|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.9978586723768736, |
|
"eval_steps": 500, |
|
"global_step": 233, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 4.4375, |
|
"learning_rate": 4.977309719247571e-06, |
|
"log_odds_chosen": 0.08039750903844833, |
|
"log_odds_ratio": -0.7017658352851868, |
|
"logits/chosen": -3.249631404876709, |
|
"logits/rejected": -3.259352445602417, |
|
"logps/chosen": -0.7289968729019165, |
|
"logps/rejected": -0.780512273311615, |
|
"loss": 0.616, |
|
"nll_loss": 0.46984538435935974, |
|
"rewards/accuracies": 0.5249999761581421, |
|
"rewards/chosen": -0.14579936861991882, |
|
"rewards/margins": 0.01030308473855257, |
|
"rewards/rejected": -0.15610246360301971, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"grad_norm": 4.125, |
|
"learning_rate": 4.909650756062782e-06, |
|
"log_odds_chosen": 0.06598430871963501, |
|
"log_odds_ratio": -0.7187274694442749, |
|
"logits/chosen": -3.1759872436523438, |
|
"logits/rejected": -3.1876461505889893, |
|
"logps/chosen": -0.7303891777992249, |
|
"logps/rejected": -0.7664512395858765, |
|
"loss": 0.5601, |
|
"nll_loss": 0.42730289697647095, |
|
"rewards/accuracies": 0.5625, |
|
"rewards/chosen": -0.14607784152030945, |
|
"rewards/margins": 0.007212400436401367, |
|
"rewards/rejected": -0.15329024195671082, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"grad_norm": 4.03125, |
|
"learning_rate": 4.7982512711416995e-06, |
|
"log_odds_chosen": 0.08186231553554535, |
|
"log_odds_ratio": -0.6992891430854797, |
|
"logits/chosen": -3.130770206451416, |
|
"logits/rejected": -3.1523356437683105, |
|
"logps/chosen": -0.7080240845680237, |
|
"logps/rejected": -0.7284458875656128, |
|
"loss": 0.5702, |
|
"nll_loss": 0.4340079426765442, |
|
"rewards/accuracies": 0.5625, |
|
"rewards/chosen": -0.14160482585430145, |
|
"rewards/margins": 0.004084363579750061, |
|
"rewards/rejected": -0.14568917453289032, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"grad_norm": 3.9375, |
|
"learning_rate": 4.64513341295515e-06, |
|
"log_odds_chosen": -0.018928013741970062, |
|
"log_odds_ratio": -0.7584706544876099, |
|
"logits/chosen": -3.2494730949401855, |
|
"logits/rejected": -3.2488815784454346, |
|
"logps/chosen": -0.7149000763893127, |
|
"logps/rejected": -0.7123271226882935, |
|
"loss": 0.5923, |
|
"nll_loss": 0.4148195683956146, |
|
"rewards/accuracies": 0.48750001192092896, |
|
"rewards/chosen": -0.14298002421855927, |
|
"rewards/margins": -0.0005145778995938599, |
|
"rewards/rejected": -0.14246544241905212, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"grad_norm": 4.21875, |
|
"learning_rate": 4.453076611255507e-06, |
|
"log_odds_chosen": 0.14689764380455017, |
|
"log_odds_ratio": -0.6600848436355591, |
|
"logits/chosen": -3.2663073539733887, |
|
"logits/rejected": -3.295487880706787, |
|
"logps/chosen": -0.6975507736206055, |
|
"logps/rejected": -0.7772414684295654, |
|
"loss": 0.5526, |
|
"nll_loss": 0.36552542448043823, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": -0.1395101547241211, |
|
"rewards/margins": 0.015938155353069305, |
|
"rewards/rejected": -0.155448317527771, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"grad_norm": 4.125, |
|
"learning_rate": 4.22556712424355e-06, |
|
"log_odds_chosen": 0.11583586037158966, |
|
"log_odds_ratio": -0.6816480755805969, |
|
"logits/chosen": -3.224388837814331, |
|
"logits/rejected": -3.2325282096862793, |
|
"logps/chosen": -0.6803846955299377, |
|
"logps/rejected": -0.7500395178794861, |
|
"loss": 0.5742, |
|
"nll_loss": 0.4383344054222107, |
|
"rewards/accuracies": 0.5625, |
|
"rewards/chosen": -0.13607695698738098, |
|
"rewards/margins": 0.013930967077612877, |
|
"rewards/rejected": -0.15000788867473602, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"grad_norm": 4.15625, |
|
"learning_rate": 3.9667347552265945e-06, |
|
"log_odds_chosen": 0.0658905878663063, |
|
"log_odds_ratio": -0.7092221975326538, |
|
"logits/chosen": -3.305513381958008, |
|
"logits/rejected": -3.320878505706787, |
|
"logps/chosen": -0.7039017677307129, |
|
"logps/rejected": -0.7465838193893433, |
|
"loss": 0.6091, |
|
"nll_loss": 0.4511532187461853, |
|
"rewards/accuracies": 0.581250011920929, |
|
"rewards/chosen": -0.14078037440776825, |
|
"rewards/margins": 0.008536392822861671, |
|
"rewards/rejected": -0.14931677281856537, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"grad_norm": 4.28125, |
|
"learning_rate": 3.68127788750129e-06, |
|
"log_odds_chosen": 0.1722445785999298, |
|
"log_odds_ratio": -0.6515191793441772, |
|
"logits/chosen": -3.323869228363037, |
|
"logits/rejected": -3.344327211380005, |
|
"logps/chosen": -0.659850001335144, |
|
"logps/rejected": -0.7550663948059082, |
|
"loss": 0.5858, |
|
"nll_loss": 0.40397030115127563, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": -0.13197001814842224, |
|
"rewards/margins": 0.019043272361159325, |
|
"rewards/rejected": -0.15101328492164612, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"grad_norm": 3.765625, |
|
"learning_rate": 3.3743781982447533e-06, |
|
"log_odds_chosen": 0.14569935202598572, |
|
"log_odds_ratio": -0.6760041117668152, |
|
"logits/chosen": -3.3209376335144043, |
|
"logits/rejected": -3.326855421066284, |
|
"logps/chosen": -0.6928235292434692, |
|
"logps/rejected": -0.7839463949203491, |
|
"loss": 0.5665, |
|
"nll_loss": 0.4432339072227478, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": -0.13856470584869385, |
|
"rewards/margins": 0.018224570900201797, |
|
"rewards/rejected": -0.15678928792476654, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"grad_norm": 3.65625, |
|
"learning_rate": 3.0516065995466336e-06, |
|
"log_odds_chosen": 0.16237546503543854, |
|
"log_odds_ratio": -0.665057361125946, |
|
"logits/chosen": -3.32145619392395, |
|
"logits/rejected": -3.345324993133545, |
|
"logps/chosen": -0.6528733968734741, |
|
"logps/rejected": -0.732820987701416, |
|
"loss": 0.5857, |
|
"nll_loss": 0.45770755410194397, |
|
"rewards/accuracies": 0.581250011920929, |
|
"rewards/chosen": -0.13057467341423035, |
|
"rewards/margins": 0.015989527106285095, |
|
"rewards/rejected": -0.14656421542167664, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"grad_norm": 4.0, |
|
"learning_rate": 2.7188221139616303e-06, |
|
"log_odds_chosen": 0.09346068650484085, |
|
"log_odds_ratio": -0.6963472366333008, |
|
"logits/chosen": -3.2813632488250732, |
|
"logits/rejected": -3.3325469493865967, |
|
"logps/chosen": -0.7247427105903625, |
|
"logps/rejected": -0.7714471817016602, |
|
"loss": 0.5606, |
|
"nll_loss": 0.42204198241233826, |
|
"rewards/accuracies": 0.5625, |
|
"rewards/chosen": -0.14494851231575012, |
|
"rewards/margins": 0.009340907447040081, |
|
"rewards/rejected": -0.15428945422172546, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 3.515625, |
|
"learning_rate": 2.3820655202161237e-06, |
|
"log_odds_chosen": 0.18920771777629852, |
|
"log_odds_ratio": -0.6524609327316284, |
|
"logits/chosen": -3.2978363037109375, |
|
"logits/rejected": -3.3177380561828613, |
|
"logps/chosen": -0.6312097311019897, |
|
"logps/rejected": -0.7275902032852173, |
|
"loss": 0.5719, |
|
"nll_loss": 0.44832339882850647, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": -0.12624196708202362, |
|
"rewards/margins": 0.019276071339845657, |
|
"rewards/rejected": -0.14551803469657898, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"grad_norm": 4.0625, |
|
"learning_rate": 2.0474496996359676e-06, |
|
"log_odds_chosen": 0.12192361056804657, |
|
"log_odds_ratio": -0.6797535419464111, |
|
"logits/chosen": -3.3045902252197266, |
|
"logits/rejected": -3.324050188064575, |
|
"logps/chosen": -0.6812834739685059, |
|
"logps/rejected": -0.7312651872634888, |
|
"loss": 0.5646, |
|
"nll_loss": 0.4076351225376129, |
|
"rewards/accuracies": 0.5687500238418579, |
|
"rewards/chosen": -0.13625670969486237, |
|
"rewards/margins": 0.009996327571570873, |
|
"rewards/rejected": -0.14625301957130432, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"grad_norm": 4.15625, |
|
"learning_rate": 1.7210486737516947e-06, |
|
"log_odds_chosen": 0.13080953061580658, |
|
"log_odds_ratio": -0.6898539066314697, |
|
"logits/chosen": -3.2814488410949707, |
|
"logits/rejected": -3.303652286529541, |
|
"logps/chosen": -0.6504844427108765, |
|
"logps/rejected": -0.7248982787132263, |
|
"loss": 0.5673, |
|
"nll_loss": 0.38752514123916626, |
|
"rewards/accuracies": 0.6187499761581421, |
|
"rewards/chosen": -0.13009688258171082, |
|
"rewards/margins": 0.014882771298289299, |
|
"rewards/rejected": -0.14497967064380646, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"grad_norm": 4.15625, |
|
"learning_rate": 1.4087873472954638e-06, |
|
"log_odds_chosen": 0.13777169585227966, |
|
"log_odds_ratio": -0.6831523776054382, |
|
"logits/chosen": -3.3204522132873535, |
|
"logits/rejected": -3.313605785369873, |
|
"logps/chosen": -0.710480809211731, |
|
"logps/rejected": -0.7831675410270691, |
|
"loss": 0.5371, |
|
"nll_loss": 0.39932551980018616, |
|
"rewards/accuracies": 0.6187499761581421, |
|
"rewards/chosen": -0.14209619164466858, |
|
"rewards/margins": 0.014537332579493523, |
|
"rewards/rejected": -0.15663352608680725, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"grad_norm": 3.671875, |
|
"learning_rate": 1.116333957999608e-06, |
|
"log_odds_chosen": 0.07410892844200134, |
|
"log_odds_ratio": -0.7069286108016968, |
|
"logits/chosen": -3.2931294441223145, |
|
"logits/rejected": -3.299168109893799, |
|
"logps/chosen": -0.7256664037704468, |
|
"logps/rejected": -0.7714512348175049, |
|
"loss": 0.5714, |
|
"nll_loss": 0.40651097893714905, |
|
"rewards/accuracies": 0.5062500238418579, |
|
"rewards/chosen": -0.14513328671455383, |
|
"rewards/margins": 0.009156969375908375, |
|
"rewards/rejected": -0.15429024398326874, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"grad_norm": 3.84375, |
|
"learning_rate": 8.48997185472226e-07, |
|
"log_odds_chosen": 0.15166924893856049, |
|
"log_odds_ratio": -0.6775228381156921, |
|
"logits/chosen": -3.263190507888794, |
|
"logits/rejected": -3.2975189685821533, |
|
"logps/chosen": -0.6769393682479858, |
|
"logps/rejected": -0.7557265758514404, |
|
"loss": 0.5598, |
|
"nll_loss": 0.4182821214199066, |
|
"rewards/accuracies": 0.5625, |
|
"rewards/chosen": -0.1353878676891327, |
|
"rewards/margins": 0.01575743779540062, |
|
"rewards/rejected": -0.1511453092098236, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"grad_norm": 4.5625, |
|
"learning_rate": 6.11629786852592e-07, |
|
"log_odds_chosen": 0.19579073786735535, |
|
"log_odds_ratio": -0.6576439142227173, |
|
"logits/chosen": -3.271582841873169, |
|
"logits/rejected": -3.2877395153045654, |
|
"logps/chosen": -0.6766762733459473, |
|
"logps/rejected": -0.7755132913589478, |
|
"loss": 0.5697, |
|
"nll_loss": 0.4091164469718933, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": -0.13533525168895721, |
|
"rewards/margins": 0.01976742222905159, |
|
"rewards/rejected": -0.1551026701927185, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"grad_norm": 3.71875, |
|
"learning_rate": 4.0854050847362966e-07, |
|
"log_odds_chosen": 0.14539101719856262, |
|
"log_odds_ratio": -0.6700536012649536, |
|
"logits/chosen": -3.2805087566375732, |
|
"logits/rejected": -3.3020291328430176, |
|
"logps/chosen": -0.6793787479400635, |
|
"logps/rejected": -0.7553696036338806, |
|
"loss": 0.5451, |
|
"nll_loss": 0.44966059923171997, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": -0.13587573170661926, |
|
"rewards/margins": 0.0151981757953763, |
|
"rewards/rejected": -0.15107391774654388, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"grad_norm": 3.703125, |
|
"learning_rate": 2.4341587253072035e-07, |
|
"log_odds_chosen": 0.21404138207435608, |
|
"log_odds_ratio": -0.6436953544616699, |
|
"logits/chosen": -3.2684574127197266, |
|
"logits/rejected": -3.2645530700683594, |
|
"logps/chosen": -0.645655632019043, |
|
"logps/rejected": -0.754039466381073, |
|
"loss": 0.55, |
|
"nll_loss": 0.39011189341545105, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": -0.12913110852241516, |
|
"rewards/margins": 0.02167678065598011, |
|
"rewards/rejected": -0.15080790221691132, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"grad_norm": 4.125, |
|
"learning_rate": 1.1925325850281416e-07, |
|
"log_odds_chosen": 0.0769767314195633, |
|
"log_odds_ratio": -0.7074853181838989, |
|
"logits/chosen": -3.295799970626831, |
|
"logits/rejected": -3.325038194656372, |
|
"logps/chosen": -0.6680190563201904, |
|
"logps/rejected": -0.6921139359474182, |
|
"loss": 0.5338, |
|
"nll_loss": 0.34862151741981506, |
|
"rewards/accuracies": 0.53125, |
|
"rewards/chosen": -0.1336037963628769, |
|
"rewards/margins": 0.0048189712688326836, |
|
"rewards/rejected": -0.13842277228832245, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"grad_norm": 3.96875, |
|
"learning_rate": 3.830649404690939e-08, |
|
"log_odds_chosen": 0.10884630680084229, |
|
"log_odds_ratio": -0.691432535648346, |
|
"logits/chosen": -3.277416706085205, |
|
"logits/rejected": -3.29689359664917, |
|
"logps/chosen": -0.69825679063797, |
|
"logps/rejected": -0.7786489725112915, |
|
"loss": 0.5774, |
|
"nll_loss": 0.40830549597740173, |
|
"rewards/accuracies": 0.5375000238418579, |
|
"rewards/chosen": -0.1396513432264328, |
|
"rewards/margins": 0.016078442335128784, |
|
"rewards/rejected": -0.15572980046272278, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"grad_norm": 4.375, |
|
"learning_rate": 2.044943012210754e-09, |
|
"log_odds_chosen": 0.10921859741210938, |
|
"log_odds_ratio": -0.6928341388702393, |
|
"logits/chosen": -3.2992472648620605, |
|
"logits/rejected": -3.3132872581481934, |
|
"logps/chosen": -0.682590901851654, |
|
"logps/rejected": -0.74882972240448, |
|
"loss": 0.5663, |
|
"nll_loss": 0.4056808352470398, |
|
"rewards/accuracies": 0.59375, |
|
"rewards/chosen": -0.1365181803703308, |
|
"rewards/margins": 0.013247755356132984, |
|
"rewards/rejected": -0.14976593852043152, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 233, |
|
"total_flos": 0.0, |
|
"train_loss": 0.5700738102581367, |
|
"train_runtime": 3228.7929, |
|
"train_samples_per_second": 4.623, |
|
"train_steps_per_second": 0.072 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 233, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 500, |
|
"total_flos": 0.0, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|