| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9968652037617555, | |
| "eval_steps": 500, | |
| "global_step": 159, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.125e-08, | |
| "logits/chosen": -2.6344056129455566, | |
| "logits/rejected": -2.5906338691711426, | |
| "logps/chosen": -158.64126586914062, | |
| "logps/pi_response": -86.09246826171875, | |
| "logps/ref_response": -86.09246826171875, | |
| "logps/rejected": -148.42047119140625, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.691999673843384, | |
| "logits/rejected": -2.6960792541503906, | |
| "logps/chosen": -233.6039276123047, | |
| "logps/pi_response": -119.84125518798828, | |
| "logps/ref_response": -118.89913940429688, | |
| "logps/rejected": -252.53822326660156, | |
| "loss": 0.692, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.012611959129571915, | |
| "rewards/margins": 0.0004196475783828646, | |
| "rewards/rejected": -0.013031606562435627, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.990353313429303e-07, | |
| "logits/chosen": -2.6983652114868164, | |
| "logits/rejected": -2.669889211654663, | |
| "logps/chosen": -250.45095825195312, | |
| "logps/pi_response": -146.63462829589844, | |
| "logps/ref_response": -131.12376403808594, | |
| "logps/rejected": -277.07781982421875, | |
| "loss": 0.6793, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.18149897456169128, | |
| "rewards/margins": 0.05194617435336113, | |
| "rewards/rejected": -0.23344513773918152, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.882681251368548e-07, | |
| "logits/chosen": -2.5047459602355957, | |
| "logits/rejected": -2.444232940673828, | |
| "logps/chosen": -243.5947265625, | |
| "logps/pi_response": -132.4476318359375, | |
| "logps/ref_response": -113.8835220336914, | |
| "logps/rejected": -249.1999969482422, | |
| "loss": 0.667, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.1728166788816452, | |
| "rewards/margins": 0.11773113906383514, | |
| "rewards/rejected": -0.29054778814315796, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6604720940421207e-07, | |
| "logits/chosen": -2.314911365509033, | |
| "logits/rejected": -2.292646884918213, | |
| "logps/chosen": -254.43429565429688, | |
| "logps/pi_response": -164.00875854492188, | |
| "logps/ref_response": -110.53662109375, | |
| "logps/rejected": -328.1941833496094, | |
| "loss": 0.6434, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.5214705467224121, | |
| "rewards/margins": 0.17735092341899872, | |
| "rewards/rejected": -0.698821485042572, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3344075855595097e-07, | |
| "logits/chosen": -2.232452392578125, | |
| "logits/rejected": -2.202117443084717, | |
| "logps/chosen": -283.18255615234375, | |
| "logps/pi_response": -181.59396362304688, | |
| "logps/ref_response": -120.1254653930664, | |
| "logps/rejected": -310.2931213378906, | |
| "loss": 0.6299, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.5738414525985718, | |
| "rewards/margins": 0.15463915467262268, | |
| "rewards/rejected": -0.7284805178642273, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.920161866827889e-07, | |
| "logits/chosen": -2.1293296813964844, | |
| "logits/rejected": -2.094069004058838, | |
| "logps/chosen": -288.9293212890625, | |
| "logps/pi_response": -199.8803253173828, | |
| "logps/ref_response": -107.5267333984375, | |
| "logps/rejected": -332.07403564453125, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.7527658939361572, | |
| "rewards/margins": 0.31268495321273804, | |
| "rewards/rejected": -1.0654507875442505, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4376480090239047e-07, | |
| "logits/chosen": -2.1092419624328613, | |
| "logits/rejected": -2.055610418319702, | |
| "logps/chosen": -297.887451171875, | |
| "logps/pi_response": -207.2319793701172, | |
| "logps/ref_response": -112.31196594238281, | |
| "logps/rejected": -314.06011962890625, | |
| "loss": 0.6006, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.7968708276748657, | |
| "rewards/margins": 0.35347780585289, | |
| "rewards/rejected": -1.150348424911499, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.910060778827554e-07, | |
| "logits/chosen": -2.1043503284454346, | |
| "logits/rejected": -2.0406947135925293, | |
| "logps/chosen": -277.8828430175781, | |
| "logps/pi_response": -199.2911376953125, | |
| "logps/ref_response": -99.59819793701172, | |
| "logps/rejected": -338.6634826660156, | |
| "loss": 0.6184, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.8253669738769531, | |
| "rewards/margins": 0.35502859950065613, | |
| "rewards/rejected": -1.1803953647613525, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.3627616503391812e-07, | |
| "logits/chosen": -2.1398611068725586, | |
| "logits/rejected": -2.0961031913757324, | |
| "logps/chosen": -299.2795715332031, | |
| "logps/pi_response": -207.5061798095703, | |
| "logps/ref_response": -100.86174011230469, | |
| "logps/rejected": -342.63848876953125, | |
| "loss": 0.6061, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.9135547876358032, | |
| "rewards/margins": 0.3614444434642792, | |
| "rewards/rejected": -1.2749992609024048, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8220596619089573e-07, | |
| "logits/chosen": -2.041874647140503, | |
| "logits/rejected": -2.030878782272339, | |
| "logps/chosen": -306.0923767089844, | |
| "logps/pi_response": -226.0547332763672, | |
| "logps/ref_response": -112.6884765625, | |
| "logps/rejected": -394.3197326660156, | |
| "loss": 0.5803, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.9776135683059692, | |
| "rewards/margins": 0.42801570892333984, | |
| "rewards/rejected": -1.4056293964385986, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3139467229135998e-07, | |
| "logits/chosen": -2.08709979057312, | |
| "logits/rejected": -2.1020517349243164, | |
| "logps/chosen": -311.7897033691406, | |
| "logps/pi_response": -208.67361450195312, | |
| "logps/ref_response": -100.60249328613281, | |
| "logps/rejected": -340.69842529296875, | |
| "loss": 0.6037, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.9508616328239441, | |
| "rewards/margins": 0.34958645701408386, | |
| "rewards/rejected": -1.3004481792449951, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.628481651367875e-08, | |
| "logits/chosen": -2.1009232997894287, | |
| "logits/rejected": -2.0271968841552734, | |
| "logps/chosen": -340.5748596191406, | |
| "logps/pi_response": -246.27499389648438, | |
| "logps/ref_response": -130.39340209960938, | |
| "logps/rejected": -368.299560546875, | |
| "loss": 0.5991, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -1.0281145572662354, | |
| "rewards/margins": 0.2725009322166443, | |
| "rewards/rejected": -1.3006153106689453, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.904486005914027e-08, | |
| "logits/chosen": -2.131903886795044, | |
| "logits/rejected": -2.1143031120300293, | |
| "logps/chosen": -329.1007995605469, | |
| "logps/pi_response": -242.9829559326172, | |
| "logps/ref_response": -113.8158950805664, | |
| "logps/rejected": -387.0042419433594, | |
| "loss": 0.5844, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -1.1540590524673462, | |
| "rewards/margins": 0.3786469101905823, | |
| "rewards/rejected": -1.5327059030532837, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1464952759020856e-08, | |
| "logits/chosen": -2.1207737922668457, | |
| "logits/rejected": -2.0526301860809326, | |
| "logps/chosen": -363.8649597167969, | |
| "logps/pi_response": -283.3131408691406, | |
| "logps/ref_response": -119.49796295166016, | |
| "logps/rejected": -411.9598083496094, | |
| "loss": 0.5928, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -1.3548119068145752, | |
| "rewards/margins": 0.4335576891899109, | |
| "rewards/rejected": -1.7883697748184204, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.8708793644441086e-09, | |
| "logits/chosen": -2.069753646850586, | |
| "logits/rejected": -2.039280652999878, | |
| "logps/chosen": -339.0675354003906, | |
| "logps/pi_response": -243.1265869140625, | |
| "logps/ref_response": -101.72931671142578, | |
| "logps/rejected": -374.07086181640625, | |
| "loss": 0.5747, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.2384488582611084, | |
| "rewards/margins": 0.3813491463661194, | |
| "rewards/rejected": -1.6197980642318726, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 159, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6180852314211288, | |
| "train_runtime": 4551.3166, | |
| "train_samples_per_second": 4.477, | |
| "train_steps_per_second": 0.035 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 159, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |