| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0020920502092050207, | |
| "grad_norm": 9.16837234811457, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -2.7655434608459473, | |
| "logits/rejected": -2.7158141136169434, | |
| "logps/chosen": -269.8756103515625, | |
| "logps/rejected": -360.6216125488281, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02092050209205021, | |
| "grad_norm": 9.188352376621864, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.5924127101898193, | |
| "logits/rejected": -2.562790632247925, | |
| "logps/chosen": -264.7654724121094, | |
| "logps/rejected": -251.53240966796875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.00027513509849086404, | |
| "rewards/margins": -9.524480992695317e-05, | |
| "rewards/rejected": 0.00037037988658994436, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04184100418410042, | |
| "grad_norm": 8.405862905672732, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.6532669067382812, | |
| "logits/rejected": -2.605334997177124, | |
| "logps/chosen": -281.41314697265625, | |
| "logps/rejected": -296.80389404296875, | |
| "loss": 0.6924, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.0009061357122845948, | |
| "rewards/margins": 0.0009388698963448405, | |
| "rewards/rejected": -3.2734242267906666e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06276150627615062, | |
| "grad_norm": 8.298184796125266, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.6643333435058594, | |
| "logits/rejected": -2.5918784141540527, | |
| "logps/chosen": -299.40911865234375, | |
| "logps/rejected": -263.7054748535156, | |
| "loss": 0.6884, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": 0.012471171095967293, | |
| "rewards/margins": 0.010278983041644096, | |
| "rewards/rejected": 0.002192187588661909, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08368200836820083, | |
| "grad_norm": 8.153488217762229, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.556988477706909, | |
| "logits/rejected": -2.5218937397003174, | |
| "logps/chosen": -263.11932373046875, | |
| "logps/rejected": -245.16162109375, | |
| "loss": 0.6774, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.039762504398822784, | |
| "rewards/margins": 0.047586839646101, | |
| "rewards/rejected": -0.007824338972568512, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.10460251046025104, | |
| "grad_norm": 8.472043005988137, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -2.497899055480957, | |
| "logits/rejected": -2.457843780517578, | |
| "logps/chosen": -279.7335205078125, | |
| "logps/rejected": -297.64471435546875, | |
| "loss": 0.6611, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 6.177672185003757e-05, | |
| "rewards/margins": 0.04885866492986679, | |
| "rewards/rejected": -0.0487968847155571, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12552301255230125, | |
| "grad_norm": 11.234854782109899, | |
| "learning_rate": 4.990398100856366e-07, | |
| "logits/chosen": -2.5127627849578857, | |
| "logits/rejected": -2.4409728050231934, | |
| "logps/chosen": -273.99835205078125, | |
| "logps/rejected": -313.34991455078125, | |
| "loss": 0.6384, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.02334701642394066, | |
| "rewards/margins": 0.1286955624818802, | |
| "rewards/rejected": -0.15204259753227234, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14644351464435146, | |
| "grad_norm": 10.848666524907674, | |
| "learning_rate": 4.967775735898179e-07, | |
| "logits/chosen": -2.360724449157715, | |
| "logits/rejected": -2.3582727909088135, | |
| "logps/chosen": -271.3426513671875, | |
| "logps/rejected": -283.6495666503906, | |
| "loss": 0.6157, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.05673746392130852, | |
| "rewards/margins": 0.21231010556221008, | |
| "rewards/rejected": -0.2690475583076477, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16736401673640167, | |
| "grad_norm": 19.300722351787282, | |
| "learning_rate": 4.931986719649298e-07, | |
| "logits/chosen": -2.557610273361206, | |
| "logits/rejected": -2.4776089191436768, | |
| "logps/chosen": -337.05694580078125, | |
| "logps/rejected": -305.25653076171875, | |
| "loss": 0.6024, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.1427958607673645, | |
| "rewards/margins": 0.26870062947273254, | |
| "rewards/rejected": -0.41149649024009705, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18828451882845187, | |
| "grad_norm": 20.757494940175373, | |
| "learning_rate": 4.883222001996351e-07, | |
| "logits/chosen": -2.0054640769958496, | |
| "logits/rejected": -1.9451837539672852, | |
| "logps/chosen": -310.46173095703125, | |
| "logps/rejected": -353.8874816894531, | |
| "loss": 0.5716, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.5131224393844604, | |
| "rewards/margins": 0.5017130970954895, | |
| "rewards/rejected": -1.0148354768753052, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "grad_norm": 18.659168826580355, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -1.321213960647583, | |
| "logits/rejected": -1.129945158958435, | |
| "logps/chosen": -307.0208435058594, | |
| "logps/rejected": -361.28350830078125, | |
| "loss": 0.5659, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.466524213552475, | |
| "rewards/margins": 0.5754273533821106, | |
| "rewards/rejected": -1.0419515371322632, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "eval_logits/chosen": -0.9821491241455078, | |
| "eval_logits/rejected": -0.7858694791793823, | |
| "eval_logps/chosen": -334.3749084472656, | |
| "eval_logps/rejected": -392.64556884765625, | |
| "eval_loss": 0.5701435804367065, | |
| "eval_rewards/accuracies": 0.71484375, | |
| "eval_rewards/chosen": -0.7174500226974487, | |
| "eval_rewards/margins": 0.5823848843574524, | |
| "eval_rewards/rejected": -1.299834966659546, | |
| "eval_runtime": 99.7714, | |
| "eval_samples_per_second": 20.046, | |
| "eval_steps_per_second": 0.321, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2301255230125523, | |
| "grad_norm": 17.55736534565649, | |
| "learning_rate": 4.747874028753375e-07, | |
| "logits/chosen": -1.0365899801254272, | |
| "logits/rejected": -0.571634829044342, | |
| "logps/chosen": -372.1576843261719, | |
| "logps/rejected": -384.3282165527344, | |
| "loss": 0.57, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.676292896270752, | |
| "rewards/margins": 0.5884906053543091, | |
| "rewards/rejected": -1.2647836208343506, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2510460251046025, | |
| "grad_norm": 20.044772578276955, | |
| "learning_rate": 4.662012913161997e-07, | |
| "logits/chosen": -0.6921079754829407, | |
| "logits/rejected": -0.4483819603919983, | |
| "logps/chosen": -354.0092468261719, | |
| "logps/rejected": -391.1457824707031, | |
| "loss": 0.5521, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.7282991409301758, | |
| "rewards/margins": 0.5842112898826599, | |
| "rewards/rejected": -1.31251060962677, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2719665271966527, | |
| "grad_norm": 20.68475231351549, | |
| "learning_rate": 4.5646165232345103e-07, | |
| "logits/chosen": -0.3410087823867798, | |
| "logits/rejected": -0.04661134257912636, | |
| "logps/chosen": -377.7615966796875, | |
| "logps/rejected": -426.6678771972656, | |
| "loss": 0.5404, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.9577744603157043, | |
| "rewards/margins": 0.6243709325790405, | |
| "rewards/rejected": -1.5821453332901, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2928870292887029, | |
| "grad_norm": 17.900468089200906, | |
| "learning_rate": 4.456204510851956e-07, | |
| "logits/chosen": -0.7499442100524902, | |
| "logits/rejected": -0.48287004232406616, | |
| "logps/chosen": -378.3758850097656, | |
| "logps/rejected": -425.58148193359375, | |
| "loss": 0.5392, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6240967512130737, | |
| "rewards/margins": 0.6252509355545044, | |
| "rewards/rejected": -1.2493478059768677, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3138075313807531, | |
| "grad_norm": 28.14974497379761, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": -0.2267804592847824, | |
| "logits/rejected": 0.15516111254692078, | |
| "logps/chosen": -373.9515380859375, | |
| "logps/rejected": -421.302001953125, | |
| "loss": 0.5406, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.9071490168571472, | |
| "rewards/margins": 0.5438954830169678, | |
| "rewards/rejected": -1.4510444402694702, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33472803347280333, | |
| "grad_norm": 26.21920743231775, | |
| "learning_rate": 4.2087030056579986e-07, | |
| "logits/chosen": -0.5155479907989502, | |
| "logits/rejected": 0.20259642601013184, | |
| "logps/chosen": -330.3095703125, | |
| "logps/rejected": -401.3794250488281, | |
| "loss": 0.5368, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.5726653337478638, | |
| "rewards/margins": 0.8669228553771973, | |
| "rewards/rejected": -1.439588189125061, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35564853556485354, | |
| "grad_norm": 29.878459650293532, | |
| "learning_rate": 4.070934040463998e-07, | |
| "logits/chosen": -0.2268867939710617, | |
| "logits/rejected": 0.22177407145500183, | |
| "logps/chosen": -337.85345458984375, | |
| "logps/rejected": -385.3772888183594, | |
| "loss": 0.5435, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8853822946548462, | |
| "rewards/margins": 0.6570433974266052, | |
| "rewards/rejected": -1.542425513267517, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.37656903765690375, | |
| "grad_norm": 23.677554496063618, | |
| "learning_rate": 3.9247834624635404e-07, | |
| "logits/chosen": 0.14835922420024872, | |
| "logits/rejected": 0.698757529258728, | |
| "logps/chosen": -335.9161682128906, | |
| "logps/rejected": -375.7064514160156, | |
| "loss": 0.504, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.7958806753158569, | |
| "rewards/margins": 0.6854288578033447, | |
| "rewards/rejected": -1.4813096523284912, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.39748953974895396, | |
| "grad_norm": 23.203690544454616, | |
| "learning_rate": 3.7710310482256523e-07, | |
| "logits/chosen": 0.024832600727677345, | |
| "logits/rejected": 0.5344244837760925, | |
| "logps/chosen": -348.7262268066406, | |
| "logps/rejected": -404.1893005371094, | |
| "loss": 0.5174, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.7743078470230103, | |
| "rewards/margins": 0.6698081493377686, | |
| "rewards/rejected": -1.4441161155700684, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "grad_norm": 50.04702866429734, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": 0.0846758633852005, | |
| "logits/rejected": 0.5746791958808899, | |
| "logps/chosen": -342.2915954589844, | |
| "logps/rejected": -408.9627685546875, | |
| "loss": 0.5493, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.8239612579345703, | |
| "rewards/margins": 0.7452048659324646, | |
| "rewards/rejected": -1.5691661834716797, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "eval_logits/chosen": -0.31423458456993103, | |
| "eval_logits/rejected": 0.28053638339042664, | |
| "eval_logps/chosen": -334.3751525878906, | |
| "eval_logps/rejected": -417.25006103515625, | |
| "eval_loss": 0.5263724327087402, | |
| "eval_rewards/accuracies": 0.7734375, | |
| "eval_rewards/chosen": -0.717452347278595, | |
| "eval_rewards/margins": 0.8284271955490112, | |
| "eval_rewards/rejected": -1.545879602432251, | |
| "eval_runtime": 97.9594, | |
| "eval_samples_per_second": 20.417, | |
| "eval_steps_per_second": 0.327, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4393305439330544, | |
| "grad_norm": 24.06066475347068, | |
| "learning_rate": 3.4440382358952115e-07, | |
| "logits/chosen": -0.26085931062698364, | |
| "logits/rejected": 0.5198220014572144, | |
| "logps/chosen": -365.29010009765625, | |
| "logps/rejected": -394.47723388671875, | |
| "loss": 0.5396, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.7562572360038757, | |
| "rewards/margins": 0.675930380821228, | |
| "rewards/rejected": -1.4321876764297485, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4602510460251046, | |
| "grad_norm": 28.79899654759585, | |
| "learning_rate": 3.272542485937368e-07, | |
| "logits/chosen": 0.5355015397071838, | |
| "logits/rejected": 1.2820476293563843, | |
| "logps/chosen": -367.3331604003906, | |
| "logps/rejected": -419.95538330078125, | |
| "loss": 0.5283, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.0195591449737549, | |
| "rewards/margins": 0.7230145335197449, | |
| "rewards/rejected": -1.7425734996795654, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4811715481171548, | |
| "grad_norm": 21.063139682190887, | |
| "learning_rate": 3.096924887558854e-07, | |
| "logits/chosen": 0.5966787338256836, | |
| "logits/rejected": 1.3136788606643677, | |
| "logps/chosen": -338.0008239746094, | |
| "logps/rejected": -426.60162353515625, | |
| "loss": 0.5352, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.8904770612716675, | |
| "rewards/margins": 0.8978098630905151, | |
| "rewards/rejected": -1.7882869243621826, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.502092050209205, | |
| "grad_norm": 22.340209509704692, | |
| "learning_rate": 2.9181224366319943e-07, | |
| "logits/chosen": 0.36957186460494995, | |
| "logits/rejected": 0.9800383448600769, | |
| "logps/chosen": -344.3677062988281, | |
| "logps/rejected": -398.9412536621094, | |
| "loss": 0.5049, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.8063153028488159, | |
| "rewards/margins": 0.6925672292709351, | |
| "rewards/rejected": -1.498882532119751, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5230125523012552, | |
| "grad_norm": 25.32395453596937, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": 0.6675643920898438, | |
| "logits/rejected": 1.5727179050445557, | |
| "logps/chosen": -386.45343017578125, | |
| "logps/rejected": -431.385986328125, | |
| "loss": 0.5053, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.8825046420097351, | |
| "rewards/margins": 0.835895836353302, | |
| "rewards/rejected": -1.7184005975723267, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5439330543933054, | |
| "grad_norm": 32.04785964174811, | |
| "learning_rate": 2.55479083351317e-07, | |
| "logits/chosen": 0.7512863874435425, | |
| "logits/rejected": 1.690630316734314, | |
| "logps/chosen": -394.2430114746094, | |
| "logps/rejected": -439.67205810546875, | |
| "loss": 0.5095, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.9994521141052246, | |
| "rewards/margins": 0.8580014109611511, | |
| "rewards/rejected": -1.8574535846710205, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5648535564853556, | |
| "grad_norm": 24.904083578928375, | |
| "learning_rate": 2.3722002126275822e-07, | |
| "logits/chosen": 0.7148577570915222, | |
| "logits/rejected": 1.4790935516357422, | |
| "logps/chosen": -385.87371826171875, | |
| "logps/rejected": -443.53839111328125, | |
| "loss": 0.5171, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.0889328718185425, | |
| "rewards/margins": 0.76996910572052, | |
| "rewards/rejected": -1.8589019775390625, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5857740585774058, | |
| "grad_norm": 25.263400006388935, | |
| "learning_rate": 2.19029145890313e-07, | |
| "logits/chosen": 0.6911333799362183, | |
| "logits/rejected": 1.6273902654647827, | |
| "logps/chosen": -368.5780334472656, | |
| "logps/rejected": -429.0997009277344, | |
| "loss": 0.5258, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0685292482376099, | |
| "rewards/margins": 0.8003263473510742, | |
| "rewards/rejected": -1.8688554763793945, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.606694560669456, | |
| "grad_norm": 26.846743184768357, | |
| "learning_rate": 2.0100351342479216e-07, | |
| "logits/chosen": 0.6784590482711792, | |
| "logits/rejected": 1.4638246297836304, | |
| "logps/chosen": -365.30206298828125, | |
| "logps/rejected": -430.20660400390625, | |
| "loss": 0.5113, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.1421018838882446, | |
| "rewards/margins": 0.7403939366340637, | |
| "rewards/rejected": -1.8824958801269531, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "grad_norm": 29.645083802417286, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": 0.45342984795570374, | |
| "logits/rejected": 1.6018460988998413, | |
| "logps/chosen": -403.8066711425781, | |
| "logps/rejected": -440.07098388671875, | |
| "loss": 0.493, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.0277760028839111, | |
| "rewards/margins": 0.7721818685531616, | |
| "rewards/rejected": -1.7999578714370728, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "eval_logits/chosen": 0.5289936065673828, | |
| "eval_logits/rejected": 1.2686660289764404, | |
| "eval_logps/chosen": -356.9808044433594, | |
| "eval_logps/rejected": -444.52154541015625, | |
| "eval_loss": 0.5118551850318909, | |
| "eval_rewards/accuracies": 0.76171875, | |
| "eval_rewards/chosen": -0.9435087442398071, | |
| "eval_rewards/margins": 0.8750853538513184, | |
| "eval_rewards/rejected": -1.8185943365097046, | |
| "eval_runtime": 97.9113, | |
| "eval_samples_per_second": 20.427, | |
| "eval_steps_per_second": 0.327, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6485355648535565, | |
| "grad_norm": 26.615301380561615, | |
| "learning_rate": 1.6583128063291573e-07, | |
| "logits/chosen": 0.6738792657852173, | |
| "logits/rejected": 1.4009501934051514, | |
| "logps/chosen": -398.75958251953125, | |
| "logps/rejected": -437.30633544921875, | |
| "loss": 0.4932, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.0345656871795654, | |
| "rewards/margins": 0.753504753112793, | |
| "rewards/rejected": -1.7880703210830688, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6694560669456067, | |
| "grad_norm": 32.234049319731376, | |
| "learning_rate": 1.488723393865766e-07, | |
| "logits/chosen": 0.8933159112930298, | |
| "logits/rejected": 1.8882709741592407, | |
| "logps/chosen": -401.068115234375, | |
| "logps/rejected": -431.6943359375, | |
| "loss": 0.4861, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.0735524892807007, | |
| "rewards/margins": 0.7948076128959656, | |
| "rewards/rejected": -1.868360161781311, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6903765690376569, | |
| "grad_norm": 34.15386792642874, | |
| "learning_rate": 1.3245295796480788e-07, | |
| "logits/chosen": 0.35191336274147034, | |
| "logits/rejected": 1.2871559858322144, | |
| "logps/chosen": -360.3251037597656, | |
| "logps/rejected": -431.16534423828125, | |
| "loss": 0.496, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.9808635711669922, | |
| "rewards/margins": 0.7355011701583862, | |
| "rewards/rejected": -1.7163646221160889, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7112970711297071, | |
| "grad_norm": 36.00524525047531, | |
| "learning_rate": 1.1666074087171627e-07, | |
| "logits/chosen": 0.36807161569595337, | |
| "logits/rejected": 1.3383424282073975, | |
| "logps/chosen": -374.64703369140625, | |
| "logps/rejected": -469.4212951660156, | |
| "loss": 0.5019, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.8456922769546509, | |
| "rewards/margins": 1.047476053237915, | |
| "rewards/rejected": -1.8931680917739868, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7322175732217573, | |
| "grad_norm": 28.054514541029086, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": 0.6923775672912598, | |
| "logits/rejected": 1.5702488422393799, | |
| "logps/chosen": -356.87725830078125, | |
| "logps/rejected": -427.8857421875, | |
| "loss": 0.4787, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.9867467880249023, | |
| "rewards/margins": 0.8851081728935242, | |
| "rewards/rejected": -1.8718547821044922, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7531380753138075, | |
| "grad_norm": 24.593262334348687, | |
| "learning_rate": 8.729103716819111e-08, | |
| "logits/chosen": 0.7163609266281128, | |
| "logits/rejected": 1.9110931158065796, | |
| "logps/chosen": -414.01123046875, | |
| "logps/rejected": -465.447998046875, | |
| "loss": 0.515, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.1131541728973389, | |
| "rewards/margins": 0.928746223449707, | |
| "rewards/rejected": -2.041900157928467, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7740585774058577, | |
| "grad_norm": 24.621281841757344, | |
| "learning_rate": 7.387025063449081e-08, | |
| "logits/chosen": 1.1743080615997314, | |
| "logits/rejected": 1.871708631515503, | |
| "logps/chosen": -397.70159912109375, | |
| "logps/rejected": -431.24530029296875, | |
| "loss": 0.5066, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.2236586809158325, | |
| "rewards/margins": 0.7059282064437866, | |
| "rewards/rejected": -1.9295867681503296, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7949790794979079, | |
| "grad_norm": 26.620813748068915, | |
| "learning_rate": 6.138919252022435e-08, | |
| "logits/chosen": 1.1752651929855347, | |
| "logits/rejected": 1.894327163696289, | |
| "logps/chosen": -367.64251708984375, | |
| "logps/rejected": -479.59814453125, | |
| "loss": 0.505, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.281599760055542, | |
| "rewards/margins": 0.9278450012207031, | |
| "rewards/rejected": -2.209444522857666, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8158995815899581, | |
| "grad_norm": 24.66450917199271, | |
| "learning_rate": 4.991445467064689e-08, | |
| "logits/chosen": 0.6577990651130676, | |
| "logits/rejected": 1.2567031383514404, | |
| "logps/chosen": -414.7494201660156, | |
| "logps/rejected": -473.61859130859375, | |
| "loss": 0.4898, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.091284990310669, | |
| "rewards/margins": 0.8050578832626343, | |
| "rewards/rejected": -1.8963428735733032, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "grad_norm": 25.534386282276298, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": 0.7974405884742737, | |
| "logits/rejected": 1.8116296529769897, | |
| "logps/chosen": -378.2085876464844, | |
| "logps/rejected": -467.15380859375, | |
| "loss": 0.5014, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.076610803604126, | |
| "rewards/margins": 0.9254143834114075, | |
| "rewards/rejected": -2.0020251274108887, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "eval_logits/chosen": 0.5168243050575256, | |
| "eval_logits/rejected": 1.2876626253128052, | |
| "eval_logps/chosen": -357.22344970703125, | |
| "eval_logps/rejected": -452.98858642578125, | |
| "eval_loss": 0.5055292248725891, | |
| "eval_rewards/accuracies": 0.78125, | |
| "eval_rewards/chosen": -0.9459351897239685, | |
| "eval_rewards/margins": 0.957329511642456, | |
| "eval_rewards/rejected": -1.9032648801803589, | |
| "eval_runtime": 98.0802, | |
| "eval_samples_per_second": 20.391, | |
| "eval_steps_per_second": 0.326, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8577405857740585, | |
| "grad_norm": 26.77180405179113, | |
| "learning_rate": 3.022313472693447e-08, | |
| "logits/chosen": 0.6059648990631104, | |
| "logits/rejected": 1.6763797998428345, | |
| "logps/chosen": -400.90740966796875, | |
| "logps/rejected": -458.0049743652344, | |
| "loss": 0.5081, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -1.019736409187317, | |
| "rewards/margins": 0.8720607757568359, | |
| "rewards/rejected": -1.8917968273162842, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8786610878661087, | |
| "grad_norm": 27.43056085728676, | |
| "learning_rate": 2.2111614344599684e-08, | |
| "logits/chosen": 0.3962455987930298, | |
| "logits/rejected": 1.2819584608078003, | |
| "logps/chosen": -398.86578369140625, | |
| "logps/rejected": -457.84796142578125, | |
| "loss": 0.495, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -1.0178868770599365, | |
| "rewards/margins": 0.8161880373954773, | |
| "rewards/rejected": -1.8340747356414795, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.899581589958159, | |
| "grad_norm": 28.218529380366814, | |
| "learning_rate": 1.521597710086439e-08, | |
| "logits/chosen": 0.9617902636528015, | |
| "logits/rejected": 1.7479569911956787, | |
| "logps/chosen": -389.6866760253906, | |
| "logps/rejected": -455.4461975097656, | |
| "loss": 0.4738, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -1.1064532995224, | |
| "rewards/margins": 0.8703316450119019, | |
| "rewards/rejected": -1.9767850637435913, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9205020920502092, | |
| "grad_norm": 31.300447561945898, | |
| "learning_rate": 9.57301420397924e-09, | |
| "logits/chosen": 0.5907198786735535, | |
| "logits/rejected": 1.4784181118011475, | |
| "logps/chosen": -381.84088134765625, | |
| "logps/rejected": -447.17059326171875, | |
| "loss": 0.5009, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9751038551330566, | |
| "rewards/margins": 0.8055868148803711, | |
| "rewards/rejected": -1.7806905508041382, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9414225941422594, | |
| "grad_norm": 24.39105540939906, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": 0.6014220714569092, | |
| "logits/rejected": 1.414150357246399, | |
| "logps/chosen": -400.45184326171875, | |
| "logps/rejected": -494.45965576171875, | |
| "loss": 0.4975, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -1.0672663450241089, | |
| "rewards/margins": 0.806206226348877, | |
| "rewards/rejected": -1.8734725713729858, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9623430962343096, | |
| "grad_norm": 27.720726174975443, | |
| "learning_rate": 2.158697848236607e-09, | |
| "logits/chosen": 0.7587317824363708, | |
| "logits/rejected": 1.6242033243179321, | |
| "logps/chosen": -371.478759765625, | |
| "logps/rejected": -423.82684326171875, | |
| "loss": 0.5041, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.994895339012146, | |
| "rewards/margins": 0.8169215321540833, | |
| "rewards/rejected": -1.811816930770874, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9832635983263598, | |
| "grad_norm": 26.198352977438653, | |
| "learning_rate": 4.269029751107489e-10, | |
| "logits/chosen": 0.5299074053764343, | |
| "logits/rejected": 1.6109470129013062, | |
| "logps/chosen": -373.5680847167969, | |
| "logps/rejected": -463.97283935546875, | |
| "loss": 0.4888, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.9487025141716003, | |
| "rewards/margins": 0.8672690391540527, | |
| "rewards/rejected": -1.8159716129302979, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5398832953624646, | |
| "train_runtime": 10917.977, | |
| "train_samples_per_second": 5.599, | |
| "train_steps_per_second": 0.044 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |