{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9984301412872841, "eval_steps": 100, "global_step": 477, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0020931449502878076, "grad_norm": 0.8892470002174377, "learning_rate": 1.0416666666666667e-07, "logits/chosen": -2.5115435123443604, "logits/rejected": -2.418633460998535, "logps/chosen": -296.6790771484375, "logps/rejected": -343.70672607421875, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.020931449502878074, "grad_norm": 0.8503940105438232, "learning_rate": 1.0416666666666667e-06, "logits/chosen": -2.2245330810546875, "logits/rejected": -2.1629724502563477, "logps/chosen": -283.80975341796875, "logps/rejected": -255.17117309570312, "loss": 0.6926, "rewards/accuracies": 0.4861111044883728, "rewards/chosen": 0.008690375834703445, "rewards/margins": 0.0015612257411703467, "rewards/rejected": 0.0071291509084403515, "step": 10 }, { "epoch": 0.04186289900575615, "grad_norm": 0.8254271149635315, "learning_rate": 2.0833333333333334e-06, "logits/chosen": -2.335510015487671, "logits/rejected": -2.22532057762146, "logps/chosen": -286.57855224609375, "logps/rejected": -261.207763671875, "loss": 0.6874, "rewards/accuracies": 0.65625, "rewards/chosen": 0.03880625218153, "rewards/margins": 0.011353857815265656, "rewards/rejected": 0.027452394366264343, "step": 20 }, { "epoch": 0.06279434850863422, "grad_norm": 0.8504378795623779, "learning_rate": 3.125e-06, "logits/chosen": -2.2984628677368164, "logits/rejected": -2.184295892715454, "logps/chosen": -261.04315185546875, "logps/rejected": -227.68020629882812, "loss": 0.6805, "rewards/accuracies": 0.699999988079071, "rewards/chosen": 0.06454791873693466, "rewards/margins": 0.028396595269441605, "rewards/rejected": 0.03615132346749306, "step": 30 }, { "epoch": 0.0837257980115123, "grad_norm": 0.9619700908660889, "learning_rate": 4.166666666666667e-06, "logits/chosen": -2.1841602325439453, "logits/rejected": -2.1042842864990234, "logps/chosen": -244.4292449951172, "logps/rejected": -248.12057495117188, "loss": 0.664, "rewards/accuracies": 0.6875, "rewards/chosen": 0.03205280378460884, "rewards/margins": 0.06447319686412811, "rewards/rejected": -0.03242039680480957, "step": 40 }, { "epoch": 0.10465724751439037, "grad_norm": 1.198764443397522, "learning_rate": 4.999731868769027e-06, "logits/chosen": -2.2160000801086426, "logits/rejected": -2.09934663772583, "logps/chosen": -269.432861328125, "logps/rejected": -274.974365234375, "loss": 0.6454, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.08960571140050888, "rewards/margins": 0.10663382709026337, "rewards/rejected": -0.19623954594135284, "step": 50 }, { "epoch": 0.12558869701726844, "grad_norm": 1.695876121520996, "learning_rate": 4.9903533134293035e-06, "logits/chosen": -2.236078977584839, "logits/rejected": -2.1389355659484863, "logps/chosen": -262.9930725097656, "logps/rejected": -276.64984130859375, "loss": 0.6264, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.264637291431427, "rewards/margins": 0.2180159091949463, "rewards/rejected": -0.4826532006263733, "step": 60 }, { "epoch": 0.14652014652014653, "grad_norm": 2.115797519683838, "learning_rate": 4.967625656594782e-06, "logits/chosen": -2.1270174980163574, "logits/rejected": -2.063481569290161, "logps/chosen": -319.60888671875, "logps/rejected": -334.22705078125, "loss": 0.607, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.4378367066383362, "rewards/margins": 0.2902434766292572, "rewards/rejected": -0.728080153465271, "step": 70 }, { "epoch": 0.1674515960230246, "grad_norm": 2.7295830249786377, "learning_rate": 4.93167072587771e-06, "logits/chosen": -2.275411605834961, "logits/rejected": -2.0854580402374268, "logps/chosen": -358.58154296875, "logps/rejected": -320.63385009765625, "loss": 0.6069, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.49991363286972046, "rewards/margins": 0.31041082739830017, "rewards/rejected": -0.8103245496749878, "step": 80 }, { "epoch": 0.18838304552590268, "grad_norm": 2.6313765048980713, "learning_rate": 4.882681251368549e-06, "logits/chosen": -2.1408169269561768, "logits/rejected": -2.067002058029175, "logps/chosen": -289.2994079589844, "logps/rejected": -331.0093994140625, "loss": 0.576, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.5190187692642212, "rewards/margins": 0.3863602876663208, "rewards/rejected": -0.9053791165351868, "step": 90 }, { "epoch": 0.20931449502878074, "grad_norm": 3.2287309169769287, "learning_rate": 4.8209198325401815e-06, "logits/chosen": -2.0671892166137695, "logits/rejected": -1.958622694015503, "logps/chosen": -370.7698059082031, "logps/rejected": -413.80120849609375, "loss": 0.5646, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.933598518371582, "rewards/margins": 0.5866175889968872, "rewards/rejected": -1.5202162265777588, "step": 100 }, { "epoch": 0.20931449502878074, "eval_logits/chosen": -1.9878015518188477, "eval_logits/rejected": -1.9267165660858154, "eval_logps/chosen": -338.9564514160156, "eval_logps/rejected": -395.69635009765625, "eval_loss": 0.573878288269043, "eval_rewards/accuracies": 0.71875, "eval_rewards/chosen": -0.9252709150314331, "eval_rewards/margins": 0.5563679933547974, "eval_rewards/rejected": -1.4816389083862305, "eval_runtime": 110.707, "eval_samples_per_second": 18.066, "eval_steps_per_second": 0.289, "step": 100 }, { "epoch": 0.2302459445316588, "grad_norm": 3.1588852405548096, "learning_rate": 4.746717530629565e-06, "logits/chosen": -2.0595319271087646, "logits/rejected": -1.9567897319793701, "logps/chosen": -360.9635009765625, "logps/rejected": -388.3221130371094, "loss": 0.5789, "rewards/accuracies": 0.65625, "rewards/chosen": -0.860044002532959, "rewards/margins": 0.4338661730289459, "rewards/rejected": -1.2939101457595825, "step": 110 }, { "epoch": 0.25117739403453687, "grad_norm": 2.8980185985565186, "learning_rate": 4.660472094042121e-06, "logits/chosen": -1.8865253925323486, "logits/rejected": -1.8495429754257202, "logps/chosen": -350.013916015625, "logps/rejected": -394.4359130859375, "loss": 0.5592, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.9938249588012695, "rewards/margins": 0.5278288125991821, "rewards/rejected": -1.5216538906097412, "step": 120 }, { "epoch": 0.272108843537415, "grad_norm": 3.334541082382202, "learning_rate": 4.5626458262912745e-06, "logits/chosen": -1.8878917694091797, "logits/rejected": -1.8218252658843994, "logps/chosen": -396.32659912109375, "logps/rejected": -439.26513671875, "loss": 0.5373, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.3216043710708618, "rewards/margins": 0.5591233968734741, "rewards/rejected": -1.880727767944336, "step": 130 }, { "epoch": 0.29304029304029305, "grad_norm": 3.480584144592285, "learning_rate": 4.453763107901676e-06, "logits/chosen": -1.8928722143173218, "logits/rejected": -1.7833354473114014, "logps/chosen": -396.819091796875, "logps/rejected": -430.72674560546875, "loss": 0.549, "rewards/accuracies": 0.75, "rewards/chosen": -1.0090841054916382, "rewards/margins": 0.7491954565048218, "rewards/rejected": -1.75827956199646, "step": 140 }, { "epoch": 0.3139717425431711, "grad_norm": 3.2153046131134033, "learning_rate": 4.33440758555951e-06, "logits/chosen": -1.8725719451904297, "logits/rejected": -1.7752315998077393, "logps/chosen": -358.0649719238281, "logps/rejected": -379.94464111328125, "loss": 0.5479, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -0.9543037414550781, "rewards/margins": 0.5243933796882629, "rewards/rejected": -1.4786970615386963, "step": 150 }, { "epoch": 0.3349031920460492, "grad_norm": 3.572585344314575, "learning_rate": 4.205219043576955e-06, "logits/chosen": -1.8374134302139282, "logits/rejected": -1.6999791860580444, "logps/chosen": -347.92791748046875, "logps/rejected": -397.92236328125, "loss": 0.5582, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.8851444125175476, "rewards/margins": 0.7162662744522095, "rewards/rejected": -1.6014108657836914, "step": 160 }, { "epoch": 0.35583464154892724, "grad_norm": 4.003506183624268, "learning_rate": 4.066889974440757e-06, "logits/chosen": -1.4849274158477783, "logits/rejected": -1.420831561088562, "logps/chosen": -422.58807373046875, "logps/rejected": -454.4142150878906, "loss": 0.5563, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -1.7905734777450562, "rewards/margins": 0.5325244665145874, "rewards/rejected": -2.3230979442596436, "step": 170 }, { "epoch": 0.37676609105180536, "grad_norm": 3.9669156074523926, "learning_rate": 3.92016186682789e-06, "logits/chosen": -1.6295009851455688, "logits/rejected": -1.550588846206665, "logps/chosen": -370.55133056640625, "logps/rejected": -407.09735107421875, "loss": 0.5112, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.166177749633789, "rewards/margins": 0.6077830195426941, "rewards/rejected": -1.773960828781128, "step": 180 }, { "epoch": 0.3976975405546834, "grad_norm": 4.69191312789917, "learning_rate": 3.7658212309857576e-06, "logits/chosen": -1.5681777000427246, "logits/rejected": -1.5323166847229004, "logps/chosen": -348.3360900878906, "logps/rejected": -402.2555847167969, "loss": 0.5366, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.0483043193817139, "rewards/margins": 0.5953764915466309, "rewards/rejected": -1.6436808109283447, "step": 190 }, { "epoch": 0.4186289900575615, "grad_norm": 3.6216626167297363, "learning_rate": 3.604695382782159e-06, "logits/chosen": -1.514801263809204, "logits/rejected": -1.5004067420959473, "logps/chosen": -343.3397216796875, "logps/rejected": -422.3468322753906, "loss": 0.5524, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -1.099816918373108, "rewards/margins": 0.5951756834983826, "rewards/rejected": -1.6949926614761353, "step": 200 }, { "epoch": 0.4186289900575615, "eval_logits/chosen": -1.5800749063491821, "eval_logits/rejected": -1.5103871822357178, "eval_logps/chosen": -331.1845397949219, "eval_logps/rejected": -401.48101806640625, "eval_loss": 0.5318315029144287, "eval_rewards/accuracies": 0.76171875, "eval_rewards/chosen": -0.8475518226623535, "eval_rewards/margins": 0.6919333338737488, "eval_rewards/rejected": -1.5394850969314575, "eval_runtime": 110.7567, "eval_samples_per_second": 18.058, "eval_steps_per_second": 0.289, "step": 200 }, { "epoch": 0.43956043956043955, "grad_norm": 4.658122539520264, "learning_rate": 3.437648009023905e-06, "logits/chosen": -1.5987634658813477, "logits/rejected": -1.4259040355682373, "logps/chosen": -384.1799011230469, "logps/rejected": -397.27679443359375, "loss": 0.5351, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.9452459216117859, "rewards/margins": 0.6696128845214844, "rewards/rejected": -1.614858865737915, "step": 210 }, { "epoch": 0.4604918890633176, "grad_norm": 4.130067348480225, "learning_rate": 3.265574537815398e-06, "logits/chosen": -1.387083649635315, "logits/rejected": -1.3616085052490234, "logps/chosen": -346.65240478515625, "logps/rejected": -429.26007080078125, "loss": 0.5306, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.253117561340332, "rewards/margins": 0.7737840414047241, "rewards/rejected": -2.0269014835357666, "step": 220 }, { "epoch": 0.48142333856619574, "grad_norm": 3.835031032562256, "learning_rate": 3.089397338773569e-06, "logits/chosen": -1.5156991481781006, "logits/rejected": -1.389380693435669, "logps/chosen": -364.519287109375, "logps/rejected": -443.21649169921875, "loss": 0.5296, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.2139475345611572, "rewards/margins": 0.9700647592544556, "rewards/rejected": -2.1840121746063232, "step": 230 }, { "epoch": 0.5023547880690737, "grad_norm": 3.9292032718658447, "learning_rate": 2.9100607788275547e-06, "logits/chosen": -1.4860846996307373, "logits/rejected": -1.3620604276657104, "logps/chosen": -396.4831848144531, "logps/rejected": -455.0039978027344, "loss": 0.4992, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -1.2688108682632446, "rewards/margins": 0.8345935940742493, "rewards/rejected": -2.1034045219421387, "step": 240 }, { "epoch": 0.5232862375719518, "grad_norm": 3.9835152626037598, "learning_rate": 2.72852616010567e-06, "logits/chosen": -1.3683382272720337, "logits/rejected": -1.203263282775879, "logps/chosen": -439.71734619140625, "logps/rejected": -500.26995849609375, "loss": 0.5193, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.8103727102279663, "rewards/margins": 0.9737499356269836, "rewards/rejected": -2.7841227054595947, "step": 250 }, { "epoch": 0.54421768707483, "grad_norm": 4.053774356842041, "learning_rate": 2.5457665670441937e-06, "logits/chosen": -1.3506557941436768, "logits/rejected": -1.3202815055847168, "logps/chosen": -425.1206970214844, "logps/rejected": -482.37060546875, "loss": 0.5157, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.7851899862289429, "rewards/margins": 0.7175502181053162, "rewards/rejected": -2.5027403831481934, "step": 260 }, { "epoch": 0.565149136577708, "grad_norm": 3.4300835132598877, "learning_rate": 2.3627616503391813e-06, "logits/chosen": -1.3187822103500366, "logits/rejected": -1.2432146072387695, "logps/chosen": -435.5782775878906, "logps/rejected": -497.5511169433594, "loss": 0.5274, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -1.6222509145736694, "rewards/margins": 0.8602613210678101, "rewards/rejected": -2.4825127124786377, "step": 270 }, { "epoch": 0.5860805860805861, "grad_norm": 4.350644588470459, "learning_rate": 2.1804923757009885e-06, "logits/chosen": -1.173998236656189, "logits/rejected": -1.0935077667236328, "logps/chosen": -396.106201171875, "logps/rejected": -447.62359619140625, "loss": 0.5269, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -1.5709511041641235, "rewards/margins": 0.7292364835739136, "rewards/rejected": -2.300187587738037, "step": 280 }, { "epoch": 0.6070120355834642, "grad_norm": 3.7417469024658203, "learning_rate": 1.9999357655598894e-06, "logits/chosen": -1.29354727268219, "logits/rejected": -1.1772781610488892, "logps/chosen": -411.03814697265625, "logps/rejected": -484.0711975097656, "loss": 0.5207, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -1.7882314920425415, "rewards/margins": 0.787774920463562, "rewards/rejected": -2.5760064125061035, "step": 290 }, { "epoch": 0.6279434850863422, "grad_norm": 5.170950889587402, "learning_rate": 1.8220596619089576e-06, "logits/chosen": -1.2801265716552734, "logits/rejected": -1.1654409170150757, "logps/chosen": -476.93023681640625, "logps/rejected": -508.8740234375, "loss": 0.4977, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.7797939777374268, "rewards/margins": 0.7282718420028687, "rewards/rejected": -2.508065700531006, "step": 300 }, { "epoch": 0.6279434850863422, "eval_logits/chosen": -1.187766194343567, "eval_logits/rejected": -1.1155990362167358, "eval_logps/chosen": -434.6387939453125, "eval_logps/rejected": -531.358642578125, "eval_loss": 0.5100404620170593, "eval_rewards/accuracies": 0.77734375, "eval_rewards/chosen": -1.882094383239746, "eval_rewards/margins": 0.9561675786972046, "eval_rewards/rejected": -2.838261842727661, "eval_runtime": 110.6981, "eval_samples_per_second": 18.067, "eval_steps_per_second": 0.289, "step": 300 }, { "epoch": 0.6488749345892203, "grad_norm": 4.2551469802856445, "learning_rate": 1.647817538357072e-06, "logits/chosen": -1.2920277118682861, "logits/rejected": -1.1881086826324463, "logps/chosen": -491.37213134765625, "logps/rejected": -541.3331298828125, "loss": 0.4999, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -2.065791130065918, "rewards/margins": 1.0087900161743164, "rewards/rejected": -3.0745811462402344, "step": 310 }, { "epoch": 0.6698063840920984, "grad_norm": 5.16890811920166, "learning_rate": 1.4781433892011132e-06, "logits/chosen": -1.2393107414245605, "logits/rejected": -1.1026521921157837, "logps/chosen": -445.18408203125, "logps/rejected": -523.7984619140625, "loss": 0.4901, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -2.0319957733154297, "rewards/margins": 1.0875332355499268, "rewards/rejected": -3.1195292472839355, "step": 320 }, { "epoch": 0.6907378335949764, "grad_norm": 4.174256324768066, "learning_rate": 1.3139467229135999e-06, "logits/chosen": -1.3440932035446167, "logits/rejected": -1.3159363269805908, "logps/chosen": -418.71905517578125, "logps/rejected": -499.96612548828125, "loss": 0.5117, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.6654974222183228, "rewards/margins": 0.8459057807922363, "rewards/rejected": -2.5114030838012695, "step": 330 }, { "epoch": 0.7116692830978545, "grad_norm": 4.793274402618408, "learning_rate": 1.1561076868822756e-06, "logits/chosen": -1.2274696826934814, "logits/rejected": -1.1634485721588135, "logps/chosen": -472.03167724609375, "logps/rejected": -538.4515380859375, "loss": 0.5043, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -2.003690719604492, "rewards/margins": 0.8766867518424988, "rewards/rejected": -2.8803775310516357, "step": 340 }, { "epoch": 0.7326007326007326, "grad_norm": 4.15632963180542, "learning_rate": 1.0054723495346484e-06, "logits/chosen": -1.2941699028015137, "logits/rejected": -1.1815686225891113, "logps/chosen": -464.3369140625, "logps/rejected": -537.1565551757812, "loss": 0.467, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -2.2074246406555176, "rewards/margins": 0.9847939610481262, "rewards/rejected": -3.192218542098999, "step": 350 }, { "epoch": 0.7535321821036107, "grad_norm": 3.9081270694732666, "learning_rate": 8.628481651367876e-07, "logits/chosen": -1.2783113718032837, "logits/rejected": -1.1381876468658447, "logps/chosen": -522.1004028320312, "logps/rejected": -580.3312377929688, "loss": 0.5229, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -2.310837745666504, "rewards/margins": 1.0041909217834473, "rewards/rejected": -3.315028429031372, "step": 360 }, { "epoch": 0.7744636316064888, "grad_norm": 4.175972938537598, "learning_rate": 7.289996455765749e-07, "logits/chosen": -1.305639386177063, "logits/rejected": -1.1913295984268188, "logps/chosen": -472.50103759765625, "logps/rejected": -539.5281372070312, "loss": 0.515, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -2.2103240489959717, "rewards/margins": 0.9552016258239746, "rewards/rejected": -3.1655256748199463, "step": 370 }, { "epoch": 0.7953950811093669, "grad_norm": 4.144151210784912, "learning_rate": 6.046442623320145e-07, "logits/chosen": -1.1470415592193604, "logits/rejected": -1.1134142875671387, "logps/chosen": -463.9144592285156, "logps/rejected": -589.644287109375, "loss": 0.4996, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -2.329439401626587, "rewards/margins": 1.0926482677459717, "rewards/rejected": -3.4220874309539795, "step": 380 }, { "epoch": 0.8163265306122449, "grad_norm": 3.9941041469573975, "learning_rate": 4.904486005914027e-07, "logits/chosen": -1.3255889415740967, "logits/rejected": -1.2115201950073242, "logps/chosen": -541.8289794921875, "logps/rejected": -589.1190185546875, "loss": 0.4868, "rewards/accuracies": 0.71875, "rewards/chosen": -2.216815710067749, "rewards/margins": 0.8514670133590698, "rewards/rejected": -3.0682828426361084, "step": 390 }, { "epoch": 0.837257980115123, "grad_norm": 4.9559149742126465, "learning_rate": 3.8702478614051353e-07, "logits/chosen": -1.2322378158569336, "logits/rejected": -1.1362693309783936, "logps/chosen": -454.14727783203125, "logps/rejected": -534.5038452148438, "loss": 0.5096, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.1453731060028076, "rewards/margins": 0.9312618374824524, "rewards/rejected": -3.0766348838806152, "step": 400 }, { "epoch": 0.837257980115123, "eval_logits/chosen": -1.2569077014923096, "eval_logits/rejected": -1.1850428581237793, "eval_logps/chosen": -448.560302734375, "eval_logps/rejected": -549.236328125, "eval_loss": 0.5035256743431091, "eval_rewards/accuracies": 0.765625, "eval_rewards/chosen": -2.0213088989257812, "eval_rewards/margins": 0.9957298040390015, "eval_rewards/rejected": -3.017038583755493, "eval_runtime": 110.7002, "eval_samples_per_second": 18.067, "eval_steps_per_second": 0.289, "step": 400 }, { "epoch": 0.858189429618001, "grad_norm": 4.206878185272217, "learning_rate": 2.9492720416985004e-07, "logits/chosen": -1.3825483322143555, "logits/rejected": -1.2790623903274536, "logps/chosen": -492.615234375, "logps/rejected": -567.6419067382812, "loss": 0.5101, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -2.1133949756622314, "rewards/margins": 1.0342962741851807, "rewards/rejected": -3.147691249847412, "step": 410 }, { "epoch": 0.8791208791208791, "grad_norm": 4.6710124015808105, "learning_rate": 2.1464952759020857e-07, "logits/chosen": -1.2218900918960571, "logits/rejected": -1.1765925884246826, "logps/chosen": -447.31451416015625, "logps/rejected": -551.3411865234375, "loss": 0.5055, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -2.0722880363464355, "rewards/margins": 0.9839819073677063, "rewards/rejected": -3.056270122528076, "step": 420 }, { "epoch": 0.9000523286237572, "grad_norm": 3.861823081970215, "learning_rate": 1.4662207078575685e-07, "logits/chosen": -1.2683221101760864, "logits/rejected": -1.1284939050674438, "logps/chosen": -473.9938049316406, "logps/rejected": -547.7625122070312, "loss": 0.478, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -1.9979991912841797, "rewards/margins": 0.9907487630844116, "rewards/rejected": -2.988748073577881, "step": 430 }, { "epoch": 0.9209837781266352, "grad_norm": 4.48820686340332, "learning_rate": 9.120948298936422e-08, "logits/chosen": -1.4051973819732666, "logits/rejected": -1.2804924249649048, "logps/chosen": -447.3819274902344, "logps/rejected": -505.71221923828125, "loss": 0.4913, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -2.0068001747131348, "rewards/margins": 0.9424559473991394, "rewards/rejected": -2.94925594329834, "step": 440 }, { "epoch": 0.9419152276295133, "grad_norm": 4.315399169921875, "learning_rate": 4.870879364444109e-08, "logits/chosen": -1.1320288181304932, "logits/rejected": -1.0678789615631104, "logps/chosen": -466.1886291503906, "logps/rejected": -553.8982543945312, "loss": 0.4949, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -2.0924811363220215, "rewards/margins": 0.9840418696403503, "rewards/rejected": -3.0765230655670166, "step": 450 }, { "epoch": 0.9628466771323915, "grad_norm": 4.0728840827941895, "learning_rate": 1.93478202307823e-08, "logits/chosen": -1.290541410446167, "logits/rejected": -1.133880376815796, "logps/chosen": -460.96710205078125, "logps/rejected": -513.3797607421875, "loss": 0.5042, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -2.039780855178833, "rewards/margins": 0.937634289264679, "rewards/rejected": -2.977414608001709, "step": 460 }, { "epoch": 0.9837781266352695, "grad_norm": 3.986497640609741, "learning_rate": 3.283947088983663e-09, "logits/chosen": -1.2861273288726807, "logits/rejected": -1.1986534595489502, "logps/chosen": -484.498046875, "logps/rejected": -541.9881591796875, "loss": 0.4914, "rewards/accuracies": 0.6875, "rewards/chosen": -2.0398573875427246, "rewards/margins": 0.8533897399902344, "rewards/rejected": -2.893247127532959, "step": 470 }, { "epoch": 0.9984301412872841, "step": 477, "total_flos": 0.0, "train_loss": 0.5421361258444796, "train_runtime": 7660.497, "train_samples_per_second": 7.98, "train_steps_per_second": 0.062 } ], "logging_steps": 10, "max_steps": 477, "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": 4, "trial_name": null, "trial_params": null }