| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 626, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003194888178913738, | |
| "grad_norm": 26.70895145343874, | |
| "learning_rate": 5.3191489361702125e-09, | |
| "logits/chosen": -1.4921875, | |
| "logits/rejected": -1.4921875, | |
| "logps/chosen": -247.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03194888178913738, | |
| "grad_norm": 27.524419104861234, | |
| "learning_rate": 5.3191489361702123e-08, | |
| "logits/chosen": -1.5390625, | |
| "logits/rejected": -1.59375, | |
| "logps/chosen": -252.0, | |
| "logps/rejected": -290.0, | |
| "loss": 0.6921, | |
| "rewards/accuracies": 0.1805555522441864, | |
| "rewards/chosen": 0.004180908203125, | |
| "rewards/margins": 0.0069580078125, | |
| "rewards/rejected": -0.0027618408203125, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06389776357827476, | |
| "grad_norm": 25.84743681377418, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.6015625, | |
| "logps/chosen": -244.0, | |
| "logps/rejected": -284.0, | |
| "loss": 0.6878, | |
| "rewards/accuracies": 0.16249999403953552, | |
| "rewards/chosen": -0.00750732421875, | |
| "rewards/margins": -0.00750732421875, | |
| "rewards/rejected": -6.109476089477539e-06, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09584664536741214, | |
| "grad_norm": 26.132304612422136, | |
| "learning_rate": 1.5957446808510638e-07, | |
| "logits/chosen": -1.5625, | |
| "logits/rejected": -1.5625, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -282.0, | |
| "loss": 0.6743, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.038818359375, | |
| "rewards/margins": 0.0244140625, | |
| "rewards/rejected": 0.014404296875, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12779552715654952, | |
| "grad_norm": 24.407986537974015, | |
| "learning_rate": 2.127659574468085e-07, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.5859375, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -308.0, | |
| "loss": 0.6435, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.1591796875, | |
| "rewards/margins": 0.1240234375, | |
| "rewards/rejected": 0.03515625, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1597444089456869, | |
| "grad_norm": 21.9426563067355, | |
| "learning_rate": 2.659574468085106e-07, | |
| "logits/chosen": -1.5625, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -239.0, | |
| "logps/rejected": -286.0, | |
| "loss": 0.6137, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.310546875, | |
| "rewards/margins": 0.2275390625, | |
| "rewards/rejected": 0.08251953125, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19169329073482427, | |
| "grad_norm": 20.96670132168061, | |
| "learning_rate": 3.1914893617021275e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.5732, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.5, | |
| "rewards/margins": 0.376953125, | |
| "rewards/rejected": 0.12109375, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22364217252396165, | |
| "grad_norm": 20.027837530897173, | |
| "learning_rate": 3.7234042553191484e-07, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.625, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.5562, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.6953125, | |
| "rewards/margins": 0.5390625, | |
| "rewards/rejected": 0.1552734375, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.25559105431309903, | |
| "grad_norm": 20.104841517252364, | |
| "learning_rate": 4.25531914893617e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -274.0, | |
| "loss": 0.5644, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.6875, | |
| "rewards/margins": 0.40234375, | |
| "rewards/rejected": 0.28515625, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28753993610223644, | |
| "grad_norm": 18.79087036985414, | |
| "learning_rate": 4.787234042553192e-07, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.6015625, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -284.0, | |
| "loss": 0.5118, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.87109375, | |
| "rewards/margins": 0.5859375, | |
| "rewards/rejected": 0.287109375, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3194888178913738, | |
| "grad_norm": 22.066743357299675, | |
| "learning_rate": 4.964497041420119e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -231.0, | |
| "logps/rejected": -288.0, | |
| "loss": 0.5409, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 1.0078125, | |
| "rewards/margins": 0.890625, | |
| "rewards/rejected": 0.12060546875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3514376996805112, | |
| "grad_norm": 19.3322147725973, | |
| "learning_rate": 4.905325443786982e-07, | |
| "logits/chosen": -1.546875, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -280.0, | |
| "loss": 0.5656, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.69921875, | |
| "rewards/margins": 0.458984375, | |
| "rewards/rejected": 0.2392578125, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38338658146964855, | |
| "grad_norm": 18.958252526138903, | |
| "learning_rate": 4.846153846153846e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.5077, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.8671875, | |
| "rewards/margins": 0.75390625, | |
| "rewards/rejected": 0.11376953125, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41533546325878595, | |
| "grad_norm": 24.459723181791798, | |
| "learning_rate": 4.78698224852071e-07, | |
| "logits/chosen": -1.5625, | |
| "logits/rejected": -1.625, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.4762, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.86328125, | |
| "rewards/margins": 0.81640625, | |
| "rewards/rejected": 0.04638671875, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4472843450479233, | |
| "grad_norm": 24.8748418291271, | |
| "learning_rate": 4.727810650887574e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -282.0, | |
| "loss": 0.5377, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.7890625, | |
| "rewards/margins": 0.640625, | |
| "rewards/rejected": 0.1484375, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4792332268370607, | |
| "grad_norm": 21.92108677224862, | |
| "learning_rate": 4.668639053254438e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -302.0, | |
| "loss": 0.5036, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.82421875, | |
| "rewards/margins": 0.79296875, | |
| "rewards/rejected": 0.029541015625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5111821086261981, | |
| "grad_norm": 23.160240290542404, | |
| "learning_rate": 4.6094674556213014e-07, | |
| "logits/chosen": -1.6328125, | |
| "logits/rejected": -1.6640625, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -290.0, | |
| "loss": 0.5215, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.84765625, | |
| "rewards/margins": 0.984375, | |
| "rewards/rejected": -0.1337890625, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5431309904153354, | |
| "grad_norm": 15.682473129649416, | |
| "learning_rate": 4.5502958579881655e-07, | |
| "logits/chosen": -1.6171875, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -230.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.4507, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.9296875, | |
| "rewards/margins": 0.9609375, | |
| "rewards/rejected": -0.02880859375, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5750798722044729, | |
| "grad_norm": 18.270260505484856, | |
| "learning_rate": 4.491124260355029e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.609375, | |
| "logps/chosen": -229.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.4635, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 1.0078125, | |
| "rewards/margins": 1.0234375, | |
| "rewards/rejected": -0.01531982421875, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6070287539936102, | |
| "grad_norm": 20.98631167244082, | |
| "learning_rate": 4.4319526627218936e-07, | |
| "logits/chosen": -1.6171875, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -230.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.4752, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.9296875, | |
| "rewards/margins": 0.953125, | |
| "rewards/rejected": -0.021484375, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6389776357827476, | |
| "grad_norm": 22.61128808819513, | |
| "learning_rate": 4.372781065088757e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.6953125, | |
| "logps/chosen": -233.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.4676, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.90234375, | |
| "rewards/margins": 1.0, | |
| "rewards/rejected": -0.1005859375, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.670926517571885, | |
| "grad_norm": 20.472996932111148, | |
| "learning_rate": 4.313609467455621e-07, | |
| "logits/chosen": -1.5625, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.4827, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.890625, | |
| "rewards/margins": 0.98828125, | |
| "rewards/rejected": -0.1015625, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7028753993610224, | |
| "grad_norm": 25.186112119569938, | |
| "learning_rate": 4.2544378698224847e-07, | |
| "logits/chosen": -1.5390625, | |
| "logits/rejected": -1.5859375, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -286.0, | |
| "loss": 0.5209, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.76171875, | |
| "rewards/margins": 0.92578125, | |
| "rewards/rejected": -0.1640625, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.7348242811501597, | |
| "grad_norm": 15.532080715300815, | |
| "learning_rate": 4.195266272189349e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.578125, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -290.0, | |
| "loss": 0.4544, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.7578125, | |
| "rewards/margins": 0.8515625, | |
| "rewards/rejected": -0.095703125, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7667731629392971, | |
| "grad_norm": 21.562659615610453, | |
| "learning_rate": 4.1360946745562133e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.4312, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.82421875, | |
| "rewards/margins": 0.96484375, | |
| "rewards/rejected": -0.1416015625, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7987220447284346, | |
| "grad_norm": 19.697962308480218, | |
| "learning_rate": 4.076923076923077e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -286.0, | |
| "loss": 0.4683, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.7421875, | |
| "rewards/margins": 0.97265625, | |
| "rewards/rejected": -0.23046875, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.8306709265175719, | |
| "grad_norm": 19.922270403408632, | |
| "learning_rate": 4.017751479289941e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.4589, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.84765625, | |
| "rewards/margins": 1.1796875, | |
| "rewards/rejected": -0.33203125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.8626198083067093, | |
| "grad_norm": 23.03342264376932, | |
| "learning_rate": 3.9585798816568044e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.4974, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.609375, | |
| "rewards/margins": 0.8125, | |
| "rewards/rejected": -0.203125, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8945686900958466, | |
| "grad_norm": 18.556977289134693, | |
| "learning_rate": 3.8994082840236685e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.6640625, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -280.0, | |
| "loss": 0.4835, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.66796875, | |
| "rewards/margins": 0.859375, | |
| "rewards/rejected": -0.1904296875, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.9265175718849841, | |
| "grad_norm": 20.159977656950176, | |
| "learning_rate": 3.840236686390532e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -227.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.4328, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.79296875, | |
| "rewards/margins": 1.046875, | |
| "rewards/rejected": -0.251953125, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.9584664536741214, | |
| "grad_norm": 18.20653927593905, | |
| "learning_rate": 3.7810650887573966e-07, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.59375, | |
| "logps/chosen": -239.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.4848, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.68359375, | |
| "rewards/margins": 1.09375, | |
| "rewards/rejected": -0.412109375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9904153354632588, | |
| "grad_norm": 22.32128048324979, | |
| "learning_rate": 3.72189349112426e-07, | |
| "logits/chosen": -1.578125, | |
| "logits/rejected": -1.59375, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -290.0, | |
| "loss": 0.5057, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.60546875, | |
| "rewards/margins": 0.86328125, | |
| "rewards/rejected": -0.255859375, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -1.609375, | |
| "eval_logits/rejected": -1.6171875, | |
| "eval_logps/chosen": -237.0, | |
| "eval_logps/rejected": -296.0, | |
| "eval_loss": 0.43648436665534973, | |
| "eval_rewards/accuracies": 0.7857142686843872, | |
| "eval_rewards/chosen": 0.58203125, | |
| "eval_rewards/margins": 0.9375, | |
| "eval_rewards/rejected": -0.35546875, | |
| "eval_runtime": 12.3846, | |
| "eval_samples_per_second": 16.149, | |
| "eval_steps_per_second": 0.565, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.0223642172523961, | |
| "grad_norm": 15.294291429304797, | |
| "learning_rate": 3.662721893491124e-07, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.6015625, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -288.0, | |
| "loss": 0.3827, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.53515625, | |
| "rewards/margins": 0.9765625, | |
| "rewards/rejected": -0.443359375, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.0543130990415335, | |
| "grad_norm": 17.823605067222292, | |
| "learning_rate": 3.603550295857988e-07, | |
| "logits/chosen": -1.546875, | |
| "logits/rejected": -1.59375, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.3641, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 0.609375, | |
| "rewards/margins": 1.171875, | |
| "rewards/rejected": -0.5546875, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0862619808306708, | |
| "grad_norm": 17.184762171933475, | |
| "learning_rate": 3.544378698224852e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.3731, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.6171875, | |
| "rewards/margins": 1.1328125, | |
| "rewards/rejected": -0.515625, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.1182108626198084, | |
| "grad_norm": 17.160818312302805, | |
| "learning_rate": 3.485207100591716e-07, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -227.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.3685, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 0.7265625, | |
| "rewards/margins": 1.4296875, | |
| "rewards/rejected": -0.703125, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.1501597444089458, | |
| "grad_norm": 17.582344513710463, | |
| "learning_rate": 3.42603550295858e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -290.0, | |
| "loss": 0.3573, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.5546875, | |
| "rewards/margins": 1.046875, | |
| "rewards/rejected": -0.494140625, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.182108626198083, | |
| "grad_norm": 13.601144206004982, | |
| "learning_rate": 3.366863905325444e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.3262, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 0.75390625, | |
| "rewards/margins": 1.328125, | |
| "rewards/rejected": -0.57421875, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.2140575079872205, | |
| "grad_norm": 14.002962198108948, | |
| "learning_rate": 3.3076923076923075e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -284.0, | |
| "loss": 0.3248, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.71484375, | |
| "rewards/margins": 1.1953125, | |
| "rewards/rejected": -0.484375, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.2460063897763578, | |
| "grad_norm": 16.239124276037153, | |
| "learning_rate": 3.2485207100591716e-07, | |
| "logits/chosen": -1.6171875, | |
| "logits/rejected": -1.609375, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.3467, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.84765625, | |
| "rewards/margins": 1.4453125, | |
| "rewards/rejected": -0.59375, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2779552715654952, | |
| "grad_norm": 16.58669423900164, | |
| "learning_rate": 3.189349112426035e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -288.0, | |
| "loss": 0.292, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 0.8515625, | |
| "rewards/margins": 1.4375, | |
| "rewards/rejected": -0.58984375, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.3099041533546325, | |
| "grad_norm": 14.052814099588554, | |
| "learning_rate": 3.130177514792899e-07, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.3533, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.6875, | |
| "rewards/margins": 1.390625, | |
| "rewards/rejected": -0.703125, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.34185303514377, | |
| "grad_norm": 13.679395410389454, | |
| "learning_rate": 3.071005917159763e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -304.0, | |
| "loss": 0.3461, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.61328125, | |
| "rewards/margins": 1.4765625, | |
| "rewards/rejected": -0.86328125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.3738019169329074, | |
| "grad_norm": 18.066697781388108, | |
| "learning_rate": 3.011834319526627e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.6953125, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.3133, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.765625, | |
| "rewards/margins": 1.578125, | |
| "rewards/rejected": -0.8125, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.4057507987220448, | |
| "grad_norm": 12.738714916763403, | |
| "learning_rate": 2.952662721893491e-07, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -1.625, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.315, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.67578125, | |
| "rewards/margins": 1.515625, | |
| "rewards/rejected": -0.83984375, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.4376996805111821, | |
| "grad_norm": 18.372879735041497, | |
| "learning_rate": 2.893491124260355e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.3155, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.60546875, | |
| "rewards/margins": 1.5390625, | |
| "rewards/rejected": -0.92578125, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.4696485623003195, | |
| "grad_norm": 16.824060938394247, | |
| "learning_rate": 2.834319526627219e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.609375, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.3328, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.5859375, | |
| "rewards/margins": 1.34375, | |
| "rewards/rejected": -0.7578125, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.5015974440894568, | |
| "grad_norm": 15.847190331363642, | |
| "learning_rate": 2.7751479289940824e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.3183, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 0.73828125, | |
| "rewards/margins": 1.5390625, | |
| "rewards/rejected": -0.80078125, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.5335463258785942, | |
| "grad_norm": 16.6941890115174, | |
| "learning_rate": 2.715976331360947e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.6171875, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.3562, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 0.66796875, | |
| "rewards/margins": 1.4453125, | |
| "rewards/rejected": -0.78125, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.5654952076677318, | |
| "grad_norm": 17.58079153854403, | |
| "learning_rate": 2.6568047337278106e-07, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -234.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.3033, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.7578125, | |
| "rewards/margins": 1.5859375, | |
| "rewards/rejected": -0.828125, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5974440894568689, | |
| "grad_norm": 12.797873931195115, | |
| "learning_rate": 2.5976331360946746e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -294.0, | |
| "loss": 0.3326, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.76171875, | |
| "rewards/margins": 1.6171875, | |
| "rewards/rejected": -0.85546875, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.6293929712460065, | |
| "grad_norm": 16.7014213517602, | |
| "learning_rate": 2.538461538461538e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -296.0, | |
| "loss": 0.342, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.474609375, | |
| "rewards/margins": 1.3046875, | |
| "rewards/rejected": -0.83203125, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.6613418530351438, | |
| "grad_norm": 16.319840171294768, | |
| "learning_rate": 2.479289940828402e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -233.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.3033, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.734375, | |
| "rewards/margins": 1.5703125, | |
| "rewards/rejected": -0.828125, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.6932907348242812, | |
| "grad_norm": 17.196203110672243, | |
| "learning_rate": 2.420118343195266e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.3623, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.6796875, | |
| "rewards/margins": 1.5234375, | |
| "rewards/rejected": -0.84375, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.7252396166134185, | |
| "grad_norm": 20.669442243399377, | |
| "learning_rate": 2.36094674556213e-07, | |
| "logits/chosen": -1.6484375, | |
| "logits/rejected": -1.6015625, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -302.0, | |
| "loss": 0.35, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 0.8203125, | |
| "rewards/margins": 1.765625, | |
| "rewards/rejected": -0.94921875, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.7571884984025559, | |
| "grad_norm": 17.765027713074762, | |
| "learning_rate": 2.301775147928994e-07, | |
| "logits/chosen": -1.6015625, | |
| "logits/rejected": -1.6640625, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -302.0, | |
| "loss": 0.3214, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 0.5859375, | |
| "rewards/margins": 1.4921875, | |
| "rewards/rejected": -0.90234375, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.7891373801916934, | |
| "grad_norm": 18.979661871107176, | |
| "learning_rate": 2.242603550295858e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.625, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -292.0, | |
| "loss": 0.35, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.51171875, | |
| "rewards/margins": 1.3203125, | |
| "rewards/rejected": -0.8125, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.8210862619808306, | |
| "grad_norm": 14.344211337956814, | |
| "learning_rate": 2.1834319526627217e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -234.0, | |
| "logps/rejected": -304.0, | |
| "loss": 0.3197, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.44921875, | |
| "rewards/margins": 1.6328125, | |
| "rewards/rejected": -1.1796875, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.8530351437699681, | |
| "grad_norm": 18.365794275487666, | |
| "learning_rate": 2.1242603550295858e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -239.0, | |
| "logps/rejected": -300.0, | |
| "loss": 0.3314, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.359375, | |
| "rewards/margins": 1.4609375, | |
| "rewards/rejected": -1.109375, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.8849840255591053, | |
| "grad_norm": 16.19312631132269, | |
| "learning_rate": 2.0650887573964496e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.578125, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.2839, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.5625, | |
| "rewards/margins": 1.625, | |
| "rewards/rejected": -1.0625, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.9169329073482428, | |
| "grad_norm": 12.485140486001677, | |
| "learning_rate": 2.0059171597633133e-07, | |
| "logits/chosen": -1.6015625, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -235.0, | |
| "logps/rejected": -308.0, | |
| "loss": 0.2658, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.57421875, | |
| "rewards/margins": 1.8046875, | |
| "rewards/rejected": -1.234375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.9488817891373802, | |
| "grad_norm": 12.432747016689122, | |
| "learning_rate": 1.9467455621301774e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -236.0, | |
| "logps/rejected": -308.0, | |
| "loss": 0.2782, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.71875, | |
| "rewards/margins": 1.875, | |
| "rewards/rejected": -1.15625, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.9808306709265175, | |
| "grad_norm": 18.34996934176123, | |
| "learning_rate": 1.8875739644970412e-07, | |
| "logits/chosen": -1.6015625, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -233.0, | |
| "logps/rejected": -302.0, | |
| "loss": 0.3063, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 0.71484375, | |
| "rewards/margins": 1.9140625, | |
| "rewards/rejected": -1.203125, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -1.640625, | |
| "eval_logits/rejected": -1.6484375, | |
| "eval_logps/chosen": -239.0, | |
| "eval_logps/rejected": -300.0, | |
| "eval_loss": 0.4112890660762787, | |
| "eval_rewards/accuracies": 0.8035714030265808, | |
| "eval_rewards/chosen": 0.40234375, | |
| "eval_rewards/margins": 1.265625, | |
| "eval_rewards/rejected": -0.859375, | |
| "eval_runtime": 12.3166, | |
| "eval_samples_per_second": 16.238, | |
| "eval_steps_per_second": 0.568, | |
| "step": 626 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 939, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |