| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9960474308300395, | |
| "eval_steps": 500, | |
| "global_step": 189, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 4737.779382861946, | |
| "learning_rate": 2.6315789473684208e-08, | |
| "logits/chosen": -4.638427734375, | |
| "logits/rejected": -4.891327857971191, | |
| "logps/chosen": -198.52749633789062, | |
| "logps/rejected": -147.3392791748047, | |
| "loss": 2.1269, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 3019.619079402316, | |
| "learning_rate": 2.631578947368421e-07, | |
| "logits/chosen": -4.496801376342773, | |
| "logits/rejected": -4.816222190856934, | |
| "logps/chosen": -224.27357482910156, | |
| "logps/rejected": -168.04739379882812, | |
| "loss": 1.9212, | |
| "rewards/accuracies": 0.5381944179534912, | |
| "rewards/chosen": 0.3541475236415863, | |
| "rewards/margins": 0.37169286608695984, | |
| "rewards/rejected": -0.017545383423566818, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1863.4195630562826, | |
| "learning_rate": 4.999573126145131e-07, | |
| "logits/chosen": -4.533459663391113, | |
| "logits/rejected": -4.848563194274902, | |
| "logps/chosen": -220.4309539794922, | |
| "logps/rejected": -180.72413635253906, | |
| "loss": 1.1783, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": 5.777209281921387, | |
| "rewards/margins": 6.331713676452637, | |
| "rewards/rejected": -0.5545047521591187, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1882.3555396757283, | |
| "learning_rate": 4.948524419003415e-07, | |
| "logits/chosen": -4.54370641708374, | |
| "logits/rejected": -4.812285423278809, | |
| "logps/chosen": -213.49411010742188, | |
| "logps/rejected": -177.16848754882812, | |
| "loss": 1.3518, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": 8.2464017868042, | |
| "rewards/margins": 12.621076583862305, | |
| "rewards/rejected": -4.3746747970581055, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1979.651061288252, | |
| "learning_rate": 4.81409414945389e-07, | |
| "logits/chosen": -4.619187831878662, | |
| "logits/rejected": -4.8958845138549805, | |
| "logps/chosen": -221.00082397460938, | |
| "logps/rejected": -184.62203979492188, | |
| "loss": 1.4689, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": 7.427975654602051, | |
| "rewards/margins": 15.529205322265625, | |
| "rewards/rejected": -8.101228713989258, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 2122.235483955093, | |
| "learning_rate": 4.6008601790947314e-07, | |
| "logits/chosen": -4.608691215515137, | |
| "logits/rejected": -4.925226211547852, | |
| "logps/chosen": -210.32058715820312, | |
| "logps/rejected": -179.0367431640625, | |
| "loss": 1.3821, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": 6.428221225738525, | |
| "rewards/margins": 16.42898941040039, | |
| "rewards/rejected": -10.00076961517334, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1692.2277360562514, | |
| "learning_rate": 4.3160839350405605e-07, | |
| "logits/chosen": -4.665585994720459, | |
| "logits/rejected": -4.9272074699401855, | |
| "logps/chosen": -205.7926788330078, | |
| "logps/rejected": -178.56011962890625, | |
| "loss": 1.3465, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": 7.834652900695801, | |
| "rewards/margins": 16.5399169921875, | |
| "rewards/rejected": -8.705263137817383, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 1810.8868167381333, | |
| "learning_rate": 3.9694631307311825e-07, | |
| "logits/chosen": -4.6464009284973145, | |
| "logits/rejected": -4.913968086242676, | |
| "logps/chosen": -207.1618194580078, | |
| "logps/rejected": -182.61012268066406, | |
| "loss": 1.3564, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": 8.192334175109863, | |
| "rewards/margins": 16.401655197143555, | |
| "rewards/rejected": -8.209321975708008, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1395.6821844604426, | |
| "learning_rate": 3.572801521931522e-07, | |
| "logits/chosen": -4.674800395965576, | |
| "logits/rejected": -4.932587623596191, | |
| "logps/chosen": -202.7789764404297, | |
| "logps/rejected": -184.74395751953125, | |
| "loss": 1.312, | |
| "rewards/accuracies": 0.846875011920929, | |
| "rewards/chosen": 5.3585076332092285, | |
| "rewards/margins": 16.119762420654297, | |
| "rewards/rejected": -10.761255264282227, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 1761.719146022038, | |
| "learning_rate": 3.139606943986089e-07, | |
| "logits/chosen": -4.721759796142578, | |
| "logits/rejected": -4.953747272491455, | |
| "logps/chosen": -199.81448364257812, | |
| "logps/rejected": -178.44004821777344, | |
| "loss": 1.3425, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": 7.849789619445801, | |
| "rewards/margins": 15.739909172058105, | |
| "rewards/rejected": -7.890120029449463, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1641.4466240114464, | |
| "learning_rate": 2.684631318687185e-07, | |
| "logits/chosen": -4.7313385009765625, | |
| "logits/rejected": -4.984685897827148, | |
| "logps/chosen": -213.2564239501953, | |
| "logps/rejected": -190.69088745117188, | |
| "loss": 1.3623, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 5.4120659828186035, | |
| "rewards/margins": 16.620161056518555, | |
| "rewards/rejected": -11.208093643188477, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1346.9601711684072, | |
| "learning_rate": 2.2233682952712483e-07, | |
| "logits/chosen": -4.668034553527832, | |
| "logits/rejected": -4.953825950622559, | |
| "logps/chosen": -216.8499298095703, | |
| "logps/rejected": -186.10470581054688, | |
| "loss": 1.1234, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 7.529428005218506, | |
| "rewards/margins": 15.865753173828125, | |
| "rewards/rejected": -8.336324691772461, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1999.3900490416042, | |
| "learning_rate": 1.7715256327766884e-07, | |
| "logits/chosen": -4.720789909362793, | |
| "logits/rejected": -5.025943279266357, | |
| "logps/chosen": -207.7978973388672, | |
| "logps/rejected": -178.0445098876953, | |
| "loss": 1.1185, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": 8.364091873168945, | |
| "rewards/margins": 16.011329650878906, | |
| "rewards/rejected": -7.647237300872803, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 1568.9083661238265, | |
| "learning_rate": 1.3444902911492174e-07, | |
| "logits/chosen": -4.702408790588379, | |
| "logits/rejected": -4.98063325881958, | |
| "logps/chosen": -215.88174438476562, | |
| "logps/rejected": -188.39645385742188, | |
| "loss": 1.2748, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": 7.107934474945068, | |
| "rewards/margins": 16.361108779907227, | |
| "rewards/rejected": -9.253173828125, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1390.666498149427, | |
| "learning_rate": 9.56804446775518e-08, | |
| "logits/chosen": -4.6129560470581055, | |
| "logits/rejected": -4.923257350921631, | |
| "logps/chosen": -208.4700469970703, | |
| "logps/rejected": -178.78623962402344, | |
| "loss": 1.0987, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": 6.485724449157715, | |
| "rewards/margins": 17.193899154663086, | |
| "rewards/rejected": -10.708174705505371, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1296.5821049110084, | |
| "learning_rate": 6.216702761078166e-08, | |
| "logits/chosen": -4.699868202209473, | |
| "logits/rejected": -4.9864583015441895, | |
| "logps/chosen": -196.650146484375, | |
| "logps/rejected": -168.93551635742188, | |
| "loss": 1.0856, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": 7.758223056793213, | |
| "rewards/margins": 17.15636444091797, | |
| "rewards/rejected": -9.398139953613281, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 1769.0071097352081, | |
| "learning_rate": 3.5050037137906885e-08, | |
| "logits/chosen": -4.634187698364258, | |
| "logits/rejected": -4.958773612976074, | |
| "logps/chosen": -211.03591918945312, | |
| "logps/rejected": -176.72067260742188, | |
| "loss": 1.1749, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": 8.515033721923828, | |
| "rewards/margins": 16.594724655151367, | |
| "rewards/rejected": -8.079689979553223, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1606.699013433802, | |
| "learning_rate": 1.5252909846235894e-08, | |
| "logits/chosen": -4.62954044342041, | |
| "logits/rejected": -4.913142204284668, | |
| "logps/chosen": -209.8083953857422, | |
| "logps/rejected": -184.52127075195312, | |
| "loss": 1.2059, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 9.714839935302734, | |
| "rewards/margins": 18.106443405151367, | |
| "rewards/rejected": -8.391606330871582, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1242.5060745172418, | |
| "learning_rate": 3.4498131616493565e-09, | |
| "logits/chosen": -4.616083145141602, | |
| "logits/rejected": -4.87780237197876, | |
| "logps/chosen": -215.27685546875, | |
| "logps/rejected": -193.55332946777344, | |
| "loss": 1.1132, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 6.19677209854126, | |
| "rewards/margins": 13.694157600402832, | |
| "rewards/rejected": -7.497385501861572, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 189, | |
| "total_flos": 0.0, | |
| "train_loss": 1.2810401298381664, | |
| "train_runtime": 5417.1403, | |
| "train_samples_per_second": 8.959, | |
| "train_steps_per_second": 0.035 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 189, | |
| "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 | |
| } | |