| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.51566991968441, | |
| "global_step": 120000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004000105276443632, | |
| "loss": 10.1199, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00040004211053127486, | |
| "loss": 9.997, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000400094748522194, | |
| "loss": 9.9386, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00040016844138622554, | |
| "loss": 9.8988, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004002631888001141, | |
| "loss": 9.8579, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004003789903482477, | |
| "loss": 9.8159, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004005158455226594, | |
| "loss": 9.7867, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004006737537230326, | |
| "loss": 9.7605, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004008527142566991, | |
| "loss": 9.7357, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004010527263386479, | |
| "loss": 9.7138, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040127378909152016, | |
| "loss": 9.6894, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000401515901545621, | |
| "loss": 9.6634, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040177906263891804, | |
| "loss": 9.6451, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040206327121705167, | |
| "loss": 9.6279, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040236852603333685, | |
| "loss": 9.6038, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004026948257487631, | |
| "loss": 9.5874, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040304216893201697, | |
| "loss": 9.5729, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004034105540594666, | |
| "loss": 9.547, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004037999795151858, | |
| "loss": 9.5348, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004042104435909525, | |
| "loss": 9.5207, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004046419444862573, | |
| "loss": 9.5061, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004050944803083139, | |
| "loss": 9.493, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004055680490720661, | |
| "loss": 9.4782, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004060626487001964, | |
| "loss": 9.4636, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004065782770231313, | |
| "loss": 9.4546, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000407114931779062, | |
| "loss": 9.4453, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00040767261061393917, | |
| "loss": 9.4174, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00040825131108149573, | |
| "loss": 9.4159, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00040885103064325357, | |
| "loss": 9.3993, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00040947176666852707, | |
| "loss": 9.3953, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00041011351643444917, | |
| "loss": 9.3854, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004107762771259713, | |
| "loss": 9.3679, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004114600458358809, | |
| "loss": 9.3595, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00041216481956481664, | |
| "loss": 9.3504, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00041289059522127414, | |
| "loss": 9.3417, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004136373696216229, | |
| "loss": 9.3275, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004144051394901274, | |
| "loss": 9.3201, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004151939014589469, | |
| "loss": 9.3123, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004160036520681667, | |
| "loss": 9.3084, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004168343877657965, | |
| "loss": 9.2954, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00041768179413688954, | |
| "loss": 9.2862, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00041855438410810103, | |
| "loss": 9.283, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00041944794797888797, | |
| "loss": 9.2711, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00042036248182962185, | |
| "loss": 9.2726, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004212979816486783, | |
| "loss": 9.2621, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00042225444333247354, | |
| "loss": 9.2527, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004232318626854678, | |
| "loss": 9.2453, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004242302354201949, | |
| "loss": 9.2314, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000425249557157276, | |
| "loss": 9.2337, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00042628982342543184, | |
| "loss": 9.2276, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004273456715498305, | |
| "loss": 9.2181, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00042842770843401837, | |
| "loss": 9.2142, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004295306759082608, | |
| "loss": 9.2052, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00043065456913437584, | |
| "loss": 9.1994, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00043179938318238693, | |
| "loss": 9.2017, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004329651130305402, | |
| "loss": 9.1991, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004341517535653445, | |
| "loss": 9.1921, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00043535929958157804, | |
| "loss": 9.1786, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004365877457823183, | |
| "loss": 9.1766, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00043783708677896244, | |
| "loss": 9.1614, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004391073170912519, | |
| "loss": 9.1717, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004403984311473017, | |
| "loss": 9.1551, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004417104232836127, | |
| "loss": 9.1542, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044304328774510786, | |
| "loss": 9.1525, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000444397018685155, | |
| "loss": 9.1443, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044577161016558405, | |
| "loss": 9.1301, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004471670561567286, | |
| "loss": 9.1343, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044858335053743655, | |
| "loss": 9.1287, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004500204870951062, | |
| "loss": 9.1189, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045147845952571257, | |
| "loss": 9.1171, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004529498156216581, | |
| "loss": 9.1105, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004544493364218305, | |
| "loss": 9.0969, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045596967366771067, | |
| "loss": 9.1014, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045751082069031036, | |
| "loss": 9.0951, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045907277072936015, | |
| "loss": 9.0867, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00046065551693333547, | |
| "loss": 9.0872, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00046225905235949306, | |
| "loss": 9.0708, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004638833699738953, | |
| "loss": 9.0716, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00046552846265143777, | |
| "loss": 9.071, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00046719432317589814, | |
| "loss": 9.0618, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004688724595049813, | |
| "loss": 9.0518, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004705797299630679, | |
| "loss": 9.0442, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004723077461105934, | |
| "loss": 9.0477, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004740565003675777, | |
| "loss": 9.0397, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004758259850630858, | |
| "loss": 9.0355, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00047761619243523283, | |
| "loss": 9.0248, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004794271146312465, | |
| "loss": 9.0137, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00048125874370748105, | |
| "loss": 9.0205, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00048311107162946065, | |
| "loss": 9.0008, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00048498409027191575, | |
| "loss": 8.9975, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004868682714790542, | |
| "loss": 8.9992, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004887825434734695, | |
| "loss": 8.9777, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004907174813103439, | |
| "loss": 8.9871, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004926730765020346, | |
| "loss": 8.9765, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004946393879009196, | |
| "loss": 8.9754, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004966361687980866, | |
| "loss": 8.9678, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000498653581087638, | |
| "loss": 8.9677, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0005006916159201579, | |
| "loss": 8.9644, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005027502643557748, | |
| "loss": 8.9642, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005048295173641828, | |
| "loss": 8.9569, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005069293658247036, | |
| "loss": 8.9605, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005090498005263129, | |
| "loss": 8.9431, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000511190812167682, | |
| "loss": 8.9431, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000513352391357226, | |
| "loss": 8.9342, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005155345286131357, | |
| "loss": 8.9324, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005177372143634305, | |
| "loss": 8.9382, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005199604389459836, | |
| "loss": 8.9424, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005222041926085837, | |
| "loss": 8.9157, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005244684655089597, | |
| "loss": 8.9236, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005267532477148378, | |
| "loss": 8.9246, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005290585292039816, | |
| "loss": 8.9268, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005313842998642265, | |
| "loss": 8.9203, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005337305494935388, | |
| "loss": 8.9095, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005360972678000522, | |
| "loss": 8.9061, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005384724576463773, | |
| "loss": 8.9117, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005408799798596632, | |
| "loss": 8.9043, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005433079393890421, | |
| "loss": 8.9012, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005457563255842242, | |
| "loss": 8.8969, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005482251277053145, | |
| "loss": 8.9013, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005507143349228714, | |
| "loss": 8.8912, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005532239363179401, | |
| "loss": 8.8896, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005557539208821075, | |
| "loss": 8.8838, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005583042775175479, | |
| "loss": 8.889, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005608749950370764, | |
| "loss": 8.888, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005634530562276738, | |
| "loss": 8.8814, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005660643599338256, | |
| "loss": 8.8755, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005686959904843206, | |
| "loss": 8.8667, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005713479363354621, | |
| "loss": 8.8748, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000574020185854441, | |
| "loss": 8.862, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005767127273193853, | |
| "loss": 8.8534, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005794255489194114, | |
| "loss": 8.8655, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005821586387546804, | |
| "loss": 8.8574, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005849119848364386, | |
| "loss": 8.8531, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005876855750870848, | |
| "loss": 8.8479, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005904653779220791, | |
| "loss": 8.8405, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005932793188544346, | |
| "loss": 8.8435, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005961134672522114, | |
| "loss": 8.8425, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005989678106833648, | |
| "loss": 8.8389, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0006018423366272695, | |
| "loss": 8.8525, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0006047370324747583, | |
| "loss": 8.8273, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0006076518855281984, | |
| "loss": 8.8306, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000610586883001531, | |
| "loss": 8.8437, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000613542012020336, | |
| "loss": 8.8236, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0006165172596218869, | |
| "loss": 8.8274, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0006194975859987236, | |
| "loss": 8.8275, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006225129310954997, | |
| "loss": 8.8211, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006255483554239195, | |
| "loss": 8.8177, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000628603845669035, | |
| "loss": 8.8223, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006316793884278832, | |
| "loss": 8.8123, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006347749702095389, | |
| "loss": 8.8107, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006378905774351747, | |
| "loss": 8.8122, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006410261964381238, | |
| "loss": 8.811, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000644181813463934, | |
| "loss": 8.813, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006473574146704329, | |
| "loss": 8.8057, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006505369586176524, | |
| "loss": 8.8033, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006537523865622775, | |
| "loss": 8.795, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0006569877567060931, | |
| "loss": 8.7938, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006602430548570907, | |
| "loss": 8.7969, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000663518266735847, | |
| "loss": 8.7966, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006668133779755819, | |
| "loss": 8.7936, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006701283741222287, | |
| "loss": 8.7888, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006734632406344993, | |
| "loss": 8.7829, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006768179628839337, | |
| "loss": 8.7789, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006801925261549872, | |
| "loss": 8.778, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006835698944044951, | |
| "loss": 8.7897, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006869839962045932, | |
| "loss": 8.779, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006904178944328165, | |
| "loss": 8.7697, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0006938715740263026, | |
| "loss": 8.7818, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0006973450198354252, | |
| "loss": 8.7667, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007008382166238496, | |
| "loss": 8.7759, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007043511490686036, | |
| "loss": 8.7797, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007078838017601421, | |
| "loss": 8.7644, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007114361592024231, | |
| "loss": 8.7678, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007150082058129618, | |
| "loss": 8.7672, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007185819184105553, | |
| "loss": 8.7672, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007221750927446872, | |
| "loss": 8.7573, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007258059161614535, | |
| "loss": 8.7584, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007294563657132755, | |
| "loss": 8.7442, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007331264253873856, | |
| "loss": 8.7595, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007368160790850002, | |
| "loss": 8.7564, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007405253106213833, | |
| "loss": 8.7517, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007442541037259286, | |
| "loss": 8.7583, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007480024420422077, | |
| "loss": 8.7426, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007517703091280727, | |
| "loss": 8.7519, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007555576884556992, | |
| "loss": 8.7393, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007593645634116821, | |
| "loss": 8.7262, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007631717371015569, | |
| "loss": 8.7383, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007670174558631893, | |
| "loss": 8.7365, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007708826199846926, | |
| "loss": 8.7385, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007747672125114589, | |
| "loss": 8.7385, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007786712164036449, | |
| "loss": 8.7164, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007825946145362667, | |
| "loss": 8.7262, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007865373896992697, | |
| "loss": 8.728, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007904995245975929, | |
| "loss": 8.7281, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007944810018512619, | |
| "loss": 8.7179, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000798481803995452, | |
| "loss": 8.7264, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008025019134805696, | |
| "loss": 8.7229, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008065210677225022, | |
| "loss": 8.7246, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000810579642586285, | |
| "loss": 8.7199, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008146574717236045, | |
| "loss": 8.7209, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008187545372469861, | |
| "loss": 8.7075, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008228708211845768, | |
| "loss": 8.7101, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0008270063054802209, | |
| "loss": 8.7144, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008311609719935404, | |
| "loss": 8.7173, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008353348025000144, | |
| "loss": 8.7183, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008395277786910574, | |
| "loss": 8.7107, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008437398821741025, | |
| "loss": 8.7113, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008479710944726774, | |
| "loss": 8.7085, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008522000980575213, | |
| "loss": 8.7115, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008564693769110079, | |
| "loss": 8.7055, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008607577087418623, | |
| "loss": 8.6935, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008650650747392373, | |
| "loss": 8.7042, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008693914560087938, | |
| "loss": 8.6849, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008737368335727785, | |
| "loss": 8.6876, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0008781011883701138, | |
| "loss": 8.6922, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0008824845012564749, | |
| "loss": 8.6922, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000886886753004381, | |
| "loss": 8.6853, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000891307924303272, | |
| "loss": 8.6936, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0008957257484203587, | |
| "loss": 8.6995, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009001846062028449, | |
| "loss": 8.6841, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009046623252050388, | |
| "loss": 8.6735, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009091588857853411, | |
| "loss": 8.6888, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009136742682195071, | |
| "loss": 8.6788, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009182084527007278, | |
| "loss": 8.6817, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009227614193397203, | |
| "loss": 8.671, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009273331481648092, | |
| "loss": 8.6738, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009319236191220222, | |
| "loss": 8.6657, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009365328120751783, | |
| "loss": 8.6599, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009411375208451828, | |
| "loss": 8.6747, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009457840036964961, | |
| "loss": 8.6698, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009504491477449178, | |
| "loss": 8.6765, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009551329325267026, | |
| "loss": 8.6732, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009598353374963477, | |
| "loss": 8.6654, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009645563420266623, | |
| "loss": 8.6614, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009692959254088748, | |
| "loss": 8.6672, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009740540668527146, | |
| "loss": 8.6508, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009788307454865058, | |
| "loss": 8.6533, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009836259403572592, | |
| "loss": 8.656, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009884155160084767, | |
| "loss": 8.65, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009932475878516138, | |
| "loss": 8.6559, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009980981126919714, | |
| "loss": 8.646, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010029670692526266, | |
| "loss": 8.6504, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.001007854436175815, | |
| "loss": 8.6437, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00101276019202301, | |
| "loss": 8.6483, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010176843152750244, | |
| "loss": 8.6381, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.001022602026387454, | |
| "loss": 8.6419, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010275627280027944, | |
| "loss": 8.6381, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010325417320913577, | |
| "loss": 8.6274, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010375390168126473, | |
| "loss": 8.6264, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0010425545602459826, | |
| "loss": 8.6279, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010475883403905893, | |
| "loss": 8.636, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.001052640335165696, | |
| "loss": 8.6305, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.001057710522410639, | |
| "loss": 8.6259, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.001062798879884943, | |
| "loss": 8.6288, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010679053852684361, | |
| "loss": 8.6286, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010730300161613388, | |
| "loss": 8.6203, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010781469914207427, | |
| "loss": 8.618, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010833077154690767, | |
| "loss": 8.6183, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010884864974642153, | |
| "loss": 8.6244, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0010936833146893334, | |
| "loss": 8.6129, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.001098898144348496, | |
| "loss": 8.622, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001104130963566756, | |
| "loss": 8.6104, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001109381749390256, | |
| "loss": 8.603, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00111465047878633, | |
| "loss": 8.6093, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001119937128643592, | |
| "loss": 8.5969, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0011252416757720606, | |
| "loss": 8.5992, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0011305374403745901, | |
| "loss": 8.6047, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001135877622966507, | |
| "loss": 8.5958, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001141235632906355, | |
| "loss": 8.5948, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0011466114466911256, | |
| "loss": 8.5896, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001152005040739713, | |
| "loss": 8.5887, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0011574163913930131, | |
| "loss": 8.5862, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0011628454749140395, | |
| "loss": 8.5949, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011682922674880192, | |
| "loss": 8.588, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011737567452224911, | |
| "loss": 8.5918, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011792388841474245, | |
| "loss": 8.5904, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011847111175024606, | |
| "loss": 8.5739, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011902284185834888, | |
| "loss": 8.5756, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0011957633086016797, | |
| "loss": 8.568, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0012013157632781366, | |
| "loss": 8.5696, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00120688575825691, | |
| "loss": 8.5768, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0012124732691051188, | |
| "loss": 8.5696, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0012180782713130424, | |
| "loss": 8.5687, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0012237007402942333, | |
| "loss": 8.56, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0012293406513856284, | |
| "loss": 8.56, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012349979798476525, | |
| "loss": 8.5602, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012406442840364133, | |
| "loss": 8.5551, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012463362859392122, | |
| "loss": 8.5556, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012520456306609733, | |
| "loss": 8.5508, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012577722931575563, | |
| "loss": 8.549, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012634874855592566, | |
| "loss": 8.5582, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.001269248621894795, | |
| "loss": 8.554, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012750270005439136, | |
| "loss": 8.5453, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012808225961596451, | |
| "loss": 8.5545, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012866353833195041, | |
| "loss": 8.543, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0012924653365255934, | |
| "loss": 8.5454, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.001298312430204715, | |
| "loss": 8.5496, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013041766387084808, | |
| "loss": 8.5348, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013100579363134381, | |
| "loss": 8.5311, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.001315956297221161, | |
| "loss": 8.5378, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013218716955583822, | |
| "loss": 8.5304, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013278041053770978, | |
| "loss": 8.5166, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013337237114701053, | |
| "loss": 8.5324, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.001339689981377656, | |
| "loss": 8.5196, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013456731846064624, | |
| "loss": 8.5191, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013516732949110932, | |
| "loss": 8.5285, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013576902859719474, | |
| "loss": 8.5143, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0013637241313953895, | |
| "loss": 8.519, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0013697748047138431, | |
| "loss": 8.4964, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0013758422793859176, | |
| "loss": 8.5216, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.001381926528796519, | |
| "loss": 8.5098, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0013880275262569807, | |
| "loss": 8.511, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0013941452450051631, | |
| "loss": 8.5124, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.001400248944656608, | |
| "loss": 8.5054, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0014063999421298785, | |
| "loss": 8.5016, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0014125675802999262, | |
| "loss": 8.501, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.001418751832112295, | |
| "loss": 8.505, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0014249526704396467, | |
| "loss": 8.4953, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0014311700680818915, | |
| "loss": 8.5074, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014374039977662987, | |
| "loss": 8.4991, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014436544321476206, | |
| "loss": 8.5087, | |
| "step": 64200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014499213438082127, | |
| "loss": 8.4871, | |
| "step": 64400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014562047052581514, | |
| "loss": 8.4954, | |
| "step": 64600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014624729492123557, | |
| "loss": 8.4791, | |
| "step": 64800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014687890455785963, | |
| "loss": 8.4853, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014751215089706584, | |
| "loss": 8.4855, | |
| "step": 65200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0014814703116110776, | |
| "loss": 8.4707, | |
| "step": 65400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.001487835425650709, | |
| "loss": 8.4743, | |
| "step": 65600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.001494216823168866, | |
| "loss": 8.4717, | |
| "step": 65800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0015006144761734279, | |
| "loss": 8.4823, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.001507028356600975, | |
| "loss": 8.4708, | |
| "step": 66200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015134584363168998, | |
| "loss": 8.4649, | |
| "step": 66400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.001519904687115537, | |
| "loss": 8.4695, | |
| "step": 66600 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015263347286438994, | |
| "loss": 8.4759, | |
| "step": 66800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015328131562056986, | |
| "loss": 8.4655, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015393076699503766, | |
| "loss": 8.4752, | |
| "step": 67200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015458182413896245, | |
| "loss": 8.4535, | |
| "step": 67400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.001552344841964707, | |
| "loss": 8.4535, | |
| "step": 67600 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015588874430465648, | |
| "loss": 8.4519, | |
| "step": 67800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.001565446015935959, | |
| "loss": 8.4568, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00157202053186359, | |
| "loss": 8.4524, | |
| "step": 68200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015786109619902212, | |
| "loss": 8.4589, | |
| "step": 68400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0015852172774068075, | |
| "loss": 8.4559, | |
| "step": 68600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0015918062988814347, | |
| "loss": 8.446, | |
| "step": 68800 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0015984442188082624, | |
| "loss": 8.439, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016050979370261006, | |
| "loss": 8.4504, | |
| "step": 69200 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016117674243482875, | |
| "loss": 8.4487, | |
| "step": 69400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016184526515189961, | |
| "loss": 8.4472, | |
| "step": 69600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016251535892133542, | |
| "loss": 8.433, | |
| "step": 69800 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.001631870208037572, | |
| "loss": 8.426, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016386024785290804, | |
| "loss": 8.4311, | |
| "step": 70200 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016453503711566474, | |
| "loss": 8.432, | |
| "step": 70400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016520800001573153, | |
| "loss": 8.4282, | |
| "step": 70600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0016588589704489114, | |
| "loss": 8.4386, | |
| "step": 70800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016656534740210893, | |
| "loss": 8.4158, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016724634810696363, | |
| "loss": 8.4242, | |
| "step": 71200 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016792889617223312, | |
| "loss": 8.4279, | |
| "step": 71400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016861298860390735, | |
| "loss": 8.4242, | |
| "step": 71600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016929862240120247, | |
| "loss": 8.4271, | |
| "step": 71800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0016998579455657307, | |
| "loss": 8.4265, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017067450205572581, | |
| "loss": 8.421, | |
| "step": 72200 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017136474187763266, | |
| "loss": 8.4156, | |
| "step": 72400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017205304834985446, | |
| "loss": 8.421, | |
| "step": 72600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017274633610356825, | |
| "loss": 8.4256, | |
| "step": 72800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017344114709189774, | |
| "loss": 8.4191, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0017413747826704132, | |
| "loss": 8.4015, | |
| "step": 73200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.001748318335641869, | |
| "loss": 8.4129, | |
| "step": 73400 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017553118838016506, | |
| "loss": 8.4179, | |
| "step": 73600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017623205421495314, | |
| "loss": 8.4142, | |
| "step": 73800 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017693442799418986, | |
| "loss": 8.4005, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017763830663689965, | |
| "loss": 8.41, | |
| "step": 74200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017834368705550597, | |
| "loss": 8.4162, | |
| "step": 74400 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.001790505661558443, | |
| "loss": 8.4081, | |
| "step": 74600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0017975894083717692, | |
| "loss": 8.4027, | |
| "step": 74800 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0018046880799220469, | |
| "loss": 8.4097, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0018118016450708232, | |
| "loss": 8.4077, | |
| "step": 75200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0018189300726143137, | |
| "loss": 8.4086, | |
| "step": 75400 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0018260375781495742, | |
| "loss": 8.4084, | |
| "step": 75600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.001833195562689592, | |
| "loss": 8.4166, | |
| "step": 75800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018403683157795104, | |
| "loss": 8.4121, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018475558059559121, | |
| "loss": 8.404, | |
| "step": 76200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.001854758001690741, | |
| "loss": 8.4035, | |
| "step": 76400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018619748713914318, | |
| "loss": 8.4044, | |
| "step": 76600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018692063834010522, | |
| "loss": 8.3948, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018764525059984417, | |
| "loss": 8.3935, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.001883713207398349, | |
| "loss": 8.4028, | |
| "step": 77200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018909884557515733, | |
| "loss": 8.401, | |
| "step": 77400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0018982417342748425, | |
| "loss": 8.4085, | |
| "step": 77600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0019055459083963232, | |
| "loss": 8.3979, | |
| "step": 77800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00191282790468081, | |
| "loss": 8.4027, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019201608770511077, | |
| "loss": 8.401, | |
| "step": 78200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.001927508236496343, | |
| "loss": 8.4054, | |
| "step": 78400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019348699507871943, | |
| "loss": 8.408, | |
| "step": 78600 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019422459876313608, | |
| "loss": 8.4064, | |
| "step": 78800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019496363146737205, | |
| "loss": 8.4066, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.001957040899496469, | |
| "loss": 8.4061, | |
| "step": 79200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019644597096192574, | |
| "loss": 8.411, | |
| "step": 79400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00197189271249934, | |
| "loss": 8.3999, | |
| "step": 79600 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.001979339875531708, | |
| "loss": 8.4002, | |
| "step": 79800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.001986801166049247, | |
| "loss": 8.4058, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019942391393900083, | |
| "loss": 8.4172, | |
| "step": 80200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002001728516400637, | |
| "loss": 8.4089, | |
| "step": 80400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002009231922688247, | |
| "loss": 8.4087, | |
| "step": 80600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002016749325339009, | |
| "loss": 8.4115, | |
| "step": 80800 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020242806913776997, | |
| "loss": 8.4317, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020318259877678373, | |
| "loss": 8.4049, | |
| "step": 81200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002039385181411845, | |
| "loss": 8.4129, | |
| "step": 81400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002046958239151178, | |
| "loss": 8.424, | |
| "step": 81600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020545451277664776, | |
| "loss": 8.4146, | |
| "step": 81800 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020621458139777164, | |
| "loss": 8.4233, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.002069722158008656, | |
| "loss": 8.4199, | |
| "step": 82200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020773502707586607, | |
| "loss": 8.4134, | |
| "step": 82400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0020849920810694245, | |
| "loss": 8.415, | |
| "step": 82600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0020926475554200047, | |
| "loss": 8.4301, | |
| "step": 82800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0021003166602295217, | |
| "loss": 8.4128, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002107999361857309, | |
| "loss": 8.4284, | |
| "step": 83200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002115695626603048, | |
| "loss": 8.422, | |
| "step": 83400 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002123405420706933, | |
| "loss": 8.4173, | |
| "step": 83600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002131128710349813, | |
| "loss": 8.4245, | |
| "step": 83800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002138865461653332, | |
| "loss": 8.427, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0021465768564397046, | |
| "loss": 8.4287, | |
| "step": 84200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002154340362309423, | |
| "loss": 8.4361, | |
| "step": 84400 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.002162078310516678, | |
| "loss": 8.4299, | |
| "step": 84600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0021698684354139377, | |
| "loss": 8.4447, | |
| "step": 84800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0021776718520393184, | |
| "loss": 8.4399, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0021854885261629875, | |
| "loss": 8.4469, | |
| "step": 85200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0021933184234969594, | |
| "loss": 8.4328, | |
| "step": 85400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022011615096952444, | |
| "loss": 8.4504, | |
| "step": 85600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.002209017750354, | |
| "loss": 8.4383, | |
| "step": 85800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022168871110116815, | |
| "loss": 8.4472, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022247695571491945, | |
| "loss": 8.448, | |
| "step": 86200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022326650541900405, | |
| "loss": 8.4451, | |
| "step": 86400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022405339926133165, | |
| "loss": 8.4415, | |
| "step": 86600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022484554226809986, | |
| "loss": 8.4439, | |
| "step": 86800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022563897997535266, | |
| "loss": 8.4497, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0022643370890266133, | |
| "loss": 8.452, | |
| "step": 87200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0022722972556393217, | |
| "loss": 8.4677, | |
| "step": 87400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0022802702646742383, | |
| "loss": 8.4595, | |
| "step": 87600 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002288256081157608, | |
| "loss": 8.4595, | |
| "step": 87800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002296254670059502, | |
| "loss": 8.4642, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0023042659962939603, | |
| "loss": 8.4709, | |
| "step": 88200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0023122900247191545, | |
| "loss": 8.4679, | |
| "step": 88400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002320286505209589, | |
| "loss": 8.4603, | |
| "step": 88600 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0023283357692971242, | |
| "loss": 8.4662, | |
| "step": 88800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002336397629992889, | |
| "loss": 8.4653, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002344472051933384, | |
| "loss": 8.4833, | |
| "step": 89200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.002352558999700007, | |
| "loss": 8.4974, | |
| "step": 89400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.002360658437819213, | |
| "loss": 8.4881, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0023687703307626647, | |
| "loss": 8.4878, | |
| "step": 89800 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0023768946429473976, | |
| "loss": 8.4846, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.002385031338735963, | |
| "loss": 8.4866, | |
| "step": 90200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0023931803824365962, | |
| "loss": 8.4847, | |
| "step": 90400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.002401300900956714, | |
| "loss": 8.4934, | |
| "step": 90600 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.002409474471896992, | |
| "loss": 8.4872, | |
| "step": 90800 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0024176602835290807, | |
| "loss": 8.4977, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0024258582999457665, | |
| "loss": 8.4967, | |
| "step": 91200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0024340684851863, | |
| "loss": 8.505, | |
| "step": 91400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.002442290803236551, | |
| "loss": 8.5126, | |
| "step": 91600 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0024505252180291688, | |
| "loss": 8.5033, | |
| "step": 91800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0024587304311256865, | |
| "loss": 8.513, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0024669888709567232, | |
| "loss": 8.5082, | |
| "step": 92200 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0024752592991915973, | |
| "loss": 8.517, | |
| "step": 92400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0024835416795519205, | |
| "loss": 8.5293, | |
| "step": 92600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.002491835975706881, | |
| "loss": 8.5094, | |
| "step": 92800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0025001421512733943, | |
| "loss": 8.5139, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0025084601698162666, | |
| "loss": 8.5099, | |
| "step": 93200 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0025167899948483575, | |
| "loss": 8.5185, | |
| "step": 93400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0025251315898307336, | |
| "loss": 8.5143, | |
| "step": 93600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.002533484918172837, | |
| "loss": 8.5277, | |
| "step": 93800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0025418499432326358, | |
| "loss": 8.5231, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.002550184715947826, | |
| "loss": 8.5436, | |
| "step": 94200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0025585729662869474, | |
| "loss": 8.5373, | |
| "step": 94400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002566972803294579, | |
| "loss": 8.5347, | |
| "step": 94600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00257538419012468, | |
| "loss": 8.5544, | |
| "step": 94800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0025838070898805453, | |
| "loss": 8.5339, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002592241465614974, | |
| "loss": 8.5405, | |
| "step": 95200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002600687280330416, | |
| "loss": 8.5501, | |
| "step": 95400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0026091444969791513, | |
| "loss": 8.5344, | |
| "step": 95600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002617613078463441, | |
| "loss": 8.5477, | |
| "step": 95800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002626092987635699, | |
| "loss": 8.5443, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002634541703276827, | |
| "loss": 8.5398, | |
| "step": 96200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.002643044100010169, | |
| "loss": 8.5523, | |
| "step": 96400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002651557712877833, | |
| "loss": 8.5562, | |
| "step": 96600 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0026600825045346955, | |
| "loss": 8.5525, | |
| "step": 96800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0026686184375866043, | |
| "loss": 8.5728, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002677165474590528, | |
| "loss": 8.5631, | |
| "step": 97200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002685723578054729, | |
| "loss": 8.5658, | |
| "step": 97400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0026942927104389334, | |
| "loss": 8.566, | |
| "step": 97600 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002702872834154482, | |
| "loss": 8.5716, | |
| "step": 97800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0027114639115645017, | |
| "loss": 8.5697, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002720022867925799, | |
| "loss": 8.5726, | |
| "step": 98200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0027286356853246747, | |
| "loss": 8.5718, | |
| "step": 98400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0027372593434088002, | |
| "loss": 8.5716, | |
| "step": 98600 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.002745893804350339, | |
| "loss": 8.5767, | |
| "step": 98800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00275453903027407, | |
| "loss": 8.5957, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0027631949832575475, | |
| "loss": 8.5881, | |
| "step": 99200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.002771861625331276, | |
| "loss": 8.5835, | |
| "step": 99400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.002780495505581529, | |
| "loss": 8.5905, | |
| "step": 99600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.002789183358769584, | |
| "loss": 8.5938, | |
| "step": 99800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0027978817870494, | |
| "loss": 8.5906, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0028065907522651585, | |
| "loss": 8.5938, | |
| "step": 100200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.002815310216214826, | |
| "loss": 8.5887, | |
| "step": 100400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00282404014065031, | |
| "loss": 8.5922, | |
| "step": 100600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0028327804872776367, | |
| "loss": 8.5926, | |
| "step": 100800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.002841531217757113, | |
| "loss": 8.5978, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0028502922937035, | |
| "loss": 8.5984, | |
| "step": 101200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0028590636766861726, | |
| "loss": 8.6046, | |
| "step": 101400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0028678453282293013, | |
| "loss": 8.6093, | |
| "step": 101600 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0028766372098120076, | |
| "loss": 8.6083, | |
| "step": 101800 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0028854392828685377, | |
| "loss": 8.6057, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0028942515087884407, | |
| "loss": 8.6146, | |
| "step": 102200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00290307384891672, | |
| "loss": 8.608, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00291190626455402, | |
| "loss": 8.6081, | |
| "step": 102600 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0029207044797924615, | |
| "loss": 8.6164, | |
| "step": 102800 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0029295568802797795, | |
| "loss": 8.6008, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0029384192401078115, | |
| "loss": 8.6166, | |
| "step": 103200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00294729152040165, | |
| "loss": 8.5962, | |
| "step": 103400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.002956173682242877, | |
| "loss": 8.6129, | |
| "step": 103600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.002965065686669722, | |
| "loss": 8.6092, | |
| "step": 103800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0029739674946772463, | |
| "loss": 8.6189, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.002982879067217503, | |
| "loss": 8.612, | |
| "step": 104200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0029918003651997144, | |
| "loss": 8.6135, | |
| "step": 104400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.003000731349490442, | |
| "loss": 8.6182, | |
| "step": 104600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0030096719809137584, | |
| "loss": 8.6423, | |
| "step": 104800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.003018622220251419, | |
| "loss": 8.6145, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0030275372054660438, | |
| "loss": 8.6249, | |
| "step": 105200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0030365064952603237, | |
| "loss": 8.6265, | |
| "step": 105400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0030454852752588536, | |
| "loss": 8.6304, | |
| "step": 105600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0030544735060760494, | |
| "loss": 8.6309, | |
| "step": 105800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0030634711482848704, | |
| "loss": 8.6258, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003072478162416994, | |
| "loss": 8.6328, | |
| "step": 106200 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003081494508962985, | |
| "loss": 8.6298, | |
| "step": 106400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0030905201483724717, | |
| "loss": 8.639, | |
| "step": 106600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0030995550410543226, | |
| "loss": 8.6212, | |
| "step": 106800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0031085991473768114, | |
| "loss": 8.6374, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003117652427667799, | |
| "loss": 8.6326, | |
| "step": 107200 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0031267148422149046, | |
| "loss": 8.6291, | |
| "step": 107400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003135740971163656, | |
| "loss": 8.6375, | |
| "step": 107600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0031448214897512507, | |
| "loss": 8.6226, | |
| "step": 107800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003153911023417371, | |
| "loss": 8.6359, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.003163009532290608, | |
| "loss": 8.6491, | |
| "step": 108200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0031721169764601844, | |
| "loss": 8.6405, | |
| "step": 108400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0031812333159761293, | |
| "loss": 8.632, | |
| "step": 108600 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.003190358510849451, | |
| "loss": 8.6363, | |
| "step": 108800 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0031994925210523124, | |
| "loss": 8.6316, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0032086353065182106, | |
| "loss": 8.6423, | |
| "step": 109200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.003217786827142146, | |
| "loss": 8.6274, | |
| "step": 109400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.003226947042780804, | |
| "loss": 8.6366, | |
| "step": 109600 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.003236070047437989, | |
| "loss": 8.6388, | |
| "step": 109800 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.003245247489550804, | |
| "loss": 8.6364, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0032544335062216403, | |
| "loss": 8.6422, | |
| "step": 110200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0032636280571558636, | |
| "loss": 8.618, | |
| "step": 110400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.003272831102021408, | |
| "loss": 8.6276, | |
| "step": 110600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.003282042600448948, | |
| "loss": 8.6454, | |
| "step": 110800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0032912625120320753, | |
| "loss": 8.6388, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033004907963274733, | |
| "loss": 8.6339, | |
| "step": 111200 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.003309727412855108, | |
| "loss": 8.6243, | |
| "step": 111400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033189723210983865, | |
| "loss": 8.6264, | |
| "step": 111600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033282254805043487, | |
| "loss": 8.6401, | |
| "step": 111800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.003337440523277331, | |
| "loss": 8.6366, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033467100224565524, | |
| "loss": 8.6338, | |
| "step": 112200 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.003355987651126521, | |
| "loss": 8.6377, | |
| "step": 112400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033652733685907424, | |
| "loss": 8.6414, | |
| "step": 112600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0033745671341172496, | |
| "loss": 8.6264, | |
| "step": 112800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0033838689069387654, | |
| "loss": 8.6289, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00339317864625289, | |
| "loss": 8.6244, | |
| "step": 113200 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.003402496311222283, | |
| "loss": 8.6287, | |
| "step": 113400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0034118218609748346, | |
| "loss": 8.6251, | |
| "step": 113600 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.003421155254603846, | |
| "loss": 8.6214, | |
| "step": 113800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0034304964511682147, | |
| "loss": 8.6303, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0034398454096926092, | |
| "loss": 8.6369, | |
| "step": 114200 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.003449202089167651, | |
| "loss": 8.6236, | |
| "step": 114400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0034585196077173436, | |
| "loss": 8.6251, | |
| "step": 114600 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.003467891567838331, | |
| "loss": 8.6295, | |
| "step": 114800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.003477271125884973, | |
| "loss": 8.6219, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0034866582407136653, | |
| "loss": 8.6271, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.003496052871147656, | |
| "loss": 8.6372, | |
| "step": 115400 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035054549759772242, | |
| "loss": 8.6238, | |
| "step": 115600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035148645139598637, | |
| "loss": 8.6207, | |
| "step": 115800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035242814438204637, | |
| "loss": 8.6099, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035337057242514833, | |
| "loss": 8.6142, | |
| "step": 116200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035431373139131472, | |
| "loss": 8.6033, | |
| "step": 116400 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035525761714336104, | |
| "loss": 8.6178, | |
| "step": 116600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035619750070819923, | |
| "loss": 8.6138, | |
| "step": 116800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035714282402552104, | |
| "loss": 8.6143, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035808886171885554, | |
| "loss": 8.6034, | |
| "step": 117200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0035903560963839124, | |
| "loss": 8.6156, | |
| "step": 117400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0035998306363120057, | |
| "loss": 8.6148, | |
| "step": 117600 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036093121954125906, | |
| "loss": 8.6039, | |
| "step": 117800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.003618800732094636, | |
| "loss": 8.6107, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036282962047364973, | |
| "loss": 8.6094, | |
| "step": 118200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036377985716861084, | |
| "loss": 8.616, | |
| "step": 118400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.003647307791261164, | |
| "loss": 8.6135, | |
| "step": 118600 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.003656823821749292, | |
| "loss": 8.6062, | |
| "step": 118800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036662989906407328, | |
| "loss": 8.6029, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036758284841655496, | |
| "loss": 8.6011, | |
| "step": 119200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0036853646634968946, | |
| "loss": 8.5993, | |
| "step": 119400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.003694907486804143, | |
| "loss": 8.6029, | |
| "step": 119600 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00370445691222752, | |
| "loss": 8.6018, | |
| "step": 119800 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.003714012897878298, | |
| "loss": 8.5978, | |
| "step": 120000 | |
| } | |
| ], | |
| "max_steps": 1000000, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.9125959786496e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |