| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 250, | |
| "global_step": 1672, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.050239234449760764, | |
| "grad_norm": 2.739856243133545, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -0.5215702056884766, | |
| "logits/rejected": -0.5743793249130249, | |
| "logps/chosen": -55.2584228515625, | |
| "logps/rejected": -27.386730194091797, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.4627976417541504, | |
| "rewards/chosen": -0.0007224595756269991, | |
| "rewards/margins": 0.00034199823858216405, | |
| "rewards/rejected": -0.001064457930624485, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.10047846889952153, | |
| "grad_norm": 5.56309175491333, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": -0.5266280174255371, | |
| "logits/rejected": -0.5796422362327576, | |
| "logps/chosen": -61.078453063964844, | |
| "logps/rejected": -28.424898147583008, | |
| "loss": 0.677, | |
| "rewards/accuracies": 0.644345223903656, | |
| "rewards/chosen": 0.01867148093879223, | |
| "rewards/margins": 0.034289903938770294, | |
| "rewards/rejected": -0.015618422068655491, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.14952153110047847, | |
| "eval_logits/chosen": -0.6673032641410828, | |
| "eval_logits/rejected": -0.6792050004005432, | |
| "eval_logps/chosen": -2.12101674079895, | |
| "eval_logps/rejected": -7.310378074645996, | |
| "eval_loss": 0.6611143946647644, | |
| "eval_rewards/accuracies": 0.878000020980835, | |
| "eval_rewards/chosen": -0.0029554006177932024, | |
| "eval_rewards/margins": 0.06584452092647552, | |
| "eval_rewards/rejected": -0.06879992038011551, | |
| "eval_runtime": 90.3233, | |
| "eval_samples_per_second": 5.536, | |
| "eval_steps_per_second": 2.768, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1507177033492823, | |
| "grad_norm": 1.9286553859710693, | |
| "learning_rate": 9.441489361702127e-07, | |
| "logits/chosen": -0.5305180549621582, | |
| "logits/rejected": -0.5869216918945312, | |
| "logps/chosen": -55.35295867919922, | |
| "logps/rejected": -28.797080993652344, | |
| "loss": 0.6057, | |
| "rewards/accuracies": 0.8898809552192688, | |
| "rewards/chosen": 0.11930066347122192, | |
| "rewards/margins": 0.22872012853622437, | |
| "rewards/rejected": -0.10941947251558304, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.20095693779904306, | |
| "grad_norm": 2.7939305305480957, | |
| "learning_rate": 8.882978723404255e-07, | |
| "logits/chosen": -0.5447388887405396, | |
| "logits/rejected": -0.5977838039398193, | |
| "logps/chosen": -50.50016784667969, | |
| "logps/rejected": -30.050025939941406, | |
| "loss": 0.5101, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 0.2724921405315399, | |
| "rewards/margins": 0.6402528285980225, | |
| "rewards/rejected": -0.36776062846183777, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.2511961722488038, | |
| "grad_norm": 2.6437010765075684, | |
| "learning_rate": 8.324468085106383e-07, | |
| "logits/chosen": -0.5628253817558289, | |
| "logits/rejected": -0.6165186762809753, | |
| "logps/chosen": -47.81046676635742, | |
| "logps/rejected": -33.11960220336914, | |
| "loss": 0.4141, | |
| "rewards/accuracies": 0.8913690447807312, | |
| "rewards/chosen": 0.3376307785511017, | |
| "rewards/margins": 1.0564574003219604, | |
| "rewards/rejected": -0.7188266515731812, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.29904306220095694, | |
| "eval_logits/chosen": -0.7404754161834717, | |
| "eval_logits/rejected": -0.7604221105575562, | |
| "eval_logps/chosen": -3.4617385864257812, | |
| "eval_logps/rejected": -19.556419372558594, | |
| "eval_loss": 0.3327634930610657, | |
| "eval_rewards/accuracies": 0.8960000276565552, | |
| "eval_rewards/chosen": -0.13702762126922607, | |
| "eval_rewards/margins": 1.1563764810562134, | |
| "eval_rewards/rejected": -1.2934041023254395, | |
| "eval_runtime": 90.3284, | |
| "eval_samples_per_second": 5.535, | |
| "eval_steps_per_second": 2.768, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3014354066985646, | |
| "grad_norm": 2.6485891342163086, | |
| "learning_rate": 7.765957446808509e-07, | |
| "logits/chosen": -0.5817141532897949, | |
| "logits/rejected": -0.6466739177703857, | |
| "logps/chosen": -46.02522277832031, | |
| "logps/rejected": -36.878692626953125, | |
| "loss": 0.3166, | |
| "rewards/accuracies": 0.9032738208770752, | |
| "rewards/chosen": 0.3111897110939026, | |
| "rewards/margins": 1.5069682598114014, | |
| "rewards/rejected": -1.1957786083221436, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.35167464114832536, | |
| "grad_norm": 3.9122705459594727, | |
| "learning_rate": 7.207446808510638e-07, | |
| "logits/chosen": -0.6006772518157959, | |
| "logits/rejected": -0.6688662767410278, | |
| "logps/chosen": -47.613197326660156, | |
| "logps/rejected": -42.94892120361328, | |
| "loss": 0.2404, | |
| "rewards/accuracies": 0.9092261791229248, | |
| "rewards/chosen": 0.2409990429878235, | |
| "rewards/margins": 2.0163769721984863, | |
| "rewards/rejected": -1.7753779888153076, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.4019138755980861, | |
| "grad_norm": 7.711537837982178, | |
| "learning_rate": 6.648936170212765e-07, | |
| "logits/chosen": -0.6143811345100403, | |
| "logits/rejected": -0.6911322474479675, | |
| "logps/chosen": -54.04777526855469, | |
| "logps/rejected": -50.569236755371094, | |
| "loss": 0.1869, | |
| "rewards/accuracies": 0.9330357313156128, | |
| "rewards/chosen": 0.23013024032115936, | |
| "rewards/margins": 2.570420026779175, | |
| "rewards/rejected": -2.340289831161499, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.4485645933014354, | |
| "eval_logits/chosen": -0.827215313911438, | |
| "eval_logits/rejected": -0.8525059819221497, | |
| "eval_logps/chosen": -5.357245445251465, | |
| "eval_logps/rejected": -34.60578918457031, | |
| "eval_loss": 0.19428196549415588, | |
| "eval_rewards/accuracies": 0.9280000329017639, | |
| "eval_rewards/chosen": -0.3265782594680786, | |
| "eval_rewards/margins": 2.4717628955841064, | |
| "eval_rewards/rejected": -2.7983410358428955, | |
| "eval_runtime": 90.3138, | |
| "eval_samples_per_second": 5.536, | |
| "eval_steps_per_second": 2.768, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.45215311004784686, | |
| "grad_norm": 4.005836486816406, | |
| "learning_rate": 6.090425531914894e-07, | |
| "logits/chosen": -0.6301328539848328, | |
| "logits/rejected": -0.7088668346405029, | |
| "logps/chosen": -42.79747772216797, | |
| "logps/rejected": -50.45292663574219, | |
| "loss": 0.1786, | |
| "rewards/accuracies": 0.9345238208770752, | |
| "rewards/chosen": 0.04424615204334259, | |
| "rewards/margins": 2.7416493892669678, | |
| "rewards/rejected": -2.6974034309387207, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5023923444976076, | |
| "grad_norm": 3.9601917266845703, | |
| "learning_rate": 5.531914893617021e-07, | |
| "logits/chosen": -0.6196833252906799, | |
| "logits/rejected": -0.7095603346824646, | |
| "logps/chosen": -47.29566192626953, | |
| "logps/rejected": -54.954986572265625, | |
| "loss": 0.1614, | |
| "rewards/accuracies": 0.9345238208770752, | |
| "rewards/chosen": 0.14783993363380432, | |
| "rewards/margins": 3.119297742843628, | |
| "rewards/rejected": -2.9714577198028564, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.5526315789473685, | |
| "grad_norm": 4.151618480682373, | |
| "learning_rate": 4.973404255319149e-07, | |
| "logits/chosen": -0.5776461958885193, | |
| "logits/rejected": -0.6859668493270874, | |
| "logps/chosen": -56.877925872802734, | |
| "logps/rejected": -61.5712890625, | |
| "loss": 0.1234, | |
| "rewards/accuracies": 0.9508928656578064, | |
| "rewards/chosen": 0.2916085124015808, | |
| "rewards/margins": 3.574091672897339, | |
| "rewards/rejected": -3.2824831008911133, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.5980861244019139, | |
| "eval_logits/chosen": -0.833606481552124, | |
| "eval_logits/rejected": -0.8610065579414368, | |
| "eval_logps/chosen": -5.521287441253662, | |
| "eval_logps/rejected": -39.60647201538086, | |
| "eval_loss": 0.1578649878501892, | |
| "eval_rewards/accuracies": 0.9380000233650208, | |
| "eval_rewards/chosen": -0.3429825007915497, | |
| "eval_rewards/margins": 2.9554266929626465, | |
| "eval_rewards/rejected": -3.2984092235565186, | |
| "eval_runtime": 90.3107, | |
| "eval_samples_per_second": 5.536, | |
| "eval_steps_per_second": 2.768, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6028708133971292, | |
| "grad_norm": 2.2310285568237305, | |
| "learning_rate": 4.4148936170212766e-07, | |
| "logits/chosen": -0.6338366866111755, | |
| "logits/rejected": -0.7232056260108948, | |
| "logps/chosen": -42.76348114013672, | |
| "logps/rejected": -56.131858825683594, | |
| "loss": 0.1267, | |
| "rewards/accuracies": 0.9449405074119568, | |
| "rewards/chosen": 0.11895173043012619, | |
| "rewards/margins": 3.4666144847869873, | |
| "rewards/rejected": -3.3476624488830566, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.65311004784689, | |
| "grad_norm": 0.4909895062446594, | |
| "learning_rate": 3.856382978723404e-07, | |
| "logits/chosen": -0.612333357334137, | |
| "logits/rejected": -0.7189978957176208, | |
| "logps/chosen": -52.86572265625, | |
| "logps/rejected": -62.92350387573242, | |
| "loss": 0.0975, | |
| "rewards/accuracies": 0.9702380895614624, | |
| "rewards/chosen": 0.2107280194759369, | |
| "rewards/margins": 3.751209259033203, | |
| "rewards/rejected": -3.5404813289642334, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.7033492822966507, | |
| "grad_norm": 1.8026127815246582, | |
| "learning_rate": 3.2978723404255315e-07, | |
| "logits/chosen": -0.6208314895629883, | |
| "logits/rejected": -0.7236352562904358, | |
| "logps/chosen": -48.80342102050781, | |
| "logps/rejected": -60.991233825683594, | |
| "loss": 0.122, | |
| "rewards/accuracies": 0.949404776096344, | |
| "rewards/chosen": 0.10279477387666702, | |
| "rewards/margins": 3.68986439704895, | |
| "rewards/rejected": -3.587069511413574, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.7476076555023924, | |
| "eval_logits/chosen": -0.8265290856361389, | |
| "eval_logits/rejected": -0.8553086519241333, | |
| "eval_logps/chosen": -5.278414249420166, | |
| "eval_logps/rejected": -42.231048583984375, | |
| "eval_loss": 0.143920436501503, | |
| "eval_rewards/accuracies": 0.9360000491142273, | |
| "eval_rewards/chosen": -0.31869521737098694, | |
| "eval_rewards/margins": 3.2421717643737793, | |
| "eval_rewards/rejected": -3.5608673095703125, | |
| "eval_runtime": 90.3194, | |
| "eval_samples_per_second": 5.536, | |
| "eval_steps_per_second": 2.768, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7535885167464115, | |
| "grad_norm": 0.7733187079429626, | |
| "learning_rate": 2.7393617021276595e-07, | |
| "logits/chosen": -0.595526933670044, | |
| "logits/rejected": -0.7032517790794373, | |
| "logps/chosen": -53.06210708618164, | |
| "logps/rejected": -64.58065795898438, | |
| "loss": 0.0987, | |
| "rewards/accuracies": 0.9598214626312256, | |
| "rewards/chosen": 0.2242652177810669, | |
| "rewards/margins": 3.922701835632324, | |
| "rewards/rejected": -3.698436975479126, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8038277511961722, | |
| "grad_norm": 13.78078556060791, | |
| "learning_rate": 2.1808510638297872e-07, | |
| "logits/chosen": -0.6165376305580139, | |
| "logits/rejected": -0.7192720174789429, | |
| "logps/chosen": -43.811336517333984, | |
| "logps/rejected": -62.044307708740234, | |
| "loss": 0.0933, | |
| "rewards/accuracies": 0.9717261791229248, | |
| "rewards/chosen": 0.13936470448970795, | |
| "rewards/margins": 3.9077537059783936, | |
| "rewards/rejected": -3.7683887481689453, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.854066985645933, | |
| "grad_norm": 8.148602485656738, | |
| "learning_rate": 1.622340425531915e-07, | |
| "logits/chosen": -0.5712071657180786, | |
| "logits/rejected": -0.6922958493232727, | |
| "logps/chosen": -53.978843688964844, | |
| "logps/rejected": -67.05591583251953, | |
| "loss": 0.0821, | |
| "rewards/accuracies": 0.973214328289032, | |
| "rewards/chosen": 0.26229721307754517, | |
| "rewards/margins": 4.126736164093018, | |
| "rewards/rejected": -3.864438772201538, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 0.8971291866028708, | |
| "eval_logits/chosen": -0.8240648508071899, | |
| "eval_logits/rejected": -0.8534601330757141, | |
| "eval_logps/chosen": -5.354381084442139, | |
| "eval_logps/rejected": -43.72341537475586, | |
| "eval_loss": 0.13976776599884033, | |
| "eval_rewards/accuracies": 0.9340000152587891, | |
| "eval_rewards/chosen": -0.3262918293476105, | |
| "eval_rewards/margins": 3.383812189102173, | |
| "eval_rewards/rejected": -3.71010422706604, | |
| "eval_runtime": 90.2962, | |
| "eval_samples_per_second": 5.537, | |
| "eval_steps_per_second": 2.769, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9043062200956937, | |
| "grad_norm": 0.7219157218933105, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -0.5762205719947815, | |
| "logits/rejected": -0.6943542957305908, | |
| "logps/chosen": -50.31645965576172, | |
| "logps/rejected": -66.44419860839844, | |
| "loss": 0.0927, | |
| "rewards/accuracies": 0.9657738208770752, | |
| "rewards/chosen": 0.20835421979427338, | |
| "rewards/margins": 4.124646186828613, | |
| "rewards/rejected": -3.916292190551758, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 0.9545454545454546, | |
| "grad_norm": 8.767385482788086, | |
| "learning_rate": 5.053191489361702e-08, | |
| "logits/chosen": -0.5936585664749146, | |
| "logits/rejected": -0.6990638971328735, | |
| "logps/chosen": -51.6839714050293, | |
| "logps/rejected": -65.34242248535156, | |
| "loss": 0.1066, | |
| "rewards/accuracies": 0.9553571343421936, | |
| "rewards/chosen": 0.15223945677280426, | |
| "rewards/margins": 4.063188076019287, | |
| "rewards/rejected": -3.9109482765197754, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1672, | |
| "total_flos": 3.5989630324781875e+17, | |
| "train_loss": 0.25213732730828975, | |
| "train_runtime": 8848.1085, | |
| "train_samples_per_second": 1.512, | |
| "train_steps_per_second": 0.189 | |
| } | |
| ], | |
| "logging_steps": 84, | |
| "max_steps": 1672, | |
| "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": 3.5989630324781875e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |