| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.996784565916399, | |
| "eval_steps": 500, | |
| "global_step": 699, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 26.875, | |
| "learning_rate": 5.000000000000001e-07, | |
| "log_odds_chosen": 0.06212496757507324, | |
| "log_odds_ratio": -0.7014996409416199, | |
| "logits/chosen": -2.1857399940490723, | |
| "logits/rejected": -2.1817708015441895, | |
| "logps/chosen": -0.9498230814933777, | |
| "logps/rejected": -0.9784062504768372, | |
| "loss": 1.3118, | |
| "nll_loss": 1.3238964080810547, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 25.375, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "log_odds_chosen": 0.16793885827064514, | |
| "log_odds_ratio": -0.6694984436035156, | |
| "logits/chosen": -2.263418674468994, | |
| "logits/rejected": -2.186417818069458, | |
| "logps/chosen": -0.8150558471679688, | |
| "logps/rejected": -0.9238722920417786, | |
| "loss": 1.205, | |
| "nll_loss": 1.1971018314361572, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 7.125, | |
| "learning_rate": 1.5e-06, | |
| "log_odds_chosen": 0.00997834000736475, | |
| "log_odds_ratio": -0.7417184114456177, | |
| "logits/chosen": -2.3210864067077637, | |
| "logits/rejected": -2.2824535369873047, | |
| "logps/chosen": -0.7758495807647705, | |
| "logps/rejected": -0.7887269258499146, | |
| "loss": 1.0752, | |
| "nll_loss": 1.1008635759353638, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 6.46875, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "log_odds_chosen": 0.12381462007761002, | |
| "log_odds_ratio": -0.6990920305252075, | |
| "logits/chosen": -2.387354612350464, | |
| "logits/rejected": -2.278778553009033, | |
| "logps/chosen": -0.6582767963409424, | |
| "logps/rejected": -0.7136567831039429, | |
| "loss": 0.9839, | |
| "nll_loss": 0.9438871145248413, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 2.5e-06, | |
| "log_odds_chosen": 0.10633399337530136, | |
| "log_odds_ratio": -0.702051043510437, | |
| "logits/chosen": -2.3278417587280273, | |
| "logits/rejected": -2.264376163482666, | |
| "logps/chosen": -0.6687366366386414, | |
| "logps/rejected": -0.7335516214370728, | |
| "loss": 0.9634, | |
| "nll_loss": 0.9985629320144653, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 3e-06, | |
| "log_odds_chosen": 0.03238314390182495, | |
| "log_odds_ratio": -0.7480851411819458, | |
| "logits/chosen": -2.3533706665039062, | |
| "logits/rejected": -2.2710163593292236, | |
| "logps/chosen": -0.7110857963562012, | |
| "logps/rejected": -0.7359805107116699, | |
| "loss": 0.9691, | |
| "nll_loss": 0.9716413617134094, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 3.5e-06, | |
| "log_odds_chosen": 0.03486456722021103, | |
| "log_odds_ratio": -0.7344475984573364, | |
| "logits/chosen": -2.3200392723083496, | |
| "logits/rejected": -2.2653918266296387, | |
| "logps/chosen": -0.6566824316978455, | |
| "logps/rejected": -0.6836172938346863, | |
| "loss": 0.9322, | |
| "nll_loss": 0.9105528593063354, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 6.375, | |
| "learning_rate": 4.000000000000001e-06, | |
| "log_odds_chosen": 0.06068553403019905, | |
| "log_odds_ratio": -0.7511457204818726, | |
| "logits/chosen": -2.2898621559143066, | |
| "logits/rejected": -2.282437324523926, | |
| "logps/chosen": -0.6991879343986511, | |
| "logps/rejected": -0.7374504208564758, | |
| "loss": 0.9628, | |
| "nll_loss": 0.9841943979263306, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 4.5e-06, | |
| "log_odds_chosen": 0.18408647179603577, | |
| "log_odds_ratio": -0.6827269792556763, | |
| "logits/chosen": -2.3683393001556396, | |
| "logits/rejected": -2.2830727100372314, | |
| "logps/chosen": -0.6308820843696594, | |
| "logps/rejected": -0.7424929141998291, | |
| "loss": 0.8944, | |
| "nll_loss": 0.9193886518478394, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 5e-06, | |
| "log_odds_chosen": 0.035386841744184494, | |
| "log_odds_ratio": -0.7439508438110352, | |
| "logits/chosen": -2.35438871383667, | |
| "logits/rejected": -2.2569315433502197, | |
| "logps/chosen": -0.6244274973869324, | |
| "logps/rejected": -0.6488394141197205, | |
| "loss": 0.8782, | |
| "nll_loss": 0.8835960626602173, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 5.875, | |
| "learning_rate": 4.996562390352354e-06, | |
| "log_odds_chosen": 0.13479042053222656, | |
| "log_odds_ratio": -0.7032457590103149, | |
| "logits/chosen": -2.338320255279541, | |
| "logits/rejected": -2.259917974472046, | |
| "logps/chosen": -0.6392095685005188, | |
| "logps/rejected": -0.7337976098060608, | |
| "loss": 0.904, | |
| "nll_loss": 0.9141031503677368, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 5.125, | |
| "learning_rate": 4.986259015137485e-06, | |
| "log_odds_chosen": 0.1813308447599411, | |
| "log_odds_ratio": -0.6845273971557617, | |
| "logits/chosen": -2.3334765434265137, | |
| "logits/rejected": -2.208571434020996, | |
| "logps/chosen": -0.6667813062667847, | |
| "logps/rejected": -0.7594255208969116, | |
| "loss": 0.8882, | |
| "nll_loss": 0.9459765553474426, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 4.96911820954103e-06, | |
| "log_odds_chosen": 0.11342030763626099, | |
| "log_odds_ratio": -0.7043333649635315, | |
| "logits/chosen": -2.245288372039795, | |
| "logits/rejected": -2.161513328552246, | |
| "logps/chosen": -0.6394560933113098, | |
| "logps/rejected": -0.7009039521217346, | |
| "loss": 0.8905, | |
| "nll_loss": 0.8281471133232117, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 5.75, | |
| "learning_rate": 4.945187112281936e-06, | |
| "log_odds_chosen": 0.0401572659611702, | |
| "log_odds_ratio": -0.7479963302612305, | |
| "logits/chosen": -2.3580965995788574, | |
| "logits/rejected": -2.3082127571105957, | |
| "logps/chosen": -0.655213475227356, | |
| "logps/rejected": -0.6817822456359863, | |
| "loss": 0.9124, | |
| "nll_loss": 0.9263037443161011, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 4.9145315359768575e-06, | |
| "log_odds_chosen": 0.08960182219743729, | |
| "log_odds_ratio": -0.7190832495689392, | |
| "logits/chosen": -2.3376193046569824, | |
| "logits/rejected": -2.2543139457702637, | |
| "logps/chosen": -0.6171378493309021, | |
| "logps/rejected": -0.6685695648193359, | |
| "loss": 0.8818, | |
| "nll_loss": 0.8352767825126648, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 4.877235786149681e-06, | |
| "log_odds_chosen": 0.001317462301813066, | |
| "log_odds_ratio": -0.7520455121994019, | |
| "logits/chosen": -2.294182300567627, | |
| "logits/rejected": -2.249718189239502, | |
| "logps/chosen": -0.5758674144744873, | |
| "logps/rejected": -0.5995661020278931, | |
| "loss": 0.8668, | |
| "nll_loss": 0.8348628282546997, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 4.833402429383947e-06, | |
| "log_odds_chosen": 0.007824589498341084, | |
| "log_odds_ratio": -0.7456755638122559, | |
| "logits/chosen": -2.394925832748413, | |
| "logits/rejected": -2.3120040893554688, | |
| "logps/chosen": -0.6642740368843079, | |
| "logps/rejected": -0.6849616169929504, | |
| "loss": 0.8936, | |
| "nll_loss": 0.8833368420600891, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 4.783152011255739e-06, | |
| "log_odds_chosen": 0.06460268795490265, | |
| "log_odds_ratio": -0.7266359925270081, | |
| "logits/chosen": -2.2752573490142822, | |
| "logits/rejected": -2.2238731384277344, | |
| "logps/chosen": -0.6234402656555176, | |
| "logps/rejected": -0.6678077578544617, | |
| "loss": 0.8939, | |
| "nll_loss": 0.8869989514350891, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 4.726622724822781e-06, | |
| "log_odds_chosen": 0.07030437141656876, | |
| "log_odds_ratio": -0.7147783041000366, | |
| "logits/chosen": -2.2865307331085205, | |
| "logits/rejected": -2.2341747283935547, | |
| "logps/chosen": -0.6607599854469299, | |
| "logps/rejected": -0.7076841592788696, | |
| "loss": 0.8779, | |
| "nll_loss": 0.8933170437812805, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 5.125, | |
| "learning_rate": 4.663970030581408e-06, | |
| "log_odds_chosen": 0.11411430686712265, | |
| "log_odds_ratio": -0.7219884395599365, | |
| "logits/chosen": -2.235715389251709, | |
| "logits/rejected": -2.195624828338623, | |
| "logps/chosen": -0.5864480137825012, | |
| "logps/rejected": -0.6477882266044617, | |
| "loss": 0.8489, | |
| "nll_loss": 0.8421682119369507, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 4.59536622893656e-06, | |
| "log_odds_chosen": -0.019357014447450638, | |
| "log_odds_ratio": -0.7726019024848938, | |
| "logits/chosen": -2.271784782409668, | |
| "logits/rejected": -2.188135862350464, | |
| "logps/chosen": -0.6931017637252808, | |
| "logps/rejected": -0.7073479890823364, | |
| "loss": 0.8628, | |
| "nll_loss": 0.8622463345527649, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 4.520999986360555e-06, | |
| "log_odds_chosen": 0.07649385929107666, | |
| "log_odds_ratio": -0.7158172726631165, | |
| "logits/chosen": -2.250640392303467, | |
| "logits/rejected": -2.2138304710388184, | |
| "logps/chosen": -0.6532458662986755, | |
| "logps/rejected": -0.6902228593826294, | |
| "loss": 0.8572, | |
| "nll_loss": 0.868952751159668, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 4.441075816543745e-06, | |
| "log_odds_chosen": 0.030604243278503418, | |
| "log_odds_ratio": -0.7345600128173828, | |
| "logits/chosen": -2.3321053981781006, | |
| "logits/rejected": -2.2295913696289062, | |
| "logps/chosen": -0.6273137331008911, | |
| "logps/rejected": -0.6507210731506348, | |
| "loss": 0.8789, | |
| "nll_loss": 0.8177094459533691, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 4.355813517963924e-06, | |
| "log_odds_chosen": 0.40961089730262756, | |
| "log_odds_ratio": -0.5910171270370483, | |
| "logits/chosen": -2.3318493366241455, | |
| "logits/rejected": -2.2643091678619385, | |
| "logps/chosen": -0.4916546940803528, | |
| "logps/rejected": -0.6644417643547058, | |
| "loss": 0.758, | |
| "nll_loss": 0.740388810634613, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 6.25, | |
| "learning_rate": 4.265447569421234e-06, | |
| "log_odds_chosen": 0.3989773392677307, | |
| "log_odds_ratio": -0.590238630771637, | |
| "logits/chosen": -2.3867685794830322, | |
| "logits/rejected": -2.3056893348693848, | |
| "logps/chosen": -0.5096999406814575, | |
| "logps/rejected": -0.6727738976478577, | |
| "loss": 0.6881, | |
| "nll_loss": 0.7256556153297424, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 6.0625, | |
| "learning_rate": 4.170226485200899e-06, | |
| "log_odds_chosen": 0.48682594299316406, | |
| "log_odds_ratio": -0.5600502490997314, | |
| "logits/chosen": -2.3787410259246826, | |
| "logits/rejected": -2.30033016204834, | |
| "logps/chosen": -0.47443705797195435, | |
| "logps/rejected": -0.6933677792549133, | |
| "loss": 0.725, | |
| "nll_loss": 0.6927866339683533, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 5.875, | |
| "learning_rate": 4.070412131637139e-06, | |
| "log_odds_chosen": 0.38443347811698914, | |
| "log_odds_ratio": -0.6165519952774048, | |
| "logits/chosen": -2.3166797161102295, | |
| "logits/rejected": -2.23835825920105, | |
| "logps/chosen": -0.4729565978050232, | |
| "logps/rejected": -0.623404324054718, | |
| "loss": 0.719, | |
| "nll_loss": 0.689362645149231, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 3.966279006957781e-06, | |
| "log_odds_chosen": 0.4762851297855377, | |
| "log_odds_ratio": -0.5657437443733215, | |
| "logits/chosen": -2.3141160011291504, | |
| "logits/rejected": -2.2335238456726074, | |
| "logps/chosen": -0.5018330812454224, | |
| "logps/rejected": -0.7071572542190552, | |
| "loss": 0.7292, | |
| "nll_loss": 0.7584232687950134, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 5.625, | |
| "learning_rate": 3.858113486390056e-06, | |
| "log_odds_chosen": 0.461935818195343, | |
| "log_odds_ratio": -0.5941855907440186, | |
| "logits/chosen": -2.339001178741455, | |
| "logits/rejected": -2.2689056396484375, | |
| "logps/chosen": -0.48000845313072205, | |
| "logps/rejected": -0.7079610228538513, | |
| "loss": 0.7325, | |
| "nll_loss": 0.7081630229949951, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 6.375, | |
| "learning_rate": 3.7462130346036e-06, | |
| "log_odds_chosen": 0.4929097592830658, | |
| "log_odds_ratio": -0.563489556312561, | |
| "logits/chosen": -2.3332343101501465, | |
| "logits/rejected": -2.2668535709381104, | |
| "logps/chosen": -0.4893871247768402, | |
| "logps/rejected": -0.7047632336616516, | |
| "loss": 0.6818, | |
| "nll_loss": 0.7156692147254944, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 5.625, | |
| "learning_rate": 3.6308853876565232e-06, | |
| "log_odds_chosen": 0.4562360346317291, | |
| "log_odds_ratio": -0.5689755082130432, | |
| "logits/chosen": -2.3754940032958984, | |
| "logits/rejected": -2.27487850189209, | |
| "logps/chosen": -0.4727197289466858, | |
| "logps/rejected": -0.6664949655532837, | |
| "loss": 0.7119, | |
| "nll_loss": 0.6855541467666626, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 5.375, | |
| "learning_rate": 3.512447706694254e-06, | |
| "log_odds_chosen": 0.3677065372467041, | |
| "log_odds_ratio": -0.6210616230964661, | |
| "logits/chosen": -2.3326685428619385, | |
| "logits/rejected": -2.284008502960205, | |
| "logps/chosen": -0.5006684064865112, | |
| "logps/rejected": -0.6665970087051392, | |
| "loss": 0.7148, | |
| "nll_loss": 0.7629369497299194, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 3.3912257057285684e-06, | |
| "log_odds_chosen": 0.5201979279518127, | |
| "log_odds_ratio": -0.5558315515518188, | |
| "logits/chosen": -2.35974383354187, | |
| "logits/rejected": -2.2589011192321777, | |
| "logps/chosen": -0.44807687401771545, | |
| "logps/rejected": -0.6775830984115601, | |
| "loss": 0.7127, | |
| "nll_loss": 0.6746495962142944, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 3.2675527558954897e-06, | |
| "log_odds_chosen": 0.4437088966369629, | |
| "log_odds_ratio": -0.5744566321372986, | |
| "logits/chosen": -2.3658008575439453, | |
| "logits/rejected": -2.302891969680786, | |
| "logps/chosen": -0.5020125508308411, | |
| "logps/rejected": -0.703687310218811, | |
| "loss": 0.7293, | |
| "nll_loss": 0.7448792457580566, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 6.21875, | |
| "learning_rate": 3.1417689686554144e-06, | |
| "log_odds_chosen": 0.4869805872440338, | |
| "log_odds_ratio": -0.5718838572502136, | |
| "logits/chosen": -2.3735198974609375, | |
| "logits/rejected": -2.3357536792755127, | |
| "logps/chosen": -0.4905489385128021, | |
| "logps/rejected": -0.7079612612724304, | |
| "loss": 0.7228, | |
| "nll_loss": 0.7263522148132324, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 5.25, | |
| "learning_rate": 3.0142202604567724e-06, | |
| "log_odds_chosen": 0.4622114300727844, | |
| "log_odds_ratio": -0.580187201499939, | |
| "logits/chosen": -2.4162421226501465, | |
| "logits/rejected": -2.32672119140625, | |
| "logps/chosen": -0.4755636155605316, | |
| "logps/rejected": -0.6753752827644348, | |
| "loss": 0.6993, | |
| "nll_loss": 0.7006024122238159, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 5.90625, | |
| "learning_rate": 2.8852574014354394e-06, | |
| "log_odds_chosen": 0.4685123562812805, | |
| "log_odds_ratio": -0.5639849901199341, | |
| "logits/chosen": -2.3573315143585205, | |
| "logits/rejected": -2.2810652256011963, | |
| "logps/chosen": -0.45586276054382324, | |
| "logps/rejected": -0.650667667388916, | |
| "loss": 0.6937, | |
| "nll_loss": 0.6792919039726257, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 2.7552350507661063e-06, | |
| "log_odds_chosen": 0.32628968358039856, | |
| "log_odds_ratio": -0.6266478300094604, | |
| "logits/chosen": -2.3865950107574463, | |
| "logits/rejected": -2.2999796867370605, | |
| "logps/chosen": -0.5150719285011292, | |
| "logps/rejected": -0.6419684290885925, | |
| "loss": 0.7454, | |
| "nll_loss": 0.7746638059616089, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 2.6245107813184286e-06, | |
| "log_odds_chosen": 0.4124643802642822, | |
| "log_odds_ratio": -0.6084017157554626, | |
| "logits/chosen": -2.300006151199341, | |
| "logits/rejected": -2.2486844062805176, | |
| "logps/chosen": -0.500686526298523, | |
| "logps/rejected": -0.7045504450798035, | |
| "loss": 0.7104, | |
| "nll_loss": 0.7621539235115051, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 2.493444096300273e-06, | |
| "log_odds_chosen": 0.3271673321723938, | |
| "log_odds_ratio": -0.6363809704780579, | |
| "logits/chosen": -2.332576274871826, | |
| "logits/rejected": -2.2507784366607666, | |
| "logps/chosen": -0.5043616890907288, | |
| "logps/rejected": -0.6402055621147156, | |
| "loss": 0.7316, | |
| "nll_loss": 0.7292400598526001, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 2.3623954405923636e-06, | |
| "log_odds_chosen": 0.35860806703567505, | |
| "log_odds_ratio": -0.6040534973144531, | |
| "logits/chosen": -2.3241639137268066, | |
| "logits/rejected": -2.2687346935272217, | |
| "logps/chosen": -0.4667055010795593, | |
| "logps/rejected": -0.6222271919250488, | |
| "loss": 0.7055, | |
| "nll_loss": 0.6745852828025818, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 2.2317252094932383e-06, | |
| "log_odds_chosen": 0.4658966064453125, | |
| "log_odds_ratio": -0.5793284177780151, | |
| "logits/chosen": -2.3523142337799072, | |
| "logits/rejected": -2.2792649269104004, | |
| "logps/chosen": -0.4811071753501892, | |
| "logps/rejected": -0.6847606897354126, | |
| "loss": 0.6927, | |
| "nll_loss": 0.7262173295021057, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 5.75, | |
| "learning_rate": 2.1017927576005657e-06, | |
| "log_odds_chosen": 0.4586619436740875, | |
| "log_odds_ratio": -0.5749837756156921, | |
| "logits/chosen": -2.309678554534912, | |
| "logits/rejected": -2.2428524494171143, | |
| "logps/chosen": -0.4688965380191803, | |
| "logps/rejected": -0.6654346585273743, | |
| "loss": 0.6911, | |
| "nll_loss": 0.7090088129043579, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.9729554105544816e-06, | |
| "log_odds_chosen": 0.39126816391944885, | |
| "log_odds_ratio": -0.5965334177017212, | |
| "logits/chosen": -2.246513605117798, | |
| "logits/rejected": -2.1855998039245605, | |
| "logps/chosen": -0.5077515840530396, | |
| "logps/rejected": -0.685882031917572, | |
| "loss": 0.73, | |
| "nll_loss": 0.696818470954895, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.8455674823607312e-06, | |
| "log_odds_chosen": 0.2512452006340027, | |
| "log_odds_ratio": -0.6640199422836304, | |
| "logits/chosen": -2.342719078063965, | |
| "logits/rejected": -2.274512529373169, | |
| "logps/chosen": -0.521196722984314, | |
| "logps/rejected": -0.6386003494262695, | |
| "loss": 0.7091, | |
| "nll_loss": 0.7769466638565063, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 1.7199793009960766e-06, | |
| "log_odds_chosen": 0.4366021156311035, | |
| "log_odds_ratio": -0.5971043705940247, | |
| "logits/chosen": -2.2552707195281982, | |
| "logits/rejected": -2.2170627117156982, | |
| "logps/chosen": -0.5236691832542419, | |
| "logps/rejected": -0.7121491432189941, | |
| "loss": 0.7011, | |
| "nll_loss": 0.7103704214096069, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.5965362449756317e-06, | |
| "log_odds_chosen": 0.5843161344528198, | |
| "log_odds_ratio": -0.5287169218063354, | |
| "logits/chosen": -2.3433125019073486, | |
| "logits/rejected": -2.2500529289245605, | |
| "logps/chosen": -0.43613916635513306, | |
| "logps/rejected": -0.6702120304107666, | |
| "loss": 0.6759, | |
| "nll_loss": 0.6880279183387756, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 6.5, | |
| "learning_rate": 1.4755777935316412e-06, | |
| "log_odds_chosen": 0.8127249479293823, | |
| "log_odds_ratio": -0.46383658051490784, | |
| "logits/chosen": -2.3992159366607666, | |
| "logits/rejected": -2.2700352668762207, | |
| "logps/chosen": -0.4094364047050476, | |
| "logps/rejected": -0.7208075523376465, | |
| "loss": 0.6001, | |
| "nll_loss": 0.5964864492416382, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 1.3574365930158272e-06, | |
| "log_odds_chosen": 0.7765734791755676, | |
| "log_odds_ratio": -0.4649999141693115, | |
| "logits/chosen": -2.294707775115967, | |
| "logits/rejected": -2.244535446166992, | |
| "logps/chosen": -0.3731532692909241, | |
| "logps/rejected": -0.6663237810134888, | |
| "loss": 0.5958, | |
| "nll_loss": 0.5779340863227844, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 5.34375, | |
| "learning_rate": 1.242437542092731e-06, | |
| "log_odds_chosen": 0.8575220108032227, | |
| "log_odds_ratio": -0.42505908012390137, | |
| "logits/chosen": -2.2913150787353516, | |
| "logits/rejected": -2.2416951656341553, | |
| "logps/chosen": -0.37527984380722046, | |
| "logps/rejected": -0.6905524730682373, | |
| "loss": 0.6326, | |
| "nll_loss": 0.623663604259491, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 1.1308968982398893e-06, | |
| "log_odds_chosen": 0.912135899066925, | |
| "log_odds_ratio": -0.41568368673324585, | |
| "logits/chosen": -2.313683271408081, | |
| "logits/rejected": -2.226602554321289, | |
| "logps/chosen": -0.3589690625667572, | |
| "logps/rejected": -0.7188132405281067, | |
| "loss": 0.5746, | |
| "nll_loss": 0.5657839179039001, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.0231214080120354e-06, | |
| "log_odds_chosen": 0.8424990773200989, | |
| "log_odds_ratio": -0.4574614465236664, | |
| "logits/chosen": -2.2810280323028564, | |
| "logits/rejected": -2.2405149936676025, | |
| "logps/chosen": -0.39062148332595825, | |
| "logps/rejected": -0.705338716506958, | |
| "loss": 0.616, | |
| "nll_loss": 0.6053126454353333, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 9.194074634611577e-07, | |
| "log_odds_chosen": 0.8043780326843262, | |
| "log_odds_ratio": -0.4463415741920471, | |
| "logits/chosen": -2.3421316146850586, | |
| "logits/rejected": -2.27047061920166, | |
| "logps/chosen": -0.3879227340221405, | |
| "logps/rejected": -0.681617259979248, | |
| "loss": 0.5906, | |
| "nll_loss": 0.6051921844482422, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 8.200402870323634e-07, | |
| "log_odds_chosen": 0.8168398141860962, | |
| "log_odds_ratio": -0.4651332497596741, | |
| "logits/chosen": -2.2850687503814697, | |
| "logits/rejected": -2.203339099884033, | |
| "logps/chosen": -0.4002053141593933, | |
| "logps/rejected": -0.7145139575004578, | |
| "loss": 0.6095, | |
| "nll_loss": 0.5440846681594849, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 7.252931471771322e-07, | |
| "log_odds_chosen": 0.957280158996582, | |
| "log_odds_ratio": -0.4218501150608063, | |
| "logits/chosen": -2.3870935440063477, | |
| "logits/rejected": -2.3037381172180176, | |
| "logps/chosen": -0.3641536235809326, | |
| "logps/rejected": -0.7135647535324097, | |
| "loss": 0.5756, | |
| "nll_loss": 0.5520345568656921, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 6.0, | |
| "learning_rate": 6.354266068411078e-07, | |
| "log_odds_chosen": 0.8184865713119507, | |
| "log_odds_ratio": -0.4773890972137451, | |
| "logits/chosen": -2.2534213066101074, | |
| "logits/rejected": -2.22419810295105, | |
| "logps/chosen": -0.41363048553466797, | |
| "logps/rejected": -0.7267066240310669, | |
| "loss": 0.6069, | |
| "nll_loss": 0.6007322072982788, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 5.50687806893139e-07, | |
| "log_odds_chosen": 0.8616418838500977, | |
| "log_odds_ratio": -0.44923096895217896, | |
| "logits/chosen": -2.276895761489868, | |
| "logits/rejected": -2.2783918380737305, | |
| "logps/chosen": -0.37030458450317383, | |
| "logps/rejected": -0.666776180267334, | |
| "loss": 0.614, | |
| "nll_loss": 0.6462346911430359, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 6.1875, | |
| "learning_rate": 4.7130978646620807e-07, | |
| "log_odds_chosen": 0.9556438326835632, | |
| "log_odds_ratio": -0.4390975534915924, | |
| "logits/chosen": -2.3128373622894287, | |
| "logits/rejected": -2.2027499675750732, | |
| "logps/chosen": -0.3874739706516266, | |
| "logps/rejected": -0.7368007898330688, | |
| "loss": 0.5845, | |
| "nll_loss": 0.6006309986114502, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 5.5, | |
| "learning_rate": 3.975108420793819e-07, | |
| "log_odds_chosen": 0.8426518440246582, | |
| "log_odds_ratio": -0.44841188192367554, | |
| "logits/chosen": -2.2656946182250977, | |
| "logits/rejected": -2.2316298484802246, | |
| "logps/chosen": -0.3601939082145691, | |
| "logps/rejected": -0.6829022765159607, | |
| "loss": 0.6035, | |
| "nll_loss": 0.5817385911941528, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 3.294939273032272e-07, | |
| "log_odds_chosen": 0.9172550439834595, | |
| "log_odds_ratio": -0.41558733582496643, | |
| "logits/chosen": -2.251704692840576, | |
| "logits/rejected": -2.228672742843628, | |
| "logps/chosen": -0.3531130254268646, | |
| "logps/rejected": -0.703464925289154, | |
| "loss": 0.5694, | |
| "nll_loss": 0.554794430732727, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 2.6744609461969523e-07, | |
| "log_odds_chosen": 0.877754807472229, | |
| "log_odds_ratio": -0.4350369870662689, | |
| "logits/chosen": -2.3534913063049316, | |
| "logits/rejected": -2.2918014526367188, | |
| "logps/chosen": -0.37197771668434143, | |
| "logps/rejected": -0.7188286781311035, | |
| "loss": 0.594, | |
| "nll_loss": 0.5847772359848022, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 2.1153798101138405e-07, | |
| "log_odds_chosen": 0.9580931663513184, | |
| "log_odds_ratio": -0.4062252938747406, | |
| "logits/chosen": -2.2987186908721924, | |
| "logits/rejected": -2.2561445236206055, | |
| "logps/chosen": -0.37130284309387207, | |
| "logps/rejected": -0.7186514735221863, | |
| "loss": 0.6113, | |
| "nll_loss": 0.6211617588996887, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 11.5625, | |
| "learning_rate": 1.61923338694871e-07, | |
| "log_odds_chosen": 0.8980989456176758, | |
| "log_odds_ratio": -0.44537553191185, | |
| "logits/chosen": -2.3041529655456543, | |
| "logits/rejected": -2.246042490005493, | |
| "logps/chosen": -0.3840414583683014, | |
| "logps/rejected": -0.7412872910499573, | |
| "loss": 0.6259, | |
| "nll_loss": 0.6146137714385986, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.1873861228862998e-07, | |
| "log_odds_chosen": 0.9331363439559937, | |
| "log_odds_ratio": -0.4274202883243561, | |
| "logits/chosen": -2.2929160594940186, | |
| "logits/rejected": -2.2195255756378174, | |
| "logps/chosen": -0.3945469558238983, | |
| "logps/rejected": -0.7453508377075195, | |
| "loss": 0.6133, | |
| "nll_loss": 0.621927797794342, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "grad_norm": 6.21875, | |
| "learning_rate": 8.210256357836065e-08, | |
| "log_odds_chosen": 1.0095536708831787, | |
| "log_odds_ratio": -0.3956521451473236, | |
| "logits/chosen": -2.263673782348633, | |
| "logits/rejected": -2.20683217048645, | |
| "logps/chosen": -0.3427670896053314, | |
| "logps/rejected": -0.7156583070755005, | |
| "loss": 0.574, | |
| "nll_loss": 0.5451396703720093, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 5.21159449116615e-08, | |
| "log_odds_chosen": 0.770971417427063, | |
| "log_odds_ratio": -0.4887896180152893, | |
| "logits/chosen": -2.284029006958008, | |
| "logits/rejected": -2.214066982269287, | |
| "logps/chosen": -0.4013218879699707, | |
| "logps/rejected": -0.6818682551383972, | |
| "loss": 0.5932, | |
| "nll_loss": 0.5923025012016296, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 2.8861222120235845e-08, | |
| "log_odds_chosen": 0.8828635215759277, | |
| "log_odds_ratio": -0.4312991201877594, | |
| "logits/chosen": -2.24355411529541, | |
| "logits/rejected": -2.219613552093506, | |
| "logps/chosen": -0.34176358580589294, | |
| "logps/rejected": -0.6538265943527222, | |
| "loss": 0.5864, | |
| "nll_loss": 0.58399498462677, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 1.2402347731620412e-08, | |
| "log_odds_chosen": 0.7922724485397339, | |
| "log_odds_ratio": -0.4506749212741852, | |
| "logits/chosen": -2.275207996368408, | |
| "logits/rejected": -2.178473949432373, | |
| "logps/chosen": -0.40824776887893677, | |
| "logps/rejected": -0.7051072120666504, | |
| "loss": 0.5958, | |
| "nll_loss": 0.5631005167961121, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 2.7845850941254914e-09, | |
| "log_odds_chosen": 0.8703590631484985, | |
| "log_odds_ratio": -0.4304262101650238, | |
| "logits/chosen": -2.343256711959839, | |
| "logits/rejected": -2.2737724781036377, | |
| "logps/chosen": -0.366679310798645, | |
| "logps/rejected": -0.6560879349708557, | |
| "loss": 0.6025, | |
| "nll_loss": 0.5677012205123901, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 699, | |
| "total_flos": 0.0, | |
| "train_loss": 0.7502862380468454, | |
| "train_runtime": 21621.4849, | |
| "train_samples_per_second": 2.071, | |
| "train_steps_per_second": 0.032 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 699, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |