| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.986666666666667, | |
| "eval_steps": 500, | |
| "global_step": 315, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.047407407407407405, | |
| "grad_norm": 1152.0, | |
| "learning_rate": 7.8125e-06, | |
| "log_odds_chosen": -1.0631844997406006, | |
| "log_odds_ratio": -11.005602836608887, | |
| "logps/chosen": -22.707393646240234, | |
| "logps/rejected": -21.644441604614258, | |
| "loss": 467.0638, | |
| "nll_loss": 9.838674545288086, | |
| "rewards/accuracies": 0.4781250059604645, | |
| "rewards/chosen": -11.353696823120117, | |
| "rewards/margins": -0.5314784646034241, | |
| "rewards/rejected": -10.822220802307129, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09481481481481481, | |
| "grad_norm": 600.0, | |
| "learning_rate": 1.5625e-05, | |
| "log_odds_chosen": -3.1253325939178467, | |
| "log_odds_ratio": -11.045626640319824, | |
| "logps/chosen": -21.99468231201172, | |
| "logps/rejected": -18.869693756103516, | |
| "loss": 435.4043, | |
| "nll_loss": 8.75256633758545, | |
| "rewards/accuracies": 0.44062501192092896, | |
| "rewards/chosen": -10.99734115600586, | |
| "rewards/margins": -1.5624935626983643, | |
| "rewards/rejected": -9.434846878051758, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14222222222222222, | |
| "grad_norm": 464.0, | |
| "learning_rate": 2.34375e-05, | |
| "log_odds_chosen": -1.283685564994812, | |
| "log_odds_ratio": -11.318685531616211, | |
| "logps/chosen": -21.985923767089844, | |
| "logps/rejected": -20.700777053833008, | |
| "loss": 412.295, | |
| "nll_loss": 8.000757217407227, | |
| "rewards/accuracies": 0.515625, | |
| "rewards/chosen": -10.992961883544922, | |
| "rewards/margins": -0.6425749063491821, | |
| "rewards/rejected": -10.350388526916504, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.18962962962962962, | |
| "grad_norm": 3360.0, | |
| "learning_rate": 3.125e-05, | |
| "log_odds_chosen": -4.3174591064453125, | |
| "log_odds_ratio": -9.530021667480469, | |
| "logps/chosen": -17.268482208251953, | |
| "logps/rejected": -12.952882766723633, | |
| "loss": 332.5653, | |
| "nll_loss": 6.284921646118164, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -8.634241104125977, | |
| "rewards/margins": -2.157799482345581, | |
| "rewards/rejected": -6.476441383361816, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23703703703703705, | |
| "grad_norm": 241.0, | |
| "learning_rate": 3.90625e-05, | |
| "log_odds_chosen": -0.13406293094158173, | |
| "log_odds_ratio": -1.9662774801254272, | |
| "logps/chosen": -4.281245708465576, | |
| "logps/rejected": -4.128834247589111, | |
| "loss": 101.8704, | |
| "nll_loss": 2.7933926582336426, | |
| "rewards/accuracies": 0.5531250238418579, | |
| "rewards/chosen": -2.140622854232788, | |
| "rewards/margins": -0.07620568573474884, | |
| "rewards/rejected": -2.0644171237945557, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.28444444444444444, | |
| "grad_norm": 242.0, | |
| "learning_rate": 4.6875e-05, | |
| "log_odds_chosen": 0.10939224064350128, | |
| "log_odds_ratio": -0.8504419326782227, | |
| "logps/chosen": -1.855107307434082, | |
| "logps/rejected": -1.9498792886734009, | |
| "loss": 64.8284, | |
| "nll_loss": 2.0649564266204834, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": -0.927553653717041, | |
| "rewards/margins": 0.04738599434494972, | |
| "rewards/rejected": -0.9749396443367004, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.33185185185185184, | |
| "grad_norm": 282.0, | |
| "learning_rate": 4.998613757348784e-05, | |
| "log_odds_chosen": 0.3194878101348877, | |
| "log_odds_ratio": -0.7833901643753052, | |
| "logps/chosen": -1.7884143590927124, | |
| "logps/rejected": -2.071709156036377, | |
| "loss": 58.4951, | |
| "nll_loss": 1.9425550699234009, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.8942071795463562, | |
| "rewards/margins": 0.14164744317531586, | |
| "rewards/rejected": -1.0358545780181885, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.37925925925925924, | |
| "grad_norm": 214.0, | |
| "learning_rate": 4.990147841143462e-05, | |
| "log_odds_chosen": 0.5328894853591919, | |
| "log_odds_ratio": -0.6567676663398743, | |
| "logps/chosen": -1.6281368732452393, | |
| "logps/rejected": -2.09126877784729, | |
| "loss": 53.2976, | |
| "nll_loss": 1.8886241912841797, | |
| "rewards/accuracies": 0.6656249761581421, | |
| "rewards/chosen": -0.8140684366226196, | |
| "rewards/margins": 0.2315659075975418, | |
| "rewards/rejected": -1.045634388923645, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 354.0, | |
| "learning_rate": 4.97401218720448e-05, | |
| "log_odds_chosen": 1.3014285564422607, | |
| "log_odds_ratio": -0.5872281789779663, | |
| "logps/chosen": -1.680598497390747, | |
| "logps/rejected": -2.899627208709717, | |
| "loss": 46.0017, | |
| "nll_loss": 1.8771553039550781, | |
| "rewards/accuracies": 0.6968749761581421, | |
| "rewards/chosen": -0.8402992486953735, | |
| "rewards/margins": 0.6095143556594849, | |
| "rewards/rejected": -1.4498136043548584, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4740740740740741, | |
| "grad_norm": 226.0, | |
| "learning_rate": 4.9502564938797946e-05, | |
| "log_odds_chosen": 0.6567279696464539, | |
| "log_odds_ratio": -0.6511496305465698, | |
| "logps/chosen": -1.4998167753219604, | |
| "logps/rejected": -2.0936269760131836, | |
| "loss": 49.0928, | |
| "nll_loss": 1.7653518915176392, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.7499083876609802, | |
| "rewards/margins": 0.29690515995025635, | |
| "rewards/rejected": -1.0468134880065918, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5214814814814814, | |
| "grad_norm": 202.0, | |
| "learning_rate": 4.918953929490768e-05, | |
| "log_odds_chosen": 1.189807653427124, | |
| "log_odds_ratio": -0.617165744304657, | |
| "logps/chosen": -1.565733551979065, | |
| "logps/rejected": -2.669482707977295, | |
| "loss": 44.8712, | |
| "nll_loss": 1.786768913269043, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.7828667759895325, | |
| "rewards/margins": 0.5518745183944702, | |
| "rewards/rejected": -1.3347413539886475, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5688888888888889, | |
| "grad_norm": 278.0, | |
| "learning_rate": 4.88020090697132e-05, | |
| "log_odds_chosen": 1.5694297552108765, | |
| "log_odds_ratio": -0.5945903062820435, | |
| "logps/chosen": -1.5673797130584717, | |
| "logps/rejected": -3.0414257049560547, | |
| "loss": 44.6048, | |
| "nll_loss": 1.8120968341827393, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.7836898565292358, | |
| "rewards/margins": 0.7370229959487915, | |
| "rewards/rejected": -1.5207128524780273, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6162962962962963, | |
| "grad_norm": 160.0, | |
| "learning_rate": 4.834116786912897e-05, | |
| "log_odds_chosen": 1.5031869411468506, | |
| "log_odds_ratio": -0.663598895072937, | |
| "logps/chosen": -1.6163688898086548, | |
| "logps/rejected": -3.024693250656128, | |
| "loss": 46.3428, | |
| "nll_loss": 1.8134605884552002, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8081844449043274, | |
| "rewards/margins": 0.7041621208190918, | |
| "rewards/rejected": -1.512346625328064, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6637037037037037, | |
| "grad_norm": 298.0, | |
| "learning_rate": 4.7808435099299045e-05, | |
| "log_odds_chosen": 2.1792948246002197, | |
| "log_odds_ratio": -0.5366494059562683, | |
| "logps/chosen": -1.6390451192855835, | |
| "logps/rejected": -3.6916420459747314, | |
| "loss": 38.6914, | |
| "nll_loss": 1.7929494380950928, | |
| "rewards/accuracies": 0.753125011920929, | |
| "rewards/chosen": -0.8195225596427917, | |
| "rewards/margins": 1.0262985229492188, | |
| "rewards/rejected": -1.8458210229873657, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 300.0, | |
| "learning_rate": 4.720545159477922e-05, | |
| "log_odds_chosen": 1.6375625133514404, | |
| "log_odds_ratio": -0.6377304196357727, | |
| "logps/chosen": -1.5810494422912598, | |
| "logps/rejected": -3.109715223312378, | |
| "loss": 42.8578, | |
| "nll_loss": 1.7745224237442017, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.7905247211456299, | |
| "rewards/margins": 0.7643328905105591, | |
| "rewards/rejected": -1.554857611656189, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7585185185185185, | |
| "grad_norm": 208.0, | |
| "learning_rate": 4.653407456471222e-05, | |
| "log_odds_chosen": 1.206108808517456, | |
| "log_odds_ratio": -0.6781158447265625, | |
| "logps/chosen": -1.5513617992401123, | |
| "logps/rejected": -2.6515350341796875, | |
| "loss": 44.2435, | |
| "nll_loss": 1.7461738586425781, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.7756808996200562, | |
| "rewards/margins": 0.5500864386558533, | |
| "rewards/rejected": -1.3257675170898438, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8059259259259259, | |
| "grad_norm": 282.0, | |
| "learning_rate": 4.579637187256222e-05, | |
| "log_odds_chosen": 0.8860191106796265, | |
| "log_odds_ratio": -0.5883544683456421, | |
| "logps/chosen": -1.4292150735855103, | |
| "logps/rejected": -2.2102513313293457, | |
| "loss": 42.4359, | |
| "nll_loss": 1.656638741493225, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7146075367927551, | |
| "rewards/margins": 0.3905182480812073, | |
| "rewards/rejected": -1.1051256656646729, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 1400.0, | |
| "learning_rate": 4.499461566702685e-05, | |
| "log_odds_chosen": 2.8823413848876953, | |
| "log_odds_ratio": -0.8564308285713196, | |
| "logps/chosen": -2.120701789855957, | |
| "logps/rejected": -4.8557209968566895, | |
| "loss": 47.9855, | |
| "nll_loss": 1.9995596408843994, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.0603508949279785, | |
| "rewards/margins": 1.3675094842910767, | |
| "rewards/rejected": -2.4278604984283447, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.9007407407407407, | |
| "grad_norm": 366.0, | |
| "learning_rate": 4.413127538374411e-05, | |
| "log_odds_chosen": 1.6698837280273438, | |
| "log_odds_ratio": -0.7055229544639587, | |
| "logps/chosen": -1.6337862014770508, | |
| "logps/rejected": -3.1812427043914795, | |
| "loss": 41.8057, | |
| "nll_loss": 1.781404733657837, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": -0.8168931007385254, | |
| "rewards/margins": 0.7737283110618591, | |
| "rewards/rejected": -1.5906213521957397, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9481481481481482, | |
| "grad_norm": 364.0, | |
| "learning_rate": 4.320901013934887e-05, | |
| "log_odds_chosen": 1.8235197067260742, | |
| "log_odds_ratio": -0.5255730748176575, | |
| "logps/chosen": -1.462209939956665, | |
| "logps/rejected": -3.1428921222686768, | |
| "loss": 34.1059, | |
| "nll_loss": 1.6777896881103516, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7311049699783325, | |
| "rewards/margins": 0.840340793132782, | |
| "rewards/rejected": -1.5714460611343384, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9955555555555555, | |
| "grad_norm": 268.0, | |
| "learning_rate": 4.223066054130568e-05, | |
| "log_odds_chosen": 1.5804908275604248, | |
| "log_odds_ratio": -0.580092191696167, | |
| "logps/chosen": -1.4532772302627563, | |
| "logps/rejected": -2.918117046356201, | |
| "loss": 37.4817, | |
| "nll_loss": 1.6894712448120117, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": -0.7266386151313782, | |
| "rewards/margins": 0.7324199080467224, | |
| "rewards/rejected": -1.4590585231781006, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.0429629629629629, | |
| "grad_norm": 286.0, | |
| "learning_rate": 4.1199239938743797e-05, | |
| "log_odds_chosen": 2.7626733779907227, | |
| "log_odds_ratio": -0.424811989068985, | |
| "logps/chosen": -1.4315791130065918, | |
| "logps/rejected": -3.964046001434326, | |
| "loss": 24.2139, | |
| "nll_loss": 1.5723967552185059, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.7157895565032959, | |
| "rewards/margins": 1.2662335634231567, | |
| "rewards/rejected": -1.982023000717163, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0903703703703704, | |
| "grad_norm": 254.0, | |
| "learning_rate": 4.0117925141242174e-05, | |
| "log_odds_chosen": 3.299778461456299, | |
| "log_odds_ratio": -0.5258711576461792, | |
| "logps/chosen": -1.3833136558532715, | |
| "logps/rejected": -4.405320644378662, | |
| "loss": 25.4402, | |
| "nll_loss": 1.6238991022109985, | |
| "rewards/accuracies": 0.784375011920929, | |
| "rewards/chosen": -0.6916568279266357, | |
| "rewards/margins": 1.5110036134719849, | |
| "rewards/rejected": -2.202660322189331, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1377777777777778, | |
| "grad_norm": 528.0, | |
| "learning_rate": 3.899004663415084e-05, | |
| "log_odds_chosen": 2.667694568634033, | |
| "log_odds_ratio": -0.40380731225013733, | |
| "logps/chosen": -1.2623659372329712, | |
| "logps/rejected": -3.6577308177948, | |
| "loss": 23.1023, | |
| "nll_loss": 1.5065176486968994, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -0.6311829686164856, | |
| "rewards/margins": 1.1976823806762695, | |
| "rewards/rejected": -1.8288654088974, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1851851851851851, | |
| "grad_norm": 342.0, | |
| "learning_rate": 3.781907832058587e-05, | |
| "log_odds_chosen": 2.9293477535247803, | |
| "log_odds_ratio": -0.45116695761680603, | |
| "logps/chosen": -1.4864990711212158, | |
| "logps/rejected": -4.153409957885742, | |
| "loss": 25.0104, | |
| "nll_loss": 1.6183675527572632, | |
| "rewards/accuracies": 0.8031250238418579, | |
| "rewards/chosen": -0.7432495355606079, | |
| "rewards/margins": 1.3334555625915527, | |
| "rewards/rejected": -2.076704978942871, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.2325925925925927, | |
| "grad_norm": 268.0, | |
| "learning_rate": 3.660862682169282e-05, | |
| "log_odds_chosen": 3.8303322792053223, | |
| "log_odds_ratio": -0.37618154287338257, | |
| "logps/chosen": -1.4800456762313843, | |
| "logps/rejected": -5.02761697769165, | |
| "loss": 18.1555, | |
| "nll_loss": 1.5957032442092896, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -0.7400228381156921, | |
| "rewards/margins": 1.7737855911254883, | |
| "rewards/rejected": -2.513808488845825, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 458.0, | |
| "learning_rate": 3.5362420368134356e-05, | |
| "log_odds_chosen": 2.9902822971343994, | |
| "log_odds_ratio": -0.43725720047950745, | |
| "logps/chosen": -1.473851203918457, | |
| "logps/rejected": -4.198307991027832, | |
| "loss": 22.4211, | |
| "nll_loss": 1.5640933513641357, | |
| "rewards/accuracies": 0.8218749761581421, | |
| "rewards/chosen": -0.7369256019592285, | |
| "rewards/margins": 1.362228274345398, | |
| "rewards/rejected": -2.099153995513916, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.3274074074074074, | |
| "grad_norm": 201.0, | |
| "learning_rate": 3.408429731701635e-05, | |
| "log_odds_chosen": 2.826463460922241, | |
| "log_odds_ratio": -0.39234691858291626, | |
| "logps/chosen": -1.3628809452056885, | |
| "logps/rejected": -3.936453342437744, | |
| "loss": 21.0249, | |
| "nll_loss": 1.5445901155471802, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.6814404726028442, | |
| "rewards/margins": 1.2867863178253174, | |
| "rewards/rejected": -1.968226671218872, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.374814814814815, | |
| "grad_norm": 784.0, | |
| "learning_rate": 3.2778194329621104e-05, | |
| "log_odds_chosen": 3.861825942993164, | |
| "log_odds_ratio": -0.3445121943950653, | |
| "logps/chosen": -1.6509920358657837, | |
| "logps/rejected": -5.243237495422363, | |
| "loss": 18.864, | |
| "nll_loss": 1.7057406902313232, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -0.8254960179328918, | |
| "rewards/margins": 1.7961229085922241, | |
| "rewards/rejected": -2.6216187477111816, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.4222222222222223, | |
| "grad_norm": 316.0, | |
| "learning_rate": 3.144813424636031e-05, | |
| "log_odds_chosen": 4.123653888702393, | |
| "log_odds_ratio": -0.4155045449733734, | |
| "logps/chosen": -1.6837345361709595, | |
| "logps/rejected": -5.53621768951416, | |
| "loss": 18.678, | |
| "nll_loss": 1.6780169010162354, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.8418672680854797, | |
| "rewards/margins": 1.9262415170669556, | |
| "rewards/rejected": -2.76810884475708, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4696296296296296, | |
| "grad_norm": 466.0, | |
| "learning_rate": 3.0098213696293542e-05, | |
| "log_odds_chosen": 3.181971311569214, | |
| "log_odds_ratio": -0.4872185289859772, | |
| "logps/chosen": -1.4469890594482422, | |
| "logps/rejected": -4.361490726470947, | |
| "loss": 21.8554, | |
| "nll_loss": 1.5998506546020508, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -0.7234945297241211, | |
| "rewards/margins": 1.457250952720642, | |
| "rewards/rejected": -2.1807453632354736, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.5170370370370372, | |
| "grad_norm": 412.0, | |
| "learning_rate": 2.8732590479375165e-05, | |
| "log_odds_chosen": 4.313605308532715, | |
| "log_odds_ratio": -0.4044560492038727, | |
| "logps/chosen": -1.7111812829971313, | |
| "logps/rejected": -5.7456865310668945, | |
| "loss": 14.9335, | |
| "nll_loss": 1.6823270320892334, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -0.8555906414985657, | |
| "rewards/margins": 2.0172524452209473, | |
| "rewards/rejected": -2.8728432655334473, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5644444444444443, | |
| "grad_norm": 270.0, | |
| "learning_rate": 2.7355470760292956e-05, | |
| "log_odds_chosen": 3.449075698852539, | |
| "log_odds_ratio": -0.4021359980106354, | |
| "logps/chosen": -1.360383152961731, | |
| "logps/rejected": -4.50086784362793, | |
| "loss": 17.5439, | |
| "nll_loss": 1.5542360544204712, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -0.6801915764808655, | |
| "rewards/margins": 1.5702422857284546, | |
| "rewards/rejected": -2.250433921813965, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.6118518518518519, | |
| "grad_norm": 334.0, | |
| "learning_rate": 2.597109611334169e-05, | |
| "log_odds_chosen": 4.0982818603515625, | |
| "log_odds_ratio": -0.4611207842826843, | |
| "logps/chosen": -1.5969505310058594, | |
| "logps/rejected": -5.419460773468018, | |
| "loss": 18.2855, | |
| "nll_loss": 1.6615581512451172, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.7984752655029297, | |
| "rewards/margins": 1.911254644393921, | |
| "rewards/rejected": -2.709730386734009, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6592592592592592, | |
| "grad_norm": 424.0, | |
| "learning_rate": 2.458373045823404e-05, | |
| "log_odds_chosen": 3.9468979835510254, | |
| "log_odds_ratio": -0.36181649565696716, | |
| "logps/chosen": -1.4357500076293945, | |
| "logps/rejected": -5.089698791503906, | |
| "loss": 14.1876, | |
| "nll_loss": 1.5684595108032227, | |
| "rewards/accuracies": 0.8218749761581421, | |
| "rewards/chosen": -0.7178750038146973, | |
| "rewards/margins": 1.8269745111465454, | |
| "rewards/rejected": -2.544849395751953, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.7066666666666666, | |
| "grad_norm": 692.0, | |
| "learning_rate": 2.3197646927086697e-05, | |
| "log_odds_chosen": 3.9691085815429688, | |
| "log_odds_ratio": -0.35701385140419006, | |
| "logps/chosen": -1.550195336341858, | |
| "logps/rejected": -5.20673942565918, | |
| "loss": 14.9335, | |
| "nll_loss": 1.635414719581604, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -0.775097668170929, | |
| "rewards/margins": 1.8282718658447266, | |
| "rewards/rejected": -2.60336971282959, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.7540740740740741, | |
| "grad_norm": 304.0, | |
| "learning_rate": 2.1817114703032176e-05, | |
| "log_odds_chosen": 4.004181861877441, | |
| "log_odds_ratio": -0.4124390184879303, | |
| "logps/chosen": -1.5889790058135986, | |
| "logps/rejected": -5.336648464202881, | |
| "loss": 13.5162, | |
| "nll_loss": 1.580910563468933, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -0.7944895029067993, | |
| "rewards/margins": 1.8738348484039307, | |
| "rewards/rejected": -2.6683242321014404, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.8014814814814815, | |
| "grad_norm": 344.0, | |
| "learning_rate": 2.0446385870993467e-05, | |
| "log_odds_chosen": 3.7639663219451904, | |
| "log_odds_ratio": -0.3424893021583557, | |
| "logps/chosen": -1.3625359535217285, | |
| "logps/rejected": -4.8325395584106445, | |
| "loss": 10.6844, | |
| "nll_loss": 1.4781030416488647, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.6812679767608643, | |
| "rewards/margins": 1.735001564025879, | |
| "rewards/rejected": -2.4162697792053223, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.8488888888888888, | |
| "grad_norm": 632.0, | |
| "learning_rate": 1.9089682321121834e-05, | |
| "log_odds_chosen": 4.872841835021973, | |
| "log_odds_ratio": -0.3931121528148651, | |
| "logps/chosen": -1.8755972385406494, | |
| "logps/rejected": -6.473967552185059, | |
| "loss": 14.5199, | |
| "nll_loss": 1.8242985010147095, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": -0.9377986192703247, | |
| "rewards/margins": 2.299185037612915, | |
| "rewards/rejected": -3.2369837760925293, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8962962962962964, | |
| "grad_norm": 438.0, | |
| "learning_rate": 1.775118274523545e-05, | |
| "log_odds_chosen": 3.7074005603790283, | |
| "log_odds_ratio": -0.5268596410751343, | |
| "logps/chosen": -1.7662988901138306, | |
| "logps/rejected": -5.216193199157715, | |
| "loss": 20.4733, | |
| "nll_loss": 1.7086811065673828, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.8831494450569153, | |
| "rewards/margins": 1.724947214126587, | |
| "rewards/rejected": -2.6080965995788574, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9437037037037037, | |
| "grad_norm": 576.0, | |
| "learning_rate": 1.643500976631037e-05, | |
| "log_odds_chosen": 3.051617383956909, | |
| "log_odds_ratio": -0.38788479566574097, | |
| "logps/chosen": -1.2543895244598389, | |
| "logps/rejected": -4.006674289703369, | |
| "loss": 16.8108, | |
| "nll_loss": 1.5027841329574585, | |
| "rewards/accuracies": 0.8218749761581421, | |
| "rewards/chosen": -0.6271947622299194, | |
| "rewards/margins": 1.3761423826217651, | |
| "rewards/rejected": -2.0033371448516846, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.991111111111111, | |
| "grad_norm": 386.0, | |
| "learning_rate": 1.514521724066537e-05, | |
| "log_odds_chosen": 3.231032609939575, | |
| "log_odds_ratio": -0.43392324447631836, | |
| "logps/chosen": -1.4704937934875488, | |
| "logps/rejected": -4.4187726974487305, | |
| "loss": 17.4563, | |
| "nll_loss": 1.565903902053833, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.7352468967437744, | |
| "rewards/margins": 1.4741394519805908, | |
| "rewards/rejected": -2.2093863487243652, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.0385185185185186, | |
| "grad_norm": 278.0, | |
| "learning_rate": 1.3885777771950348e-05, | |
| "log_odds_chosen": 5.03135871887207, | |
| "log_odds_ratio": -0.3135664165019989, | |
| "logps/chosen": -1.407021164894104, | |
| "logps/rejected": -6.015364646911621, | |
| "loss": 0.5226, | |
| "nll_loss": 1.4759743213653564, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.703510582447052, | |
| "rewards/margins": 2.3041718006134033, | |
| "rewards/rejected": -3.0076823234558105, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.0859259259259257, | |
| "grad_norm": 378.0, | |
| "learning_rate": 1.2660570475395683e-05, | |
| "log_odds_chosen": 5.138548374176025, | |
| "log_odds_ratio": -0.25997596979141235, | |
| "logps/chosen": -1.37165367603302, | |
| "logps/rejected": -6.125999450683594, | |
| "loss": -0.8241, | |
| "nll_loss": 1.4749200344085693, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.68582683801651, | |
| "rewards/margins": 2.3771729469299316, | |
| "rewards/rejected": -3.062999725341797, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.1333333333333333, | |
| "grad_norm": 396.0, | |
| "learning_rate": 1.1473369030008974e-05, | |
| "log_odds_chosen": 6.11403751373291, | |
| "log_odds_ratio": -0.1846025437116623, | |
| "logps/chosen": -1.402966856956482, | |
| "logps/rejected": -7.111693382263184, | |
| "loss": -8.5304, | |
| "nll_loss": 1.5103604793548584, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.701483428478241, | |
| "rewards/margins": 2.854363441467285, | |
| "rewards/rejected": -3.555846691131592, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.180740740740741, | |
| "grad_norm": 368.0, | |
| "learning_rate": 1.0327830055518842e-05, | |
| "log_odds_chosen": 6.256018161773682, | |
| "log_odds_ratio": -0.18493634462356567, | |
| "logps/chosen": -1.3649917840957642, | |
| "logps/rejected": -7.209232330322266, | |
| "loss": -7.4943, | |
| "nll_loss": 1.4895808696746826, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.6824958920478821, | |
| "rewards/margins": 2.9221205711364746, | |
| "rewards/rejected": -3.604616165161133, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.228148148148148, | |
| "grad_norm": 328.0, | |
| "learning_rate": 9.227481849865235e-06, | |
| "log_odds_chosen": 6.152120113372803, | |
| "log_odds_ratio": -0.22947028279304504, | |
| "logps/chosen": -1.2814245223999023, | |
| "logps/rejected": -7.014758110046387, | |
| "loss": -6.2174, | |
| "nll_loss": 1.4657459259033203, | |
| "rewards/accuracies": 0.9156249761581421, | |
| "rewards/chosen": -0.6407122611999512, | |
| "rewards/margins": 2.866666793823242, | |
| "rewards/rejected": -3.5073790550231934, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.2755555555555556, | |
| "grad_norm": 440.0, | |
| "learning_rate": 8.175713521924978e-06, | |
| "log_odds_chosen": 6.160048484802246, | |
| "log_odds_ratio": -0.1592252552509308, | |
| "logps/chosen": -1.235910177230835, | |
| "logps/rejected": -6.972237586975098, | |
| "loss": -8.9307, | |
| "nll_loss": 1.388727068901062, | |
| "rewards/accuracies": 0.921875, | |
| "rewards/chosen": -0.6179550886154175, | |
| "rewards/margins": 2.8681633472442627, | |
| "rewards/rejected": -3.486118793487549, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.322962962962963, | |
| "grad_norm": 412.0, | |
| "learning_rate": 7.1757645529443665e-06, | |
| "log_odds_chosen": 6.512506008148193, | |
| "log_odds_ratio": -0.16299432516098022, | |
| "logps/chosen": -1.2822771072387695, | |
| "logps/rejected": -7.385122776031494, | |
| "loss": -10.2451, | |
| "nll_loss": 1.4544556140899658, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.6411385536193848, | |
| "rewards/margins": 3.051422595977783, | |
| "rewards/rejected": -3.692561388015747, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.3703703703703702, | |
| "grad_norm": 366.0, | |
| "learning_rate": 6.230714818829733e-06, | |
| "log_odds_chosen": 6.763820648193359, | |
| "log_odds_ratio": -0.15657763183116913, | |
| "logps/chosen": -1.2465496063232422, | |
| "logps/rejected": -7.550148010253906, | |
| "loss": -8.9088, | |
| "nll_loss": 1.4477818012237549, | |
| "rewards/accuracies": 0.9281250238418579, | |
| "rewards/chosen": -0.6232748031616211, | |
| "rewards/margins": 3.151798963546753, | |
| "rewards/rejected": -3.775074005126953, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.417777777777778, | |
| "grad_norm": 348.0, | |
| "learning_rate": 5.343475104027743e-06, | |
| "log_odds_chosen": 6.812985897064209, | |
| "log_odds_ratio": -0.13997027277946472, | |
| "logps/chosen": -1.285111665725708, | |
| "logps/rejected": -7.66811990737915, | |
| "loss": -11.2052, | |
| "nll_loss": 1.4328773021697998, | |
| "rewards/accuracies": 0.9468749761581421, | |
| "rewards/chosen": -0.642555832862854, | |
| "rewards/margins": 3.1915042400360107, | |
| "rewards/rejected": -3.834059953689575, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.4651851851851854, | |
| "grad_norm": 632.0, | |
| "learning_rate": 4.516778136213037e-06, | |
| "log_odds_chosen": 6.276130676269531, | |
| "log_odds_ratio": -0.18893949687480927, | |
| "logps/chosen": -1.3154428005218506, | |
| "logps/rejected": -7.166337013244629, | |
| "loss": -5.5648, | |
| "nll_loss": 1.4617598056793213, | |
| "rewards/accuracies": 0.9156249761581421, | |
| "rewards/chosen": -0.6577214002609253, | |
| "rewards/margins": 2.9254469871520996, | |
| "rewards/rejected": -3.5831685066223145, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.5125925925925925, | |
| "grad_norm": 644.0, | |
| "learning_rate": 3.7531701693965554e-06, | |
| "log_odds_chosen": 6.7295122146606445, | |
| "log_odds_ratio": -0.2116066962480545, | |
| "logps/chosen": -1.275740623474121, | |
| "logps/rejected": -7.586668968200684, | |
| "loss": -9.4219, | |
| "nll_loss": 1.4439576864242554, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.6378703117370605, | |
| "rewards/margins": 3.155463933944702, | |
| "rewards/rejected": -3.793334484100342, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 314.0, | |
| "learning_rate": 3.055003141378948e-06, | |
| "log_odds_chosen": 6.7960004806518555, | |
| "log_odds_ratio": -0.18422751128673553, | |
| "logps/chosen": -1.1799513101577759, | |
| "logps/rejected": -7.515390872955322, | |
| "loss": -11.538, | |
| "nll_loss": 1.3993252515792847, | |
| "rewards/accuracies": 0.934374988079071, | |
| "rewards/chosen": -0.5899756550788879, | |
| "rewards/margins": 3.167720079421997, | |
| "rewards/rejected": -3.757695436477661, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.6074074074074076, | |
| "grad_norm": 398.0, | |
| "learning_rate": 2.424427429704365e-06, | |
| "log_odds_chosen": 6.602663516998291, | |
| "log_odds_ratio": -0.18241380155086517, | |
| "logps/chosen": -1.3822739124298096, | |
| "logps/rejected": -7.560347557067871, | |
| "loss": -8.7863, | |
| "nll_loss": 1.474333643913269, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.6911369562149048, | |
| "rewards/margins": 3.0890371799468994, | |
| "rewards/rejected": -3.7801737785339355, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.6548148148148147, | |
| "grad_norm": 700.0, | |
| "learning_rate": 1.8633852284264508e-06, | |
| "log_odds_chosen": 6.423117637634277, | |
| "log_odds_ratio": -0.18244215846061707, | |
| "logps/chosen": -1.2519899606704712, | |
| "logps/rejected": -7.237380027770996, | |
| "loss": -7.6169, | |
| "nll_loss": 1.4630454778671265, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.6259949803352356, | |
| "rewards/margins": 2.9926953315734863, | |
| "rewards/rejected": -3.618690013885498, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.7022222222222223, | |
| "grad_norm": 376.0, | |
| "learning_rate": 1.3736045660864034e-06, | |
| "log_odds_chosen": 6.383296966552734, | |
| "log_odds_ratio": -0.22375822067260742, | |
| "logps/chosen": -1.2126858234405518, | |
| "logps/rejected": -7.17873477935791, | |
| "loss": -6.2815, | |
| "nll_loss": 1.423060417175293, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.6063429117202759, | |
| "rewards/margins": 2.9830245971679688, | |
| "rewards/rejected": -3.589367389678955, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.74962962962963, | |
| "grad_norm": 288.0, | |
| "learning_rate": 9.565939833279192e-07, | |
| "log_odds_chosen": 5.9918012619018555, | |
| "log_odds_ratio": -0.24828758835792542, | |
| "logps/chosen": -1.4515832662582397, | |
| "logps/rejected": -7.042089939117432, | |
| "loss": -2.5592, | |
| "nll_loss": 1.5330219268798828, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.7257916331291199, | |
| "rewards/margins": 2.7952535152435303, | |
| "rewards/rejected": -3.521044969558716, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.797037037037037, | |
| "grad_norm": 452.0, | |
| "learning_rate": 6.136378865420872e-07, | |
| "log_odds_chosen": 6.23110294342041, | |
| "log_odds_ratio": -0.2531821131706238, | |
| "logps/chosen": -1.3049705028533936, | |
| "logps/rejected": -7.137003421783447, | |
| "loss": -4.6802, | |
| "nll_loss": 1.456378698348999, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.6524852514266968, | |
| "rewards/margins": 2.9160165786743164, | |
| "rewards/rejected": -3.5685017108917236, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.8444444444444446, | |
| "grad_norm": 354.0, | |
| "learning_rate": 3.45792591853214e-07, | |
| "log_odds_chosen": 6.851447105407715, | |
| "log_odds_ratio": -0.21110931038856506, | |
| "logps/chosen": -1.1985723972320557, | |
| "logps/rejected": -7.616302490234375, | |
| "loss": -9.172, | |
| "nll_loss": 1.4328322410583496, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.5992861986160278, | |
| "rewards/margins": 3.20886492729187, | |
| "rewards/rejected": -3.8081512451171875, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.891851851851852, | |
| "grad_norm": 468.0, | |
| "learning_rate": 1.538830716302092e-07, | |
| "log_odds_chosen": 6.681235313415527, | |
| "log_odds_ratio": -0.16963359713554382, | |
| "logps/chosen": -1.2472527027130127, | |
| "logps/rejected": -7.471646308898926, | |
| "loss": -10.8492, | |
| "nll_loss": 1.403349757194519, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.6236263513565063, | |
| "rewards/margins": 3.112197160720825, | |
| "rewards/rejected": -3.735823154449463, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.9392592592592592, | |
| "grad_norm": 384.0, | |
| "learning_rate": 3.8500413544415025e-08, | |
| "log_odds_chosen": 6.875144958496094, | |
| "log_odds_ratio": -0.12083999067544937, | |
| "logps/chosen": -1.2005863189697266, | |
| "logps/rejected": -7.624534606933594, | |
| "loss": -11.3909, | |
| "nll_loss": 1.4330816268920898, | |
| "rewards/accuracies": 0.953125, | |
| "rewards/chosen": -0.6002931594848633, | |
| "rewards/margins": 3.2119739055633545, | |
| "rewards/rejected": -3.812267303466797, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "grad_norm": 354.0, | |
| "learning_rate": 0.0, | |
| "log_odds_chosen": 6.552358150482178, | |
| "log_odds_ratio": -0.21020980179309845, | |
| "logps/chosen": -1.2049369812011719, | |
| "logps/rejected": -7.295576572418213, | |
| "loss": -9.7319, | |
| "nll_loss": 1.3733375072479248, | |
| "rewards/accuracies": 0.903124988079071, | |
| "rewards/chosen": -0.6024684906005859, | |
| "rewards/margins": 3.0453197956085205, | |
| "rewards/rejected": -3.6477882862091064, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "step": 315, | |
| "total_flos": 0.0, | |
| "train_loss": 43.15912766380916, | |
| "train_runtime": 7320.6498, | |
| "train_samples_per_second": 2.766, | |
| "train_steps_per_second": 0.043 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 315, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100000, | |
| "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": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |