{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 100.0, "eval_steps": 500, "global_step": 21100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.23696682464454977, "grad_norm": 8.05551815032959, "learning_rate": 4.9883886255924175e-05, "loss": 8.334, "step": 50 }, { "epoch": 0.47393364928909953, "grad_norm": 7.429960250854492, "learning_rate": 4.9765402843601896e-05, "loss": 6.6018, "step": 100 }, { "epoch": 0.7109004739336493, "grad_norm": 6.704407691955566, "learning_rate": 4.9646919431279625e-05, "loss": 6.3637, "step": 150 }, { "epoch": 0.9478672985781991, "grad_norm": 6.104093551635742, "learning_rate": 4.9528436018957346e-05, "loss": 6.0119, "step": 200 }, { "epoch": 1.1848341232227488, "grad_norm": 8.27985954284668, "learning_rate": 4.9409952606635075e-05, "loss": 5.4826, "step": 250 }, { "epoch": 1.4218009478672986, "grad_norm": 8.599013328552246, "learning_rate": 4.92914691943128e-05, "loss": 5.2738, "step": 300 }, { "epoch": 1.6587677725118484, "grad_norm": 6.247586250305176, "learning_rate": 4.9172985781990525e-05, "loss": 5.2822, "step": 350 }, { "epoch": 1.8957345971563981, "grad_norm": 6.061827182769775, "learning_rate": 4.9054502369668246e-05, "loss": 5.1185, "step": 400 }, { "epoch": 2.132701421800948, "grad_norm": 5.755828857421875, "learning_rate": 4.8936018957345974e-05, "loss": 4.9572, "step": 450 }, { "epoch": 2.3696682464454977, "grad_norm": 5.827480792999268, "learning_rate": 4.88175355450237e-05, "loss": 4.7363, "step": 500 }, { "epoch": 2.6066350710900474, "grad_norm": 6.81092643737793, "learning_rate": 4.8699052132701424e-05, "loss": 4.7752, "step": 550 }, { "epoch": 2.843601895734597, "grad_norm": 6.485591888427734, "learning_rate": 4.858056872037915e-05, "loss": 4.6314, "step": 600 }, { "epoch": 3.080568720379147, "grad_norm": 6.661850929260254, "learning_rate": 4.8462085308056874e-05, "loss": 4.5587, "step": 650 }, { "epoch": 3.3175355450236967, "grad_norm": 5.563061237335205, "learning_rate": 4.8343601895734596e-05, "loss": 4.42, "step": 700 }, { "epoch": 3.5545023696682465, "grad_norm": 5.746306896209717, "learning_rate": 4.8225118483412324e-05, "loss": 4.1655, "step": 750 }, { "epoch": 3.7914691943127963, "grad_norm": 6.550754070281982, "learning_rate": 4.810663507109005e-05, "loss": 4.3413, "step": 800 }, { "epoch": 4.028436018957346, "grad_norm": 6.368891716003418, "learning_rate": 4.7988151658767774e-05, "loss": 4.4649, "step": 850 }, { "epoch": 4.265402843601896, "grad_norm": 6.2703351974487305, "learning_rate": 4.7869668246445496e-05, "loss": 3.9531, "step": 900 }, { "epoch": 4.502369668246446, "grad_norm": 6.344620704650879, "learning_rate": 4.7751184834123224e-05, "loss": 3.987, "step": 950 }, { "epoch": 4.739336492890995, "grad_norm": 5.440389633178711, "learning_rate": 4.763270142180095e-05, "loss": 3.9812, "step": 1000 }, { "epoch": 4.976303317535545, "grad_norm": 5.8826985359191895, "learning_rate": 4.7514218009478674e-05, "loss": 4.0952, "step": 1050 }, { "epoch": 5.213270142180095, "grad_norm": 5.59810209274292, "learning_rate": 4.73957345971564e-05, "loss": 3.7354, "step": 1100 }, { "epoch": 5.450236966824645, "grad_norm": 6.954726219177246, "learning_rate": 4.7277251184834124e-05, "loss": 3.6271, "step": 1150 }, { "epoch": 5.687203791469194, "grad_norm": 6.965663909912109, "learning_rate": 4.715876777251185e-05, "loss": 3.7573, "step": 1200 }, { "epoch": 5.924170616113744, "grad_norm": 6.535094261169434, "learning_rate": 4.7040284360189574e-05, "loss": 3.893, "step": 1250 }, { "epoch": 6.161137440758294, "grad_norm": 6.765582084655762, "learning_rate": 4.69218009478673e-05, "loss": 3.4036, "step": 1300 }, { "epoch": 6.398104265402844, "grad_norm": 6.829212665557861, "learning_rate": 4.680331753554503e-05, "loss": 3.5051, "step": 1350 }, { "epoch": 6.6350710900473935, "grad_norm": 7.073272228240967, "learning_rate": 4.668483412322275e-05, "loss": 3.4106, "step": 1400 }, { "epoch": 6.872037914691943, "grad_norm": 6.111814975738525, "learning_rate": 4.6566350710900473e-05, "loss": 3.5722, "step": 1450 }, { "epoch": 7.109004739336493, "grad_norm": 7.566643714904785, "learning_rate": 4.64478672985782e-05, "loss": 3.3311, "step": 1500 }, { "epoch": 7.345971563981043, "grad_norm": 7.837409973144531, "learning_rate": 4.632938388625593e-05, "loss": 3.0516, "step": 1550 }, { "epoch": 7.5829383886255926, "grad_norm": 6.301836013793945, "learning_rate": 4.621090047393365e-05, "loss": 3.0299, "step": 1600 }, { "epoch": 7.819905213270142, "grad_norm": 7.4018402099609375, "learning_rate": 4.609241706161137e-05, "loss": 3.4025, "step": 1650 }, { "epoch": 8.056872037914692, "grad_norm": 7.050328254699707, "learning_rate": 4.59739336492891e-05, "loss": 3.1522, "step": 1700 }, { "epoch": 8.293838862559241, "grad_norm": 6.999985218048096, "learning_rate": 4.585545023696683e-05, "loss": 2.9214, "step": 1750 }, { "epoch": 8.530805687203792, "grad_norm": 7.938794136047363, "learning_rate": 4.573696682464455e-05, "loss": 2.9386, "step": 1800 }, { "epoch": 8.76777251184834, "grad_norm": 8.065939903259277, "learning_rate": 4.561848341232228e-05, "loss": 2.8978, "step": 1850 }, { "epoch": 9.004739336492891, "grad_norm": 6.599789619445801, "learning_rate": 4.55e-05, "loss": 2.9907, "step": 1900 }, { "epoch": 9.24170616113744, "grad_norm": 7.754159927368164, "learning_rate": 4.538151658767772e-05, "loss": 2.6728, "step": 1950 }, { "epoch": 9.47867298578199, "grad_norm": 5.911088943481445, "learning_rate": 4.526303317535545e-05, "loss": 2.6302, "step": 2000 }, { "epoch": 9.71563981042654, "grad_norm": 8.413790702819824, "learning_rate": 4.514454976303318e-05, "loss": 2.9362, "step": 2050 }, { "epoch": 9.95260663507109, "grad_norm": 6.573764801025391, "learning_rate": 4.502606635071091e-05, "loss": 2.5738, "step": 2100 }, { "epoch": 10.189573459715639, "grad_norm": 7.7850141525268555, "learning_rate": 4.490758293838862e-05, "loss": 2.5018, "step": 2150 }, { "epoch": 10.42654028436019, "grad_norm": 7.518754959106445, "learning_rate": 4.478909952606635e-05, "loss": 2.5022, "step": 2200 }, { "epoch": 10.663507109004739, "grad_norm": 6.728266716003418, "learning_rate": 4.467061611374408e-05, "loss": 2.363, "step": 2250 }, { "epoch": 10.90047393364929, "grad_norm": 8.154065132141113, "learning_rate": 4.45521327014218e-05, "loss": 2.5758, "step": 2300 }, { "epoch": 11.137440758293838, "grad_norm": 7.86083459854126, "learning_rate": 4.443364928909953e-05, "loss": 2.3456, "step": 2350 }, { "epoch": 11.374407582938389, "grad_norm": 7.525182723999023, "learning_rate": 4.431516587677725e-05, "loss": 2.1671, "step": 2400 }, { "epoch": 11.611374407582938, "grad_norm": 8.167869567871094, "learning_rate": 4.419668246445498e-05, "loss": 2.2758, "step": 2450 }, { "epoch": 11.848341232227488, "grad_norm": 9.416289329528809, "learning_rate": 4.40781990521327e-05, "loss": 2.353, "step": 2500 }, { "epoch": 12.085308056872037, "grad_norm": 8.906156539916992, "learning_rate": 4.395971563981043e-05, "loss": 2.1967, "step": 2550 }, { "epoch": 12.322274881516588, "grad_norm": 8.13156795501709, "learning_rate": 4.384123222748816e-05, "loss": 2.05, "step": 2600 }, { "epoch": 12.559241706161137, "grad_norm": 7.436127185821533, "learning_rate": 4.372274881516588e-05, "loss": 1.9814, "step": 2650 }, { "epoch": 12.796208530805687, "grad_norm": 7.9628214836120605, "learning_rate": 4.36042654028436e-05, "loss": 2.1599, "step": 2700 }, { "epoch": 13.033175355450236, "grad_norm": 6.870182991027832, "learning_rate": 4.348578199052133e-05, "loss": 2.1117, "step": 2750 }, { "epoch": 13.270142180094787, "grad_norm": 7.705275535583496, "learning_rate": 4.336729857819906e-05, "loss": 1.7504, "step": 2800 }, { "epoch": 13.507109004739336, "grad_norm": 7.287713050842285, "learning_rate": 4.324881516587678e-05, "loss": 1.8817, "step": 2850 }, { "epoch": 13.744075829383887, "grad_norm": 8.195038795471191, "learning_rate": 4.31303317535545e-05, "loss": 1.9218, "step": 2900 }, { "epoch": 13.981042654028435, "grad_norm": 8.265512466430664, "learning_rate": 4.301184834123223e-05, "loss": 1.9988, "step": 2950 }, { "epoch": 14.218009478672986, "grad_norm": 8.913329124450684, "learning_rate": 4.289336492890996e-05, "loss": 1.6557, "step": 3000 }, { "epoch": 14.454976303317535, "grad_norm": 8.75406551361084, "learning_rate": 4.277488151658768e-05, "loss": 1.6558, "step": 3050 }, { "epoch": 14.691943127962086, "grad_norm": 9.669599533081055, "learning_rate": 4.265639810426541e-05, "loss": 1.7633, "step": 3100 }, { "epoch": 14.928909952606634, "grad_norm": 7.665035247802734, "learning_rate": 4.253791469194313e-05, "loss": 1.7764, "step": 3150 }, { "epoch": 15.165876777251185, "grad_norm": 8.728958129882812, "learning_rate": 4.241943127962085e-05, "loss": 1.5575, "step": 3200 }, { "epoch": 15.402843601895734, "grad_norm": 10.094686508178711, "learning_rate": 4.230094786729858e-05, "loss": 1.5309, "step": 3250 }, { "epoch": 15.639810426540285, "grad_norm": 8.494065284729004, "learning_rate": 4.218246445497631e-05, "loss": 1.6156, "step": 3300 }, { "epoch": 15.876777251184834, "grad_norm": 9.575804710388184, "learning_rate": 4.2063981042654035e-05, "loss": 1.5968, "step": 3350 }, { "epoch": 16.113744075829384, "grad_norm": 6.980827808380127, "learning_rate": 4.194549763033176e-05, "loss": 1.5025, "step": 3400 }, { "epoch": 16.350710900473935, "grad_norm": 11.093771934509277, "learning_rate": 4.182701421800948e-05, "loss": 1.402, "step": 3450 }, { "epoch": 16.587677725118482, "grad_norm": 9.239128112792969, "learning_rate": 4.170853080568721e-05, "loss": 1.4107, "step": 3500 }, { "epoch": 16.824644549763033, "grad_norm": 9.813140869140625, "learning_rate": 4.159004739336493e-05, "loss": 1.4569, "step": 3550 }, { "epoch": 17.061611374407583, "grad_norm": 6.551173686981201, "learning_rate": 4.1471563981042657e-05, "loss": 1.4321, "step": 3600 }, { "epoch": 17.298578199052134, "grad_norm": 7.340089321136475, "learning_rate": 4.135308056872038e-05, "loss": 1.2886, "step": 3650 }, { "epoch": 17.53554502369668, "grad_norm": 6.799836158752441, "learning_rate": 4.1234597156398106e-05, "loss": 1.2473, "step": 3700 }, { "epoch": 17.77251184834123, "grad_norm": 9.836413383483887, "learning_rate": 4.111611374407583e-05, "loss": 1.3547, "step": 3750 }, { "epoch": 18.009478672985782, "grad_norm": 6.786983013153076, "learning_rate": 4.0997630331753556e-05, "loss": 1.3666, "step": 3800 }, { "epoch": 18.246445497630333, "grad_norm": 6.9445109367370605, "learning_rate": 4.0879146919431285e-05, "loss": 1.163, "step": 3850 }, { "epoch": 18.48341232227488, "grad_norm": 10.60429859161377, "learning_rate": 4.0760663507109006e-05, "loss": 1.1638, "step": 3900 }, { "epoch": 18.72037914691943, "grad_norm": 6.641298770904541, "learning_rate": 4.064218009478673e-05, "loss": 1.2068, "step": 3950 }, { "epoch": 18.95734597156398, "grad_norm": 7.5319695472717285, "learning_rate": 4.0523696682464456e-05, "loss": 1.2355, "step": 4000 }, { "epoch": 19.194312796208532, "grad_norm": 6.43670654296875, "learning_rate": 4.0405213270142185e-05, "loss": 1.0902, "step": 4050 }, { "epoch": 19.43127962085308, "grad_norm": 7.7062835693359375, "learning_rate": 4.0286729857819906e-05, "loss": 1.0694, "step": 4100 }, { "epoch": 19.66824644549763, "grad_norm": 9.379792213439941, "learning_rate": 4.0168246445497634e-05, "loss": 1.0934, "step": 4150 }, { "epoch": 19.90521327014218, "grad_norm": 10.263785362243652, "learning_rate": 4.0049763033175356e-05, "loss": 1.1875, "step": 4200 }, { "epoch": 20.14218009478673, "grad_norm": 7.680784225463867, "learning_rate": 3.9931279620853084e-05, "loss": 1.0024, "step": 4250 }, { "epoch": 20.379146919431278, "grad_norm": 5.845160007476807, "learning_rate": 3.9812796208530806e-05, "loss": 1.0054, "step": 4300 }, { "epoch": 20.61611374407583, "grad_norm": 7.681657791137695, "learning_rate": 3.9694312796208534e-05, "loss": 1.0188, "step": 4350 }, { "epoch": 20.85308056872038, "grad_norm": 6.548282146453857, "learning_rate": 3.957582938388626e-05, "loss": 1.0767, "step": 4400 }, { "epoch": 21.09004739336493, "grad_norm": 8.785778045654297, "learning_rate": 3.945734597156398e-05, "loss": 0.9832, "step": 4450 }, { "epoch": 21.327014218009477, "grad_norm": 7.1846771240234375, "learning_rate": 3.9338862559241706e-05, "loss": 0.9377, "step": 4500 }, { "epoch": 21.563981042654028, "grad_norm": 8.198999404907227, "learning_rate": 3.9220379146919434e-05, "loss": 0.9162, "step": 4550 }, { "epoch": 21.80094786729858, "grad_norm": 7.712432861328125, "learning_rate": 3.910189573459716e-05, "loss": 0.9446, "step": 4600 }, { "epoch": 22.03791469194313, "grad_norm": 8.10291576385498, "learning_rate": 3.8983412322274884e-05, "loss": 0.9245, "step": 4650 }, { "epoch": 22.274881516587676, "grad_norm": 6.943071365356445, "learning_rate": 3.8864928909952606e-05, "loss": 0.8313, "step": 4700 }, { "epoch": 22.511848341232227, "grad_norm": 7.750855445861816, "learning_rate": 3.8746445497630334e-05, "loss": 0.8521, "step": 4750 }, { "epoch": 22.748815165876778, "grad_norm": 6.502680778503418, "learning_rate": 3.862796208530806e-05, "loss": 0.8819, "step": 4800 }, { "epoch": 22.98578199052133, "grad_norm": 9.593765258789062, "learning_rate": 3.8509478672985784e-05, "loss": 0.8965, "step": 4850 }, { "epoch": 23.222748815165875, "grad_norm": 6.064005374908447, "learning_rate": 3.839099526066351e-05, "loss": 0.7502, "step": 4900 }, { "epoch": 23.459715639810426, "grad_norm": 8.366140365600586, "learning_rate": 3.8272511848341234e-05, "loss": 0.7663, "step": 4950 }, { "epoch": 23.696682464454977, "grad_norm": 7.638758659362793, "learning_rate": 3.8154028436018955e-05, "loss": 0.8177, "step": 5000 }, { "epoch": 23.933649289099527, "grad_norm": 9.369016647338867, "learning_rate": 3.8035545023696684e-05, "loss": 0.8435, "step": 5050 }, { "epoch": 24.170616113744074, "grad_norm": 7.946335315704346, "learning_rate": 3.791706161137441e-05, "loss": 0.7265, "step": 5100 }, { "epoch": 24.407582938388625, "grad_norm": 6.578442573547363, "learning_rate": 3.779857819905214e-05, "loss": 0.7346, "step": 5150 }, { "epoch": 24.644549763033176, "grad_norm": 8.496620178222656, "learning_rate": 3.7680094786729855e-05, "loss": 0.7574, "step": 5200 }, { "epoch": 24.881516587677726, "grad_norm": 7.400977611541748, "learning_rate": 3.7561611374407583e-05, "loss": 0.7754, "step": 5250 }, { "epoch": 25.118483412322274, "grad_norm": 8.459925651550293, "learning_rate": 3.744312796208531e-05, "loss": 0.6973, "step": 5300 }, { "epoch": 25.355450236966824, "grad_norm": 8.262625694274902, "learning_rate": 3.732464454976303e-05, "loss": 0.65, "step": 5350 }, { "epoch": 25.592417061611375, "grad_norm": 9.051399230957031, "learning_rate": 3.720616113744076e-05, "loss": 0.6762, "step": 5400 }, { "epoch": 25.829383886255926, "grad_norm": 8.525854110717773, "learning_rate": 3.708767772511848e-05, "loss": 0.6978, "step": 5450 }, { "epoch": 26.066350710900473, "grad_norm": 7.894672870635986, "learning_rate": 3.696919431279621e-05, "loss": 0.6949, "step": 5500 }, { "epoch": 26.303317535545023, "grad_norm": 9.261528968811035, "learning_rate": 3.685071090047393e-05, "loss": 0.5926, "step": 5550 }, { "epoch": 26.540284360189574, "grad_norm": 7.603061676025391, "learning_rate": 3.673222748815166e-05, "loss": 0.6085, "step": 5600 }, { "epoch": 26.777251184834125, "grad_norm": 8.329363822937012, "learning_rate": 3.661374407582939e-05, "loss": 0.6553, "step": 5650 }, { "epoch": 27.01421800947867, "grad_norm": 5.974715232849121, "learning_rate": 3.649526066350711e-05, "loss": 0.6389, "step": 5700 }, { "epoch": 27.251184834123222, "grad_norm": 5.819736957550049, "learning_rate": 3.637677725118483e-05, "loss": 0.5514, "step": 5750 }, { "epoch": 27.488151658767773, "grad_norm": 8.34289264678955, "learning_rate": 3.625829383886256e-05, "loss": 0.5706, "step": 5800 }, { "epoch": 27.725118483412324, "grad_norm": 8.434243202209473, "learning_rate": 3.613981042654029e-05, "loss": 0.5867, "step": 5850 }, { "epoch": 27.96208530805687, "grad_norm": 7.840861797332764, "learning_rate": 3.602132701421801e-05, "loss": 0.5927, "step": 5900 }, { "epoch": 28.19905213270142, "grad_norm": 6.7204155921936035, "learning_rate": 3.590284360189573e-05, "loss": 0.5281, "step": 5950 }, { "epoch": 28.436018957345972, "grad_norm": 8.39271354675293, "learning_rate": 3.578436018957346e-05, "loss": 0.5377, "step": 6000 }, { "epoch": 28.672985781990523, "grad_norm": 6.774899005889893, "learning_rate": 3.566587677725119e-05, "loss": 0.5254, "step": 6050 }, { "epoch": 28.90995260663507, "grad_norm": 6.437871932983398, "learning_rate": 3.554739336492891e-05, "loss": 0.5482, "step": 6100 }, { "epoch": 29.14691943127962, "grad_norm": 7.66685676574707, "learning_rate": 3.542890995260664e-05, "loss": 0.5282, "step": 6150 }, { "epoch": 29.38388625592417, "grad_norm": 8.003113746643066, "learning_rate": 3.531042654028436e-05, "loss": 0.4543, "step": 6200 }, { "epoch": 29.620853080568722, "grad_norm": 5.657834529876709, "learning_rate": 3.519194312796208e-05, "loss": 0.4595, "step": 6250 }, { "epoch": 29.85781990521327, "grad_norm": 8.523530960083008, "learning_rate": 3.507345971563981e-05, "loss": 0.5251, "step": 6300 }, { "epoch": 30.09478672985782, "grad_norm": 6.704988479614258, "learning_rate": 3.495497630331754e-05, "loss": 0.4879, "step": 6350 }, { "epoch": 30.33175355450237, "grad_norm": 6.956995010375977, "learning_rate": 3.483649289099527e-05, "loss": 0.4438, "step": 6400 }, { "epoch": 30.56872037914692, "grad_norm": 6.2612690925598145, "learning_rate": 3.471800947867299e-05, "loss": 0.4504, "step": 6450 }, { "epoch": 30.805687203791468, "grad_norm": 7.727517127990723, "learning_rate": 3.459952606635071e-05, "loss": 0.476, "step": 6500 }, { "epoch": 31.04265402843602, "grad_norm": 6.553852081298828, "learning_rate": 3.448104265402844e-05, "loss": 0.471, "step": 6550 }, { "epoch": 31.27962085308057, "grad_norm": 8.569219589233398, "learning_rate": 3.436255924170616e-05, "loss": 0.3958, "step": 6600 }, { "epoch": 31.51658767772512, "grad_norm": 9.081930160522461, "learning_rate": 3.424407582938389e-05, "loss": 0.4247, "step": 6650 }, { "epoch": 31.753554502369667, "grad_norm": 7.493210315704346, "learning_rate": 3.412559241706161e-05, "loss": 0.425, "step": 6700 }, { "epoch": 31.990521327014218, "grad_norm": 7.720577239990234, "learning_rate": 3.400710900473934e-05, "loss": 0.4443, "step": 6750 }, { "epoch": 32.22748815165877, "grad_norm": 6.875296115875244, "learning_rate": 3.388862559241706e-05, "loss": 0.3706, "step": 6800 }, { "epoch": 32.464454976303315, "grad_norm": 7.385213851928711, "learning_rate": 3.377014218009479e-05, "loss": 0.3716, "step": 6850 }, { "epoch": 32.70142180094787, "grad_norm": 6.193999767303467, "learning_rate": 3.365165876777252e-05, "loss": 0.4035, "step": 6900 }, { "epoch": 32.93838862559242, "grad_norm": 7.366576671600342, "learning_rate": 3.353317535545024e-05, "loss": 0.3896, "step": 6950 }, { "epoch": 33.175355450236964, "grad_norm": 6.549801826477051, "learning_rate": 3.341469194312796e-05, "loss": 0.3705, "step": 7000 }, { "epoch": 33.41232227488152, "grad_norm": 6.870905876159668, "learning_rate": 3.329620853080569e-05, "loss": 0.3593, "step": 7050 }, { "epoch": 33.649289099526065, "grad_norm": 7.719451904296875, "learning_rate": 3.317772511848342e-05, "loss": 0.3762, "step": 7100 }, { "epoch": 33.88625592417062, "grad_norm": 6.358677387237549, "learning_rate": 3.305924170616114e-05, "loss": 0.3473, "step": 7150 }, { "epoch": 34.12322274881517, "grad_norm": 5.549831390380859, "learning_rate": 3.294075829383887e-05, "loss": 0.3598, "step": 7200 }, { "epoch": 34.360189573459714, "grad_norm": 6.222386360168457, "learning_rate": 3.282227488151659e-05, "loss": 0.3294, "step": 7250 }, { "epoch": 34.59715639810427, "grad_norm": 5.687737941741943, "learning_rate": 3.2703791469194317e-05, "loss": 0.3377, "step": 7300 }, { "epoch": 34.834123222748815, "grad_norm": 6.7961530685424805, "learning_rate": 3.258530805687204e-05, "loss": 0.3465, "step": 7350 }, { "epoch": 35.07109004739336, "grad_norm": 7.967190742492676, "learning_rate": 3.2466824644549767e-05, "loss": 0.3241, "step": 7400 }, { "epoch": 35.308056872037916, "grad_norm": 6.925683975219727, "learning_rate": 3.2348341232227495e-05, "loss": 0.3078, "step": 7450 }, { "epoch": 35.54502369668246, "grad_norm": 7.531118869781494, "learning_rate": 3.222985781990521e-05, "loss": 0.3151, "step": 7500 }, { "epoch": 35.78199052132702, "grad_norm": 7.131246089935303, "learning_rate": 3.211137440758294e-05, "loss": 0.3135, "step": 7550 }, { "epoch": 36.018957345971565, "grad_norm": 7.393649101257324, "learning_rate": 3.1992890995260666e-05, "loss": 0.3178, "step": 7600 }, { "epoch": 36.25592417061611, "grad_norm": 5.593276500701904, "learning_rate": 3.1874407582938395e-05, "loss": 0.2701, "step": 7650 }, { "epoch": 36.492890995260666, "grad_norm": 7.743380546569824, "learning_rate": 3.1755924170616116e-05, "loss": 0.277, "step": 7700 }, { "epoch": 36.72985781990521, "grad_norm": 7.773705005645752, "learning_rate": 3.163744075829384e-05, "loss": 0.2922, "step": 7750 }, { "epoch": 36.96682464454976, "grad_norm": 4.5125932693481445, "learning_rate": 3.1518957345971566e-05, "loss": 0.293, "step": 7800 }, { "epoch": 37.203791469194314, "grad_norm": 6.160907745361328, "learning_rate": 3.140047393364929e-05, "loss": 0.2603, "step": 7850 }, { "epoch": 37.44075829383886, "grad_norm": 5.3235578536987305, "learning_rate": 3.1281990521327016e-05, "loss": 0.2532, "step": 7900 }, { "epoch": 37.677725118483416, "grad_norm": 5.948000431060791, "learning_rate": 3.1163507109004744e-05, "loss": 0.2782, "step": 7950 }, { "epoch": 37.91469194312796, "grad_norm": 6.24420690536499, "learning_rate": 3.1045023696682466e-05, "loss": 0.2673, "step": 8000 }, { "epoch": 38.15165876777251, "grad_norm": 6.408054351806641, "learning_rate": 3.092654028436019e-05, "loss": 0.2523, "step": 8050 }, { "epoch": 38.388625592417064, "grad_norm": 6.102354526519775, "learning_rate": 3.0808056872037916e-05, "loss": 0.2531, "step": 8100 }, { "epoch": 38.62559241706161, "grad_norm": 5.659121036529541, "learning_rate": 3.0689573459715644e-05, "loss": 0.2503, "step": 8150 }, { "epoch": 38.86255924170616, "grad_norm": 7.54925012588501, "learning_rate": 3.0571090047393366e-05, "loss": 0.2603, "step": 8200 }, { "epoch": 39.09952606635071, "grad_norm": 5.3906402587890625, "learning_rate": 3.045260663507109e-05, "loss": 0.239, "step": 8250 }, { "epoch": 39.33649289099526, "grad_norm": 5.1655497550964355, "learning_rate": 3.0334123222748816e-05, "loss": 0.2312, "step": 8300 }, { "epoch": 39.573459715639814, "grad_norm": 6.628402233123779, "learning_rate": 3.0215639810426544e-05, "loss": 0.2418, "step": 8350 }, { "epoch": 39.81042654028436, "grad_norm": 6.020391941070557, "learning_rate": 3.0097156398104266e-05, "loss": 0.2277, "step": 8400 }, { "epoch": 40.04739336492891, "grad_norm": 5.597630500793457, "learning_rate": 2.997867298578199e-05, "loss": 0.2301, "step": 8450 }, { "epoch": 40.28436018957346, "grad_norm": 6.318883895874023, "learning_rate": 2.986018957345972e-05, "loss": 0.2111, "step": 8500 }, { "epoch": 40.52132701421801, "grad_norm": 4.044495105743408, "learning_rate": 2.9741706161137444e-05, "loss": 0.221, "step": 8550 }, { "epoch": 40.758293838862556, "grad_norm": 5.434823513031006, "learning_rate": 2.9623222748815165e-05, "loss": 0.2216, "step": 8600 }, { "epoch": 40.99526066350711, "grad_norm": 5.392724514007568, "learning_rate": 2.9504739336492894e-05, "loss": 0.229, "step": 8650 }, { "epoch": 41.23222748815166, "grad_norm": 3.498589277267456, "learning_rate": 2.938625592417062e-05, "loss": 0.1955, "step": 8700 }, { "epoch": 41.46919431279621, "grad_norm": 4.551211833953857, "learning_rate": 2.926777251184834e-05, "loss": 0.211, "step": 8750 }, { "epoch": 41.70616113744076, "grad_norm": 5.960665702819824, "learning_rate": 2.914928909952607e-05, "loss": 0.2063, "step": 8800 }, { "epoch": 41.943127962085306, "grad_norm": 6.883888244628906, "learning_rate": 2.9030805687203794e-05, "loss": 0.2106, "step": 8850 }, { "epoch": 42.18009478672986, "grad_norm": 5.70069694519043, "learning_rate": 2.8912322274881522e-05, "loss": 0.1939, "step": 8900 }, { "epoch": 42.41706161137441, "grad_norm": 4.06521463394165, "learning_rate": 2.879383886255924e-05, "loss": 0.2021, "step": 8950 }, { "epoch": 42.654028436018955, "grad_norm": 5.1299028396606445, "learning_rate": 2.867535545023697e-05, "loss": 0.199, "step": 9000 }, { "epoch": 42.89099526066351, "grad_norm": 4.64950704574585, "learning_rate": 2.8556872037914693e-05, "loss": 0.1931, "step": 9050 }, { "epoch": 43.127962085308056, "grad_norm": 6.922746658325195, "learning_rate": 2.8438388625592415e-05, "loss": 0.1893, "step": 9100 }, { "epoch": 43.36492890995261, "grad_norm": 7.4043660163879395, "learning_rate": 2.8319905213270143e-05, "loss": 0.1881, "step": 9150 }, { "epoch": 43.60189573459716, "grad_norm": 4.761287212371826, "learning_rate": 2.8201421800947868e-05, "loss": 0.1886, "step": 9200 }, { "epoch": 43.838862559241704, "grad_norm": 5.436069488525391, "learning_rate": 2.8082938388625597e-05, "loss": 0.1977, "step": 9250 }, { "epoch": 44.07582938388626, "grad_norm": 5.317116737365723, "learning_rate": 2.7964454976303318e-05, "loss": 0.1934, "step": 9300 }, { "epoch": 44.312796208530806, "grad_norm": 7.17617654800415, "learning_rate": 2.7845971563981043e-05, "loss": 0.1826, "step": 9350 }, { "epoch": 44.54976303317535, "grad_norm": 5.663304805755615, "learning_rate": 2.772748815165877e-05, "loss": 0.1762, "step": 9400 }, { "epoch": 44.78672985781991, "grad_norm": 6.618777751922607, "learning_rate": 2.7609004739336493e-05, "loss": 0.1755, "step": 9450 }, { "epoch": 45.023696682464454, "grad_norm": 4.825708389282227, "learning_rate": 2.7490521327014218e-05, "loss": 0.1878, "step": 9500 }, { "epoch": 45.26066350710901, "grad_norm": 5.505741596221924, "learning_rate": 2.7372037914691946e-05, "loss": 0.1712, "step": 9550 }, { "epoch": 45.497630331753555, "grad_norm": 7.058389663696289, "learning_rate": 2.725355450236967e-05, "loss": 0.1592, "step": 9600 }, { "epoch": 45.7345971563981, "grad_norm": 5.179974555969238, "learning_rate": 2.7135071090047393e-05, "loss": 0.17, "step": 9650 }, { "epoch": 45.97156398104266, "grad_norm": 6.574582099914551, "learning_rate": 2.7016587677725118e-05, "loss": 0.1745, "step": 9700 }, { "epoch": 46.208530805687204, "grad_norm": 6.780065059661865, "learning_rate": 2.6898104265402846e-05, "loss": 0.1594, "step": 9750 }, { "epoch": 46.44549763033175, "grad_norm": 4.467445373535156, "learning_rate": 2.677962085308057e-05, "loss": 0.1454, "step": 9800 }, { "epoch": 46.682464454976305, "grad_norm": 5.493533134460449, "learning_rate": 2.6661137440758293e-05, "loss": 0.1703, "step": 9850 }, { "epoch": 46.91943127962085, "grad_norm": 5.029524326324463, "learning_rate": 2.654265402843602e-05, "loss": 0.1709, "step": 9900 }, { "epoch": 47.156398104265406, "grad_norm": 4.597728729248047, "learning_rate": 2.6424170616113746e-05, "loss": 0.1525, "step": 9950 }, { "epoch": 47.39336492890995, "grad_norm": 4.824158668518066, "learning_rate": 2.6305687203791467e-05, "loss": 0.1528, "step": 10000 }, { "epoch": 47.6303317535545, "grad_norm": 5.1738080978393555, "learning_rate": 2.6187203791469196e-05, "loss": 0.1563, "step": 10050 }, { "epoch": 47.867298578199055, "grad_norm": 3.893418073654175, "learning_rate": 2.606872037914692e-05, "loss": 0.1507, "step": 10100 }, { "epoch": 48.1042654028436, "grad_norm": 4.902078628540039, "learning_rate": 2.595023696682465e-05, "loss": 0.1491, "step": 10150 }, { "epoch": 48.34123222748815, "grad_norm": 3.625319480895996, "learning_rate": 2.583175355450237e-05, "loss": 0.1415, "step": 10200 }, { "epoch": 48.5781990521327, "grad_norm": 6.172499656677246, "learning_rate": 2.5713270142180096e-05, "loss": 0.1371, "step": 10250 }, { "epoch": 48.81516587677725, "grad_norm": 5.210315704345703, "learning_rate": 2.5594786729857824e-05, "loss": 0.1447, "step": 10300 }, { "epoch": 49.052132701421804, "grad_norm": 6.0598978996276855, "learning_rate": 2.5476303317535542e-05, "loss": 0.1501, "step": 10350 }, { "epoch": 49.28909952606635, "grad_norm": 4.792999267578125, "learning_rate": 2.535781990521327e-05, "loss": 0.1368, "step": 10400 }, { "epoch": 49.5260663507109, "grad_norm": 5.838809013366699, "learning_rate": 2.5239336492890995e-05, "loss": 0.1442, "step": 10450 }, { "epoch": 49.76303317535545, "grad_norm": 4.270689964294434, "learning_rate": 2.5120853080568724e-05, "loss": 0.1276, "step": 10500 }, { "epoch": 50.0, "grad_norm": 6.24560022354126, "learning_rate": 2.5002369668246445e-05, "loss": 0.1549, "step": 10550 }, { "epoch": 50.23696682464455, "grad_norm": 4.995229244232178, "learning_rate": 2.488388625592417e-05, "loss": 0.1208, "step": 10600 }, { "epoch": 50.4739336492891, "grad_norm": 4.177917003631592, "learning_rate": 2.47654028436019e-05, "loss": 0.1328, "step": 10650 }, { "epoch": 50.71090047393365, "grad_norm": 4.9936323165893555, "learning_rate": 2.4646919431279624e-05, "loss": 0.1274, "step": 10700 }, { "epoch": 50.9478672985782, "grad_norm": 4.75553560256958, "learning_rate": 2.4528436018957345e-05, "loss": 0.1384, "step": 10750 }, { "epoch": 51.18483412322275, "grad_norm": 4.1424560546875, "learning_rate": 2.4409952606635073e-05, "loss": 0.1223, "step": 10800 }, { "epoch": 51.4218009478673, "grad_norm": 6.769973278045654, "learning_rate": 2.4291469194312795e-05, "loss": 0.1239, "step": 10850 }, { "epoch": 51.65876777251185, "grad_norm": 3.4160497188568115, "learning_rate": 2.4172985781990523e-05, "loss": 0.135, "step": 10900 }, { "epoch": 51.8957345971564, "grad_norm": 3.5807864665985107, "learning_rate": 2.4054502369668248e-05, "loss": 0.1383, "step": 10950 }, { "epoch": 52.132701421800945, "grad_norm": 4.546731472015381, "learning_rate": 2.3936018957345973e-05, "loss": 0.1185, "step": 11000 }, { "epoch": 52.3696682464455, "grad_norm": 4.606685161590576, "learning_rate": 2.3817535545023698e-05, "loss": 0.1137, "step": 11050 }, { "epoch": 52.60663507109005, "grad_norm": 4.739546298980713, "learning_rate": 2.3699052132701423e-05, "loss": 0.119, "step": 11100 }, { "epoch": 52.843601895734594, "grad_norm": 4.913407325744629, "learning_rate": 2.3580568720379148e-05, "loss": 0.127, "step": 11150 }, { "epoch": 53.08056872037915, "grad_norm": 4.719569206237793, "learning_rate": 2.3462085308056873e-05, "loss": 0.1289, "step": 11200 }, { "epoch": 53.317535545023695, "grad_norm": 4.4446940422058105, "learning_rate": 2.3343601895734598e-05, "loss": 0.1097, "step": 11250 }, { "epoch": 53.55450236966825, "grad_norm": 5.161761283874512, "learning_rate": 2.3225118483412323e-05, "loss": 0.1152, "step": 11300 }, { "epoch": 53.791469194312796, "grad_norm": 3.8273890018463135, "learning_rate": 2.3106635071090048e-05, "loss": 0.1163, "step": 11350 }, { "epoch": 54.02843601895734, "grad_norm": 3.5801877975463867, "learning_rate": 2.2988151658767773e-05, "loss": 0.1063, "step": 11400 }, { "epoch": 54.2654028436019, "grad_norm": 3.7107162475585938, "learning_rate": 2.28696682464455e-05, "loss": 0.111, "step": 11450 }, { "epoch": 54.502369668246445, "grad_norm": 4.708107948303223, "learning_rate": 2.2751184834123223e-05, "loss": 0.1074, "step": 11500 }, { "epoch": 54.73933649289099, "grad_norm": 3.315948724746704, "learning_rate": 2.263270142180095e-05, "loss": 0.1138, "step": 11550 }, { "epoch": 54.976303317535546, "grad_norm": 5.415637493133545, "learning_rate": 2.2514218009478673e-05, "loss": 0.115, "step": 11600 }, { "epoch": 55.21327014218009, "grad_norm": 3.2730207443237305, "learning_rate": 2.2395734597156398e-05, "loss": 0.1104, "step": 11650 }, { "epoch": 55.45023696682465, "grad_norm": 4.767010688781738, "learning_rate": 2.2277251184834126e-05, "loss": 0.1022, "step": 11700 }, { "epoch": 55.687203791469194, "grad_norm": 5.740176677703857, "learning_rate": 2.2158767772511848e-05, "loss": 0.1016, "step": 11750 }, { "epoch": 55.92417061611374, "grad_norm": 4.717856407165527, "learning_rate": 2.2040284360189576e-05, "loss": 0.106, "step": 11800 }, { "epoch": 56.161137440758296, "grad_norm": 5.084784030914307, "learning_rate": 2.19218009478673e-05, "loss": 0.0996, "step": 11850 }, { "epoch": 56.39810426540284, "grad_norm": 4.658839225769043, "learning_rate": 2.1803317535545026e-05, "loss": 0.1003, "step": 11900 }, { "epoch": 56.63507109004739, "grad_norm": 5.200915813446045, "learning_rate": 2.168483412322275e-05, "loss": 0.1072, "step": 11950 }, { "epoch": 56.872037914691944, "grad_norm": 4.565636157989502, "learning_rate": 2.1566350710900472e-05, "loss": 0.105, "step": 12000 }, { "epoch": 57.10900473933649, "grad_norm": 4.767094135284424, "learning_rate": 2.14478672985782e-05, "loss": 0.0997, "step": 12050 }, { "epoch": 57.345971563981045, "grad_norm": 4.945188045501709, "learning_rate": 2.1329383886255926e-05, "loss": 0.0995, "step": 12100 }, { "epoch": 57.58293838862559, "grad_norm": 5.004721164703369, "learning_rate": 2.121090047393365e-05, "loss": 0.0968, "step": 12150 }, { "epoch": 57.81990521327014, "grad_norm": 4.6731696128845215, "learning_rate": 2.1092417061611375e-05, "loss": 0.1005, "step": 12200 }, { "epoch": 58.056872037914694, "grad_norm": 6.432486057281494, "learning_rate": 2.09739336492891e-05, "loss": 0.0982, "step": 12250 }, { "epoch": 58.29383886255924, "grad_norm": 3.2889387607574463, "learning_rate": 2.0855450236966825e-05, "loss": 0.0935, "step": 12300 }, { "epoch": 58.53080568720379, "grad_norm": 2.5075995922088623, "learning_rate": 2.073696682464455e-05, "loss": 0.0911, "step": 12350 }, { "epoch": 58.76777251184834, "grad_norm": 4.071348667144775, "learning_rate": 2.0618483412322275e-05, "loss": 0.096, "step": 12400 }, { "epoch": 59.00473933649289, "grad_norm": 2.9908299446105957, "learning_rate": 2.05e-05, "loss": 0.1048, "step": 12450 }, { "epoch": 59.241706161137444, "grad_norm": 4.291813850402832, "learning_rate": 2.0381516587677725e-05, "loss": 0.0804, "step": 12500 }, { "epoch": 59.47867298578199, "grad_norm": 4.400552749633789, "learning_rate": 2.026303317535545e-05, "loss": 0.0881, "step": 12550 }, { "epoch": 59.71563981042654, "grad_norm": 4.662256240844727, "learning_rate": 2.014454976303318e-05, "loss": 0.0856, "step": 12600 }, { "epoch": 59.95260663507109, "grad_norm": 4.559081554412842, "learning_rate": 2.00260663507109e-05, "loss": 0.0928, "step": 12650 }, { "epoch": 60.18957345971564, "grad_norm": 5.430093765258789, "learning_rate": 1.990758293838863e-05, "loss": 0.0857, "step": 12700 }, { "epoch": 60.426540284360186, "grad_norm": 4.222429275512695, "learning_rate": 1.978909952606635e-05, "loss": 0.0849, "step": 12750 }, { "epoch": 60.66350710900474, "grad_norm": 3.7678847312927246, "learning_rate": 1.9670616113744078e-05, "loss": 0.0907, "step": 12800 }, { "epoch": 60.90047393364929, "grad_norm": 4.276619911193848, "learning_rate": 1.9552132701421803e-05, "loss": 0.088, "step": 12850 }, { "epoch": 61.13744075829384, "grad_norm": 3.183441162109375, "learning_rate": 1.9433649289099525e-05, "loss": 0.084, "step": 12900 }, { "epoch": 61.37440758293839, "grad_norm": 3.7540106773376465, "learning_rate": 1.9315165876777253e-05, "loss": 0.0838, "step": 12950 }, { "epoch": 61.611374407582936, "grad_norm": 4.090873718261719, "learning_rate": 1.9196682464454978e-05, "loss": 0.0857, "step": 13000 }, { "epoch": 61.84834123222749, "grad_norm": 5.003512382507324, "learning_rate": 1.9078199052132703e-05, "loss": 0.0821, "step": 13050 }, { "epoch": 62.08530805687204, "grad_norm": 4.456465721130371, "learning_rate": 1.8959715639810428e-05, "loss": 0.0837, "step": 13100 }, { "epoch": 62.322274881516584, "grad_norm": 4.660475254058838, "learning_rate": 1.8841232227488153e-05, "loss": 0.0812, "step": 13150 }, { "epoch": 62.55924170616114, "grad_norm": 4.613561630249023, "learning_rate": 1.8722748815165878e-05, "loss": 0.083, "step": 13200 }, { "epoch": 62.796208530805686, "grad_norm": 4.390279293060303, "learning_rate": 1.8604265402843603e-05, "loss": 0.0905, "step": 13250 }, { "epoch": 63.03317535545024, "grad_norm": 4.213273048400879, "learning_rate": 1.8485781990521328e-05, "loss": 0.0755, "step": 13300 }, { "epoch": 63.27014218009479, "grad_norm": 3.4082424640655518, "learning_rate": 1.8367298578199053e-05, "loss": 0.0698, "step": 13350 }, { "epoch": 63.507109004739334, "grad_norm": 3.142331123352051, "learning_rate": 1.8248815165876778e-05, "loss": 0.0774, "step": 13400 }, { "epoch": 63.74407582938389, "grad_norm": 3.847400426864624, "learning_rate": 1.8130331753554503e-05, "loss": 0.0819, "step": 13450 }, { "epoch": 63.981042654028435, "grad_norm": 3.6660194396972656, "learning_rate": 1.8011848341232228e-05, "loss": 0.0826, "step": 13500 }, { "epoch": 64.21800947867298, "grad_norm": 5.401995658874512, "learning_rate": 1.7893364928909953e-05, "loss": 0.0779, "step": 13550 }, { "epoch": 64.45497630331754, "grad_norm": 4.982518672943115, "learning_rate": 1.777488151658768e-05, "loss": 0.0732, "step": 13600 }, { "epoch": 64.69194312796209, "grad_norm": 4.009726047515869, "learning_rate": 1.7656398104265402e-05, "loss": 0.08, "step": 13650 }, { "epoch": 64.92890995260663, "grad_norm": 4.493566036224365, "learning_rate": 1.7537914691943127e-05, "loss": 0.0777, "step": 13700 }, { "epoch": 65.16587677725119, "grad_norm": 4.450813293457031, "learning_rate": 1.7419431279620856e-05, "loss": 0.0756, "step": 13750 }, { "epoch": 65.40284360189574, "grad_norm": 5.273355007171631, "learning_rate": 1.7300947867298577e-05, "loss": 0.0709, "step": 13800 }, { "epoch": 65.63981042654028, "grad_norm": 3.010479688644409, "learning_rate": 1.7182464454976306e-05, "loss": 0.0718, "step": 13850 }, { "epoch": 65.87677725118483, "grad_norm": 3.340604782104492, "learning_rate": 1.7063981042654027e-05, "loss": 0.0708, "step": 13900 }, { "epoch": 66.11374407582939, "grad_norm": 3.8978354930877686, "learning_rate": 1.6945497630331756e-05, "loss": 0.0719, "step": 13950 }, { "epoch": 66.35071090047393, "grad_norm": 3.0230929851531982, "learning_rate": 1.682701421800948e-05, "loss": 0.0738, "step": 14000 }, { "epoch": 66.58767772511848, "grad_norm": 2.5646421909332275, "learning_rate": 1.6708530805687205e-05, "loss": 0.0724, "step": 14050 }, { "epoch": 66.82464454976304, "grad_norm": 2.8454196453094482, "learning_rate": 1.659004739336493e-05, "loss": 0.0719, "step": 14100 }, { "epoch": 67.06161137440758, "grad_norm": 5.150911808013916, "learning_rate": 1.6471563981042652e-05, "loss": 0.0728, "step": 14150 }, { "epoch": 67.29857819905213, "grad_norm": 3.6012842655181885, "learning_rate": 1.635308056872038e-05, "loss": 0.064, "step": 14200 }, { "epoch": 67.53554502369668, "grad_norm": 2.7426695823669434, "learning_rate": 1.6234597156398105e-05, "loss": 0.0659, "step": 14250 }, { "epoch": 67.77251184834124, "grad_norm": 3.487116575241089, "learning_rate": 1.611611374407583e-05, "loss": 0.0659, "step": 14300 }, { "epoch": 68.00947867298578, "grad_norm": 4.15194034576416, "learning_rate": 1.5997630331753555e-05, "loss": 0.0655, "step": 14350 }, { "epoch": 68.24644549763033, "grad_norm": 2.8579485416412354, "learning_rate": 1.587914691943128e-05, "loss": 0.0685, "step": 14400 }, { "epoch": 68.48341232227489, "grad_norm": 2.9810163974761963, "learning_rate": 1.5760663507109005e-05, "loss": 0.0693, "step": 14450 }, { "epoch": 68.72037914691943, "grad_norm": 3.442668914794922, "learning_rate": 1.5642180094786733e-05, "loss": 0.0624, "step": 14500 }, { "epoch": 68.95734597156398, "grad_norm": 5.029447078704834, "learning_rate": 1.5523696682464455e-05, "loss": 0.07, "step": 14550 }, { "epoch": 69.19431279620854, "grad_norm": 3.7671704292297363, "learning_rate": 1.540521327014218e-05, "loss": 0.0682, "step": 14600 }, { "epoch": 69.43127962085308, "grad_norm": 5.002161502838135, "learning_rate": 1.5286729857819905e-05, "loss": 0.0626, "step": 14650 }, { "epoch": 69.66824644549763, "grad_norm": 3.129530906677246, "learning_rate": 1.516824644549763e-05, "loss": 0.0633, "step": 14700 }, { "epoch": 69.90521327014218, "grad_norm": 2.9057610034942627, "learning_rate": 1.5049763033175357e-05, "loss": 0.0715, "step": 14750 }, { "epoch": 70.14218009478672, "grad_norm": 4.254899501800537, "learning_rate": 1.4931279620853081e-05, "loss": 0.0594, "step": 14800 }, { "epoch": 70.37914691943128, "grad_norm": 3.992370367050171, "learning_rate": 1.4812796208530808e-05, "loss": 0.0604, "step": 14850 }, { "epoch": 70.61611374407583, "grad_norm": 3.8310561180114746, "learning_rate": 1.4694312796208531e-05, "loss": 0.0634, "step": 14900 }, { "epoch": 70.85308056872037, "grad_norm": 2.7978172302246094, "learning_rate": 1.4575829383886255e-05, "loss": 0.0592, "step": 14950 }, { "epoch": 71.09004739336493, "grad_norm": 3.4164352416992188, "learning_rate": 1.4457345971563981e-05, "loss": 0.0618, "step": 15000 }, { "epoch": 71.32701421800948, "grad_norm": 1.2698974609375, "learning_rate": 1.4338862559241706e-05, "loss": 0.0534, "step": 15050 }, { "epoch": 71.56398104265404, "grad_norm": 3.1819329261779785, "learning_rate": 1.4220379146919433e-05, "loss": 0.057, "step": 15100 }, { "epoch": 71.80094786729858, "grad_norm": 2.293548107147217, "learning_rate": 1.4101895734597156e-05, "loss": 0.0626, "step": 15150 }, { "epoch": 72.03791469194313, "grad_norm": 4.761962890625, "learning_rate": 1.3983412322274883e-05, "loss": 0.0616, "step": 15200 }, { "epoch": 72.27488151658768, "grad_norm": 4.062314510345459, "learning_rate": 1.3864928909952608e-05, "loss": 0.0539, "step": 15250 }, { "epoch": 72.51184834123222, "grad_norm": 3.0416853427886963, "learning_rate": 1.3746445497630334e-05, "loss": 0.0478, "step": 15300 }, { "epoch": 72.74881516587678, "grad_norm": 4.738624095916748, "learning_rate": 1.3627962085308058e-05, "loss": 0.0594, "step": 15350 }, { "epoch": 72.98578199052133, "grad_norm": 5.105005264282227, "learning_rate": 1.3509478672985781e-05, "loss": 0.0597, "step": 15400 }, { "epoch": 73.22274881516587, "grad_norm": 4.290104866027832, "learning_rate": 1.3390995260663508e-05, "loss": 0.0627, "step": 15450 }, { "epoch": 73.45971563981043, "grad_norm": 2.02573561668396, "learning_rate": 1.3272511848341232e-05, "loss": 0.0588, "step": 15500 }, { "epoch": 73.69668246445498, "grad_norm": 3.635725975036621, "learning_rate": 1.3154028436018959e-05, "loss": 0.0541, "step": 15550 }, { "epoch": 73.93364928909952, "grad_norm": 2.3941638469696045, "learning_rate": 1.3035545023696682e-05, "loss": 0.053, "step": 15600 }, { "epoch": 74.17061611374407, "grad_norm": 2.6838598251342773, "learning_rate": 1.2917061611374409e-05, "loss": 0.0557, "step": 15650 }, { "epoch": 74.40758293838863, "grad_norm": 1.7956262826919556, "learning_rate": 1.2798578199052132e-05, "loss": 0.0549, "step": 15700 }, { "epoch": 74.64454976303317, "grad_norm": 2.1158010959625244, "learning_rate": 1.2680094786729859e-05, "loss": 0.0553, "step": 15750 }, { "epoch": 74.88151658767772, "grad_norm": 3.1257917881011963, "learning_rate": 1.2561611374407584e-05, "loss": 0.0588, "step": 15800 }, { "epoch": 75.11848341232228, "grad_norm": 2.137049674987793, "learning_rate": 1.2443127962085309e-05, "loss": 0.0498, "step": 15850 }, { "epoch": 75.35545023696683, "grad_norm": 2.947986364364624, "learning_rate": 1.2324644549763034e-05, "loss": 0.0512, "step": 15900 }, { "epoch": 75.59241706161137, "grad_norm": 3.1947827339172363, "learning_rate": 1.2206161137440759e-05, "loss": 0.0546, "step": 15950 }, { "epoch": 75.82938388625593, "grad_norm": 2.4928057193756104, "learning_rate": 1.2087677725118484e-05, "loss": 0.0581, "step": 16000 }, { "epoch": 76.06635071090048, "grad_norm": 4.707916736602783, "learning_rate": 1.1969194312796209e-05, "loss": 0.059, "step": 16050 }, { "epoch": 76.30331753554502, "grad_norm": 3.632354736328125, "learning_rate": 1.1850710900473934e-05, "loss": 0.0555, "step": 16100 }, { "epoch": 76.54028436018957, "grad_norm": 4.005579948425293, "learning_rate": 1.1732227488151659e-05, "loss": 0.0518, "step": 16150 }, { "epoch": 76.77725118483413, "grad_norm": 3.95902943611145, "learning_rate": 1.1613744075829385e-05, "loss": 0.0518, "step": 16200 }, { "epoch": 77.01421800947867, "grad_norm": 1.924726128578186, "learning_rate": 1.149526066350711e-05, "loss": 0.0482, "step": 16250 }, { "epoch": 77.25118483412322, "grad_norm": 2.120603322982788, "learning_rate": 1.1376777251184835e-05, "loss": 0.0447, "step": 16300 }, { "epoch": 77.48815165876778, "grad_norm": 3.6519405841827393, "learning_rate": 1.125829383886256e-05, "loss": 0.052, "step": 16350 }, { "epoch": 77.72511848341232, "grad_norm": 3.5274534225463867, "learning_rate": 1.1139810426540285e-05, "loss": 0.0459, "step": 16400 }, { "epoch": 77.96208530805687, "grad_norm": 4.313018798828125, "learning_rate": 1.102132701421801e-05, "loss": 0.0475, "step": 16450 }, { "epoch": 78.19905213270142, "grad_norm": 2.805720329284668, "learning_rate": 1.0902843601895735e-05, "loss": 0.047, "step": 16500 }, { "epoch": 78.43601895734596, "grad_norm": 2.8536834716796875, "learning_rate": 1.078436018957346e-05, "loss": 0.0459, "step": 16550 }, { "epoch": 78.67298578199052, "grad_norm": 2.179976224899292, "learning_rate": 1.0665876777251185e-05, "loss": 0.0461, "step": 16600 }, { "epoch": 78.90995260663507, "grad_norm": 3.698432445526123, "learning_rate": 1.054739336492891e-05, "loss": 0.045, "step": 16650 }, { "epoch": 79.14691943127963, "grad_norm": 3.369558334350586, "learning_rate": 1.0428909952606636e-05, "loss": 0.0431, "step": 16700 }, { "epoch": 79.38388625592417, "grad_norm": 3.4015305042266846, "learning_rate": 1.0310426540284361e-05, "loss": 0.0468, "step": 16750 }, { "epoch": 79.62085308056872, "grad_norm": 2.809027671813965, "learning_rate": 1.0191943127962086e-05, "loss": 0.0483, "step": 16800 }, { "epoch": 79.85781990521328, "grad_norm": 3.1171741485595703, "learning_rate": 1.007345971563981e-05, "loss": 0.0475, "step": 16850 }, { "epoch": 80.09478672985782, "grad_norm": 2.8119149208068848, "learning_rate": 9.954976303317536e-06, "loss": 0.0434, "step": 16900 }, { "epoch": 80.33175355450237, "grad_norm": 3.034045696258545, "learning_rate": 9.836492890995261e-06, "loss": 0.0486, "step": 16950 }, { "epoch": 80.56872037914692, "grad_norm": 3.1837401390075684, "learning_rate": 9.718009478672986e-06, "loss": 0.0432, "step": 17000 }, { "epoch": 80.80568720379146, "grad_norm": 2.203754425048828, "learning_rate": 9.599526066350711e-06, "loss": 0.0444, "step": 17050 }, { "epoch": 81.04265402843602, "grad_norm": 3.1515495777130127, "learning_rate": 9.481042654028436e-06, "loss": 0.0478, "step": 17100 }, { "epoch": 81.27962085308057, "grad_norm": 2.4488179683685303, "learning_rate": 9.362559241706163e-06, "loss": 0.0431, "step": 17150 }, { "epoch": 81.51658767772511, "grad_norm": 3.5143661499023438, "learning_rate": 9.244075829383888e-06, "loss": 0.0439, "step": 17200 }, { "epoch": 81.75355450236967, "grad_norm": 2.8133413791656494, "learning_rate": 9.125592417061613e-06, "loss": 0.0447, "step": 17250 }, { "epoch": 81.99052132701422, "grad_norm": 3.704498767852783, "learning_rate": 9.007109004739336e-06, "loss": 0.0491, "step": 17300 }, { "epoch": 82.22748815165876, "grad_norm": 4.158998012542725, "learning_rate": 8.888625592417062e-06, "loss": 0.046, "step": 17350 }, { "epoch": 82.46445497630332, "grad_norm": 2.24314284324646, "learning_rate": 8.770142180094787e-06, "loss": 0.0416, "step": 17400 }, { "epoch": 82.70142180094787, "grad_norm": 4.9121270179748535, "learning_rate": 8.651658767772512e-06, "loss": 0.0507, "step": 17450 }, { "epoch": 82.93838862559242, "grad_norm": 1.077348232269287, "learning_rate": 8.533175355450237e-06, "loss": 0.0472, "step": 17500 }, { "epoch": 83.17535545023696, "grad_norm": 2.680873394012451, "learning_rate": 8.414691943127962e-06, "loss": 0.0446, "step": 17550 }, { "epoch": 83.41232227488152, "grad_norm": 2.237860679626465, "learning_rate": 8.296208530805687e-06, "loss": 0.0364, "step": 17600 }, { "epoch": 83.64928909952607, "grad_norm": 4.735086917877197, "learning_rate": 8.177725118483414e-06, "loss": 0.0442, "step": 17650 }, { "epoch": 83.88625592417061, "grad_norm": 3.1517481803894043, "learning_rate": 8.059241706161137e-06, "loss": 0.043, "step": 17700 }, { "epoch": 84.12322274881517, "grad_norm": 1.734181523323059, "learning_rate": 7.940758293838862e-06, "loss": 0.0431, "step": 17750 }, { "epoch": 84.36018957345972, "grad_norm": 1.2816810607910156, "learning_rate": 7.822274881516587e-06, "loss": 0.0443, "step": 17800 }, { "epoch": 84.59715639810426, "grad_norm": 2.2617380619049072, "learning_rate": 7.703791469194314e-06, "loss": 0.0414, "step": 17850 }, { "epoch": 84.83412322274881, "grad_norm": 2.536569833755493, "learning_rate": 7.585308056872039e-06, "loss": 0.0425, "step": 17900 }, { "epoch": 85.07109004739337, "grad_norm": 1.9184726476669312, "learning_rate": 7.466824644549764e-06, "loss": 0.038, "step": 17950 }, { "epoch": 85.30805687203791, "grad_norm": 2.9358866214752197, "learning_rate": 7.3483412322274886e-06, "loss": 0.0432, "step": 18000 }, { "epoch": 85.54502369668246, "grad_norm": 2.8088200092315674, "learning_rate": 7.229857819905214e-06, "loss": 0.0371, "step": 18050 }, { "epoch": 85.78199052132702, "grad_norm": 2.5315921306610107, "learning_rate": 7.111374407582938e-06, "loss": 0.0428, "step": 18100 }, { "epoch": 86.01895734597156, "grad_norm": 1.333961009979248, "learning_rate": 6.9928909952606634e-06, "loss": 0.039, "step": 18150 }, { "epoch": 86.25592417061611, "grad_norm": 4.303163528442383, "learning_rate": 6.874407582938388e-06, "loss": 0.037, "step": 18200 }, { "epoch": 86.49289099526067, "grad_norm": 1.5333720445632935, "learning_rate": 6.755924170616114e-06, "loss": 0.039, "step": 18250 }, { "epoch": 86.72985781990522, "grad_norm": 1.7607487440109253, "learning_rate": 6.637440758293839e-06, "loss": 0.0372, "step": 18300 }, { "epoch": 86.96682464454976, "grad_norm": 2.450984477996826, "learning_rate": 6.518957345971564e-06, "loss": 0.0385, "step": 18350 }, { "epoch": 87.20379146919431, "grad_norm": 1.7824897766113281, "learning_rate": 6.40047393364929e-06, "loss": 0.0343, "step": 18400 }, { "epoch": 87.44075829383887, "grad_norm": 2.242223024368286, "learning_rate": 6.281990521327015e-06, "loss": 0.0426, "step": 18450 }, { "epoch": 87.67772511848341, "grad_norm": 2.929349422454834, "learning_rate": 6.16350710900474e-06, "loss": 0.0348, "step": 18500 }, { "epoch": 87.91469194312796, "grad_norm": 2.6789379119873047, "learning_rate": 6.045023696682465e-06, "loss": 0.0397, "step": 18550 }, { "epoch": 88.15165876777252, "grad_norm": 4.260210990905762, "learning_rate": 5.9265402843601905e-06, "loss": 0.0394, "step": 18600 }, { "epoch": 88.38862559241706, "grad_norm": 2.6164708137512207, "learning_rate": 5.808056872037915e-06, "loss": 0.0373, "step": 18650 }, { "epoch": 88.62559241706161, "grad_norm": 2.758242130279541, "learning_rate": 5.68957345971564e-06, "loss": 0.0404, "step": 18700 }, { "epoch": 88.86255924170617, "grad_norm": 3.604731321334839, "learning_rate": 5.571090047393365e-06, "loss": 0.0384, "step": 18750 }, { "epoch": 89.0995260663507, "grad_norm": 2.9058897495269775, "learning_rate": 5.45260663507109e-06, "loss": 0.0344, "step": 18800 }, { "epoch": 89.33649289099526, "grad_norm": 3.103808879852295, "learning_rate": 5.334123222748815e-06, "loss": 0.0366, "step": 18850 }, { "epoch": 89.57345971563981, "grad_norm": 3.0444133281707764, "learning_rate": 5.21563981042654e-06, "loss": 0.0376, "step": 18900 }, { "epoch": 89.81042654028435, "grad_norm": 2.281196117401123, "learning_rate": 5.097156398104266e-06, "loss": 0.0362, "step": 18950 }, { "epoch": 90.04739336492891, "grad_norm": 0.9347835779190063, "learning_rate": 4.978672985781991e-06, "loss": 0.0377, "step": 19000 }, { "epoch": 90.28436018957346, "grad_norm": 4.489046096801758, "learning_rate": 4.860189573459716e-06, "loss": 0.0327, "step": 19050 }, { "epoch": 90.52132701421802, "grad_norm": 1.2588768005371094, "learning_rate": 4.741706161137441e-06, "loss": 0.0378, "step": 19100 }, { "epoch": 90.75829383886256, "grad_norm": 2.3064308166503906, "learning_rate": 4.623222748815166e-06, "loss": 0.0343, "step": 19150 }, { "epoch": 90.99526066350711, "grad_norm": 0.9740315079689026, "learning_rate": 4.504739336492892e-06, "loss": 0.0344, "step": 19200 }, { "epoch": 91.23222748815166, "grad_norm": 1.8359525203704834, "learning_rate": 4.386255924170617e-06, "loss": 0.0313, "step": 19250 }, { "epoch": 91.4691943127962, "grad_norm": 2.839707851409912, "learning_rate": 4.2677725118483416e-06, "loss": 0.0384, "step": 19300 }, { "epoch": 91.70616113744076, "grad_norm": 0.8259750604629517, "learning_rate": 4.1492890995260665e-06, "loss": 0.0362, "step": 19350 }, { "epoch": 91.94312796208531, "grad_norm": 2.452700138092041, "learning_rate": 4.0308056872037915e-06, "loss": 0.0362, "step": 19400 }, { "epoch": 92.18009478672985, "grad_norm": 3.368640184402466, "learning_rate": 3.912322274881517e-06, "loss": 0.0341, "step": 19450 }, { "epoch": 92.41706161137441, "grad_norm": 1.2565133571624756, "learning_rate": 3.793838862559242e-06, "loss": 0.0352, "step": 19500 }, { "epoch": 92.65402843601896, "grad_norm": 2.70945143699646, "learning_rate": 3.6753554502369667e-06, "loss": 0.0325, "step": 19550 }, { "epoch": 92.8909952606635, "grad_norm": 1.1319694519042969, "learning_rate": 3.556872037914692e-06, "loss": 0.0344, "step": 19600 }, { "epoch": 93.12796208530806, "grad_norm": 2.4756088256835938, "learning_rate": 3.4383886255924175e-06, "loss": 0.0328, "step": 19650 }, { "epoch": 93.36492890995261, "grad_norm": 2.462559700012207, "learning_rate": 3.319905213270142e-06, "loss": 0.0356, "step": 19700 }, { "epoch": 93.60189573459715, "grad_norm": 1.8483948707580566, "learning_rate": 3.2014218009478674e-06, "loss": 0.0346, "step": 19750 }, { "epoch": 93.8388625592417, "grad_norm": 2.1201963424682617, "learning_rate": 3.0829383886255928e-06, "loss": 0.0337, "step": 19800 }, { "epoch": 94.07582938388626, "grad_norm": 1.1958675384521484, "learning_rate": 2.9644549763033177e-06, "loss": 0.0332, "step": 19850 }, { "epoch": 94.31279620853081, "grad_norm": 1.8907455205917358, "learning_rate": 2.8459715639810427e-06, "loss": 0.0343, "step": 19900 }, { "epoch": 94.54976303317535, "grad_norm": 2.1910035610198975, "learning_rate": 2.7274881516587676e-06, "loss": 0.0336, "step": 19950 }, { "epoch": 94.7867298578199, "grad_norm": 4.0819830894470215, "learning_rate": 2.609004739336493e-06, "loss": 0.0338, "step": 20000 }, { "epoch": 95.02369668246446, "grad_norm": 1.0278799533843994, "learning_rate": 2.490521327014218e-06, "loss": 0.0311, "step": 20050 }, { "epoch": 95.260663507109, "grad_norm": 2.3462882041931152, "learning_rate": 2.3720379146919433e-06, "loss": 0.0317, "step": 20100 }, { "epoch": 95.49763033175356, "grad_norm": 2.494734525680542, "learning_rate": 2.2535545023696687e-06, "loss": 0.032, "step": 20150 }, { "epoch": 95.73459715639811, "grad_norm": 3.1034739017486572, "learning_rate": 2.1350710900473932e-06, "loss": 0.0369, "step": 20200 }, { "epoch": 95.97156398104265, "grad_norm": 1.2096375226974487, "learning_rate": 2.0165876777251186e-06, "loss": 0.0299, "step": 20250 }, { "epoch": 96.2085308056872, "grad_norm": 3.7373147010803223, "learning_rate": 1.8981042654028436e-06, "loss": 0.0366, "step": 20300 }, { "epoch": 96.44549763033176, "grad_norm": 2.0919508934020996, "learning_rate": 1.779620853080569e-06, "loss": 0.0342, "step": 20350 }, { "epoch": 96.6824644549763, "grad_norm": 4.377305507659912, "learning_rate": 1.661137440758294e-06, "loss": 0.03, "step": 20400 }, { "epoch": 96.91943127962085, "grad_norm": 3.0394375324249268, "learning_rate": 1.542654028436019e-06, "loss": 0.0297, "step": 20450 }, { "epoch": 97.1563981042654, "grad_norm": 3.33341383934021, "learning_rate": 1.424170616113744e-06, "loss": 0.0319, "step": 20500 }, { "epoch": 97.39336492890995, "grad_norm": 3.8229756355285645, "learning_rate": 1.3056872037914692e-06, "loss": 0.0362, "step": 20550 }, { "epoch": 97.6303317535545, "grad_norm": 2.007678508758545, "learning_rate": 1.1872037914691946e-06, "loss": 0.0318, "step": 20600 }, { "epoch": 97.86729857819905, "grad_norm": 1.4807387590408325, "learning_rate": 1.0687203791469195e-06, "loss": 0.0265, "step": 20650 }, { "epoch": 98.10426540284361, "grad_norm": 2.9157817363739014, "learning_rate": 9.502369668246447e-07, "loss": 0.0322, "step": 20700 }, { "epoch": 98.34123222748815, "grad_norm": 1.0535069704055786, "learning_rate": 8.317535545023697e-07, "loss": 0.0325, "step": 20750 }, { "epoch": 98.5781990521327, "grad_norm": 0.855229377746582, "learning_rate": 7.132701421800948e-07, "loss": 0.0326, "step": 20800 }, { "epoch": 98.81516587677726, "grad_norm": 3.276888370513916, "learning_rate": 5.9478672985782e-07, "loss": 0.0342, "step": 20850 }, { "epoch": 99.0521327014218, "grad_norm": 1.533959984779358, "learning_rate": 4.7630331753554506e-07, "loss": 0.0272, "step": 20900 }, { "epoch": 99.28909952606635, "grad_norm": 1.6495184898376465, "learning_rate": 3.578199052132702e-07, "loss": 0.0303, "step": 20950 }, { "epoch": 99.5260663507109, "grad_norm": 0.8474237322807312, "learning_rate": 2.393364928909953e-07, "loss": 0.0287, "step": 21000 }, { "epoch": 99.76303317535545, "grad_norm": 2.015770673751831, "learning_rate": 1.208530805687204e-07, "loss": 0.0338, "step": 21050 }, { "epoch": 100.0, "grad_norm": 0.6671549677848816, "learning_rate": 2.369668246445498e-09, "loss": 0.0298, "step": 21100 } ], "logging_steps": 50, "max_steps": 21100, "num_input_tokens_seen": 0, "num_train_epochs": 100, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2750098636800000.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }