/rejected": -259.08758544921875,
"loss": 1526.0234,
"rewards/accuracies": 0.47265625,
"rewards/chosen": -286.71014404296875,
"rewards/margins": -27.622533798217773,
"rewards/rejected": -259.08758544921875,
"slic/ce_loss": 286.71014404296875,
"slic/rank_loss": 94.79573059082031,
"step": 20
},
{
"epoch": 0.06282722513089005,
"grad_norm": 4871.51513671875,
"learning_rate": 3.020833333333333e-07,
"logits/chosen": -0.6329924464225769,
"logits/rejected": -0.6390506029129028,
"logps/chosen": -277.3640441894531,
"logps/rejected": -255.4713134765625,
"loss": 1477.4605,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -277.3640441894531,
"rewards/margins": -21.892715454101562,
"rewards/rejected": -255.4713134765625,
"slic/ce_loss": 277.3640441894531,
"slic/rank_loss": 92.00110626220703,
"step": 30
},
{
"epoch": 0.08376963350785341,
"grad_norm": 4104.8447265625,
"learning_rate": 4.0625e-07,
"logits/chosen": -0.6297017931938171,
"logits/rejected": -0.6496596932411194,
"logps/chosen": -279.53680419921875,
"logps/rejected": -251.68435668945312,
"loss": 1506.1701,
"rewards/accuracies": 0.47578126192092896,
"rewards/chosen": -279.53680419921875,
"rewards/margins": -27.85247230529785,
"rewards/rejected": -251.68435668945312,
"slic/ce_loss": 279.53680419921875,
"slic/rank_loss": 97.00569915771484,
"step": 40
},
{
"epoch": 0.10471204188481675,
"grad_norm": 3677.18603515625,
"learning_rate": 4.999932966293553e-07,
"logits/chosen": -0.6434795260429382,
"logits/rejected": -0.6600109338760376,
"logps/chosen": -273.2533874511719,
"logps/rejected": -256.40338134765625,
"loss": 1450.0907,
"rewards/accuracies": 0.48359376192092896,
"rewards/chosen": -273.2533874511719,
"rewards/margins": -16.849985122680664,
"rewards/rejected": -256.40338134765625,
"slic/ce_loss": 273.2533874511719,
"slic/rank_loss": 89.26930236816406,
"step": 50
},
{
"epoch": 0.1256544502617801,
"grad_norm": 3486.959716796875,
"learning_rate": 4.991893270335525e-07,
"logits/chosen": -0.6477519869804382,
"logits/rejected": -0.6574610471725464,
"logps/chosen": -261.81854248046875,
"logps/rejected": -248.384521484375,
"loss": 1407.8849,
"rewards/accuracies": 0.503125011920929,
"rewards/chosen": -261.81854248046875,
"rewards/margins": -13.434022903442383,
"rewards/rejected": -248.384521484375,
"slic/ce_loss": 261.81854248046875,
"slic/rank_loss": 90.15267944335938,
"step": 60
},
{
"epoch": 0.14659685863874344,
"grad_norm": 3554.351806640625,
"learning_rate": 4.970496218214204e-07,
"logits/chosen": -0.6433520913124084,
"logits/rejected": -0.6552119851112366,
"logps/chosen": -261.0797424316406,
"logps/rejected": -245.8175506591797,
"loss": 1383.9182,
"rewards/accuracies": 0.48828125,
"rewards/chosen": -261.0797424316406,
"rewards/margins": -15.262176513671875,
"rewards/rejected": -245.8175506591797,
"slic/ce_loss": 261.0797424316406,
"slic/rank_loss": 84.89978790283203,
"step": 70
},
{
"epoch": 0.16753926701570682,
"grad_norm": 3468.174072265625,
"learning_rate": 4.935856505068998e-07,
"logits/chosen": -0.6140779256820679,
"logits/rejected": -0.6229384541511536,
"logps/chosen": -262.1397705078125,
"logps/rejected": -246.156494140625,
"loss": 1382.4392,
