| { | |
| "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": 14.15949417784876, | |
| "learning_rate": 5.3191489361702125e-09, | |
| "logits/chosen": -0.875, | |
| "logits/rejected": -0.8671875, | |
| "logps/chosen": -29.375, | |
| "logps/rejected": -38.0, | |
| "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": 12.702097036550947, | |
| "learning_rate": 5.3191489361702123e-08, | |
| "logits/chosen": -0.828125, | |
| "logits/rejected": -0.91796875, | |
| "logps/chosen": -26.875, | |
| "logps/rejected": -27.25, | |
| "loss": 0.6917, | |
| "rewards/accuracies": 0.1944444477558136, | |
| "rewards/chosen": 0.00121307373046875, | |
| "rewards/margins": 0.0020751953125, | |
| "rewards/rejected": -0.0008697509765625, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06389776357827476, | |
| "grad_norm": 14.201541608200035, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -0.921875, | |
| "logits/rejected": -0.93359375, | |
| "logps/chosen": -30.75, | |
| "logps/rejected": -24.625, | |
| "loss": 0.6924, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": 0.00141143798828125, | |
| "rewards/margins": 7.724761962890625e-05, | |
| "rewards/rejected": 0.0013275146484375, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09584664536741214, | |
| "grad_norm": 12.074020986080455, | |
| "learning_rate": 1.5957446808510638e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.9453125, | |
| "logps/chosen": -32.25, | |
| "logps/rejected": -28.875, | |
| "loss": 0.6916, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.0011749267578125, | |
| "rewards/margins": 0.008056640625, | |
| "rewards/rejected": -0.00689697265625, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12779552715654952, | |
| "grad_norm": 13.073953515135278, | |
| "learning_rate": 2.127659574468085e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.890625, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -26.375, | |
| "loss": 0.6888, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.00186920166015625, | |
| "rewards/margins": 0.002593994140625, | |
| "rewards/rejected": -0.00445556640625, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1597444089456869, | |
| "grad_norm": 11.709384491164828, | |
| "learning_rate": 2.659574468085106e-07, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -27.75, | |
| "logps/rejected": -23.5, | |
| "loss": 0.6822, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.0009307861328125, | |
| "rewards/margins": 0.0206298828125, | |
| "rewards/rejected": -0.0196533203125, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19169329073482427, | |
| "grad_norm": 11.881387515429367, | |
| "learning_rate": 3.1914893617021275e-07, | |
| "logits/chosen": -0.921875, | |
| "logits/rejected": -0.90234375, | |
| "logps/chosen": -26.25, | |
| "logps/rejected": -29.0, | |
| "loss": 0.6851, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.0028839111328125, | |
| "rewards/margins": 0.01385498046875, | |
| "rewards/rejected": -0.010986328125, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22364217252396165, | |
| "grad_norm": 15.065379921321332, | |
| "learning_rate": 3.7234042553191484e-07, | |
| "logits/chosen": -0.921875, | |
| "logits/rejected": -0.921875, | |
| "logps/chosen": -42.0, | |
| "logps/rejected": -24.75, | |
| "loss": 0.6652, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.015625, | |
| "rewards/margins": 0.07958984375, | |
| "rewards/rejected": -0.09521484375, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.25559105431309903, | |
| "grad_norm": 9.160023538736992, | |
| "learning_rate": 4.25531914893617e-07, | |
| "logits/chosen": -0.94921875, | |
| "logits/rejected": -0.9140625, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -30.875, | |
| "loss": 0.6678, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.08447265625, | |
| "rewards/margins": 0.0703125, | |
| "rewards/rejected": -0.154296875, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28753993610223644, | |
| "grad_norm": 11.744213372479827, | |
| "learning_rate": 4.787234042553192e-07, | |
| "logits/chosen": -0.953125, | |
| "logits/rejected": -0.95703125, | |
| "logps/chosen": -35.0, | |
| "logps/rejected": -25.125, | |
| "loss": 0.6325, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.047607421875, | |
| "rewards/margins": 0.1240234375, | |
| "rewards/rejected": -0.076171875, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3194888178913738, | |
| "grad_norm": 12.175021032470806, | |
| "learning_rate": 4.964497041420119e-07, | |
| "logits/chosen": -0.96875, | |
| "logits/rejected": -0.90625, | |
| "logps/chosen": -29.0, | |
| "logps/rejected": -27.375, | |
| "loss": 0.6503, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.015380859375, | |
| "rewards/margins": 0.09716796875, | |
| "rewards/rejected": -0.08203125, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3514376996805112, | |
| "grad_norm": 15.178658782857307, | |
| "learning_rate": 4.905325443786982e-07, | |
| "logits/chosen": -0.99609375, | |
| "logits/rejected": -0.8984375, | |
| "logps/chosen": -31.75, | |
| "logps/rejected": -34.75, | |
| "loss": 0.6204, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.046142578125, | |
| "rewards/margins": 0.1982421875, | |
| "rewards/rejected": -0.15234375, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38338658146964855, | |
| "grad_norm": 16.16675788193662, | |
| "learning_rate": 4.846153846153846e-07, | |
| "logits/chosen": -0.96875, | |
| "logits/rejected": -0.93359375, | |
| "logps/chosen": -42.75, | |
| "logps/rejected": -36.25, | |
| "loss": 0.6309, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.2294921875, | |
| "rewards/margins": 0.2216796875, | |
| "rewards/rejected": -0.451171875, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41533546325878595, | |
| "grad_norm": 10.193866441849577, | |
| "learning_rate": 4.78698224852071e-07, | |
| "logits/chosen": -0.9453125, | |
| "logits/rejected": -0.9375, | |
| "logps/chosen": -30.25, | |
| "logps/rejected": -33.0, | |
| "loss": 0.6548, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.1484375, | |
| "rewards/margins": 0.228515625, | |
| "rewards/rejected": -0.376953125, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4472843450479233, | |
| "grad_norm": 12.44670055298775, | |
| "learning_rate": 4.727810650887574e-07, | |
| "logits/chosen": -0.94140625, | |
| "logits/rejected": -0.9375, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -33.75, | |
| "loss": 0.6025, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.1123046875, | |
| "rewards/margins": 0.373046875, | |
| "rewards/rejected": -0.486328125, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4792332268370607, | |
| "grad_norm": 9.718117808971781, | |
| "learning_rate": 4.668639053254438e-07, | |
| "logits/chosen": -0.97265625, | |
| "logits/rejected": -0.9375, | |
| "logps/chosen": -35.25, | |
| "logps/rejected": -34.75, | |
| "loss": 0.6049, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.365234375, | |
| "rewards/rejected": -0.69921875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5111821086261981, | |
| "grad_norm": 12.959238451666906, | |
| "learning_rate": 4.6094674556213014e-07, | |
| "logits/chosen": -0.9296875, | |
| "logits/rejected": -0.859375, | |
| "logps/chosen": -34.0, | |
| "logps/rejected": -32.75, | |
| "loss": 0.5895, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.3359375, | |
| "rewards/margins": 0.4140625, | |
| "rewards/rejected": -0.75, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5431309904153354, | |
| "grad_norm": 10.449440815665023, | |
| "learning_rate": 4.5502958579881655e-07, | |
| "logits/chosen": -0.8984375, | |
| "logits/rejected": -0.91015625, | |
| "logps/chosen": -30.75, | |
| "logps/rejected": -31.75, | |
| "loss": 0.6003, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.306640625, | |
| "rewards/rejected": -0.6484375, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5750798722044729, | |
| "grad_norm": 21.831604570880106, | |
| "learning_rate": 4.491124260355029e-07, | |
| "logits/chosen": -0.8984375, | |
| "logits/rejected": -0.90625, | |
| "logps/chosen": -39.0, | |
| "logps/rejected": -35.75, | |
| "loss": 0.5721, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.515625, | |
| "rewards/margins": 0.388671875, | |
| "rewards/rejected": -0.90625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6070287539936102, | |
| "grad_norm": 10.29045072439074, | |
| "learning_rate": 4.4319526627218936e-07, | |
| "logits/chosen": -0.9296875, | |
| "logits/rejected": -0.87109375, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -30.25, | |
| "loss": 0.609, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.470703125, | |
| "rewards/margins": 0.419921875, | |
| "rewards/rejected": -0.890625, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6389776357827476, | |
| "grad_norm": 20.548507017891822, | |
| "learning_rate": 4.372781065088757e-07, | |
| "logits/chosen": -0.9375, | |
| "logits/rejected": -0.83984375, | |
| "logps/chosen": -36.75, | |
| "logps/rejected": -35.0, | |
| "loss": 0.6401, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.58984375, | |
| "rewards/margins": 0.251953125, | |
| "rewards/rejected": -0.83984375, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.670926517571885, | |
| "grad_norm": 9.119895639662692, | |
| "learning_rate": 4.313609467455621e-07, | |
| "logits/chosen": -0.9140625, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -32.0, | |
| "logps/rejected": -32.5, | |
| "loss": 0.5835, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.404296875, | |
| "rewards/margins": 0.390625, | |
| "rewards/rejected": -0.796875, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7028753993610224, | |
| "grad_norm": 10.016002486299485, | |
| "learning_rate": 4.2544378698224847e-07, | |
| "logits/chosen": -0.94921875, | |
| "logits/rejected": -0.90625, | |
| "logps/chosen": -36.0, | |
| "logps/rejected": -33.25, | |
| "loss": 0.5812, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.447265625, | |
| "rewards/rejected": -0.76953125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.7348242811501597, | |
| "grad_norm": 8.739343921649029, | |
| "learning_rate": 4.195266272189349e-07, | |
| "logits/chosen": -0.9453125, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -40.75, | |
| "logps/rejected": -36.75, | |
| "loss": 0.5854, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.52734375, | |
| "rewards/margins": 0.60546875, | |
| "rewards/rejected": -1.1328125, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7667731629392971, | |
| "grad_norm": 20.107539290726088, | |
| "learning_rate": 4.1360946745562133e-07, | |
| "logits/chosen": -0.8984375, | |
| "logits/rejected": -0.87109375, | |
| "logps/chosen": -33.75, | |
| "logps/rejected": -30.125, | |
| "loss": 0.6357, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -0.5546875, | |
| "rewards/margins": 0.19140625, | |
| "rewards/rejected": -0.74609375, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7987220447284346, | |
| "grad_norm": 13.470404552518705, | |
| "learning_rate": 4.076923076923077e-07, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.92578125, | |
| "logps/chosen": -39.75, | |
| "logps/rejected": -39.75, | |
| "loss": 0.5625, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.52734375, | |
| "rewards/margins": 0.5078125, | |
| "rewards/rejected": -1.03125, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.8306709265175719, | |
| "grad_norm": 13.121143639536331, | |
| "learning_rate": 4.017751479289941e-07, | |
| "logits/chosen": -0.8828125, | |
| "logits/rejected": -0.8359375, | |
| "logps/chosen": -34.5, | |
| "logps/rejected": -34.75, | |
| "loss": 0.5635, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.72265625, | |
| "rewards/margins": 0.484375, | |
| "rewards/rejected": -1.2109375, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.8626198083067093, | |
| "grad_norm": 10.329929244037169, | |
| "learning_rate": 3.9585798816568044e-07, | |
| "logits/chosen": -0.90625, | |
| "logits/rejected": -0.859375, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -41.25, | |
| "loss": 0.595, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.796875, | |
| "rewards/margins": 0.58203125, | |
| "rewards/rejected": -1.375, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8945686900958466, | |
| "grad_norm": 16.191548100134025, | |
| "learning_rate": 3.8994082840236685e-07, | |
| "logits/chosen": -0.8515625, | |
| "logits/rejected": -0.875, | |
| "logps/chosen": -42.25, | |
| "logps/rejected": -38.0, | |
| "loss": 0.555, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6875, | |
| "rewards/margins": 0.63671875, | |
| "rewards/rejected": -1.328125, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.9265175718849841, | |
| "grad_norm": 12.02406721548161, | |
| "learning_rate": 3.840236686390532e-07, | |
| "logits/chosen": -0.94140625, | |
| "logits/rejected": -0.86328125, | |
| "logps/chosen": -42.5, | |
| "logps/rejected": -43.5, | |
| "loss": 0.5886, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.76171875, | |
| "rewards/margins": 0.53125, | |
| "rewards/rejected": -1.296875, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.9584664536741214, | |
| "grad_norm": 5.752074059086699, | |
| "learning_rate": 3.7810650887573966e-07, | |
| "logits/chosen": -0.87890625, | |
| "logits/rejected": -0.890625, | |
| "logps/chosen": -31.75, | |
| "logps/rejected": -46.75, | |
| "loss": 0.5734, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.66796875, | |
| "rewards/margins": 0.67578125, | |
| "rewards/rejected": -1.34375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9904153354632588, | |
| "grad_norm": 10.316894713261737, | |
| "learning_rate": 3.72189349112426e-07, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.89453125, | |
| "logps/chosen": -43.5, | |
| "logps/rejected": -44.75, | |
| "loss": 0.5646, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.703125, | |
| "rewards/margins": 0.5859375, | |
| "rewards/rejected": -1.2890625, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -0.9296875, | |
| "eval_logits/rejected": -0.921875, | |
| "eval_logps/chosen": -45.5, | |
| "eval_logps/rejected": -38.0, | |
| "eval_loss": 0.5726562738418579, | |
| "eval_rewards/accuracies": 0.4107142984867096, | |
| "eval_rewards/chosen": -0.65625, | |
| "eval_rewards/margins": 0.53125, | |
| "eval_rewards/rejected": -1.1875, | |
| "eval_runtime": 12.4256, | |
| "eval_samples_per_second": 16.096, | |
| "eval_steps_per_second": 0.563, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.0223642172523961, | |
| "grad_norm": 11.604345250214385, | |
| "learning_rate": 3.662721893491124e-07, | |
| "logits/chosen": -0.859375, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -44.25, | |
| "loss": 0.4945, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.51171875, | |
| "rewards/margins": 0.921875, | |
| "rewards/rejected": -1.4296875, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.0543130990415335, | |
| "grad_norm": 8.164279499540298, | |
| "learning_rate": 3.603550295857988e-07, | |
| "logits/chosen": -0.93359375, | |
| "logits/rejected": -0.8359375, | |
| "logps/chosen": -41.5, | |
| "logps/rejected": -47.0, | |
| "loss": 0.487, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.51171875, | |
| "rewards/margins": 0.98046875, | |
| "rewards/rejected": -1.4921875, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0862619808306708, | |
| "grad_norm": 10.098456650224257, | |
| "learning_rate": 3.544378698224852e-07, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.89453125, | |
| "logps/chosen": -38.25, | |
| "logps/rejected": -46.5, | |
| "loss": 0.4921, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.74609375, | |
| "rewards/margins": 0.71875, | |
| "rewards/rejected": -1.4609375, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.1182108626198084, | |
| "grad_norm": 12.312245605594391, | |
| "learning_rate": 3.485207100591716e-07, | |
| "logits/chosen": -0.8984375, | |
| "logits/rejected": -0.9140625, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -44.0, | |
| "loss": 0.4951, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.671875, | |
| "rewards/margins": 0.86328125, | |
| "rewards/rejected": -1.5390625, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.1501597444089458, | |
| "grad_norm": 7.6387271402180925, | |
| "learning_rate": 3.42603550295858e-07, | |
| "logits/chosen": -0.95703125, | |
| "logits/rejected": -0.875, | |
| "logps/chosen": -39.5, | |
| "logps/rejected": -43.5, | |
| "loss": 0.5121, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.74609375, | |
| "rewards/margins": 0.8125, | |
| "rewards/rejected": -1.5625, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.182108626198083, | |
| "grad_norm": 9.066794201428493, | |
| "learning_rate": 3.366863905325444e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.890625, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -43.25, | |
| "loss": 0.5005, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.9140625, | |
| "rewards/margins": 0.7890625, | |
| "rewards/rejected": -1.703125, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.2140575079872205, | |
| "grad_norm": 13.248069014797954, | |
| "learning_rate": 3.3076923076923075e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.91796875, | |
| "logps/chosen": -37.5, | |
| "logps/rejected": -51.0, | |
| "loss": 0.4653, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.71484375, | |
| "rewards/margins": 1.015625, | |
| "rewards/rejected": -1.7265625, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.2460063897763578, | |
| "grad_norm": 10.867663199240486, | |
| "learning_rate": 3.2485207100591716e-07, | |
| "logits/chosen": -0.94140625, | |
| "logits/rejected": -0.890625, | |
| "logps/chosen": -42.25, | |
| "logps/rejected": -45.75, | |
| "loss": 0.4909, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5625, | |
| "rewards/margins": 1.109375, | |
| "rewards/rejected": -1.671875, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2779552715654952, | |
| "grad_norm": 12.647214530771212, | |
| "learning_rate": 3.189349112426035e-07, | |
| "logits/chosen": -0.84765625, | |
| "logits/rejected": -0.921875, | |
| "logps/chosen": -29.625, | |
| "logps/rejected": -60.0, | |
| "loss": 0.4816, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.96484375, | |
| "rewards/margins": 1.0, | |
| "rewards/rejected": -1.96875, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.3099041533546325, | |
| "grad_norm": 13.831769507648124, | |
| "learning_rate": 3.130177514792899e-07, | |
| "logits/chosen": -0.89453125, | |
| "logits/rejected": -0.85546875, | |
| "logps/chosen": -32.5, | |
| "logps/rejected": -43.75, | |
| "loss": 0.5379, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -1.078125, | |
| "rewards/margins": 0.50390625, | |
| "rewards/rejected": -1.578125, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.34185303514377, | |
| "grad_norm": 12.24140890124011, | |
| "learning_rate": 3.071005917159763e-07, | |
| "logits/chosen": -0.9375, | |
| "logits/rejected": -0.9375, | |
| "logps/chosen": -41.0, | |
| "logps/rejected": -53.25, | |
| "loss": 0.4786, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -1.265625, | |
| "rewards/margins": 1.046875, | |
| "rewards/rejected": -2.3125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.3738019169329074, | |
| "grad_norm": 8.001226654192951, | |
| "learning_rate": 3.011834319526627e-07, | |
| "logits/chosen": -0.95703125, | |
| "logits/rejected": -0.8984375, | |
| "logps/chosen": -44.5, | |
| "logps/rejected": -50.0, | |
| "loss": 0.4892, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -1.0703125, | |
| "rewards/margins": 0.984375, | |
| "rewards/rejected": -2.046875, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.4057507987220448, | |
| "grad_norm": 13.90707171430186, | |
| "learning_rate": 2.952662721893491e-07, | |
| "logits/chosen": -0.95703125, | |
| "logits/rejected": -0.87890625, | |
| "logps/chosen": -44.0, | |
| "logps/rejected": -47.25, | |
| "loss": 0.4805, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -1.03125, | |
| "rewards/margins": 1.0546875, | |
| "rewards/rejected": -2.09375, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.4376996805111821, | |
| "grad_norm": 11.71134112335707, | |
| "learning_rate": 2.893491124260355e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.86328125, | |
| "logps/chosen": -46.5, | |
| "logps/rejected": -45.25, | |
| "loss": 0.4582, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -1.171875, | |
| "rewards/margins": 0.83984375, | |
| "rewards/rejected": -2.015625, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.4696485623003195, | |
| "grad_norm": 10.206570733769754, | |
| "learning_rate": 2.834319526627219e-07, | |
| "logits/chosen": -1.0, | |
| "logits/rejected": -0.921875, | |
| "logps/chosen": -44.25, | |
| "logps/rejected": -46.75, | |
| "loss": 0.4662, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -1.1015625, | |
| "rewards/margins": 0.8984375, | |
| "rewards/rejected": -2.015625, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.5015974440894568, | |
| "grad_norm": 8.481816160880287, | |
| "learning_rate": 2.7751479289940824e-07, | |
| "logits/chosen": -0.97265625, | |
| "logits/rejected": -0.8671875, | |
| "logps/chosen": -43.0, | |
| "logps/rejected": -54.0, | |
| "loss": 0.4951, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -1.2421875, | |
| "rewards/margins": 1.078125, | |
| "rewards/rejected": -2.3125, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.5335463258785942, | |
| "grad_norm": 9.450961138318995, | |
| "learning_rate": 2.715976331360947e-07, | |
| "logits/chosen": -0.96484375, | |
| "logits/rejected": -0.859375, | |
| "logps/chosen": -45.0, | |
| "logps/rejected": -50.5, | |
| "loss": 0.4609, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.8359375, | |
| "rewards/margins": 1.359375, | |
| "rewards/rejected": -2.1875, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.5654952076677318, | |
| "grad_norm": 10.079665777936816, | |
| "learning_rate": 2.6568047337278106e-07, | |
| "logits/chosen": -0.9140625, | |
| "logits/rejected": -0.8671875, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -46.25, | |
| "loss": 0.484, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -1.171875, | |
| "rewards/margins": 0.91796875, | |
| "rewards/rejected": -2.09375, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5974440894568689, | |
| "grad_norm": 5.892858649998084, | |
| "learning_rate": 2.5976331360946746e-07, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.85546875, | |
| "logps/chosen": -43.5, | |
| "logps/rejected": -44.0, | |
| "loss": 0.4725, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.96875, | |
| "rewards/margins": 0.8984375, | |
| "rewards/rejected": -1.8671875, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.6293929712460065, | |
| "grad_norm": 19.866826656455657, | |
| "learning_rate": 2.538461538461538e-07, | |
| "logits/chosen": -0.88671875, | |
| "logits/rejected": -0.875, | |
| "logps/chosen": -37.0, | |
| "logps/rejected": -49.0, | |
| "loss": 0.4713, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -1.015625, | |
| "rewards/margins": 1.09375, | |
| "rewards/rejected": -2.109375, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.6613418530351438, | |
| "grad_norm": 11.462164104029984, | |
| "learning_rate": 2.479289940828402e-07, | |
| "logits/chosen": -1.0, | |
| "logits/rejected": -0.86328125, | |
| "logps/chosen": -48.25, | |
| "logps/rejected": -46.5, | |
| "loss": 0.4437, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.69140625, | |
| "rewards/margins": 1.28125, | |
| "rewards/rejected": -1.9765625, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.6932907348242812, | |
| "grad_norm": 4.754132390092114, | |
| "learning_rate": 2.420118343195266e-07, | |
| "logits/chosen": -0.984375, | |
| "logits/rejected": -0.89453125, | |
| "logps/chosen": -43.5, | |
| "logps/rejected": -44.5, | |
| "loss": 0.4539, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -1.0234375, | |
| "rewards/margins": 1.0234375, | |
| "rewards/rejected": -2.046875, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.7252396166134185, | |
| "grad_norm": 13.855605681924747, | |
| "learning_rate": 2.36094674556213e-07, | |
| "logits/chosen": -1.0234375, | |
| "logits/rejected": -0.828125, | |
| "logps/chosen": -50.0, | |
| "logps/rejected": -45.75, | |
| "loss": 0.4537, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.81640625, | |
| "rewards/margins": 1.203125, | |
| "rewards/rejected": -2.015625, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.7571884984025559, | |
| "grad_norm": 13.694710524562387, | |
| "learning_rate": 2.301775147928994e-07, | |
| "logits/chosen": -0.98828125, | |
| "logits/rejected": -0.91796875, | |
| "logps/chosen": -41.75, | |
| "logps/rejected": -52.5, | |
| "loss": 0.4505, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -1.09375, | |
| "rewards/margins": 1.265625, | |
| "rewards/rejected": -2.359375, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.7891373801916934, | |
| "grad_norm": 12.265823828649788, | |
| "learning_rate": 2.242603550295858e-07, | |
| "logits/chosen": -0.98828125, | |
| "logits/rejected": -0.92578125, | |
| "logps/chosen": -39.75, | |
| "logps/rejected": -59.0, | |
| "loss": 0.4626, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.86328125, | |
| "rewards/margins": 1.3828125, | |
| "rewards/rejected": -2.25, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.8210862619808306, | |
| "grad_norm": 15.811792160256678, | |
| "learning_rate": 2.1834319526627217e-07, | |
| "logits/chosen": -0.95703125, | |
| "logits/rejected": -0.8984375, | |
| "logps/chosen": -40.5, | |
| "logps/rejected": -50.5, | |
| "loss": 0.4507, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.609375, | |
| "rewards/margins": 1.28125, | |
| "rewards/rejected": -1.8828125, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.8530351437699681, | |
| "grad_norm": 12.03300574763422, | |
| "learning_rate": 2.1242603550295858e-07, | |
| "logits/chosen": -1.046875, | |
| "logits/rejected": -0.91015625, | |
| "logps/chosen": -42.25, | |
| "logps/rejected": -49.25, | |
| "loss": 0.45, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.62890625, | |
| "rewards/margins": 1.3984375, | |
| "rewards/rejected": -2.03125, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.8849840255591053, | |
| "grad_norm": 6.498105360999796, | |
| "learning_rate": 2.0650887573964496e-07, | |
| "logits/chosen": -0.9921875, | |
| "logits/rejected": -0.8828125, | |
| "logps/chosen": -45.75, | |
| "logps/rejected": -52.75, | |
| "loss": 0.4347, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.94921875, | |
| "rewards/margins": 1.546875, | |
| "rewards/rejected": -2.5, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.9169329073482428, | |
| "grad_norm": 17.196261475524192, | |
| "learning_rate": 2.0059171597633133e-07, | |
| "logits/chosen": -0.91796875, | |
| "logits/rejected": -0.859375, | |
| "logps/chosen": -39.25, | |
| "logps/rejected": -50.75, | |
| "loss": 0.4452, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -1.1875, | |
| "rewards/margins": 1.171875, | |
| "rewards/rejected": -2.359375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.9488817891373802, | |
| "grad_norm": 6.682211779881807, | |
| "learning_rate": 1.9467455621301774e-07, | |
| "logits/chosen": -0.81640625, | |
| "logits/rejected": -0.93359375, | |
| "logps/chosen": -42.25, | |
| "logps/rejected": -57.25, | |
| "loss": 0.4483, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -1.2890625, | |
| "rewards/margins": 1.265625, | |
| "rewards/rejected": -2.546875, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.9808306709265175, | |
| "grad_norm": 8.358366031471173, | |
| "learning_rate": 1.8875739644970412e-07, | |
| "logits/chosen": -0.9921875, | |
| "logits/rejected": -0.90234375, | |
| "logps/chosen": -45.25, | |
| "logps/rejected": -57.75, | |
| "loss": 0.4365, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -1.296875, | |
| "rewards/margins": 1.40625, | |
| "rewards/rejected": -2.703125, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -1.03125, | |
| "eval_logits/rejected": -0.9375, | |
| "eval_logps/chosen": -51.75, | |
| "eval_logps/rejected": -49.75, | |
| "eval_loss": 0.5380077958106995, | |
| "eval_rewards/accuracies": 0.5, | |
| "eval_rewards/chosen": -1.2890625, | |
| "eval_rewards/margins": 1.078125, | |
| "eval_rewards/rejected": -2.375, | |
| "eval_runtime": 12.4486, | |
| "eval_samples_per_second": 16.066, | |
| "eval_steps_per_second": 0.562, | |
| "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 | |
| } | |