"rewards/accuracies": 0.47578126192092896,
"rewards/chosen": -262.1397705078125,
"rewards/margins": -15.983256340026855,
"rewards/rejected": -246.156494140625,
"slic/ce_loss": 262.1397705078125,
"slic/rank_loss": 83.47003173828125,
"step": 80
},
{
"epoch": 0.18848167539267016,
"grad_norm": 3276.579833984375,
"learning_rate": 4.8881598109976e-07,
"logits/chosen": -0.637958288192749,
"logits/rejected": -0.6412946581840515,
"logps/chosen": -259.00439453125,
"logps/rejected": -244.78060913085938,
"loss": 1378.6654,
"rewards/accuracies": 0.5,
"rewards/chosen": -259.00439453125,
"rewards/margins": -14.223783493041992,
"rewards/rejected": -244.78060913085938,
"slic/ce_loss": 259.00439453125,
"slic/rank_loss": 85.66192626953125,
"step": 90
},
{
"epoch": 0.2094240837696335,
"grad_norm": 3682.858642578125,
"learning_rate": 4.827661805750437e-07,
"logits/chosen": -0.6100295782089233,
"logits/rejected": -0.6224995255470276,
"logps/chosen": -260.15911865234375,
"logps/rejected": -240.53005981445312,
"loss": 1390.1286,
"rewards/accuracies": 0.49921876192092896,
"rewards/chosen": -260.15911865234375,
"rewards/margins": -19.629114151000977,
"rewards/rejected": -240.53005981445312,
"slic/ce_loss": 260.15911865234375,
"slic/rank_loss": 87.37300109863281,
"step": 100
},
{
"epoch": 0.23036649214659685,
"grad_norm": 3362.56884765625,
"learning_rate": 4.75468677825789e-07,
"logits/chosen": -0.6101893186569214,
"logits/rejected": -0.6191698312759399,
"logps/chosen": -259.8664245605469,
"logps/rejected": -244.67868041992188,
"loss": 1384.9318,
"rewards/accuracies": 0.4765625,
"rewards/chosen": -259.8664245605469,
"rewards/margins": -15.187728881835938,
"rewards/rejected": -244.67868041992188,
"slic/ce_loss": 259.8664245605469,
"slic/rank_loss": 86.36650848388672,
"step": 110
},
{
"epoch": 0.2513089005235602,
"grad_norm": 3407.73095703125,
"learning_rate": 4.669625898336438e-07,
"logits/chosen": -0.6205486059188843,
"logits/rejected": -0.62354576587677,
"logps/chosen": -264.4170227050781,
"logps/rejected": -248.2465362548828,
"loss": 1411.6574,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -264.4170227050781,
"rewards/margins": -16.170438766479492,
"rewards/rejected": -248.2465362548828,
"slic/ce_loss": 264.4170227050781,
"slic/rank_loss": 88.49736022949219,
"step": 120
},
{
"epoch": 0.27225130890052357,
"grad_norm": 3431.46142578125,
"learning_rate": 4.5729351198915705e-07,
"logits/chosen": -0.6135658025741577,
"logits/rejected": -0.6137000322341919,
"logps/chosen": -263.35980224609375,
"logps/rejected": -245.09103393554688,
"loss": 1415.0682,
"rewards/accuracies": 0.4921875,
"rewards/chosen": -263.35980224609375,
"rewards/margins": -18.268768310546875,
"rewards/rejected": -245.09103393554688,
"slic/ce_loss": 263.35980224609375,
"slic/rank_loss": 90.4072494506836,
"step": 130
},
{
"epoch": 0.2931937172774869,
"grad_norm": 3545.783935546875,
"learning_rate": 4.4651327368569684e-07,
"logits/chosen": -0.6140355467796326,
"logits/rejected": -0.6084020733833313,
"logps/chosen": -265.9866027832031,
"logps/rejected": -250.787109375,
"loss": 1405.8035,
"rewards/accuracies": 0.4984374940395355,
"rewards/chosen": -265.9866027832031,
"rewards/margins": -15.199483871459961,
"rewards/rejected": -250.787109375,
"slic/ce_loss": 265.9866027832031,
"slic/rank_loss": 85.46427917480469,
"step": 140
},
{
"epoch": 0.31413612565445026,
"grad_norm": 3279.9306640625,
"learning_rate": 4.346796604970912e-07,
"logits/chosen": -0.610595703125,
"logits/rejected": -0.6149523258209229,
"logps/chosen": -262.4765930175781,
"logps/rejected": -238.65792846679688,
"loss": 1396.4196,
"rewards/accuracies": 0.4585937559604645,
"rewards/chosen": -262.4765930175781,
"rewards/margins": -23.818683624267578,
"rewards/rejected": -238.65792846679688,
"slic/ce_loss": 262.4765930175781,
"slic/rank_loss": 86.62831115722656,
"step": 150
},
{
"epoch": 0.33507853403141363,
"grad_norm": 3371.940673828125,
"learning_rate": 4.218561044282098e-07,
"logits/chosen": -0.6132291555404663,
"logits/rejected": -0.615606963634491,
"logps/chosen": -260.2372131347656,
"logps/rejected": -246.6212158203125,
"loss": 1395.4537,
"rewards/accuracies": 0.4976562559604645,
"rewards/chosen": -260.2372131347656,
"rewards/margins": -13.615992546081543,
"rewards/rejected": -246.6212158203125,
"slic/ce_loss": 260.2372131347656,
"slic/rank_loss": 88.62622833251953,
"step": 160
},
{
"epoch": 0.35602094240837695,
"grad_norm": 3451.259521484375,
"learning_rate": 4.081113438988443e-07,
"logits/chosen": -0.6058305501937866,
"logits/rejected": -0.6137093305587769,
"logps/chosen": -264.1838684082031,
"logps/rejected": -232.7599639892578,
"loss": 1435.0535,
"rewards/accuracies": 0.46406251192092896,
"rewards/chosen": -264.1838684082031,
"rewards/margins": -31.423877716064453,
"rewards/rejected": -232.7599639892578,
"slic/ce_loss": 264.1838684082031,
"slic/rank_loss": 94.57951354980469,
"step": 170
},
{
"epoch": 0.3769633507853403,
"grad_norm": 3250.255126953125,
"learning_rate": 3.935190552834828e-07,
"logits/chosen": -0.604538083076477,
"logits/rejected": -0.6160414814949036,
"logps/chosen": -263.60711669921875,
"logps/rejected": -244.922607421875,
"loss": 1397.7748,
"rewards/accuracies": 0.48828125,
"rewards/chosen": -263.60711669921875,
"rewards/margins": -18.68451499938965,
"rewards/rejected": -244.922607421875,
"slic/ce_loss": 263.60711669921875,
"slic/rank_loss": 85.83655548095703,
"step": 180
},
{
"epoch": 0.39790575916230364,
"grad_norm": 3169.67041015625,
"learning_rate": 3.781574579820464e-07,
"logits/chosen": -0.604019820690155,
"logits/rejected": -0.6174649596214294,
"logps/chosen": -261.44183349609375,
"logps/rejected": -239.5850830078125,
"loss": 1405.417,
"rewards/accuracies": 0.4765625,
"rewards/chosen": -261.44183349609375,
"rewards/margins": -21.85675621032715,
"rewards/rejected": -239.5850830078125,
"slic/ce_loss": 261.44183349609375,
"slic/rank_loss": 89.91242980957031,
"step": 190
},
{
"epoch": 0.418848167539267,
"grad_norm": 3375.69482421875,
"learning_rate": 3.621088951385353e-07,
"logits/chosen": -0.6043376922607422,
"logits/rejected": -0.6085049510002136,
"logps/chosen": -257.392333984375,
"logps/rejected": -241.6600799560547,
"loss": 1368.2239,
"rewards/accuracies": 0.49609375,
"rewards/chosen": -257.392333984375,
"rewards/margins": -15.732267379760742,
"rewards/rejected": -241.6600799560547,
"slic/ce_loss": 257.392333984375,
"slic/rank_loss": 84.66365814208984,
"step": 200
},
{
"epoch": 0.418848167539267,
"eval_logits/chosen": -0.6125423312187195,
"eval_logits/rejected": -0.6204564571380615,
"eval_logps/chosen": -262.1701965332031,
"eval_logps/rejected": -246.15359497070312,
"eval_loss": 345.58563232421875,
"eval_rewards/accuracies": 0.4873991906642914,
"eval_rewards/chosen": -262.1701965332031,
"eval_rewards/margins": -16.016645431518555,
"eval_rewards/rejected": -246.15359497070312,
"eval_runtime": 23.996,
"eval_samples_per_second": 83.347,
"eval_slic/ce_loss": 262.1701965332031,
"eval_slic/rank_loss": 83.41545104980469,
"eval_steps_per_second": 2.625,
"step": 200
},
{
"epoch": 0.4397905759162304,
"grad_norm": 3124.86767578125,
"learning_rate": 3.454593922550693e-07,
"logits/chosen": -0.6011446714401245,
"logits/rejected": -0.6014553308486938,
"logps/chosen": -262.05340576171875,
"logps/rejected": -250.4723358154297,
"loss": 1388.7663,
"rewards/accuracies": 0.4781250059604645,
"rewards/chosen": -262.05340576171875,
"rewards/margins": -11.581093788146973,
"rewards/rejected": -250.4723358154297,
"slic/ce_loss": 262.05340576171875,
"slic/rank_loss": 85.13812255859375,
"step": 210
},
{
"epoch": 0.4607329842931937,
"grad_norm": 3453.45849609375,
"learning_rate": 3.2829819606729477e-07,
"logits/chosen": -0.595199704170227,
"logits/rejected": -0.6031100153923035,
"logps/chosen": -268.83941650390625,
"logps/rejected": -252.3756866455078,
"loss": 1436.1566,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -268.83941650390625,
"rewards/margins": -16.46368980407715,
"rewards/rejected": -252.3756866455078,
"slic/ce_loss": 268.83941650390625,
"slic/rank_loss": 90.19975280761719,
"step": 220
},
{
"epoch": 0.4816753926701571,
"grad_norm": 3144.963623046875,
"learning_rate": 3.1071729615293424e-07,
"logits/chosen": -0.5905068516731262,
"logits/rejected": -0.6006661653518677,
"logps/chosen": -256.08294677734375,
"logps/rejected": -239.11331176757812,
"loss": 1356.9873,
"rewards/accuracies": 0.48359376192092896,
"rewards/chosen": -256.08294677734375,
"rewards/margins": -16.969640731811523,
"rewards/rejected": -239.11331176757812,
"slic/ce_loss": 256.08294677734375,
"slic/rank_loss": 83.16390228271484,
"step": 230
},
{
"epoch": 0.5026178010471204,
"grad_norm": 3449.269287109375,
"learning_rate": 2.9281093183781403e-07,
"logits/chosen": -0.5943664312362671,
"logits/rejected": -0.6033399701118469,
"logps/chosen": -262.95050048828125,
"logps/rejected": -244.4707794189453,
"loss": 1402.5626,
"rewards/accuracies": 0.4742187559604645,
"rewards/chosen": -262.95050048828125,
"rewards/margins": -18.479671478271484,
"rewards/rejected": -244.4707794189453,
"slic/ce_loss": 262.95050048828125,
"slic/rank_loss": 87.6901626586914,
"step": 240
},
{
"epoch": 0.5235602094240838,
"grad_norm": 3244.26611328125,
"learning_rate": 2.7467508704251135e-07,
"logits/chosen": -0.5862177610397339,
"logits/rejected": -0.5929856300354004,
"logps/chosen": -261.1794738769531,
"logps/rejected": -238.1936492919922,
"loss": 1404.9589,
"rewards/accuracies": 0.4742187559604645,
"rewards/chosen": -261.1794738769531,
"rewards/margins": -22.985858917236328,
"rewards/rejected": -238.1936492919922,
"slic/ce_loss": 261.1794738769531,
"slic/rank_loss": 90.06024169921875,
"step": 250
},
{
"epoch": 0.5445026178010471,
"grad_norm": 3298.73388671875,
"learning_rate": 2.5640697577740815e-07,
"logits/chosen": -0.6019514203071594,
"logits/rejected": -0.5996894836425781,
"logps/chosen": -261.6252746582031,
"logps/rejected": -237.8870391845703,
"loss": 1411.4449,
"rewards/accuracies": 0.46484375,
"rewards/chosen": -261.6252746582031,
"rewards/margins": -23.738243103027344,
"rewards/rejected": -237.8870391845703,
"slic/ce_loss": 261.6252746582031,
"slic/rank_loss": 91.23591613769531,
"step": 260
},
{
"epoch": 0.5654450261780105,
"grad_norm": 3326.556884765625,
"learning_rate": 2.381045210440644e-07,
"logits/chosen": -0.5840014219284058,
"logits/rejected": -0.5949230790138245,
"logps/chosen": -254.5701446533203,
"logps/rejected": -237.69515991210938,
"loss": 1350.7662,
"rewards/accuracies": 0.48046875,
"rewards/chosen": -254.5701446533203,
"rewards/margins": -16.87495231628418,
"rewards/rejected": -237.69515991210938,
"slic/ce_loss": 254.5701446533203,
"slic/rank_loss": 83.12141418457031,
"step": 270
},
{
"epoch": 0.5863874345549738,
"grad_norm": 3088.087646484375,
"learning_rate": 2.1986582993616925e-07,
"logits/chosen": -0.5930246114730835,
"logits/rejected": -0.5939750075340271,
"logps/chosen": -253.7728271484375,
"logps/rejected": -241.112548828125,
"loss": 1342.3729,
"rewards/accuracies": 0.48906248807907104,
"rewards/chosen": -253.7728271484375,
"rewards/margins": -12.66028118133545,
"rewards/rejected": -241.112548828125,
"slic/ce_loss": 253.7728271484375,
"slic/rank_loss": 81.8204116821289,
"step": 280
},
{
"epoch": 0.6073298429319371,
"grad_norm": 3437.228759765625,
"learning_rate": 2.0178866775369774e-07,
"logits/chosen": -0.5804582238197327,
"logits/rejected": -0.5853986740112305,
"logps/chosen": -268.73333740234375,
"logps/rejected": -250.84423828125,
"loss": 1440.4078,
"rewards/accuracies": 0.48046875,
"rewards/chosen": -268.73333740234375,
"rewards/margins": -17.889144897460938,
"rewards/rejected": -250.84423828125,
"slic/ce_loss": 268.73333740234375,
"slic/rank_loss": 91.36856842041016,
"step": 290
},
{
"epoch": 0.6282722513089005,
"grad_norm": 3455.057861328125,
"learning_rate": 1.839699339491937e-07,
"logits/chosen": -0.590076744556427,
"logits/rejected": -0.5911703109741211,
"logps/chosen": -255.82763671875,
"logps/rejected": -247.8886260986328,
"loss": 1343.457,
"rewards/accuracies": 0.507031261920929,
"rewards/chosen": -255.82763671875,
"rewards/margins": -7.939053535461426,
"rewards/rejected": -247.8886260986328,
"slic/ce_loss": 255.82763671875,
"slic/rank_loss": 80.03659057617188,
"step": 300
},
{
"epoch": 0.6492146596858639,
"grad_norm": 3349.302734375,
"learning_rate": 1.6650514271527465e-07,
"logits/chosen": -0.5793516635894775,
"logits/rejected": -0.5949603319168091,
"logps/chosen": -258.7000427246094,
"logps/rejected": -238.9949188232422,
"loss": 1385.5395,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -258.7000427246094,
"rewards/margins": -19.705102920532227,
"rewards/rejected": -238.9949188232422,
"slic/ce_loss": 258.7000427246094,
"slic/rank_loss": 87.68482971191406,
"step": 310
},
{
"epoch": 0.6701570680628273,
"grad_norm": 3584.829345703125,
"learning_rate": 1.4948791099758052e-07,
"logits/chosen": -0.6042912006378174,
"logits/rejected": -0.602425217628479,
"logps/chosen": -263.06732177734375,
"logps/rejected": -240.22341918945312,
"loss": 1407.0932,
"rewards/accuracies": 0.4867187440395355,
"rewards/chosen": -263.06732177734375,
"rewards/margins": -22.843936920166016,
"rewards/rejected": -240.22341918945312,
"slic/ce_loss": 263.06732177734375,
"slic/rank_loss": 88.70594787597656,
"step": 320
},
{
"epoch": 0.6910994764397905,
"grad_norm": 3191.279296875,
"learning_rate": 1.3300945667758012e-07,
"logits/chosen": -0.5990831255912781,
"logits/rejected": -0.5979740023612976,
"logps/chosen": -257.674072265625,
"logps/rejected": -244.77487182617188,
"loss": 1362.2186,
"rewards/accuracies": 0.504687488079071,
"rewards/chosen": -257.674072265625,
"rewards/margins": -12.899203300476074,
"rewards/rejected": -244.77487182617188,
"slic/ce_loss": 257.674072265625,
"slic/rank_loss": 82.88054656982422,
"step": 330
},
{
"epoch": 0.7120418848167539,
"grad_norm": 2967.82421875,
"learning_rate": 1.1715810961514072e-07,
"logits/chosen": -0.6025289297103882,
"logits/rejected": -0.5973036885261536,
"logps/chosen": -251.06820678710938,
"logps/rejected": -234.9936065673828,
"loss": 1341.9119,
"rewards/accuracies": 0.4859375059604645,
"rewards/chosen": -251.06820678710938,
"rewards/margins": -16.074626922607422,
"rewards/rejected": -234.9936065673828,
"slic/ce_loss": 251.06820678710938,
"slic/rank_loss": 84.40972900390625,
"step": 340
},
{
"epoch": 0.7329842931937173,
"grad_norm": 4420.080078125,
"learning_rate": 1.0201883817182949e-07,
"logits/chosen": -0.5965454578399658,
"logits/rejected": -0.6024629473686218,
"logps/chosen": -265.9053955078125,
"logps/rejected": -244.15786743164062,
"loss": 1410.9908,
"rewards/accuracies": 0.4546875059604645,
"rewards/chosen": -265.9053955078125,
"rewards/margins": -21.74753189086914,
"rewards/rejected": -244.15786743164062,
"slic/ce_loss": 265.9053955078125,
"slic/rank_loss": 86.84226989746094,
"step": 350
},
{
"epoch": 0.7539267015706806,
"grad_norm": 3473.728515625,
"learning_rate": 8.76727937529367e-08,
"logits/chosen": -0.5978257060050964,
"logits/rejected": -0.6121166944503784,
"logps/chosen": -250.9609832763672,
"logps/rejected": -233.37783813476562,
"loss": 1331.2162,
"rewards/accuracies": 0.500781238079071,
"rewards/chosen": -250.9609832763672,
"rewards/margins": -17.583179473876953,
"rewards/rejected": -233.37783813476562,
"slic/ce_loss": 250.9609832763672,
"slic/rank_loss": 81.8430404663086,
"step": 360
},
{
"epoch": 0.774869109947644,
"grad_norm": 3074.599853515625,
"learning_rate": 7.419687580962222e-08,
"logits/chosen": -0.5856522917747498,
"logits/rejected": -0.5918489098548889,
"logps/chosen": -257.7669677734375,
"logps/rejected": -240.96908569335938,
"loss": 1375.6232,
"rewards/accuracies": 0.49609375,
"rewards/chosen": -257.7669677734375,
"rewards/margins": -16.797880172729492,
"rewards/rejected": -240.96908569335938,
"slic/ce_loss": 257.7669677734375,
"slic/rank_loss": 86.13883972167969,
"step": 370
},
{
"epoch": 0.7958115183246073,
"grad_norm": 3393.130859375,
"learning_rate": 6.166331963291519e-08,
"logits/chosen": -0.5945131182670593,
"logits/rejected": -0.5999140739440918,
"logps/chosen": -275.88726806640625,
"logps/rejected": -248.7072296142578,
"loss": 1463.2406,
"rewards/accuracies": 0.47187501192092896,
"rewards/chosen": -275.88726806640625,
"rewards/margins": -27.180028915405273,
"rewards/rejected": -248.7072296142578,
"slic/ce_loss": 275.88726806640625,
"slic/rank_loss": 89.92294311523438,
"step": 380
},
{
"epoch": 0.8167539267015707,
"grad_norm": 3126.19189453125,
"learning_rate": 5.013930914912476e-08,
"logits/chosen": -0.5959919095039368,
"logits/rejected": -0.596234917640686,
"logps/chosen": -252.68051147460938,
"logps/rejected": -245.86611938476562,
"loss": 1317.6791,
"rewards/accuracies": 0.514843761920929,
"rewards/chosen": -252.68051147460938,
"rewards/margins": -6.814435005187988,
"rewards/rejected": -245.86611938476562,
"slic/ce_loss": 252.68051147460938,
"slic/rank_loss": 76.7392807006836,
"step": 390
},
{
"epoch": 0.837696335078534,
"grad_norm": 3121.567626953125,
"learning_rate": 3.968661679220467e-08,
"logits/chosen": -0.5884934663772583,
"logits/rejected": -0.5969708561897278,
"logps/chosen": -265.9261779785156,
"logps/rejected": -247.37106323242188,
"loss": 1395.7896,
"rewards/accuracies": 0.46406251192092896,
"rewards/chosen": -265.9261779785156,
"rewards/margins": -18.555116653442383,
"rewards/rejected": -247.37106323242188,
"slic/ce_loss": 265.9261779785156,
"slic/rank_loss": 83.02122497558594,
"step": 400
},
{
"epoch": 0.837696335078534,
"eval_logits/chosen": -0.5998061299324036,
"eval_logits/rejected": -0.6055657267570496,
"eval_logps/chosen": -260.8731994628906,
"eval_logps/rejected": -246.98541259765625,
"eval_loss": 341.91009521484375,
"eval_rewards/accuracies": 0.49193549156188965,
"eval_rewards/chosen": -260.8731994628906,
"eval_rewards/margins": -13.887796401977539,
"eval_rewards/rejected": -246.98541259765625,
"eval_runtime": 23.8106,
"eval_samples_per_second": 83.996,
"eval_slic/ce_loss": 260.8731994628906,
"eval_slic/rank_loss": 81.03688049316406,
"eval_steps_per_second": 2.646,
"step": 400
},
{
"epoch": 0.8586387434554974,
"grad_norm": 2985.495849609375,
"learning_rate": 3.036127238347164e-08,
"logits/chosen": -0.6078251600265503,
"logits/rejected": -0.6096236109733582,
"logps/chosen": -263.81060791015625,
"logps/rejected": -248.4971923828125,
"loss": 1406.5264,
"rewards/accuracies": 0.48906248807907104,
"rewards/chosen": -263.81060791015625,
"rewards/margins": -15.313403129577637,
"rewards/rejected": -248.4971923828125,
"slic/ce_loss": 263.81060791015625,
"slic/rank_loss": 87.82101440429688,
"step": 410
},
{
"epoch": 0.8795811518324608,
"grad_norm": 3290.060791015625,
"learning_rate": 2.2213262793589482e-08,
"logits/chosen": -0.6001953482627869,
"logits/rejected": -0.6040524244308472,
"logps/chosen": -262.39385986328125,
"logps/rejected": -246.28182983398438,
"loss": 1379.8988,
"rewards/accuracies": 0.484375,
"rewards/chosen": -262.39385986328125,
"rewards/margins": -16.112010955810547,
"rewards/rejected": -246.28182983398438,
"slic/ce_loss": 262.39385986328125,
"slic/rank_loss": 82.58085632324219,
"step": 420
},
{
"epoch": 0.900523560209424,
"grad_norm": 3281.253662109375,
"learning_rate": 1.5286263996730026e-08,
"logits/chosen": -0.5900901556015015,
"logits/rejected": -0.6069929003715515,
"logps/chosen": -264.75018310546875,
"logps/rejected": -240.9290313720703,
"loss": 1414.56,
"rewards/accuracies": 0.484375,
"rewards/chosen": -264.75018310546875,
"rewards/margins": -23.821134567260742,
"rewards/rejected": -240.9290313720703,
"slic/ce_loss": 264.75018310546875,
"slic/rank_loss": 88.88983154296875,
"step": 430
},
{
"epoch": 0.9214659685863874,
"grad_norm": 3399.375,
"learning_rate": 9.617406953185136e-09,
"logits/chosen": -0.5953913927078247,
"logits/rejected": -0.6039350628852844,
"logps/chosen": -252.87686157226562,
"logps/rejected": -241.09890747070312,
"loss": 1344.8779,
"rewards/accuracies": 0.503125011920929,
"rewards/chosen": -252.87686157226562,
"rewards/margins": -11.77791976928711,
"rewards/rejected": -241.09890747070312,
"slic/ce_loss": 252.87686157226562,
"slic/rank_loss": 83.34260559082031,
"step": 440
},
{
"epoch": 0.9424083769633508,
"grad_norm": 3320.21875,
"learning_rate": 5.2370785753763356e-09,
"logits/chosen": -0.5980680584907532,
"logits/rejected": -0.6059701442718506,
"logps/chosen": -259.66754150390625,
"logps/rejected": -238.99984741210938,
"loss": 1389.4178,
"rewards/accuracies": 0.4859375059604645,
"rewards/chosen": -259.66754150390625,
"rewards/margins": -20.667722702026367,
"rewards/rejected": -238.99984741210938,
"slic/ce_loss": 259.66754150390625,
"slic/rank_loss": 87.68685150146484,
"step": 450
},
{
"epoch": 0.9633507853403142,
"grad_norm": 3521.455322265625,
"learning_rate": 2.168758844148272e-09,
"logits/chosen": -0.5775882005691528,
"logits/rejected": -0.5878804922103882,
"logps/chosen": -275.74432373046875,
"logps/rejected": -259.4674072265625,
"loss": 1447.9953,
"rewards/accuracies": 0.4867187440395355,
"rewards/chosen": -275.74432373046875,
"rewards/margins": -16.27688217163086,
"rewards/rejected": -259.4674072265625,
"slic/ce_loss": 275.74432373046875,
"slic/rank_loss": 86.25456237792969,
"step": 460
},
{
"epoch": 0.9842931937172775,
"grad_norm": 3612.229736328125,
"learning_rate": 4.288949484559934e-10,
"logits/chosen": -0.5883210897445679,
"logits/rejected": -0.5901591181755066,
"logps/chosen": -265.39093017578125,
"logps/rejected": -239.3176727294922,
"loss": 1424.0908,
"rewards/accuracies": 0.4867187440395355,
"rewards/chosen": -265.39093017578125,
"rewards/margins": -26.073261260986328,
"rewards/rejected": -239.3176727294922,
"slic/ce_loss": 265.39093017578125,
"slic/rank_loss": 90.63172912597656,
"step": 470
},
{
"epoch": 0.9989528795811519,
"step": 477,
"total_flos": 0.0,
"train_loss": 1398.719245958628,
"train_runtime": 2677.8284,
"train_samples_per_second": 22.83,
"train_steps_per_second": 0.178
}
],
"logging_steps": 10,
"max_steps": 477,
"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": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}