| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.99835255354201, | |
| "eval_steps": 500, | |
| "global_step": 1365, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.002196595277320154, | |
| "grad_norm": 1.747358798980713, | |
| "learning_rate": 2.18978102189781e-06, | |
| "loss": 3.0822, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004393190554640308, | |
| "grad_norm": 1.2177706956863403, | |
| "learning_rate": 4.37956204379562e-06, | |
| "loss": 2.7456, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006589785831960461, | |
| "grad_norm": 1.1733002662658691, | |
| "learning_rate": 6.56934306569343e-06, | |
| "loss": 2.4922, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.008786381109280615, | |
| "grad_norm": 1.492933750152588, | |
| "learning_rate": 8.75912408759124e-06, | |
| "loss": 2.1144, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.010982976386600769, | |
| "grad_norm": 1.3570741415023804, | |
| "learning_rate": 1.0948905109489049e-05, | |
| "loss": 2.6946, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.013179571663920923, | |
| "grad_norm": 1.5913143157958984, | |
| "learning_rate": 1.313868613138686e-05, | |
| "loss": 3.2832, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.015376166941241077, | |
| "grad_norm": 13.426933288574219, | |
| "learning_rate": 1.532846715328467e-05, | |
| "loss": 2.723, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01757276221856123, | |
| "grad_norm": 0.9860666394233704, | |
| "learning_rate": 1.751824817518248e-05, | |
| "loss": 2.1906, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.019769357495881382, | |
| "grad_norm": 1.4485764503479004, | |
| "learning_rate": 1.9708029197080292e-05, | |
| "loss": 2.8082, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.021965952773201538, | |
| "grad_norm": 1.2479782104492188, | |
| "learning_rate": 2.1897810218978098e-05, | |
| "loss": 2.5686, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02416254805052169, | |
| "grad_norm": 37.56756591796875, | |
| "learning_rate": 2.408759124087591e-05, | |
| "loss": 3.056, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.026359143327841845, | |
| "grad_norm": 1.0153371095657349, | |
| "learning_rate": 2.627737226277372e-05, | |
| "loss": 2.6569, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.028555738605161998, | |
| "grad_norm": 2.357285499572754, | |
| "learning_rate": 2.846715328467153e-05, | |
| "loss": 3.4058, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.030752333882482153, | |
| "grad_norm": 1.858598232269287, | |
| "learning_rate": 3.065693430656934e-05, | |
| "loss": 3.1293, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.032948929159802305, | |
| "grad_norm": 1.9496299028396606, | |
| "learning_rate": 3.284671532846715e-05, | |
| "loss": 2.4733, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03514552443712246, | |
| "grad_norm": 1.9457651376724243, | |
| "learning_rate": 3.503649635036496e-05, | |
| "loss": 3.1546, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.037342119714442616, | |
| "grad_norm": 4.610772132873535, | |
| "learning_rate": 3.722627737226277e-05, | |
| "loss": 2.3109, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.039538714991762765, | |
| "grad_norm": 2.0946030616760254, | |
| "learning_rate": 3.9416058394160584e-05, | |
| "loss": 3.0454, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04173531026908292, | |
| "grad_norm": 1.0136306285858154, | |
| "learning_rate": 4.160583941605839e-05, | |
| "loss": 2.3315, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.043931905546403076, | |
| "grad_norm": 2.042799234390259, | |
| "learning_rate": 4.3795620437956196e-05, | |
| "loss": 2.9462, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04612850082372323, | |
| "grad_norm": 1.8854206800460815, | |
| "learning_rate": 4.5985401459854016e-05, | |
| "loss": 2.6763, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04832509610104338, | |
| "grad_norm": 4.8470659255981445, | |
| "learning_rate": 4.817518248175182e-05, | |
| "loss": 2.392, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.050521691378363535, | |
| "grad_norm": 2.681057929992676, | |
| "learning_rate": 5.036496350364963e-05, | |
| "loss": 3.3371, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05271828665568369, | |
| "grad_norm": 1.8750619888305664, | |
| "learning_rate": 5.255474452554744e-05, | |
| "loss": 2.7932, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.054914881933003847, | |
| "grad_norm": 3.679814577102661, | |
| "learning_rate": 5.4744525547445253e-05, | |
| "loss": 3.7707, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.057111477210323995, | |
| "grad_norm": 1.7441668510437012, | |
| "learning_rate": 5.693430656934306e-05, | |
| "loss": 2.4032, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05930807248764415, | |
| "grad_norm": 2.74934458732605, | |
| "learning_rate": 5.912408759124087e-05, | |
| "loss": 2.5579, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.061504667764964306, | |
| "grad_norm": 2.1697113513946533, | |
| "learning_rate": 6.131386861313868e-05, | |
| "loss": 2.2688, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06370126304228446, | |
| "grad_norm": 2.6786093711853027, | |
| "learning_rate": 6.350364963503648e-05, | |
| "loss": 2.825, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06589785831960461, | |
| "grad_norm": 25.93538475036621, | |
| "learning_rate": 6.56934306569343e-05, | |
| "loss": 2.4854, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06809445359692477, | |
| "grad_norm": 2.770275115966797, | |
| "learning_rate": 6.788321167883211e-05, | |
| "loss": 2.9208, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07029104887424492, | |
| "grad_norm": 3.3287343978881836, | |
| "learning_rate": 7.007299270072992e-05, | |
| "loss": 3.2447, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07248764415156507, | |
| "grad_norm": 2.1245789527893066, | |
| "learning_rate": 7.226277372262774e-05, | |
| "loss": 2.6584, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07468423942888523, | |
| "grad_norm": 3.4754011631011963, | |
| "learning_rate": 7.445255474452554e-05, | |
| "loss": 2.2553, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07688083470620538, | |
| "grad_norm": 3.5811989307403564, | |
| "learning_rate": 7.664233576642336e-05, | |
| "loss": 2.8118, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07907742998352553, | |
| "grad_norm": 2.6096246242523193, | |
| "learning_rate": 7.883211678832117e-05, | |
| "loss": 2.1184, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08127402526084569, | |
| "grad_norm": 5.642367362976074, | |
| "learning_rate": 8.102189781021897e-05, | |
| "loss": 2.4832, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08347062053816584, | |
| "grad_norm": 2.8767082691192627, | |
| "learning_rate": 8.321167883211678e-05, | |
| "loss": 2.3821, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.085667215815486, | |
| "grad_norm": 2.144075870513916, | |
| "learning_rate": 8.540145985401459e-05, | |
| "loss": 2.2968, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08786381109280615, | |
| "grad_norm": 2.516916275024414, | |
| "learning_rate": 8.759124087591239e-05, | |
| "loss": 2.0048, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0900604063701263, | |
| "grad_norm": 2.992218255996704, | |
| "learning_rate": 8.978102189781021e-05, | |
| "loss": 2.3517, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09225700164744646, | |
| "grad_norm": 2.995621919631958, | |
| "learning_rate": 9.197080291970803e-05, | |
| "loss": 2.492, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09445359692476661, | |
| "grad_norm": 3.145153284072876, | |
| "learning_rate": 9.416058394160584e-05, | |
| "loss": 1.8786, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.09665019220208676, | |
| "grad_norm": 3.090636730194092, | |
| "learning_rate": 9.635036496350364e-05, | |
| "loss": 2.1675, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.09884678747940692, | |
| "grad_norm": 9.894902229309082, | |
| "learning_rate": 9.854014598540145e-05, | |
| "loss": 2.047, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10104338275672707, | |
| "grad_norm": 1.908050298690796, | |
| "learning_rate": 0.00010072992700729926, | |
| "loss": 1.8293, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.10323997803404723, | |
| "grad_norm": 4.571338176727295, | |
| "learning_rate": 0.00010291970802919708, | |
| "loss": 2.2743, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10543657331136738, | |
| "grad_norm": 2.4507877826690674, | |
| "learning_rate": 0.00010510948905109488, | |
| "loss": 2.2364, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.10763316858868753, | |
| "grad_norm": 2.7809364795684814, | |
| "learning_rate": 0.00010729927007299269, | |
| "loss": 2.1512, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.10982976386600769, | |
| "grad_norm": 2.3399386405944824, | |
| "learning_rate": 0.00010948905109489051, | |
| "loss": 1.8232, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11202635914332784, | |
| "grad_norm": 2.9450507164001465, | |
| "learning_rate": 0.00011167883211678831, | |
| "loss": 2.0175, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11422295442064799, | |
| "grad_norm": 3.1694955825805664, | |
| "learning_rate": 0.00011386861313868612, | |
| "loss": 1.6346, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11641954969796815, | |
| "grad_norm": 3.0246846675872803, | |
| "learning_rate": 0.00011605839416058394, | |
| "loss": 1.8694, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.1186161449752883, | |
| "grad_norm": 2.406731128692627, | |
| "learning_rate": 0.00011824817518248174, | |
| "loss": 1.5345, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12081274025260846, | |
| "grad_norm": 7.692804336547852, | |
| "learning_rate": 0.00012043795620437955, | |
| "loss": 1.8612, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.12300933552992861, | |
| "grad_norm": 5.040726184844971, | |
| "learning_rate": 0.00012262773722627736, | |
| "loss": 1.9258, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12520593080724876, | |
| "grad_norm": 1.6690394878387451, | |
| "learning_rate": 0.00012481751824817516, | |
| "loss": 1.4422, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.12740252608456892, | |
| "grad_norm": 4.254453659057617, | |
| "learning_rate": 0.00012700729927007297, | |
| "loss": 2.0041, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.12959912136188906, | |
| "grad_norm": 2.6219050884246826, | |
| "learning_rate": 0.00012919708029197077, | |
| "loss": 1.5277, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13179571663920922, | |
| "grad_norm": 3.4418587684631348, | |
| "learning_rate": 0.0001313868613138686, | |
| "loss": 1.5139, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13399231191652938, | |
| "grad_norm": 2.1308038234710693, | |
| "learning_rate": 0.00013357664233576641, | |
| "loss": 1.5121, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.13618890719384955, | |
| "grad_norm": 2.175471544265747, | |
| "learning_rate": 0.00013576642335766422, | |
| "loss": 1.3398, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.13838550247116968, | |
| "grad_norm": 3.6143200397491455, | |
| "learning_rate": 0.00013795620437956203, | |
| "loss": 1.3717, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14058209774848984, | |
| "grad_norm": 2.2427210807800293, | |
| "learning_rate": 0.00014014598540145983, | |
| "loss": 1.4886, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14277869302581, | |
| "grad_norm": 4.4549102783203125, | |
| "learning_rate": 0.00014233576642335764, | |
| "loss": 1.6368, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14497528830313014, | |
| "grad_norm": 2.3526768684387207, | |
| "learning_rate": 0.00014452554744525547, | |
| "loss": 1.2409, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.1471718835804503, | |
| "grad_norm": 4.02423620223999, | |
| "learning_rate": 0.00014671532846715328, | |
| "loss": 1.6012, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.14936847885777046, | |
| "grad_norm": 3.2113428115844727, | |
| "learning_rate": 0.00014890510948905108, | |
| "loss": 1.4285, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1515650741350906, | |
| "grad_norm": 3.3457765579223633, | |
| "learning_rate": 0.00015109489051094892, | |
| "loss": 1.5543, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.15376166941241076, | |
| "grad_norm": 14.925931930541992, | |
| "learning_rate": 0.00015328467153284672, | |
| "loss": 1.2421, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15595826468973092, | |
| "grad_norm": 1.9167107343673706, | |
| "learning_rate": 0.00015547445255474453, | |
| "loss": 1.5159, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.15815485996705106, | |
| "grad_norm": 2.6971662044525146, | |
| "learning_rate": 0.00015766423357664234, | |
| "loss": 1.5556, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.16035145524437122, | |
| "grad_norm": 5.165077209472656, | |
| "learning_rate": 0.00015985401459854014, | |
| "loss": 1.8581, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.16254805052169138, | |
| "grad_norm": 2.293755292892456, | |
| "learning_rate": 0.00016204379562043795, | |
| "loss": 1.4541, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.16474464579901152, | |
| "grad_norm": 3.0468106269836426, | |
| "learning_rate": 0.00016423357664233575, | |
| "loss": 1.443, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16694124107633168, | |
| "grad_norm": 2.405366897583008, | |
| "learning_rate": 0.00016642335766423356, | |
| "loss": 1.3887, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.16913783635365184, | |
| "grad_norm": 2.1093297004699707, | |
| "learning_rate": 0.00016861313868613137, | |
| "loss": 1.2783, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.171334431630972, | |
| "grad_norm": 1.6092991828918457, | |
| "learning_rate": 0.00017080291970802917, | |
| "loss": 0.9538, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17353102690829214, | |
| "grad_norm": 1.9024285078048706, | |
| "learning_rate": 0.00017299270072992698, | |
| "loss": 1.2746, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1757276221856123, | |
| "grad_norm": 1.9854646921157837, | |
| "learning_rate": 0.00017518248175182478, | |
| "loss": 1.3484, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17792421746293247, | |
| "grad_norm": 2.5449113845825195, | |
| "learning_rate": 0.00017737226277372262, | |
| "loss": 1.3065, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.1801208127402526, | |
| "grad_norm": 4.848789215087891, | |
| "learning_rate": 0.00017956204379562042, | |
| "loss": 1.1787, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18231740801757276, | |
| "grad_norm": 2.6889891624450684, | |
| "learning_rate": 0.00018175182481751826, | |
| "loss": 1.1507, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.18451400329489293, | |
| "grad_norm": 2.2919535636901855, | |
| "learning_rate": 0.00018394160583941606, | |
| "loss": 1.1736, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.18671059857221306, | |
| "grad_norm": 2.736064910888672, | |
| "learning_rate": 0.00018613138686131387, | |
| "loss": 1.402, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.18890719384953322, | |
| "grad_norm": 1.9482638835906982, | |
| "learning_rate": 0.00018832116788321167, | |
| "loss": 1.5493, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.19110378912685339, | |
| "grad_norm": 1.9475115537643433, | |
| "learning_rate": 0.00019051094890510948, | |
| "loss": 1.3407, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.19330038440417352, | |
| "grad_norm": 2.567296266555786, | |
| "learning_rate": 0.0001927007299270073, | |
| "loss": 1.3718, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.19549697968149368, | |
| "grad_norm": 2.2314236164093018, | |
| "learning_rate": 0.0001948905109489051, | |
| "loss": 1.1082, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.19769357495881384, | |
| "grad_norm": 3.055356979370117, | |
| "learning_rate": 0.0001970802919708029, | |
| "loss": 1.3822, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19989017023613398, | |
| "grad_norm": 2.3975324630737305, | |
| "learning_rate": 0.0001992700729927007, | |
| "loss": 1.1911, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.20208676551345414, | |
| "grad_norm": 3.026642322540283, | |
| "learning_rate": 0.0002014598540145985, | |
| "loss": 1.2064, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2042833607907743, | |
| "grad_norm": 2.4065756797790527, | |
| "learning_rate": 0.00020364963503649632, | |
| "loss": 1.1886, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.20647995606809447, | |
| "grad_norm": 2.632692813873291, | |
| "learning_rate": 0.00020583941605839415, | |
| "loss": 1.5078, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2086765513454146, | |
| "grad_norm": 2.1865198612213135, | |
| "learning_rate": 0.00020802919708029196, | |
| "loss": 1.1853, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21087314662273476, | |
| "grad_norm": 2.4859867095947266, | |
| "learning_rate": 0.00021021897810218976, | |
| "loss": 1.3652, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.21306974190005493, | |
| "grad_norm": 9.942977905273438, | |
| "learning_rate": 0.00021240875912408757, | |
| "loss": 1.3374, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.21526633717737506, | |
| "grad_norm": 2.040611505508423, | |
| "learning_rate": 0.00021459854014598537, | |
| "loss": 1.1695, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.21746293245469522, | |
| "grad_norm": 2.8881020545959473, | |
| "learning_rate": 0.0002167883211678832, | |
| "loss": 1.3527, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.21965952773201539, | |
| "grad_norm": 1.9736099243164062, | |
| "learning_rate": 0.00021897810218978101, | |
| "loss": 1.2486, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22185612300933552, | |
| "grad_norm": 3.0882883071899414, | |
| "learning_rate": 0.00022116788321167882, | |
| "loss": 1.3892, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.22405271828665568, | |
| "grad_norm": 3.0648694038391113, | |
| "learning_rate": 0.00022335766423357663, | |
| "loss": 1.2263, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.22624931356397585, | |
| "grad_norm": 1.8923087120056152, | |
| "learning_rate": 0.00022554744525547443, | |
| "loss": 1.307, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.22844590884129598, | |
| "grad_norm": 2.4342446327209473, | |
| "learning_rate": 0.00022773722627737224, | |
| "loss": 1.0324, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.23064250411861614, | |
| "grad_norm": 2.0455095767974854, | |
| "learning_rate": 0.00022992700729927004, | |
| "loss": 1.2811, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2328390993959363, | |
| "grad_norm": 2.5229058265686035, | |
| "learning_rate": 0.00023211678832116788, | |
| "loss": 1.5242, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.23503569467325644, | |
| "grad_norm": 3.1301496028900146, | |
| "learning_rate": 0.00023430656934306568, | |
| "loss": 1.313, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2372322899505766, | |
| "grad_norm": 2.648033618927002, | |
| "learning_rate": 0.0002364963503649635, | |
| "loss": 1.1697, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.23942888522789676, | |
| "grad_norm": 2.8274435997009277, | |
| "learning_rate": 0.0002386861313868613, | |
| "loss": 1.2034, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.24162548050521693, | |
| "grad_norm": 2.162809371948242, | |
| "learning_rate": 0.0002408759124087591, | |
| "loss": 1.1901, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24382207578253706, | |
| "grad_norm": 2.481313943862915, | |
| "learning_rate": 0.0002430656934306569, | |
| "loss": 1.1715, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.24601867105985722, | |
| "grad_norm": 2.6703312397003174, | |
| "learning_rate": 0.0002452554744525547, | |
| "loss": 1.2438, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2482152663371774, | |
| "grad_norm": 2.607210159301758, | |
| "learning_rate": 0.0002474452554744525, | |
| "loss": 1.1144, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.2504118616144975, | |
| "grad_norm": 2.166051149368286, | |
| "learning_rate": 0.0002496350364963503, | |
| "loss": 1.0557, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.25260845689181766, | |
| "grad_norm": 2.729848861694336, | |
| "learning_rate": 0.00025182481751824813, | |
| "loss": 1.1372, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.25480505216913785, | |
| "grad_norm": 2.4407503604888916, | |
| "learning_rate": 0.00025401459854014594, | |
| "loss": 1.3674, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.257001647446458, | |
| "grad_norm": 3.000013589859009, | |
| "learning_rate": 0.00025620437956204374, | |
| "loss": 1.2909, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2591982427237781, | |
| "grad_norm": 3.2139499187469482, | |
| "learning_rate": 0.00025839416058394155, | |
| "loss": 1.2209, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2613948380010983, | |
| "grad_norm": 2.9669885635375977, | |
| "learning_rate": 0.0002605839416058394, | |
| "loss": 1.0909, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.26359143327841844, | |
| "grad_norm": 2.2950656414031982, | |
| "learning_rate": 0.0002627737226277372, | |
| "loss": 1.0598, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.26578802855573863, | |
| "grad_norm": 2.153634548187256, | |
| "learning_rate": 0.000264963503649635, | |
| "loss": 1.1558, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.26798462383305877, | |
| "grad_norm": 1.5495681762695312, | |
| "learning_rate": 0.00026715328467153283, | |
| "loss": 0.9861, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2701812191103789, | |
| "grad_norm": 2.99607253074646, | |
| "learning_rate": 0.00026934306569343063, | |
| "loss": 1.1161, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.2723778143876991, | |
| "grad_norm": 2.21091365814209, | |
| "learning_rate": 0.00027153284671532844, | |
| "loss": 1.0399, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2745744096650192, | |
| "grad_norm": 2.5754196643829346, | |
| "learning_rate": 0.00027372262773722625, | |
| "loss": 1.2746, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.27677100494233936, | |
| "grad_norm": 2.4166269302368164, | |
| "learning_rate": 0.00027591240875912405, | |
| "loss": 0.8927, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.27896760021965955, | |
| "grad_norm": 2.2439630031585693, | |
| "learning_rate": 0.00027810218978102186, | |
| "loss": 1.1788, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.2811641954969797, | |
| "grad_norm": 3.3574883937835693, | |
| "learning_rate": 0.00028029197080291966, | |
| "loss": 1.4365, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2833607907742998, | |
| "grad_norm": 3.1181893348693848, | |
| "learning_rate": 0.00028248175182481747, | |
| "loss": 1.275, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.28555738605162, | |
| "grad_norm": 2.638686418533325, | |
| "learning_rate": 0.0002846715328467153, | |
| "loss": 1.0451, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.28775398132894014, | |
| "grad_norm": 1.788402795791626, | |
| "learning_rate": 0.00028686131386861314, | |
| "loss": 1.0938, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2899505766062603, | |
| "grad_norm": 2.1229209899902344, | |
| "learning_rate": 0.00028905109489051094, | |
| "loss": 0.9558, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.29214717188358047, | |
| "grad_norm": 1.6538920402526855, | |
| "learning_rate": 0.00029124087591240875, | |
| "loss": 0.8578, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.2943437671609006, | |
| "grad_norm": 2.7279489040374756, | |
| "learning_rate": 0.00029343065693430656, | |
| "loss": 1.0735, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.29654036243822074, | |
| "grad_norm": 2.167997360229492, | |
| "learning_rate": 0.00029562043795620436, | |
| "loss": 1.1019, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.29873695771554093, | |
| "grad_norm": 2.885739326477051, | |
| "learning_rate": 0.00029781021897810217, | |
| "loss": 1.295, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.30093355299286106, | |
| "grad_norm": 2.4096012115478516, | |
| "learning_rate": 0.0003, | |
| "loss": 1.2098, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3031301482701812, | |
| "grad_norm": 1.2121049165725708, | |
| "learning_rate": 0.00029975570032573286, | |
| "loss": 1.074, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3053267435475014, | |
| "grad_norm": 2.2965598106384277, | |
| "learning_rate": 0.00029951140065146575, | |
| "loss": 1.1367, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3075233388248215, | |
| "grad_norm": 1.3320868015289307, | |
| "learning_rate": 0.0002992671009771987, | |
| "loss": 1.156, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.30971993410214166, | |
| "grad_norm": 2.639571189880371, | |
| "learning_rate": 0.0002990228013029316, | |
| "loss": 1.2763, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.31191652937946185, | |
| "grad_norm": 2.763967514038086, | |
| "learning_rate": 0.0002987785016286645, | |
| "loss": 1.2301, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.314113124656782, | |
| "grad_norm": 2.0238993167877197, | |
| "learning_rate": 0.00029853420195439736, | |
| "loss": 1.1127, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3163097199341021, | |
| "grad_norm": 3.809309482574463, | |
| "learning_rate": 0.00029828990228013025, | |
| "loss": 1.4973, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3185063152114223, | |
| "grad_norm": 1.432543158531189, | |
| "learning_rate": 0.00029804560260586314, | |
| "loss": 1.0948, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.32070291048874244, | |
| "grad_norm": 2.1377265453338623, | |
| "learning_rate": 0.0002978013029315961, | |
| "loss": 1.1005, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.3228995057660626, | |
| "grad_norm": 1.920548439025879, | |
| "learning_rate": 0.000297557003257329, | |
| "loss": 1.0654, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.32509610104338277, | |
| "grad_norm": 1.4740238189697266, | |
| "learning_rate": 0.00029731270358306186, | |
| "loss": 1.0263, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.3272926963207029, | |
| "grad_norm": 2.2052175998687744, | |
| "learning_rate": 0.00029706840390879475, | |
| "loss": 1.0969, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.32948929159802304, | |
| "grad_norm": 2.317547559738159, | |
| "learning_rate": 0.00029682410423452764, | |
| "loss": 1.488, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3316858868753432, | |
| "grad_norm": 2.7809996604919434, | |
| "learning_rate": 0.0002965798045602606, | |
| "loss": 1.3018, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.33388248215266336, | |
| "grad_norm": 2.3972601890563965, | |
| "learning_rate": 0.0002963355048859935, | |
| "loss": 1.2042, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.33607907742998355, | |
| "grad_norm": 1.817764163017273, | |
| "learning_rate": 0.00029609120521172636, | |
| "loss": 0.9002, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3382756727073037, | |
| "grad_norm": 2.6748712062835693, | |
| "learning_rate": 0.00029584690553745925, | |
| "loss": 1.059, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.3404722679846238, | |
| "grad_norm": 3.6842854022979736, | |
| "learning_rate": 0.00029560260586319214, | |
| "loss": 1.0055, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.342668863261944, | |
| "grad_norm": 1.6820331811904907, | |
| "learning_rate": 0.00029535830618892503, | |
| "loss": 1.0711, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.34486545853926415, | |
| "grad_norm": 3.2321956157684326, | |
| "learning_rate": 0.000295114006514658, | |
| "loss": 1.1185, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3470620538165843, | |
| "grad_norm": 1.640162467956543, | |
| "learning_rate": 0.00029486970684039086, | |
| "loss": 0.9881, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.34925864909390447, | |
| "grad_norm": 2.503201723098755, | |
| "learning_rate": 0.00029462540716612375, | |
| "loss": 1.1448, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3514552443712246, | |
| "grad_norm": 2.3501734733581543, | |
| "learning_rate": 0.00029438110749185664, | |
| "loss": 1.0674, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35365183964854474, | |
| "grad_norm": 2.236201286315918, | |
| "learning_rate": 0.00029413680781758953, | |
| "loss": 1.0231, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.35584843492586493, | |
| "grad_norm": 2.7517476081848145, | |
| "learning_rate": 0.0002938925081433224, | |
| "loss": 0.9751, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.35804503020318507, | |
| "grad_norm": 1.3848055601119995, | |
| "learning_rate": 0.00029364820846905536, | |
| "loss": 0.7511, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.3602416254805052, | |
| "grad_norm": 2.3137853145599365, | |
| "learning_rate": 0.00029340390879478825, | |
| "loss": 0.9742, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3624382207578254, | |
| "grad_norm": 1.9163036346435547, | |
| "learning_rate": 0.00029315960912052114, | |
| "loss": 1.1395, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3646348160351455, | |
| "grad_norm": 3.1795966625213623, | |
| "learning_rate": 0.00029291530944625403, | |
| "loss": 1.0091, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.36683141131246566, | |
| "grad_norm": 1.7285435199737549, | |
| "learning_rate": 0.0002926710097719869, | |
| "loss": 0.987, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.36902800658978585, | |
| "grad_norm": 2.1685400009155273, | |
| "learning_rate": 0.00029242671009771986, | |
| "loss": 1.0749, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.371224601867106, | |
| "grad_norm": 2.4906914234161377, | |
| "learning_rate": 0.00029218241042345275, | |
| "loss": 1.0997, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3734211971444261, | |
| "grad_norm": 3.4433047771453857, | |
| "learning_rate": 0.00029193811074918564, | |
| "loss": 1.2508, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3756177924217463, | |
| "grad_norm": 2.1682846546173096, | |
| "learning_rate": 0.00029169381107491853, | |
| "loss": 0.95, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.37781438769906645, | |
| "grad_norm": 1.8938531875610352, | |
| "learning_rate": 0.0002914495114006514, | |
| "loss": 0.8669, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3800109829763866, | |
| "grad_norm": 1.886663794517517, | |
| "learning_rate": 0.0002912052117263843, | |
| "loss": 1.1436, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.38220757825370677, | |
| "grad_norm": 1.8953970670700073, | |
| "learning_rate": 0.00029096091205211725, | |
| "loss": 1.3218, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3844041735310269, | |
| "grad_norm": 2.0012447834014893, | |
| "learning_rate": 0.00029071661237785014, | |
| "loss": 0.9116, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.38660076880834704, | |
| "grad_norm": 1.5806304216384888, | |
| "learning_rate": 0.00029047231270358303, | |
| "loss": 1.2798, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.38879736408566723, | |
| "grad_norm": 2.0393028259277344, | |
| "learning_rate": 0.0002902280130293159, | |
| "loss": 0.72, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.39099395936298736, | |
| "grad_norm": 1.4992449283599854, | |
| "learning_rate": 0.0002899837133550488, | |
| "loss": 1.0711, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3931905546403075, | |
| "grad_norm": 1.5056639909744263, | |
| "learning_rate": 0.00028973941368078175, | |
| "loss": 1.0368, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3953871499176277, | |
| "grad_norm": 2.5978055000305176, | |
| "learning_rate": 0.00028949511400651464, | |
| "loss": 0.9809, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3975837451949478, | |
| "grad_norm": 2.0348634719848633, | |
| "learning_rate": 0.00028925081433224753, | |
| "loss": 0.9633, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.39978034047226796, | |
| "grad_norm": 1.2767269611358643, | |
| "learning_rate": 0.0002890065146579804, | |
| "loss": 0.8409, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.40197693574958815, | |
| "grad_norm": 1.583351969718933, | |
| "learning_rate": 0.0002887622149837133, | |
| "loss": 1.1328, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4041735310269083, | |
| "grad_norm": 2.2252044677734375, | |
| "learning_rate": 0.0002885179153094462, | |
| "loss": 1.5012, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.4063701263042284, | |
| "grad_norm": 2.067443609237671, | |
| "learning_rate": 0.00028827361563517914, | |
| "loss": 1.1907, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4085667215815486, | |
| "grad_norm": 2.117236852645874, | |
| "learning_rate": 0.00028802931596091203, | |
| "loss": 1.0136, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.41076331685886874, | |
| "grad_norm": 2.7158007621765137, | |
| "learning_rate": 0.0002877850162866449, | |
| "loss": 0.8449, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.41295991213618893, | |
| "grad_norm": 1.7849931716918945, | |
| "learning_rate": 0.0002875407166123778, | |
| "loss": 1.0018, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.41515650741350907, | |
| "grad_norm": 2.2555902004241943, | |
| "learning_rate": 0.0002872964169381107, | |
| "loss": 1.0679, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4173531026908292, | |
| "grad_norm": 3.6691648960113525, | |
| "learning_rate": 0.00028705211726384364, | |
| "loss": 1.035, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4195496979681494, | |
| "grad_norm": 3.3280532360076904, | |
| "learning_rate": 0.00028680781758957653, | |
| "loss": 1.1715, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.42174629324546953, | |
| "grad_norm": 1.5284361839294434, | |
| "learning_rate": 0.0002865635179153094, | |
| "loss": 1.1426, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.42394288852278966, | |
| "grad_norm": 1.8025450706481934, | |
| "learning_rate": 0.0002863192182410423, | |
| "loss": 1.1173, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.42613948380010985, | |
| "grad_norm": 1.9927115440368652, | |
| "learning_rate": 0.0002860749185667752, | |
| "loss": 1.1087, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.42833607907743, | |
| "grad_norm": 1.5994879007339478, | |
| "learning_rate": 0.0002858306188925081, | |
| "loss": 1.2334, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4305326743547501, | |
| "grad_norm": 1.5510128736495972, | |
| "learning_rate": 0.00028558631921824103, | |
| "loss": 1.4506, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.4327292696320703, | |
| "grad_norm": 1.3985286951065063, | |
| "learning_rate": 0.0002853420195439739, | |
| "loss": 0.9328, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.43492586490939045, | |
| "grad_norm": 2.2626256942749023, | |
| "learning_rate": 0.0002850977198697068, | |
| "loss": 0.9607, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.4371224601867106, | |
| "grad_norm": 1.7840726375579834, | |
| "learning_rate": 0.0002848534201954397, | |
| "loss": 1.0102, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.43931905546403077, | |
| "grad_norm": 2.102325916290283, | |
| "learning_rate": 0.0002846091205211726, | |
| "loss": 1.2043, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4415156507413509, | |
| "grad_norm": 1.9143919944763184, | |
| "learning_rate": 0.0002843648208469055, | |
| "loss": 0.9151, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.44371224601867104, | |
| "grad_norm": 1.7058296203613281, | |
| "learning_rate": 0.0002841205211726384, | |
| "loss": 1.0377, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.44590884129599123, | |
| "grad_norm": 1.8087902069091797, | |
| "learning_rate": 0.0002838762214983713, | |
| "loss": 1.0323, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.44810543657331137, | |
| "grad_norm": 2.2558298110961914, | |
| "learning_rate": 0.0002836319218241042, | |
| "loss": 1.2809, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.4503020318506315, | |
| "grad_norm": 15.062314987182617, | |
| "learning_rate": 0.0002833876221498371, | |
| "loss": 1.2362, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.4524986271279517, | |
| "grad_norm": 2.950195550918579, | |
| "learning_rate": 0.00028314332247557, | |
| "loss": 1.2492, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.4546952224052718, | |
| "grad_norm": 1.9644147157669067, | |
| "learning_rate": 0.0002828990228013029, | |
| "loss": 1.0598, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.45689181768259196, | |
| "grad_norm": 2.312228202819824, | |
| "learning_rate": 0.0002826547231270358, | |
| "loss": 1.0074, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.45908841295991215, | |
| "grad_norm": 1.9971623420715332, | |
| "learning_rate": 0.0002824104234527687, | |
| "loss": 1.0693, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.4612850082372323, | |
| "grad_norm": 3.6430881023406982, | |
| "learning_rate": 0.0002821661237785016, | |
| "loss": 1.1241, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4634816035145524, | |
| "grad_norm": 1.1851502656936646, | |
| "learning_rate": 0.0002819218241042345, | |
| "loss": 1.1803, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.4656781987918726, | |
| "grad_norm": 1.4428644180297852, | |
| "learning_rate": 0.00028167752442996737, | |
| "loss": 1.1207, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.46787479406919275, | |
| "grad_norm": 2.1165335178375244, | |
| "learning_rate": 0.0002814332247557003, | |
| "loss": 0.8893, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.4700713893465129, | |
| "grad_norm": 1.4343347549438477, | |
| "learning_rate": 0.0002811889250814332, | |
| "loss": 0.9964, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.47226798462383307, | |
| "grad_norm": 1.5154262781143188, | |
| "learning_rate": 0.0002809446254071661, | |
| "loss": 1.2375, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4744645799011532, | |
| "grad_norm": 2.090571880340576, | |
| "learning_rate": 0.000280700325732899, | |
| "loss": 1.0334, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.47666117517847334, | |
| "grad_norm": 1.6777268648147583, | |
| "learning_rate": 0.00028045602605863187, | |
| "loss": 0.9726, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.47885777045579353, | |
| "grad_norm": 1.430808424949646, | |
| "learning_rate": 0.0002802117263843648, | |
| "loss": 1.0488, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.48105436573311366, | |
| "grad_norm": 3.5004630088806152, | |
| "learning_rate": 0.0002799674267100977, | |
| "loss": 1.0473, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.48325096101043385, | |
| "grad_norm": 1.818041205406189, | |
| "learning_rate": 0.0002797231270358306, | |
| "loss": 1.087, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.485447556287754, | |
| "grad_norm": 5.02909517288208, | |
| "learning_rate": 0.0002794788273615635, | |
| "loss": 1.0509, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.4876441515650741, | |
| "grad_norm": 2.552549362182617, | |
| "learning_rate": 0.00027923452768729637, | |
| "loss": 0.8397, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.4898407468423943, | |
| "grad_norm": 1.9568651914596558, | |
| "learning_rate": 0.00027899022801302926, | |
| "loss": 1.1243, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.49203734211971445, | |
| "grad_norm": 1.5215007066726685, | |
| "learning_rate": 0.0002787459283387622, | |
| "loss": 1.1379, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.4942339373970346, | |
| "grad_norm": 1.1862854957580566, | |
| "learning_rate": 0.0002785016286644951, | |
| "loss": 0.8463, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4964305326743548, | |
| "grad_norm": 2.241152286529541, | |
| "learning_rate": 0.000278257328990228, | |
| "loss": 1.3126, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.4986271279516749, | |
| "grad_norm": 1.6096278429031372, | |
| "learning_rate": 0.00027801302931596087, | |
| "loss": 0.9576, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.500823723228995, | |
| "grad_norm": 1.6132186651229858, | |
| "learning_rate": 0.00027776872964169376, | |
| "loss": 1.0485, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.5030203185063152, | |
| "grad_norm": 1.0983858108520508, | |
| "learning_rate": 0.0002775244299674267, | |
| "loss": 0.9714, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.5052169137836353, | |
| "grad_norm": 1.8677927255630493, | |
| "learning_rate": 0.0002772801302931596, | |
| "loss": 1.1422, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5074135090609555, | |
| "grad_norm": 2.401646614074707, | |
| "learning_rate": 0.0002770358306188925, | |
| "loss": 1.1352, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.5096101043382757, | |
| "grad_norm": 1.330978512763977, | |
| "learning_rate": 0.00027679153094462537, | |
| "loss": 0.9467, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.5118066996155958, | |
| "grad_norm": 2.359093189239502, | |
| "learning_rate": 0.00027654723127035826, | |
| "loss": 1.0384, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.514003294892916, | |
| "grad_norm": 1.8499308824539185, | |
| "learning_rate": 0.00027630293159609115, | |
| "loss": 1.0706, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.5161998901702362, | |
| "grad_norm": 1.7341316938400269, | |
| "learning_rate": 0.0002760586319218241, | |
| "loss": 1.0867, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5183964854475562, | |
| "grad_norm": 1.3630411624908447, | |
| "learning_rate": 0.000275814332247557, | |
| "loss": 0.9488, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5205930807248764, | |
| "grad_norm": 1.6731019020080566, | |
| "learning_rate": 0.00027557003257328987, | |
| "loss": 1.2673, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.5227896760021966, | |
| "grad_norm": 1.4638540744781494, | |
| "learning_rate": 0.00027532573289902276, | |
| "loss": 1.0782, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5249862712795168, | |
| "grad_norm": 1.3307517766952515, | |
| "learning_rate": 0.00027508143322475565, | |
| "loss": 0.9071, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5271828665568369, | |
| "grad_norm": 1.5072880983352661, | |
| "learning_rate": 0.00027483713355048854, | |
| "loss": 1.0157, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5293794618341571, | |
| "grad_norm": 3.3565926551818848, | |
| "learning_rate": 0.0002745928338762215, | |
| "loss": 1.1135, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5315760571114773, | |
| "grad_norm": 1.6890822649002075, | |
| "learning_rate": 0.00027434853420195437, | |
| "loss": 0.9098, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.5337726523887973, | |
| "grad_norm": 1.0350748300552368, | |
| "learning_rate": 0.00027410423452768726, | |
| "loss": 0.8302, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.5359692476661175, | |
| "grad_norm": 2.94840669631958, | |
| "learning_rate": 0.00027385993485342015, | |
| "loss": 1.1717, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.5381658429434377, | |
| "grad_norm": 1.6727745532989502, | |
| "learning_rate": 0.00027361563517915304, | |
| "loss": 1.0563, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5403624382207578, | |
| "grad_norm": 1.4592437744140625, | |
| "learning_rate": 0.000273371335504886, | |
| "loss": 1.0218, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.542559033498078, | |
| "grad_norm": 1.7169549465179443, | |
| "learning_rate": 0.00027312703583061887, | |
| "loss": 0.8812, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.5447556287753982, | |
| "grad_norm": 1.4268460273742676, | |
| "learning_rate": 0.00027288273615635176, | |
| "loss": 1.0062, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.5469522240527183, | |
| "grad_norm": 1.468353271484375, | |
| "learning_rate": 0.00027263843648208465, | |
| "loss": 0.8986, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.5491488193300385, | |
| "grad_norm": 1.1650422811508179, | |
| "learning_rate": 0.00027239413680781754, | |
| "loss": 0.9799, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5513454146073586, | |
| "grad_norm": 1.9035193920135498, | |
| "learning_rate": 0.0002721498371335504, | |
| "loss": 1.004, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.5535420098846787, | |
| "grad_norm": 2.188706874847412, | |
| "learning_rate": 0.00027190553745928337, | |
| "loss": 0.9968, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.5557386051619989, | |
| "grad_norm": 1.1403478384017944, | |
| "learning_rate": 0.00027166123778501626, | |
| "loss": 1.0693, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.5579352004393191, | |
| "grad_norm": 1.2063301801681519, | |
| "learning_rate": 0.00027141693811074915, | |
| "loss": 0.9277, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.5601317957166392, | |
| "grad_norm": 1.3058998584747314, | |
| "learning_rate": 0.00027117263843648204, | |
| "loss": 1.0224, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5623283909939594, | |
| "grad_norm": 1.560442566871643, | |
| "learning_rate": 0.0002709283387622149, | |
| "loss": 1.0582, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.5645249862712796, | |
| "grad_norm": 1.3178349733352661, | |
| "learning_rate": 0.00027068403908794787, | |
| "loss": 0.8867, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.5667215815485996, | |
| "grad_norm": 1.7636605501174927, | |
| "learning_rate": 0.00027043973941368076, | |
| "loss": 1.0631, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.5689181768259198, | |
| "grad_norm": 1.6119142770767212, | |
| "learning_rate": 0.00027019543973941365, | |
| "loss": 0.9565, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.57111477210324, | |
| "grad_norm": 1.360366940498352, | |
| "learning_rate": 0.00026995114006514654, | |
| "loss": 1.0431, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5733113673805601, | |
| "grad_norm": 1.331679344177246, | |
| "learning_rate": 0.00026970684039087943, | |
| "loss": 1.0239, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.5755079626578803, | |
| "grad_norm": 1.2744184732437134, | |
| "learning_rate": 0.0002694625407166123, | |
| "loss": 1.0403, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5777045579352005, | |
| "grad_norm": 1.4823205471038818, | |
| "learning_rate": 0.00026921824104234526, | |
| "loss": 0.8674, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5799011532125206, | |
| "grad_norm": 2.1643168926239014, | |
| "learning_rate": 0.00026897394136807815, | |
| "loss": 1.0512, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5820977484898407, | |
| "grad_norm": 1.863612174987793, | |
| "learning_rate": 0.00026872964169381104, | |
| "loss": 1.4371, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5842943437671609, | |
| "grad_norm": 1.7413291931152344, | |
| "learning_rate": 0.00026848534201954393, | |
| "loss": 1.1652, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.586490939044481, | |
| "grad_norm": 1.5070873498916626, | |
| "learning_rate": 0.0002682410423452768, | |
| "loss": 1.0386, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5886875343218012, | |
| "grad_norm": 2.4088172912597656, | |
| "learning_rate": 0.00026799674267100976, | |
| "loss": 1.1527, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5908841295991214, | |
| "grad_norm": 1.5876343250274658, | |
| "learning_rate": 0.00026775244299674265, | |
| "loss": 1.1303, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.5930807248764415, | |
| "grad_norm": 2.447817087173462, | |
| "learning_rate": 0.00026750814332247554, | |
| "loss": 1.2059, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5952773201537617, | |
| "grad_norm": 1.7195405960083008, | |
| "learning_rate": 0.00026726384364820843, | |
| "loss": 0.9292, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.5974739154310819, | |
| "grad_norm": 2.7252278327941895, | |
| "learning_rate": 0.0002670195439739413, | |
| "loss": 0.9344, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.5996705107084019, | |
| "grad_norm": 1.6341739892959595, | |
| "learning_rate": 0.0002667752442996742, | |
| "loss": 0.8941, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.6018671059857221, | |
| "grad_norm": 1.8838438987731934, | |
| "learning_rate": 0.00026653094462540715, | |
| "loss": 0.9795, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.6040637012630423, | |
| "grad_norm": 1.828791856765747, | |
| "learning_rate": 0.00026628664495114004, | |
| "loss": 0.9957, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6062602965403624, | |
| "grad_norm": 1.6993612051010132, | |
| "learning_rate": 0.00026604234527687293, | |
| "loss": 0.7836, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.6084568918176826, | |
| "grad_norm": 4.5748796463012695, | |
| "learning_rate": 0.0002657980456026058, | |
| "loss": 1.0971, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.6106534870950028, | |
| "grad_norm": 2.429795265197754, | |
| "learning_rate": 0.0002655537459283387, | |
| "loss": 1.1318, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.6128500823723229, | |
| "grad_norm": 1.5458786487579346, | |
| "learning_rate": 0.00026530944625407165, | |
| "loss": 1.0223, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.615046677649643, | |
| "grad_norm": 1.653350591659546, | |
| "learning_rate": 0.00026506514657980454, | |
| "loss": 0.9198, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6172432729269632, | |
| "grad_norm": 2.094423770904541, | |
| "learning_rate": 0.00026482084690553743, | |
| "loss": 0.9482, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.6194398682042833, | |
| "grad_norm": 1.495132327079773, | |
| "learning_rate": 0.0002645765472312703, | |
| "loss": 0.8928, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6216364634816035, | |
| "grad_norm": 1.4427629709243774, | |
| "learning_rate": 0.0002643322475570032, | |
| "loss": 1.0551, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.6238330587589237, | |
| "grad_norm": 1.5103418827056885, | |
| "learning_rate": 0.0002640879478827361, | |
| "loss": 0.9083, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6260296540362438, | |
| "grad_norm": 1.409588098526001, | |
| "learning_rate": 0.00026384364820846904, | |
| "loss": 0.9435, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.628226249313564, | |
| "grad_norm": 1.2543412446975708, | |
| "learning_rate": 0.00026359934853420193, | |
| "loss": 0.8909, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6304228445908842, | |
| "grad_norm": 1.6690617799758911, | |
| "learning_rate": 0.0002633550488599348, | |
| "loss": 1.0993, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.6326194398682042, | |
| "grad_norm": 8.145515441894531, | |
| "learning_rate": 0.0002631107491856677, | |
| "loss": 0.952, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.6348160351455244, | |
| "grad_norm": 1.9644888639450073, | |
| "learning_rate": 0.0002628664495114006, | |
| "loss": 0.7753, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.6370126304228446, | |
| "grad_norm": 1.5240304470062256, | |
| "learning_rate": 0.0002626221498371335, | |
| "loss": 1.0064, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6392092257001647, | |
| "grad_norm": 3.2485849857330322, | |
| "learning_rate": 0.00026237785016286643, | |
| "loss": 1.0777, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.6414058209774849, | |
| "grad_norm": 1.6179548501968384, | |
| "learning_rate": 0.0002621335504885993, | |
| "loss": 0.9909, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.6436024162548051, | |
| "grad_norm": 1.5184112787246704, | |
| "learning_rate": 0.00026188925081433226, | |
| "loss": 1.0065, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.6457990115321252, | |
| "grad_norm": 1.6845955848693848, | |
| "learning_rate": 0.00026164495114006515, | |
| "loss": 0.9876, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.6479956068094453, | |
| "grad_norm": 2.0729987621307373, | |
| "learning_rate": 0.00026140065146579804, | |
| "loss": 1.0495, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6501922020867655, | |
| "grad_norm": 4.543087005615234, | |
| "learning_rate": 0.00026115635179153093, | |
| "loss": 1.0367, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.6523887973640856, | |
| "grad_norm": 1.5828183889389038, | |
| "learning_rate": 0.0002609120521172638, | |
| "loss": 0.8839, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.6545853926414058, | |
| "grad_norm": 2.21020245552063, | |
| "learning_rate": 0.00026066775244299676, | |
| "loss": 1.0169, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.656781987918726, | |
| "grad_norm": 1.3326387405395508, | |
| "learning_rate": 0.00026042345276872965, | |
| "loss": 1.0125, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.6589785831960461, | |
| "grad_norm": 1.628922700881958, | |
| "learning_rate": 0.00026017915309446254, | |
| "loss": 1.0483, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6611751784733663, | |
| "grad_norm": 2.096686363220215, | |
| "learning_rate": 0.00025993485342019543, | |
| "loss": 1.1018, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.6633717737506865, | |
| "grad_norm": 7.434841632843018, | |
| "learning_rate": 0.0002596905537459283, | |
| "loss": 0.9659, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.6655683690280065, | |
| "grad_norm": 1.9217203855514526, | |
| "learning_rate": 0.0002594462540716612, | |
| "loss": 0.889, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.6677649643053267, | |
| "grad_norm": 1.9974533319473267, | |
| "learning_rate": 0.00025920195439739415, | |
| "loss": 1.1628, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.6699615595826469, | |
| "grad_norm": 2.040860891342163, | |
| "learning_rate": 0.00025895765472312704, | |
| "loss": 1.0375, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6721581548599671, | |
| "grad_norm": 2.4363908767700195, | |
| "learning_rate": 0.00025871335504885993, | |
| "loss": 0.927, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.6743547501372872, | |
| "grad_norm": 1.454214334487915, | |
| "learning_rate": 0.0002584690553745928, | |
| "loss": 0.9476, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.6765513454146074, | |
| "grad_norm": 1.3225269317626953, | |
| "learning_rate": 0.0002582247557003257, | |
| "loss": 0.9812, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.6787479406919276, | |
| "grad_norm": 2.6042890548706055, | |
| "learning_rate": 0.00025798045602605865, | |
| "loss": 1.1419, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.6809445359692476, | |
| "grad_norm": 3.2416915893554688, | |
| "learning_rate": 0.00025773615635179154, | |
| "loss": 1.1792, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6831411312465678, | |
| "grad_norm": 1.830832839012146, | |
| "learning_rate": 0.00025749185667752443, | |
| "loss": 0.8569, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.685337726523888, | |
| "grad_norm": 3.2830545902252197, | |
| "learning_rate": 0.0002572475570032573, | |
| "loss": 1.103, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.6875343218012081, | |
| "grad_norm": 2.9121148586273193, | |
| "learning_rate": 0.0002570032573289902, | |
| "loss": 0.9738, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.6897309170785283, | |
| "grad_norm": 2.821794271469116, | |
| "learning_rate": 0.0002567589576547231, | |
| "loss": 0.7733, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.6919275123558485, | |
| "grad_norm": 1.7697335481643677, | |
| "learning_rate": 0.00025651465798045604, | |
| "loss": 0.7286, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6941241076331686, | |
| "grad_norm": 3.069547176361084, | |
| "learning_rate": 0.00025627035830618893, | |
| "loss": 1.0309, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.6963207029104888, | |
| "grad_norm": 1.3961061239242554, | |
| "learning_rate": 0.0002560260586319218, | |
| "loss": 1.0732, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.6985172981878089, | |
| "grad_norm": 1.3182768821716309, | |
| "learning_rate": 0.0002557817589576547, | |
| "loss": 1.126, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.700713893465129, | |
| "grad_norm": 1.6211820840835571, | |
| "learning_rate": 0.0002555374592833876, | |
| "loss": 0.8911, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.7029104887424492, | |
| "grad_norm": 1.7093966007232666, | |
| "learning_rate": 0.0002552931596091205, | |
| "loss": 0.9642, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7051070840197694, | |
| "grad_norm": 1.767195224761963, | |
| "learning_rate": 0.00025504885993485343, | |
| "loss": 0.9903, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.7073036792970895, | |
| "grad_norm": 4.666885852813721, | |
| "learning_rate": 0.0002548045602605863, | |
| "loss": 1.0818, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.7095002745744097, | |
| "grad_norm": 1.7851799726486206, | |
| "learning_rate": 0.0002545602605863192, | |
| "loss": 0.85, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.7116968698517299, | |
| "grad_norm": 2.2941346168518066, | |
| "learning_rate": 0.0002543159609120521, | |
| "loss": 1.0595, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.7138934651290499, | |
| "grad_norm": 3.1601450443267822, | |
| "learning_rate": 0.000254071661237785, | |
| "loss": 0.9059, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7160900604063701, | |
| "grad_norm": 1.9102383852005005, | |
| "learning_rate": 0.00025382736156351793, | |
| "loss": 0.9883, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.7182866556836903, | |
| "grad_norm": 3.2335336208343506, | |
| "learning_rate": 0.0002535830618892508, | |
| "loss": 1.0788, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.7204832509610104, | |
| "grad_norm": 1.418175220489502, | |
| "learning_rate": 0.0002533387622149837, | |
| "loss": 1.074, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.7226798462383306, | |
| "grad_norm": 1.4739829301834106, | |
| "learning_rate": 0.0002530944625407166, | |
| "loss": 0.9082, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.7248764415156508, | |
| "grad_norm": 2.552597999572754, | |
| "learning_rate": 0.0002528501628664495, | |
| "loss": 0.9358, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7270730367929709, | |
| "grad_norm": 1.9511799812316895, | |
| "learning_rate": 0.0002526058631921824, | |
| "loss": 1.1114, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.729269632070291, | |
| "grad_norm": 8.497620582580566, | |
| "learning_rate": 0.0002523615635179153, | |
| "loss": 0.9667, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.7314662273476112, | |
| "grad_norm": 1.145659327507019, | |
| "learning_rate": 0.0002521172638436482, | |
| "loss": 0.9959, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.7336628226249313, | |
| "grad_norm": 1.6846892833709717, | |
| "learning_rate": 0.0002518729641693811, | |
| "loss": 1.1819, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7358594179022515, | |
| "grad_norm": 1.568519949913025, | |
| "learning_rate": 0.000251628664495114, | |
| "loss": 1.0465, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7380560131795717, | |
| "grad_norm": 1.426245927810669, | |
| "learning_rate": 0.0002513843648208469, | |
| "loss": 0.9116, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.7402526084568918, | |
| "grad_norm": 1.3409892320632935, | |
| "learning_rate": 0.0002511400651465798, | |
| "loss": 1.122, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.742449203734212, | |
| "grad_norm": 1.9830353260040283, | |
| "learning_rate": 0.0002508957654723127, | |
| "loss": 0.9173, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.7446457990115322, | |
| "grad_norm": 1.6047143936157227, | |
| "learning_rate": 0.0002506514657980456, | |
| "loss": 0.9305, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.7468423942888522, | |
| "grad_norm": 1.4045612812042236, | |
| "learning_rate": 0.0002504071661237785, | |
| "loss": 1.0368, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7490389895661724, | |
| "grad_norm": 1.4184879064559937, | |
| "learning_rate": 0.0002501628664495114, | |
| "loss": 0.8933, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.7512355848434926, | |
| "grad_norm": 1.1624377965927124, | |
| "learning_rate": 0.00024991856677524427, | |
| "loss": 0.8587, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.7534321801208127, | |
| "grad_norm": 0.8981499075889587, | |
| "learning_rate": 0.0002496742671009772, | |
| "loss": 0.8911, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.7556287753981329, | |
| "grad_norm": 2.6175858974456787, | |
| "learning_rate": 0.0002494299674267101, | |
| "loss": 1.0256, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.7578253706754531, | |
| "grad_norm": 1.1921144723892212, | |
| "learning_rate": 0.000249185667752443, | |
| "loss": 0.8499, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7600219659527732, | |
| "grad_norm": 1.6792922019958496, | |
| "learning_rate": 0.0002489413680781759, | |
| "loss": 0.9529, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.7622185612300933, | |
| "grad_norm": 1.2778786420822144, | |
| "learning_rate": 0.00024869706840390877, | |
| "loss": 0.9715, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.7644151565074135, | |
| "grad_norm": 3.354434013366699, | |
| "learning_rate": 0.0002484527687296417, | |
| "loss": 1.2905, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.7666117517847336, | |
| "grad_norm": 1.641374945640564, | |
| "learning_rate": 0.0002482084690553746, | |
| "loss": 0.7536, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.7688083470620538, | |
| "grad_norm": 1.6452834606170654, | |
| "learning_rate": 0.0002479641693811075, | |
| "loss": 1.2153, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.771004942339374, | |
| "grad_norm": 1.6607787609100342, | |
| "learning_rate": 0.0002477198697068404, | |
| "loss": 1.2046, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.7732015376166941, | |
| "grad_norm": 2.1596899032592773, | |
| "learning_rate": 0.00024747557003257327, | |
| "loss": 1.0147, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.7753981328940143, | |
| "grad_norm": 1.4244449138641357, | |
| "learning_rate": 0.00024723127035830616, | |
| "loss": 0.7669, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.7775947281713345, | |
| "grad_norm": 1.389800786972046, | |
| "learning_rate": 0.0002469869706840391, | |
| "loss": 0.9349, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.7797913234486545, | |
| "grad_norm": 2.1857922077178955, | |
| "learning_rate": 0.000246742671009772, | |
| "loss": 0.7593, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7819879187259747, | |
| "grad_norm": 1.4879189729690552, | |
| "learning_rate": 0.0002464983713355049, | |
| "loss": 0.8054, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.7841845140032949, | |
| "grad_norm": 1.8027299642562866, | |
| "learning_rate": 0.00024625407166123777, | |
| "loss": 0.9185, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.786381109280615, | |
| "grad_norm": 1.6629337072372437, | |
| "learning_rate": 0.00024600977198697066, | |
| "loss": 0.9214, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.7885777045579352, | |
| "grad_norm": 6.917058944702148, | |
| "learning_rate": 0.00024576547231270354, | |
| "loss": 0.8919, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.7907742998352554, | |
| "grad_norm": 1.3103359937667847, | |
| "learning_rate": 0.0002455211726384365, | |
| "loss": 0.8292, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7929708951125755, | |
| "grad_norm": 1.3542835712432861, | |
| "learning_rate": 0.0002452768729641694, | |
| "loss": 1.0415, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.7951674903898956, | |
| "grad_norm": 2.5355348587036133, | |
| "learning_rate": 0.00024503257328990227, | |
| "loss": 1.0706, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.7973640856672158, | |
| "grad_norm": 1.975866675376892, | |
| "learning_rate": 0.00024478827361563516, | |
| "loss": 0.7955, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.7995606809445359, | |
| "grad_norm": 1.7187716960906982, | |
| "learning_rate": 0.00024454397394136804, | |
| "loss": 0.8885, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.8017572762218561, | |
| "grad_norm": 1.6666568517684937, | |
| "learning_rate": 0.000244299674267101, | |
| "loss": 0.7576, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.8039538714991763, | |
| "grad_norm": 1.430021047592163, | |
| "learning_rate": 0.00024405537459283385, | |
| "loss": 1.0997, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.8061504667764964, | |
| "grad_norm": 2.3396389484405518, | |
| "learning_rate": 0.00024381107491856677, | |
| "loss": 0.8867, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.8083470620538166, | |
| "grad_norm": 2.1217405796051025, | |
| "learning_rate": 0.00024356677524429966, | |
| "loss": 0.742, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.8105436573311368, | |
| "grad_norm": 1.1842306852340698, | |
| "learning_rate": 0.00024332247557003257, | |
| "loss": 0.8598, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.8127402526084568, | |
| "grad_norm": 2.5551540851593018, | |
| "learning_rate": 0.00024307817589576546, | |
| "loss": 0.7736, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.814936847885777, | |
| "grad_norm": 1.2459492683410645, | |
| "learning_rate": 0.00024283387622149835, | |
| "loss": 1.006, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.8171334431630972, | |
| "grad_norm": 3.0364813804626465, | |
| "learning_rate": 0.00024258957654723127, | |
| "loss": 1.0, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.8193300384404174, | |
| "grad_norm": 1.2043843269348145, | |
| "learning_rate": 0.00024234527687296416, | |
| "loss": 0.9499, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.8215266337177375, | |
| "grad_norm": 1.8204615116119385, | |
| "learning_rate": 0.00024210097719869705, | |
| "loss": 0.9867, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.8237232289950577, | |
| "grad_norm": 1.1458009481430054, | |
| "learning_rate": 0.00024185667752442996, | |
| "loss": 1.0411, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8259198242723779, | |
| "grad_norm": 1.6867096424102783, | |
| "learning_rate": 0.00024161237785016285, | |
| "loss": 0.9643, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.828116419549698, | |
| "grad_norm": 1.2090388536453247, | |
| "learning_rate": 0.00024136807817589574, | |
| "loss": 0.9849, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.8303130148270181, | |
| "grad_norm": 1.2127000093460083, | |
| "learning_rate": 0.00024112377850162866, | |
| "loss": 0.9703, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.8325096101043383, | |
| "grad_norm": 1.8187060356140137, | |
| "learning_rate": 0.00024087947882736155, | |
| "loss": 0.7177, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.8347062053816584, | |
| "grad_norm": 2.0383870601654053, | |
| "learning_rate": 0.00024063517915309446, | |
| "loss": 1.0515, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8369028006589786, | |
| "grad_norm": 1.4340029954910278, | |
| "learning_rate": 0.00024039087947882735, | |
| "loss": 0.9456, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.8390993959362988, | |
| "grad_norm": 1.5647447109222412, | |
| "learning_rate": 0.00024014657980456024, | |
| "loss": 0.9669, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8412959912136189, | |
| "grad_norm": 1.3744195699691772, | |
| "learning_rate": 0.00023990228013029316, | |
| "loss": 0.9828, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.8434925864909391, | |
| "grad_norm": 1.9642137289047241, | |
| "learning_rate": 0.00023965798045602605, | |
| "loss": 0.9943, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.8456891817682592, | |
| "grad_norm": 1.9889167547225952, | |
| "learning_rate": 0.00023941368078175893, | |
| "loss": 1.1805, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8478857770455793, | |
| "grad_norm": 16.377487182617188, | |
| "learning_rate": 0.00023916938110749185, | |
| "loss": 1.0322, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.8500823723228995, | |
| "grad_norm": 1.0606954097747803, | |
| "learning_rate": 0.00023892508143322474, | |
| "loss": 0.9012, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.8522789676002197, | |
| "grad_norm": 1.5951098203659058, | |
| "learning_rate": 0.00023868078175895763, | |
| "loss": 1.0096, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.8544755628775398, | |
| "grad_norm": 1.2166616916656494, | |
| "learning_rate": 0.00023843648208469055, | |
| "loss": 0.7532, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.85667215815486, | |
| "grad_norm": 1.3039473295211792, | |
| "learning_rate": 0.00023819218241042343, | |
| "loss": 0.8477, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8588687534321802, | |
| "grad_norm": 1.3621416091918945, | |
| "learning_rate": 0.00023794788273615632, | |
| "loss": 0.9796, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.8610653487095002, | |
| "grad_norm": 1.9069013595581055, | |
| "learning_rate": 0.00023770358306188924, | |
| "loss": 1.2839, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.8632619439868204, | |
| "grad_norm": 1.4560531377792358, | |
| "learning_rate": 0.00023745928338762213, | |
| "loss": 1.295, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.8654585392641406, | |
| "grad_norm": 1.0867977142333984, | |
| "learning_rate": 0.00023721498371335505, | |
| "loss": 0.9257, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.8676551345414607, | |
| "grad_norm": 1.4225832223892212, | |
| "learning_rate": 0.00023697068403908794, | |
| "loss": 0.9102, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8698517298187809, | |
| "grad_norm": 1.3702126741409302, | |
| "learning_rate": 0.00023672638436482082, | |
| "loss": 0.9362, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.8720483250961011, | |
| "grad_norm": 1.1921621561050415, | |
| "learning_rate": 0.00023648208469055374, | |
| "loss": 0.8919, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.8742449203734212, | |
| "grad_norm": 3.185751438140869, | |
| "learning_rate": 0.00023623778501628663, | |
| "loss": 0.864, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.8764415156507414, | |
| "grad_norm": 1.3370988368988037, | |
| "learning_rate": 0.00023599348534201952, | |
| "loss": 0.9638, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.8786381109280615, | |
| "grad_norm": 1.833447813987732, | |
| "learning_rate": 0.00023574918566775244, | |
| "loss": 1.0502, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8808347062053816, | |
| "grad_norm": 2.057830810546875, | |
| "learning_rate": 0.00023550488599348532, | |
| "loss": 0.8211, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.8830313014827018, | |
| "grad_norm": 1.0621795654296875, | |
| "learning_rate": 0.00023526058631921821, | |
| "loss": 0.9224, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.885227896760022, | |
| "grad_norm": 1.4506641626358032, | |
| "learning_rate": 0.00023501628664495113, | |
| "loss": 1.107, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.8874244920373421, | |
| "grad_norm": 0.849584698677063, | |
| "learning_rate": 0.00023477198697068402, | |
| "loss": 1.0752, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.8896210873146623, | |
| "grad_norm": 1.1948933601379395, | |
| "learning_rate": 0.0002345276872964169, | |
| "loss": 0.787, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.8918176825919825, | |
| "grad_norm": 1.3642253875732422, | |
| "learning_rate": 0.00023428338762214982, | |
| "loss": 0.848, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.8940142778693025, | |
| "grad_norm": 1.8935388326644897, | |
| "learning_rate": 0.00023403908794788271, | |
| "loss": 0.8598, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.8962108731466227, | |
| "grad_norm": 1.0388127565383911, | |
| "learning_rate": 0.00023379478827361563, | |
| "loss": 0.7879, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.8984074684239429, | |
| "grad_norm": 0.8143270015716553, | |
| "learning_rate": 0.00023355048859934852, | |
| "loss": 0.8682, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.900604063701263, | |
| "grad_norm": 1.7284828424453735, | |
| "learning_rate": 0.0002333061889250814, | |
| "loss": 1.086, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.9028006589785832, | |
| "grad_norm": 1.4692379236221313, | |
| "learning_rate": 0.00023306188925081432, | |
| "loss": 0.8044, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.9049972542559034, | |
| "grad_norm": 3.0223801136016846, | |
| "learning_rate": 0.00023281758957654721, | |
| "loss": 0.8228, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.9071938495332235, | |
| "grad_norm": 0.9196959137916565, | |
| "learning_rate": 0.0002325732899022801, | |
| "loss": 0.9209, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.9093904448105437, | |
| "grad_norm": 1.1501983404159546, | |
| "learning_rate": 0.00023232899022801302, | |
| "loss": 1.0272, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.9115870400878638, | |
| "grad_norm": 1.8854323625564575, | |
| "learning_rate": 0.0002320846905537459, | |
| "loss": 0.8936, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.9137836353651839, | |
| "grad_norm": 1.395145297050476, | |
| "learning_rate": 0.0002318403908794788, | |
| "loss": 1.0426, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.9159802306425041, | |
| "grad_norm": 1.1210417747497559, | |
| "learning_rate": 0.00023159609120521171, | |
| "loss": 0.8759, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.9181768259198243, | |
| "grad_norm": 1.911478042602539, | |
| "learning_rate": 0.0002313517915309446, | |
| "loss": 0.782, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.9203734211971444, | |
| "grad_norm": 1.1630089282989502, | |
| "learning_rate": 0.00023110749185667752, | |
| "loss": 0.8546, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.9225700164744646, | |
| "grad_norm": 1.0276585817337036, | |
| "learning_rate": 0.0002308631921824104, | |
| "loss": 0.9619, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9247666117517848, | |
| "grad_norm": 1.2841545343399048, | |
| "learning_rate": 0.0002306188925081433, | |
| "loss": 0.8925, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.9269632070291048, | |
| "grad_norm": 1.0926916599273682, | |
| "learning_rate": 0.00023037459283387621, | |
| "loss": 0.8648, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.929159802306425, | |
| "grad_norm": 1.6268560886383057, | |
| "learning_rate": 0.0002301302931596091, | |
| "loss": 1.1369, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.9313563975837452, | |
| "grad_norm": 1.269709825515747, | |
| "learning_rate": 0.000229885993485342, | |
| "loss": 0.9373, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.9335529928610653, | |
| "grad_norm": 1.6098746061325073, | |
| "learning_rate": 0.0002296416938110749, | |
| "loss": 0.8599, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9357495881383855, | |
| "grad_norm": 1.6270309686660767, | |
| "learning_rate": 0.0002293973941368078, | |
| "loss": 0.8828, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9379461834157057, | |
| "grad_norm": 1.231642723083496, | |
| "learning_rate": 0.0002291530944625407, | |
| "loss": 0.9115, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.9401427786930258, | |
| "grad_norm": 1.0608386993408203, | |
| "learning_rate": 0.0002289087947882736, | |
| "loss": 0.733, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.942339373970346, | |
| "grad_norm": 0.9125494360923767, | |
| "learning_rate": 0.0002286644951140065, | |
| "loss": 0.7058, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.9445359692476661, | |
| "grad_norm": 1.1660634279251099, | |
| "learning_rate": 0.00022842019543973938, | |
| "loss": 1.0265, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9467325645249862, | |
| "grad_norm": 1.6889920234680176, | |
| "learning_rate": 0.0002281758957654723, | |
| "loss": 0.862, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.9489291598023064, | |
| "grad_norm": 1.3840012550354004, | |
| "learning_rate": 0.0002279315960912052, | |
| "loss": 1.0728, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.9511257550796266, | |
| "grad_norm": 1.049270510673523, | |
| "learning_rate": 0.0002276872964169381, | |
| "loss": 0.9433, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.9533223503569467, | |
| "grad_norm": 2.3411548137664795, | |
| "learning_rate": 0.000227442996742671, | |
| "loss": 0.8488, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.9555189456342669, | |
| "grad_norm": 1.306808590888977, | |
| "learning_rate": 0.00022719869706840388, | |
| "loss": 0.8358, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9577155409115871, | |
| "grad_norm": 1.1646819114685059, | |
| "learning_rate": 0.0002269543973941368, | |
| "loss": 0.7485, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.9599121361889071, | |
| "grad_norm": 0.9352891445159912, | |
| "learning_rate": 0.0002267100977198697, | |
| "loss": 0.9126, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.9621087314662273, | |
| "grad_norm": 1.5916190147399902, | |
| "learning_rate": 0.00022646579804560258, | |
| "loss": 0.7953, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.9643053267435475, | |
| "grad_norm": 1.2098153829574585, | |
| "learning_rate": 0.0002262214983713355, | |
| "loss": 1.24, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.9665019220208677, | |
| "grad_norm": 3.2436654567718506, | |
| "learning_rate": 0.00022597719869706838, | |
| "loss": 1.0406, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9686985172981878, | |
| "grad_norm": 1.3033084869384766, | |
| "learning_rate": 0.00022573289902280127, | |
| "loss": 0.8385, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.970895112575508, | |
| "grad_norm": 1.251079797744751, | |
| "learning_rate": 0.0002254885993485342, | |
| "loss": 0.7956, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.9730917078528282, | |
| "grad_norm": 1.4023674726486206, | |
| "learning_rate": 0.00022524429967426708, | |
| "loss": 0.9474, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.9752883031301482, | |
| "grad_norm": 1.3533451557159424, | |
| "learning_rate": 0.000225, | |
| "loss": 1.0806, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.9774848984074684, | |
| "grad_norm": 1.7573014497756958, | |
| "learning_rate": 0.00022475570032573288, | |
| "loss": 0.9506, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9796814936847886, | |
| "grad_norm": 2.111967086791992, | |
| "learning_rate": 0.00022451140065146577, | |
| "loss": 1.0109, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.9818780889621087, | |
| "grad_norm": 2.4967286586761475, | |
| "learning_rate": 0.0002242671009771987, | |
| "loss": 1.1756, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.9840746842394289, | |
| "grad_norm": 1.1087509393692017, | |
| "learning_rate": 0.00022402280130293158, | |
| "loss": 0.8664, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.9862712795167491, | |
| "grad_norm": 1.2817955017089844, | |
| "learning_rate": 0.00022377850162866447, | |
| "loss": 1.0192, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.9884678747940692, | |
| "grad_norm": 1.390347957611084, | |
| "learning_rate": 0.00022353420195439738, | |
| "loss": 0.9234, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9906644700713894, | |
| "grad_norm": 1.1981894969940186, | |
| "learning_rate": 0.00022328990228013027, | |
| "loss": 0.9671, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.9928610653487095, | |
| "grad_norm": 1.5251400470733643, | |
| "learning_rate": 0.00022304560260586316, | |
| "loss": 0.9823, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.9950576606260296, | |
| "grad_norm": 1.4941556453704834, | |
| "learning_rate": 0.00022280130293159608, | |
| "loss": 1.0097, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.9972542559033498, | |
| "grad_norm": 1.712870478630066, | |
| "learning_rate": 0.00022255700325732897, | |
| "loss": 0.857, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.99945085118067, | |
| "grad_norm": 0.9738109707832336, | |
| "learning_rate": 0.00022231270358306186, | |
| "loss": 0.8348, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.00164744645799, | |
| "grad_norm": 1.2786909341812134, | |
| "learning_rate": 0.00022206840390879477, | |
| "loss": 1.0193, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.0038440417353103, | |
| "grad_norm": 1.1460750102996826, | |
| "learning_rate": 0.00022182410423452766, | |
| "loss": 0.9958, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.0060406370126305, | |
| "grad_norm": 9.903060913085938, | |
| "learning_rate": 0.00022157980456026058, | |
| "loss": 0.8973, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.0082372322899507, | |
| "grad_norm": 1.5211352109909058, | |
| "learning_rate": 0.00022133550488599347, | |
| "loss": 0.855, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.0104338275672706, | |
| "grad_norm": 1.4610356092453003, | |
| "learning_rate": 0.00022109120521172636, | |
| "loss": 0.9299, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.0126304228445908, | |
| "grad_norm": 1.6042611598968506, | |
| "learning_rate": 0.00022084690553745927, | |
| "loss": 0.7761, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.014827018121911, | |
| "grad_norm": 1.4318044185638428, | |
| "learning_rate": 0.00022060260586319216, | |
| "loss": 1.2622, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.0170236133992312, | |
| "grad_norm": 1.2310857772827148, | |
| "learning_rate": 0.00022035830618892505, | |
| "loss": 1.0837, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.0192202086765514, | |
| "grad_norm": 1.2733674049377441, | |
| "learning_rate": 0.00022011400651465797, | |
| "loss": 0.8861, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.0214168039538716, | |
| "grad_norm": 1.7253901958465576, | |
| "learning_rate": 0.00021986970684039086, | |
| "loss": 1.0829, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.0236133992311915, | |
| "grad_norm": 1.7851202487945557, | |
| "learning_rate": 0.00021962540716612375, | |
| "loss": 1.1121, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.0258099945085117, | |
| "grad_norm": 1.1333625316619873, | |
| "learning_rate": 0.00021938110749185666, | |
| "loss": 0.9005, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.028006589785832, | |
| "grad_norm": 1.2626473903656006, | |
| "learning_rate": 0.00021913680781758955, | |
| "loss": 0.814, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.0302031850631521, | |
| "grad_norm": 1.1382790803909302, | |
| "learning_rate": 0.00021889250814332244, | |
| "loss": 0.9896, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.0323997803404723, | |
| "grad_norm": 3.7624616622924805, | |
| "learning_rate": 0.00021864820846905536, | |
| "loss": 1.0526, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0345963756177925, | |
| "grad_norm": 3.796247720718384, | |
| "learning_rate": 0.00021840390879478825, | |
| "loss": 0.7565, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.0367929708951125, | |
| "grad_norm": 1.8124761581420898, | |
| "learning_rate": 0.00021815960912052116, | |
| "loss": 0.881, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.0389895661724327, | |
| "grad_norm": 1.2979084253311157, | |
| "learning_rate": 0.00021791530944625405, | |
| "loss": 1.1165, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.0411861614497528, | |
| "grad_norm": 1.5521516799926758, | |
| "learning_rate": 0.00021767100977198694, | |
| "loss": 0.9027, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.043382756727073, | |
| "grad_norm": 1.2450944185256958, | |
| "learning_rate": 0.00021742671009771986, | |
| "loss": 0.9687, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0455793520043932, | |
| "grad_norm": 1.5662953853607178, | |
| "learning_rate": 0.00021718241042345275, | |
| "loss": 1.2001, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.0477759472817134, | |
| "grad_norm": 1.0187841653823853, | |
| "learning_rate": 0.00021693811074918564, | |
| "loss": 0.7745, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.0499725425590336, | |
| "grad_norm": 5.503617286682129, | |
| "learning_rate": 0.00021669381107491855, | |
| "loss": 0.9993, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.0521691378363536, | |
| "grad_norm": 1.139510989189148, | |
| "learning_rate": 0.00021644951140065144, | |
| "loss": 0.9368, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.0543657331136738, | |
| "grad_norm": 1.0795385837554932, | |
| "learning_rate": 0.00021620521172638433, | |
| "loss": 0.7842, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.056562328390994, | |
| "grad_norm": 0.976901650428772, | |
| "learning_rate": 0.00021596091205211725, | |
| "loss": 0.8341, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.0587589236683141, | |
| "grad_norm": 1.6595531702041626, | |
| "learning_rate": 0.00021571661237785014, | |
| "loss": 0.9013, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.0609555189456343, | |
| "grad_norm": 1.4204063415527344, | |
| "learning_rate": 0.00021547231270358305, | |
| "loss": 0.8216, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.0631521142229543, | |
| "grad_norm": 0.8639747500419617, | |
| "learning_rate": 0.00021522801302931594, | |
| "loss": 0.9362, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.0653487095002745, | |
| "grad_norm": 0.8762627840042114, | |
| "learning_rate": 0.00021498371335504883, | |
| "loss": 0.8913, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.0675453047775947, | |
| "grad_norm": 1.7928240299224854, | |
| "learning_rate": 0.00021473941368078175, | |
| "loss": 0.8331, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.0697419000549149, | |
| "grad_norm": 3.6022591590881348, | |
| "learning_rate": 0.00021449511400651464, | |
| "loss": 0.888, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.071938495332235, | |
| "grad_norm": 1.1222225427627563, | |
| "learning_rate": 0.00021425081433224753, | |
| "loss": 1.0016, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.0741350906095553, | |
| "grad_norm": 1.3989416360855103, | |
| "learning_rate": 0.00021400651465798044, | |
| "loss": 0.8646, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.0763316858868754, | |
| "grad_norm": 1.1427356004714966, | |
| "learning_rate": 0.00021376221498371333, | |
| "loss": 0.8259, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0785282811641954, | |
| "grad_norm": 1.782548427581787, | |
| "learning_rate": 0.00021351791530944622, | |
| "loss": 1.134, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.0807248764415156, | |
| "grad_norm": 2.3628106117248535, | |
| "learning_rate": 0.00021327361563517914, | |
| "loss": 0.8043, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.0829214717188358, | |
| "grad_norm": 1.5517092943191528, | |
| "learning_rate": 0.00021302931596091203, | |
| "loss": 1.0221, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.085118066996156, | |
| "grad_norm": 1.2082037925720215, | |
| "learning_rate": 0.00021278501628664491, | |
| "loss": 0.8166, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.0873146622734762, | |
| "grad_norm": 1.522373914718628, | |
| "learning_rate": 0.00021254071661237783, | |
| "loss": 0.9166, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.0895112575507964, | |
| "grad_norm": 2.141221761703491, | |
| "learning_rate": 0.00021229641693811072, | |
| "loss": 0.8354, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.0917078528281163, | |
| "grad_norm": 1.0814484357833862, | |
| "learning_rate": 0.00021205211726384364, | |
| "loss": 0.8891, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.0939044481054365, | |
| "grad_norm": 1.5728553533554077, | |
| "learning_rate": 0.00021180781758957653, | |
| "loss": 0.9096, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.0961010433827567, | |
| "grad_norm": 1.8437684774398804, | |
| "learning_rate": 0.00021156351791530941, | |
| "loss": 0.9069, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.098297638660077, | |
| "grad_norm": 1.8091014623641968, | |
| "learning_rate": 0.00021131921824104233, | |
| "loss": 0.9404, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.100494233937397, | |
| "grad_norm": 1.0524958372116089, | |
| "learning_rate": 0.00021107491856677522, | |
| "loss": 1.0821, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.1026908292147173, | |
| "grad_norm": 0.9981535077095032, | |
| "learning_rate": 0.0002108306188925081, | |
| "loss": 0.7547, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.1048874244920373, | |
| "grad_norm": 1.9247081279754639, | |
| "learning_rate": 0.00021058631921824103, | |
| "loss": 0.9277, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.1070840197693574, | |
| "grad_norm": 0.8213505148887634, | |
| "learning_rate": 0.00021034201954397392, | |
| "loss": 0.7722, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.1092806150466776, | |
| "grad_norm": 1.3774635791778564, | |
| "learning_rate": 0.0002100977198697068, | |
| "loss": 1.1133, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.1114772103239978, | |
| "grad_norm": 1.130212664604187, | |
| "learning_rate": 0.00020985342019543972, | |
| "loss": 0.9232, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.113673805601318, | |
| "grad_norm": 1.0186429023742676, | |
| "learning_rate": 0.0002096091205211726, | |
| "loss": 0.9788, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.1158704008786382, | |
| "grad_norm": 1.8471168279647827, | |
| "learning_rate": 0.0002093648208469055, | |
| "loss": 1.0126, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.1180669961559582, | |
| "grad_norm": 1.7704755067825317, | |
| "learning_rate": 0.00020912052117263842, | |
| "loss": 0.9778, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.1202635914332784, | |
| "grad_norm": 1.1113084554672241, | |
| "learning_rate": 0.0002088762214983713, | |
| "loss": 0.9751, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.1224601867105986, | |
| "grad_norm": 1.502360224723816, | |
| "learning_rate": 0.00020863192182410422, | |
| "loss": 0.8652, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.1246567819879187, | |
| "grad_norm": 1.096174955368042, | |
| "learning_rate": 0.0002083876221498371, | |
| "loss": 0.8496, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.126853377265239, | |
| "grad_norm": 0.9662922024726868, | |
| "learning_rate": 0.00020814332247557, | |
| "loss": 1.0048, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.1290499725425591, | |
| "grad_norm": 1.0588403940200806, | |
| "learning_rate": 0.00020789902280130292, | |
| "loss": 0.8388, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.131246567819879, | |
| "grad_norm": 1.2862454652786255, | |
| "learning_rate": 0.0002076547231270358, | |
| "loss": 0.897, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.1334431630971993, | |
| "grad_norm": 2.5013811588287354, | |
| "learning_rate": 0.0002074104234527687, | |
| "loss": 0.9872, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.1356397583745195, | |
| "grad_norm": 1.9320764541625977, | |
| "learning_rate": 0.0002071661237785016, | |
| "loss": 1.0187, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.1378363536518397, | |
| "grad_norm": 1.1736301183700562, | |
| "learning_rate": 0.0002069218241042345, | |
| "loss": 0.6748, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.1400329489291599, | |
| "grad_norm": 1.287230134010315, | |
| "learning_rate": 0.0002066775244299674, | |
| "loss": 0.6739, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.14222954420648, | |
| "grad_norm": 1.5623549222946167, | |
| "learning_rate": 0.0002064332247557003, | |
| "loss": 0.9486, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.1444261394838002, | |
| "grad_norm": 1.071854591369629, | |
| "learning_rate": 0.0002061889250814332, | |
| "loss": 0.7981, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.1466227347611202, | |
| "grad_norm": 1.1585288047790527, | |
| "learning_rate": 0.0002059446254071661, | |
| "loss": 0.746, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.1488193300384404, | |
| "grad_norm": 2.46455979347229, | |
| "learning_rate": 0.000205700325732899, | |
| "loss": 0.6512, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.1510159253157606, | |
| "grad_norm": 1.350869059562683, | |
| "learning_rate": 0.0002054560260586319, | |
| "loss": 1.0259, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.1532125205930808, | |
| "grad_norm": 1.9205849170684814, | |
| "learning_rate": 0.0002052117263843648, | |
| "loss": 0.8838, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.155409115870401, | |
| "grad_norm": 0.9957300424575806, | |
| "learning_rate": 0.0002049674267100977, | |
| "loss": 1.1261, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.157605711147721, | |
| "grad_norm": 1.2502689361572266, | |
| "learning_rate": 0.00020472312703583058, | |
| "loss": 0.807, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.1598023064250411, | |
| "grad_norm": 1.0679610967636108, | |
| "learning_rate": 0.0002044788273615635, | |
| "loss": 0.9642, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.1619989017023613, | |
| "grad_norm": 1.3164231777191162, | |
| "learning_rate": 0.0002042345276872964, | |
| "loss": 0.6971, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.1641954969796815, | |
| "grad_norm": 1.2634435892105103, | |
| "learning_rate": 0.00020399022801302928, | |
| "loss": 0.6401, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.1663920922570017, | |
| "grad_norm": 1.0449302196502686, | |
| "learning_rate": 0.0002037459283387622, | |
| "loss": 0.7929, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.1685886875343219, | |
| "grad_norm": 1.0542372465133667, | |
| "learning_rate": 0.00020350162866449508, | |
| "loss": 1.004, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.170785282811642, | |
| "grad_norm": 1.2895562648773193, | |
| "learning_rate": 0.00020325732899022797, | |
| "loss": 0.9978, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.172981878088962, | |
| "grad_norm": 1.819820761680603, | |
| "learning_rate": 0.0002030130293159609, | |
| "loss": 0.9912, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.1751784733662822, | |
| "grad_norm": 1.0017704963684082, | |
| "learning_rate": 0.00020276872964169378, | |
| "loss": 0.8269, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.1773750686436024, | |
| "grad_norm": 1.1681811809539795, | |
| "learning_rate": 0.0002025244299674267, | |
| "loss": 0.7805, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.1795716639209226, | |
| "grad_norm": 1.0063347816467285, | |
| "learning_rate": 0.00020228013029315958, | |
| "loss": 0.8451, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.1817682591982428, | |
| "grad_norm": 1.3340643644332886, | |
| "learning_rate": 0.00020203583061889247, | |
| "loss": 0.8759, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.1839648544755628, | |
| "grad_norm": 1.1200175285339355, | |
| "learning_rate": 0.0002017915309446254, | |
| "loss": 1.131, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.186161449752883, | |
| "grad_norm": 1.030397653579712, | |
| "learning_rate": 0.00020154723127035828, | |
| "loss": 0.8012, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.1883580450302031, | |
| "grad_norm": 2.4552953243255615, | |
| "learning_rate": 0.00020130293159609117, | |
| "loss": 0.9271, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.1905546403075233, | |
| "grad_norm": 1.2497228384017944, | |
| "learning_rate": 0.00020105863192182408, | |
| "loss": 0.776, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.1927512355848435, | |
| "grad_norm": 1.4785922765731812, | |
| "learning_rate": 0.00020081433224755697, | |
| "loss": 0.9259, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.1949478308621637, | |
| "grad_norm": 1.197167992591858, | |
| "learning_rate": 0.00020057003257328986, | |
| "loss": 0.9017, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.197144426139484, | |
| "grad_norm": 1.6644405126571655, | |
| "learning_rate": 0.00020032573289902278, | |
| "loss": 1.0203, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.1993410214168039, | |
| "grad_norm": 1.0461171865463257, | |
| "learning_rate": 0.00020008143322475567, | |
| "loss": 0.9453, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.201537616694124, | |
| "grad_norm": 1.1469635963439941, | |
| "learning_rate": 0.00019983713355048856, | |
| "loss": 1.1816, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.2037342119714443, | |
| "grad_norm": 1.1605957746505737, | |
| "learning_rate": 0.00019959283387622147, | |
| "loss": 0.9087, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.2059308072487644, | |
| "grad_norm": 1.235202670097351, | |
| "learning_rate": 0.00019934853420195436, | |
| "loss": 1.1788, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.2081274025260846, | |
| "grad_norm": 2.015349864959717, | |
| "learning_rate": 0.00019910423452768728, | |
| "loss": 1.064, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.2103239978034046, | |
| "grad_norm": 1.7907201051712036, | |
| "learning_rate": 0.00019885993485342017, | |
| "loss": 0.8684, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.2125205930807248, | |
| "grad_norm": 1.4917091131210327, | |
| "learning_rate": 0.00019861563517915306, | |
| "loss": 1.0924, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.214717188358045, | |
| "grad_norm": 1.3111851215362549, | |
| "learning_rate": 0.00019837133550488597, | |
| "loss": 1.0327, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.2169137836353652, | |
| "grad_norm": 0.9068641066551208, | |
| "learning_rate": 0.00019812703583061886, | |
| "loss": 0.8682, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.2191103789126854, | |
| "grad_norm": 1.3541110754013062, | |
| "learning_rate": 0.00019788273615635175, | |
| "loss": 0.9349, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.2213069741900056, | |
| "grad_norm": 1.1679444313049316, | |
| "learning_rate": 0.00019763843648208467, | |
| "loss": 0.8017, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.2235035694673257, | |
| "grad_norm": 1.2594337463378906, | |
| "learning_rate": 0.00019739413680781756, | |
| "loss": 0.9463, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.2257001647446457, | |
| "grad_norm": 2.6442201137542725, | |
| "learning_rate": 0.00019714983713355045, | |
| "loss": 0.8009, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.227896760021966, | |
| "grad_norm": 2.6649391651153564, | |
| "learning_rate": 0.00019690553745928336, | |
| "loss": 1.1733, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.230093355299286, | |
| "grad_norm": 0.9813634753227234, | |
| "learning_rate": 0.00019666123778501625, | |
| "loss": 0.9458, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.2322899505766063, | |
| "grad_norm": 1.3521826267242432, | |
| "learning_rate": 0.00019641693811074917, | |
| "loss": 0.8761, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.2344865458539265, | |
| "grad_norm": 1.7125831842422485, | |
| "learning_rate": 0.00019617263843648206, | |
| "loss": 0.9374, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.2366831411312464, | |
| "grad_norm": 1.490022897720337, | |
| "learning_rate": 0.00019592833876221495, | |
| "loss": 1.1167, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.2388797364085666, | |
| "grad_norm": 1.1457475423812866, | |
| "learning_rate": 0.00019568403908794786, | |
| "loss": 0.7711, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.2410763316858868, | |
| "grad_norm": 1.2306512594223022, | |
| "learning_rate": 0.00019543973941368075, | |
| "loss": 1.0045, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.243272926963207, | |
| "grad_norm": 2.0270729064941406, | |
| "learning_rate": 0.00019519543973941364, | |
| "loss": 0.8869, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.2454695222405272, | |
| "grad_norm": 1.5920954942703247, | |
| "learning_rate": 0.00019495114006514656, | |
| "loss": 0.8605, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.2476661175178474, | |
| "grad_norm": 1.633131980895996, | |
| "learning_rate": 0.00019470684039087945, | |
| "loss": 0.9852, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.2498627127951676, | |
| "grad_norm": 1.4640898704528809, | |
| "learning_rate": 0.00019446254071661234, | |
| "loss": 0.819, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.2520593080724876, | |
| "grad_norm": 1.4026305675506592, | |
| "learning_rate": 0.00019421824104234525, | |
| "loss": 0.6648, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.2542559033498077, | |
| "grad_norm": 1.6480371952056885, | |
| "learning_rate": 0.00019397394136807814, | |
| "loss": 1.0672, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.256452498627128, | |
| "grad_norm": 1.193241834640503, | |
| "learning_rate": 0.00019372964169381103, | |
| "loss": 0.9375, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.2586490939044481, | |
| "grad_norm": 1.4845542907714844, | |
| "learning_rate": 0.00019348534201954395, | |
| "loss": 0.8896, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.2608456891817683, | |
| "grad_norm": 1.12018620967865, | |
| "learning_rate": 0.00019324104234527684, | |
| "loss": 0.9258, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.2630422844590883, | |
| "grad_norm": 1.212038516998291, | |
| "learning_rate": 0.00019299674267100975, | |
| "loss": 0.9859, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.2652388797364087, | |
| "grad_norm": 1.103137731552124, | |
| "learning_rate": 0.00019275244299674264, | |
| "loss": 0.8246, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.2674354750137287, | |
| "grad_norm": 1.1651033163070679, | |
| "learning_rate": 0.00019250814332247553, | |
| "loss": 1.1447, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.2696320702910489, | |
| "grad_norm": 1.0638104677200317, | |
| "learning_rate": 0.00019226384364820845, | |
| "loss": 0.6808, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.271828665568369, | |
| "grad_norm": 1.3430167436599731, | |
| "learning_rate": 0.00019201954397394134, | |
| "loss": 0.6916, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.2740252608456892, | |
| "grad_norm": 0.9743119478225708, | |
| "learning_rate": 0.00019177524429967423, | |
| "loss": 0.7395, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.2762218561230094, | |
| "grad_norm": 1.242250680923462, | |
| "learning_rate": 0.00019153094462540714, | |
| "loss": 0.9501, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.2784184514003294, | |
| "grad_norm": 1.5769577026367188, | |
| "learning_rate": 0.00019128664495114003, | |
| "loss": 0.7829, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.2806150466776496, | |
| "grad_norm": 0.9817864298820496, | |
| "learning_rate": 0.00019104234527687292, | |
| "loss": 0.7999, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.2828116419549698, | |
| "grad_norm": 2.2283880710601807, | |
| "learning_rate": 0.00019079804560260584, | |
| "loss": 0.7766, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.28500823723229, | |
| "grad_norm": 1.8748104572296143, | |
| "learning_rate": 0.00019055374592833873, | |
| "loss": 0.9424, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.2872048325096102, | |
| "grad_norm": 12.249737739562988, | |
| "learning_rate": 0.00019030944625407164, | |
| "loss": 0.8662, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.2894014277869301, | |
| "grad_norm": 1.352903962135315, | |
| "learning_rate": 0.00019006514657980453, | |
| "loss": 0.866, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.2915980230642505, | |
| "grad_norm": 1.704245686531067, | |
| "learning_rate": 0.00018982084690553742, | |
| "loss": 0.9203, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.2937946183415705, | |
| "grad_norm": 2.5010740756988525, | |
| "learning_rate": 0.00018957654723127034, | |
| "loss": 0.8425, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.2959912136188907, | |
| "grad_norm": 1.8178290128707886, | |
| "learning_rate": 0.00018933224755700323, | |
| "loss": 0.8844, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.2981878088962109, | |
| "grad_norm": 1.080711007118225, | |
| "learning_rate": 0.00018908794788273612, | |
| "loss": 0.6711, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.300384404173531, | |
| "grad_norm": 1.2653353214263916, | |
| "learning_rate": 0.00018884364820846903, | |
| "loss": 0.8724, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.3025809994508513, | |
| "grad_norm": 1.0701664686203003, | |
| "learning_rate": 0.00018859934853420192, | |
| "loss": 0.7468, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.3047775947281712, | |
| "grad_norm": 1.424072027206421, | |
| "learning_rate": 0.0001883550488599348, | |
| "loss": 0.91, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.3069741900054914, | |
| "grad_norm": 1.383643627166748, | |
| "learning_rate": 0.00018811074918566773, | |
| "loss": 1.0485, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.3091707852828116, | |
| "grad_norm": 1.4447416067123413, | |
| "learning_rate": 0.00018786644951140062, | |
| "loss": 0.7424, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.3113673805601318, | |
| "grad_norm": 1.198759913444519, | |
| "learning_rate": 0.0001876221498371335, | |
| "loss": 0.8248, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.313563975837452, | |
| "grad_norm": 1.6977317333221436, | |
| "learning_rate": 0.00018737785016286645, | |
| "loss": 0.7539, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.3157605711147722, | |
| "grad_norm": 1.3681342601776123, | |
| "learning_rate": 0.00018713355048859934, | |
| "loss": 0.869, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.3179571663920924, | |
| "grad_norm": 0.8428243398666382, | |
| "learning_rate": 0.00018688925081433225, | |
| "loss": 1.069, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.3201537616694123, | |
| "grad_norm": 1.107496976852417, | |
| "learning_rate": 0.00018664495114006514, | |
| "loss": 0.9182, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.3223503569467325, | |
| "grad_norm": 1.169148325920105, | |
| "learning_rate": 0.00018640065146579803, | |
| "loss": 0.8348, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.3245469522240527, | |
| "grad_norm": 1.0859352350234985, | |
| "learning_rate": 0.00018615635179153095, | |
| "loss": 1.0487, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.326743547501373, | |
| "grad_norm": 1.5463238954544067, | |
| "learning_rate": 0.00018591205211726384, | |
| "loss": 0.9265, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.328940142778693, | |
| "grad_norm": 1.1321184635162354, | |
| "learning_rate": 0.00018566775244299675, | |
| "loss": 0.9892, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.331136738056013, | |
| "grad_norm": 1.6157218217849731, | |
| "learning_rate": 0.00018542345276872964, | |
| "loss": 0.7491, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 1.0804728269577026, | |
| "learning_rate": 0.00018517915309446253, | |
| "loss": 0.6766, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.3355299286106534, | |
| "grad_norm": 1.231313705444336, | |
| "learning_rate": 0.00018493485342019545, | |
| "loss": 0.8942, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.3377265238879736, | |
| "grad_norm": 1.341189980506897, | |
| "learning_rate": 0.00018469055374592834, | |
| "loss": 0.9648, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.3399231191652938, | |
| "grad_norm": 1.6832330226898193, | |
| "learning_rate": 0.00018444625407166123, | |
| "loss": 0.9547, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.342119714442614, | |
| "grad_norm": 1.6071604490280151, | |
| "learning_rate": 0.00018420195439739414, | |
| "loss": 0.9765, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.3443163097199342, | |
| "grad_norm": 3.548957586288452, | |
| "learning_rate": 0.00018395765472312703, | |
| "loss": 0.8427, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.3465129049972542, | |
| "grad_norm": 1.1544692516326904, | |
| "learning_rate": 0.00018371335504885992, | |
| "loss": 0.9211, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.3487095002745744, | |
| "grad_norm": 1.376746654510498, | |
| "learning_rate": 0.00018346905537459284, | |
| "loss": 0.9043, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.3509060955518946, | |
| "grad_norm": 1.2523698806762695, | |
| "learning_rate": 0.00018322475570032573, | |
| "loss": 1.0389, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.3531026908292147, | |
| "grad_norm": 1.1789697408676147, | |
| "learning_rate": 0.00018298045602605864, | |
| "loss": 0.975, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.355299286106535, | |
| "grad_norm": 0.9141327738761902, | |
| "learning_rate": 0.00018273615635179153, | |
| "loss": 0.8723, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.357495881383855, | |
| "grad_norm": 1.0598615407943726, | |
| "learning_rate": 0.00018249185667752442, | |
| "loss": 1.1544, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.359692476661175, | |
| "grad_norm": 1.1385844945907593, | |
| "learning_rate": 0.00018224755700325734, | |
| "loss": 0.6705, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.3618890719384953, | |
| "grad_norm": 1.4906302690505981, | |
| "learning_rate": 0.00018200325732899023, | |
| "loss": 0.8348, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.3640856672158155, | |
| "grad_norm": 2.3867242336273193, | |
| "learning_rate": 0.00018175895765472312, | |
| "loss": 1.1355, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.3662822624931357, | |
| "grad_norm": 1.430229902267456, | |
| "learning_rate": 0.00018151465798045603, | |
| "loss": 0.8373, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.3684788577704559, | |
| "grad_norm": 0.9465194940567017, | |
| "learning_rate": 0.00018127035830618892, | |
| "loss": 0.7438, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.370675453047776, | |
| "grad_norm": 1.255829930305481, | |
| "learning_rate": 0.0001810260586319218, | |
| "loss": 0.9055, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.372872048325096, | |
| "grad_norm": 1.6284000873565674, | |
| "learning_rate": 0.00018078175895765473, | |
| "loss": 0.9432, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3750686436024162, | |
| "grad_norm": 0.9144400358200073, | |
| "learning_rate": 0.00018053745928338762, | |
| "loss": 0.7964, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.3772652388797364, | |
| "grad_norm": 0.842767059803009, | |
| "learning_rate": 0.0001802931596091205, | |
| "loss": 0.8065, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.3794618341570566, | |
| "grad_norm": 1.7188353538513184, | |
| "learning_rate": 0.00018004885993485342, | |
| "loss": 0.6765, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.3816584294343768, | |
| "grad_norm": 2.4767227172851562, | |
| "learning_rate": 0.0001798045602605863, | |
| "loss": 0.9008, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.3838550247116967, | |
| "grad_norm": 1.179839015007019, | |
| "learning_rate": 0.00017956026058631923, | |
| "loss": 0.792, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.3860516199890172, | |
| "grad_norm": 1.023276448249817, | |
| "learning_rate": 0.00017931596091205212, | |
| "loss": 0.8494, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.3882482152663371, | |
| "grad_norm": 1.0346344709396362, | |
| "learning_rate": 0.000179071661237785, | |
| "loss": 1.0668, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.3904448105436573, | |
| "grad_norm": 1.955815076828003, | |
| "learning_rate": 0.00017882736156351792, | |
| "loss": 0.9543, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.3926414058209775, | |
| "grad_norm": 0.8277578949928284, | |
| "learning_rate": 0.0001785830618892508, | |
| "loss": 0.7639, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.3948380010982977, | |
| "grad_norm": 1.2157933712005615, | |
| "learning_rate": 0.0001783387622149837, | |
| "loss": 0.8648, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.3970345963756179, | |
| "grad_norm": 0.9396011829376221, | |
| "learning_rate": 0.00017809446254071662, | |
| "loss": 1.0109, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.3992311916529379, | |
| "grad_norm": 1.1028892993927002, | |
| "learning_rate": 0.0001778501628664495, | |
| "loss": 0.8305, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.401427786930258, | |
| "grad_norm": 1.3041939735412598, | |
| "learning_rate": 0.0001776058631921824, | |
| "loss": 0.7819, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.4036243822075782, | |
| "grad_norm": 1.803083062171936, | |
| "learning_rate": 0.0001773615635179153, | |
| "loss": 0.8393, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.4058209774848984, | |
| "grad_norm": 1.051144003868103, | |
| "learning_rate": 0.0001771172638436482, | |
| "loss": 1.139, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.4080175727622186, | |
| "grad_norm": 1.7257338762283325, | |
| "learning_rate": 0.0001768729641693811, | |
| "loss": 0.8357, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.4102141680395386, | |
| "grad_norm": 1.035652995109558, | |
| "learning_rate": 0.000176628664495114, | |
| "loss": 1.0127, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.412410763316859, | |
| "grad_norm": 3.475193500518799, | |
| "learning_rate": 0.0001763843648208469, | |
| "loss": 0.7518, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.414607358594179, | |
| "grad_norm": 1.0679947137832642, | |
| "learning_rate": 0.0001761400651465798, | |
| "loss": 0.8067, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.4168039538714992, | |
| "grad_norm": 2.229825496673584, | |
| "learning_rate": 0.0001758957654723127, | |
| "loss": 0.953, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.4190005491488193, | |
| "grad_norm": 1.1916773319244385, | |
| "learning_rate": 0.0001756514657980456, | |
| "loss": 0.909, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.4211971444261395, | |
| "grad_norm": 1.1973612308502197, | |
| "learning_rate": 0.0001754071661237785, | |
| "loss": 0.8404, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.4233937397034597, | |
| "grad_norm": 1.0691834688186646, | |
| "learning_rate": 0.0001751628664495114, | |
| "loss": 0.8783, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.4255903349807797, | |
| "grad_norm": 1.3762271404266357, | |
| "learning_rate": 0.00017491856677524429, | |
| "loss": 0.9292, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.4277869302580999, | |
| "grad_norm": 0.9204732179641724, | |
| "learning_rate": 0.0001746742671009772, | |
| "loss": 0.7545, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.42998352553542, | |
| "grad_norm": 0.8237535357475281, | |
| "learning_rate": 0.0001744299674267101, | |
| "loss": 0.8127, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.4321801208127403, | |
| "grad_norm": 0.9223546981811523, | |
| "learning_rate": 0.00017418566775244298, | |
| "loss": 0.848, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.4343767160900605, | |
| "grad_norm": 1.0721185207366943, | |
| "learning_rate": 0.0001739413680781759, | |
| "loss": 0.8983, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.4365733113673804, | |
| "grad_norm": 1.3847533464431763, | |
| "learning_rate": 0.00017369706840390879, | |
| "loss": 0.8853, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.4387699066447008, | |
| "grad_norm": 1.6580514907836914, | |
| "learning_rate": 0.0001734527687296417, | |
| "loss": 0.8249, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.4409665019220208, | |
| "grad_norm": 1.5866856575012207, | |
| "learning_rate": 0.0001732084690553746, | |
| "loss": 0.975, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.443163097199341, | |
| "grad_norm": 1.0152703523635864, | |
| "learning_rate": 0.00017296416938110748, | |
| "loss": 1.0054, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.4453596924766612, | |
| "grad_norm": 3.3993537425994873, | |
| "learning_rate": 0.0001727198697068404, | |
| "loss": 1.0471, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.4475562877539814, | |
| "grad_norm": 1.275221824645996, | |
| "learning_rate": 0.00017247557003257329, | |
| "loss": 1.0855, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.4497528830313016, | |
| "grad_norm": 0.9472977519035339, | |
| "learning_rate": 0.00017223127035830618, | |
| "loss": 0.827, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.4519494783086215, | |
| "grad_norm": 1.1003230810165405, | |
| "learning_rate": 0.0001719869706840391, | |
| "loss": 0.7401, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.4541460735859417, | |
| "grad_norm": 1.442337989807129, | |
| "learning_rate": 0.00017174267100977198, | |
| "loss": 0.9002, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.456342668863262, | |
| "grad_norm": 1.229732871055603, | |
| "learning_rate": 0.00017149837133550487, | |
| "loss": 0.9264, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.458539264140582, | |
| "grad_norm": 0.9192113280296326, | |
| "learning_rate": 0.00017125407166123779, | |
| "loss": 0.7956, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.4607358594179023, | |
| "grad_norm": 1.039777159690857, | |
| "learning_rate": 0.00017100977198697068, | |
| "loss": 0.9775, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.4629324546952225, | |
| "grad_norm": 1.052749752998352, | |
| "learning_rate": 0.00017076547231270356, | |
| "loss": 1.0048, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.4651290499725427, | |
| "grad_norm": 1.6796581745147705, | |
| "learning_rate": 0.00017052117263843648, | |
| "loss": 0.8261, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.4673256452498626, | |
| "grad_norm": 1.010292649269104, | |
| "learning_rate": 0.00017027687296416937, | |
| "loss": 1.0151, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.4695222405271828, | |
| "grad_norm": 1.3808190822601318, | |
| "learning_rate": 0.00017003257328990229, | |
| "loss": 0.8036, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.471718835804503, | |
| "grad_norm": 1.405461311340332, | |
| "learning_rate": 0.00016978827361563518, | |
| "loss": 0.8817, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.4739154310818232, | |
| "grad_norm": 1.8762201070785522, | |
| "learning_rate": 0.00016954397394136806, | |
| "loss": 1.0388, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.4761120263591434, | |
| "grad_norm": 1.1670676469802856, | |
| "learning_rate": 0.00016929967426710098, | |
| "loss": 0.905, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.4783086216364634, | |
| "grad_norm": 1.347756266593933, | |
| "learning_rate": 0.00016905537459283387, | |
| "loss": 0.822, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.4805052169137836, | |
| "grad_norm": 0.9841303825378418, | |
| "learning_rate": 0.00016881107491856676, | |
| "loss": 1.0208, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.4827018121911038, | |
| "grad_norm": 1.5004286766052246, | |
| "learning_rate": 0.00016856677524429968, | |
| "loss": 0.7704, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.484898407468424, | |
| "grad_norm": 1.1298267841339111, | |
| "learning_rate": 0.00016832247557003257, | |
| "loss": 0.8899, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.4870950027457441, | |
| "grad_norm": 1.212978482246399, | |
| "learning_rate": 0.00016807817589576545, | |
| "loss": 0.8157, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.4892915980230643, | |
| "grad_norm": 1.191117286682129, | |
| "learning_rate": 0.00016783387622149837, | |
| "loss": 0.9365, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.4914881933003845, | |
| "grad_norm": 0.9882132411003113, | |
| "learning_rate": 0.00016758957654723126, | |
| "loss": 0.9006, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.4936847885777045, | |
| "grad_norm": 6.74979829788208, | |
| "learning_rate": 0.00016734527687296415, | |
| "loss": 1.1473, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.4958813838550247, | |
| "grad_norm": 1.1934260129928589, | |
| "learning_rate": 0.00016710097719869707, | |
| "loss": 0.9049, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.4980779791323449, | |
| "grad_norm": 1.019740104675293, | |
| "learning_rate": 0.00016685667752442995, | |
| "loss": 1.0111, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.500274574409665, | |
| "grad_norm": 1.3660500049591064, | |
| "learning_rate": 0.00016661237785016287, | |
| "loss": 0.6096, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.5024711696869852, | |
| "grad_norm": 1.236396074295044, | |
| "learning_rate": 0.00016636807817589576, | |
| "loss": 0.7315, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.5046677649643052, | |
| "grad_norm": 1.1221923828125, | |
| "learning_rate": 0.00016612377850162865, | |
| "loss": 0.9671, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.5068643602416256, | |
| "grad_norm": 1.4163709878921509, | |
| "learning_rate": 0.00016587947882736157, | |
| "loss": 1.0352, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.5090609555189456, | |
| "grad_norm": 1.8958035707473755, | |
| "learning_rate": 0.00016563517915309445, | |
| "loss": 0.9268, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.5112575507962658, | |
| "grad_norm": 1.3060574531555176, | |
| "learning_rate": 0.00016539087947882734, | |
| "loss": 0.9716, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.513454146073586, | |
| "grad_norm": 0.9746711254119873, | |
| "learning_rate": 0.00016514657980456026, | |
| "loss": 0.761, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.515650741350906, | |
| "grad_norm": 1.2283755540847778, | |
| "learning_rate": 0.00016490228013029315, | |
| "loss": 0.845, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.5178473366282264, | |
| "grad_norm": 1.9815654754638672, | |
| "learning_rate": 0.00016465798045602604, | |
| "loss": 1.0402, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.5200439319055463, | |
| "grad_norm": 1.644201636314392, | |
| "learning_rate": 0.00016441368078175895, | |
| "loss": 0.9276, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.5222405271828665, | |
| "grad_norm": 1.0192824602127075, | |
| "learning_rate": 0.00016416938110749184, | |
| "loss": 0.9284, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.5244371224601867, | |
| "grad_norm": 1.3996363878250122, | |
| "learning_rate": 0.00016392508143322476, | |
| "loss": 0.8212, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.526633717737507, | |
| "grad_norm": 0.9490194320678711, | |
| "learning_rate": 0.00016368078175895765, | |
| "loss": 0.8737, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.528830313014827, | |
| "grad_norm": 1.0806269645690918, | |
| "learning_rate": 0.00016343648208469054, | |
| "loss": 0.8336, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.531026908292147, | |
| "grad_norm": 0.9588049650192261, | |
| "learning_rate": 0.00016319218241042346, | |
| "loss": 0.8639, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.5332235035694675, | |
| "grad_norm": 0.9937188625335693, | |
| "learning_rate": 0.00016294788273615634, | |
| "loss": 0.8339, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.5354200988467874, | |
| "grad_norm": 1.173140287399292, | |
| "learning_rate": 0.00016270358306188923, | |
| "loss": 0.9153, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.5376166941241076, | |
| "grad_norm": 0.9692608118057251, | |
| "learning_rate": 0.00016245928338762215, | |
| "loss": 1.0605, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.5398132894014278, | |
| "grad_norm": 1.2687755823135376, | |
| "learning_rate": 0.00016221498371335504, | |
| "loss": 0.8685, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.5420098846787478, | |
| "grad_norm": 1.347936749458313, | |
| "learning_rate": 0.00016197068403908793, | |
| "loss": 0.8561, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.5442064799560682, | |
| "grad_norm": 0.875302255153656, | |
| "learning_rate": 0.00016172638436482084, | |
| "loss": 0.7182, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.5464030752333882, | |
| "grad_norm": 1.016817331314087, | |
| "learning_rate": 0.00016148208469055373, | |
| "loss": 0.7293, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.5485996705107083, | |
| "grad_norm": 1.2300879955291748, | |
| "learning_rate": 0.00016123778501628662, | |
| "loss": 0.8671, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.5507962657880285, | |
| "grad_norm": 1.3806672096252441, | |
| "learning_rate": 0.00016099348534201954, | |
| "loss": 0.8746, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.5529928610653487, | |
| "grad_norm": 1.736348032951355, | |
| "learning_rate": 0.00016074918566775243, | |
| "loss": 0.9862, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.555189456342669, | |
| "grad_norm": 1.498680591583252, | |
| "learning_rate": 0.00016050488599348534, | |
| "loss": 0.9622, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.5573860516199889, | |
| "grad_norm": 1.5123640298843384, | |
| "learning_rate": 0.00016026058631921823, | |
| "loss": 1.0653, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.5595826468973093, | |
| "grad_norm": 1.0846017599105835, | |
| "learning_rate": 0.00016001628664495112, | |
| "loss": 0.6153, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.5617792421746293, | |
| "grad_norm": 1.3977952003479004, | |
| "learning_rate": 0.00015977198697068404, | |
| "loss": 1.01, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.5639758374519495, | |
| "grad_norm": 1.300642490386963, | |
| "learning_rate": 0.00015952768729641693, | |
| "loss": 0.6798, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.5661724327292696, | |
| "grad_norm": 1.6283328533172607, | |
| "learning_rate": 0.00015928338762214982, | |
| "loss": 0.7862, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.5683690280065898, | |
| "grad_norm": 0.8933014869689941, | |
| "learning_rate": 0.00015903908794788273, | |
| "loss": 0.7287, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.57056562328391, | |
| "grad_norm": 1.3467384576797485, | |
| "learning_rate": 0.00015879478827361562, | |
| "loss": 0.9296, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.57276221856123, | |
| "grad_norm": 2.0327916145324707, | |
| "learning_rate": 0.0001585504885993485, | |
| "loss": 1.1482, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.5749588138385504, | |
| "grad_norm": 1.0758147239685059, | |
| "learning_rate": 0.00015830618892508143, | |
| "loss": 0.8874, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.5771554091158704, | |
| "grad_norm": 1.9648406505584717, | |
| "learning_rate": 0.00015806188925081432, | |
| "loss": 0.8308, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.5793520043931906, | |
| "grad_norm": 1.5760283470153809, | |
| "learning_rate": 0.00015781758957654723, | |
| "loss": 0.8988, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.5815485996705108, | |
| "grad_norm": 1.0667542219161987, | |
| "learning_rate": 0.00015757328990228012, | |
| "loss": 0.7887, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.5837451949478307, | |
| "grad_norm": 0.9665087461471558, | |
| "learning_rate": 0.000157328990228013, | |
| "loss": 1.0063, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.5859417902251511, | |
| "grad_norm": 1.3701244592666626, | |
| "learning_rate": 0.00015708469055374593, | |
| "loss": 0.9551, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.588138385502471, | |
| "grad_norm": 1.3653502464294434, | |
| "learning_rate": 0.00015684039087947882, | |
| "loss": 0.8966, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.5903349807797913, | |
| "grad_norm": 1.5704392194747925, | |
| "learning_rate": 0.0001565960912052117, | |
| "loss": 0.7833, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.5925315760571115, | |
| "grad_norm": 1.3713306188583374, | |
| "learning_rate": 0.00015635179153094462, | |
| "loss": 1.0065, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.5947281713344317, | |
| "grad_norm": 1.6414796113967896, | |
| "learning_rate": 0.0001561074918566775, | |
| "loss": 0.9069, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.5969247666117519, | |
| "grad_norm": 1.0879948139190674, | |
| "learning_rate": 0.0001558631921824104, | |
| "loss": 0.9571, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.5991213618890718, | |
| "grad_norm": 1.090989112854004, | |
| "learning_rate": 0.00015561889250814332, | |
| "loss": 0.9503, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.6013179571663922, | |
| "grad_norm": 3.652448892593384, | |
| "learning_rate": 0.0001553745928338762, | |
| "loss": 0.8992, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.6035145524437122, | |
| "grad_norm": 1.2672741413116455, | |
| "learning_rate": 0.0001551302931596091, | |
| "loss": 0.8807, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.6057111477210324, | |
| "grad_norm": 0.7997329235076904, | |
| "learning_rate": 0.000154885993485342, | |
| "loss": 0.606, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.6079077429983526, | |
| "grad_norm": 1.6741856336593628, | |
| "learning_rate": 0.0001546416938110749, | |
| "loss": 0.9221, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.6101043382756726, | |
| "grad_norm": 1.914759874343872, | |
| "learning_rate": 0.00015439739413680782, | |
| "loss": 1.154, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.612300933552993, | |
| "grad_norm": 1.01466703414917, | |
| "learning_rate": 0.0001541530944625407, | |
| "loss": 0.9454, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.614497528830313, | |
| "grad_norm": 1.3268802165985107, | |
| "learning_rate": 0.0001539087947882736, | |
| "loss": 0.9796, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.6166941241076331, | |
| "grad_norm": 1.0863169431686401, | |
| "learning_rate": 0.0001536644951140065, | |
| "loss": 0.9613, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.6188907193849533, | |
| "grad_norm": 1.5867358446121216, | |
| "learning_rate": 0.0001534201954397394, | |
| "loss": 0.9142, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.6210873146622735, | |
| "grad_norm": 2.6295602321624756, | |
| "learning_rate": 0.0001531758957654723, | |
| "loss": 0.8732, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.6232839099395937, | |
| "grad_norm": 1.5077576637268066, | |
| "learning_rate": 0.0001529315960912052, | |
| "loss": 0.943, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.6254805052169137, | |
| "grad_norm": 1.1349165439605713, | |
| "learning_rate": 0.0001526872964169381, | |
| "loss": 0.9143, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.627677100494234, | |
| "grad_norm": 1.1518207788467407, | |
| "learning_rate": 0.00015244299674267099, | |
| "loss": 1.0074, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.629873695771554, | |
| "grad_norm": 2.2564549446105957, | |
| "learning_rate": 0.0001521986970684039, | |
| "loss": 0.8593, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.6320702910488742, | |
| "grad_norm": 1.5264606475830078, | |
| "learning_rate": 0.0001519543973941368, | |
| "loss": 0.8201, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.6342668863261944, | |
| "grad_norm": 1.0632388591766357, | |
| "learning_rate": 0.00015171009771986968, | |
| "loss": 0.7329, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.6364634816035144, | |
| "grad_norm": 1.3610093593597412, | |
| "learning_rate": 0.0001514657980456026, | |
| "loss": 0.7716, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.6386600768808348, | |
| "grad_norm": 0.7864057421684265, | |
| "learning_rate": 0.0001512214983713355, | |
| "loss": 0.5905, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.6408566721581548, | |
| "grad_norm": 1.6561729907989502, | |
| "learning_rate": 0.0001509771986970684, | |
| "loss": 0.8471, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.643053267435475, | |
| "grad_norm": 1.4799500703811646, | |
| "learning_rate": 0.0001507328990228013, | |
| "loss": 0.7494, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.6452498627127952, | |
| "grad_norm": 0.9044152498245239, | |
| "learning_rate": 0.00015048859934853418, | |
| "loss": 0.7549, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.6474464579901154, | |
| "grad_norm": 0.9943232536315918, | |
| "learning_rate": 0.0001502442996742671, | |
| "loss": 0.772, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.6496430532674355, | |
| "grad_norm": 1.2595269680023193, | |
| "learning_rate": 0.00015, | |
| "loss": 0.9264, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.6518396485447555, | |
| "grad_norm": 3.0879321098327637, | |
| "learning_rate": 0.00014975570032573288, | |
| "loss": 1.0204, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.654036243822076, | |
| "grad_norm": 2.1531927585601807, | |
| "learning_rate": 0.0001495114006514658, | |
| "loss": 1.1315, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.656232839099396, | |
| "grad_norm": 1.7019902467727661, | |
| "learning_rate": 0.00014926710097719868, | |
| "loss": 0.7552, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.658429434376716, | |
| "grad_norm": 1.5576536655426025, | |
| "learning_rate": 0.00014902280130293157, | |
| "loss": 0.8689, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.6606260296540363, | |
| "grad_norm": 0.9407386779785156, | |
| "learning_rate": 0.0001487785016286645, | |
| "loss": 0.8607, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.6628226249313562, | |
| "grad_norm": 1.333757758140564, | |
| "learning_rate": 0.00014853420195439738, | |
| "loss": 0.9623, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.6650192202086767, | |
| "grad_norm": 0.9244624376296997, | |
| "learning_rate": 0.0001482899022801303, | |
| "loss": 0.8952, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.6672158154859966, | |
| "grad_norm": 1.361205816268921, | |
| "learning_rate": 0.00014804560260586318, | |
| "loss": 0.9607, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.6694124107633168, | |
| "grad_norm": 1.3580145835876465, | |
| "learning_rate": 0.00014780130293159607, | |
| "loss": 0.8567, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.671609006040637, | |
| "grad_norm": 1.1472671031951904, | |
| "learning_rate": 0.000147557003257329, | |
| "loss": 0.8411, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.6738056013179572, | |
| "grad_norm": 1.0054603815078735, | |
| "learning_rate": 0.00014731270358306188, | |
| "loss": 0.8871, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.6760021965952774, | |
| "grad_norm": 1.0102900266647339, | |
| "learning_rate": 0.00014706840390879477, | |
| "loss": 0.8451, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.6781987918725974, | |
| "grad_norm": 1.0074560642242432, | |
| "learning_rate": 0.00014682410423452768, | |
| "loss": 1.0188, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.6803953871499178, | |
| "grad_norm": 0.9837278723716736, | |
| "learning_rate": 0.00014657980456026057, | |
| "loss": 0.8967, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.6825919824272377, | |
| "grad_norm": 1.156173825263977, | |
| "learning_rate": 0.00014633550488599346, | |
| "loss": 0.7818, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.684788577704558, | |
| "grad_norm": 1.1852209568023682, | |
| "learning_rate": 0.00014609120521172638, | |
| "loss": 0.91, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.6869851729818781, | |
| "grad_norm": 1.3577544689178467, | |
| "learning_rate": 0.00014584690553745927, | |
| "loss": 0.8128, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.689181768259198, | |
| "grad_norm": 0.9448763728141785, | |
| "learning_rate": 0.00014560260586319216, | |
| "loss": 0.8271, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.6913783635365185, | |
| "grad_norm": 1.0891531705856323, | |
| "learning_rate": 0.00014535830618892507, | |
| "loss": 1.0133, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.6935749588138385, | |
| "grad_norm": 1.0757027864456177, | |
| "learning_rate": 0.00014511400651465796, | |
| "loss": 1.0497, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.6957715540911587, | |
| "grad_norm": 1.0277023315429688, | |
| "learning_rate": 0.00014486970684039088, | |
| "loss": 1.0265, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.6979681493684788, | |
| "grad_norm": 1.1304364204406738, | |
| "learning_rate": 0.00014462540716612377, | |
| "loss": 0.8129, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.700164744645799, | |
| "grad_norm": 1.291098713874817, | |
| "learning_rate": 0.00014438110749185666, | |
| "loss": 0.7548, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.7023613399231192, | |
| "grad_norm": 1.4948410987854004, | |
| "learning_rate": 0.00014413680781758957, | |
| "loss": 0.9142, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.7045579352004392, | |
| "grad_norm": 1.1901264190673828, | |
| "learning_rate": 0.00014389250814332246, | |
| "loss": 0.6685, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.7067545304777596, | |
| "grad_norm": 1.1417033672332764, | |
| "learning_rate": 0.00014364820846905535, | |
| "loss": 0.9447, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.7089511257550796, | |
| "grad_norm": 1.2960147857666016, | |
| "learning_rate": 0.00014340390879478827, | |
| "loss": 1.0054, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.7111477210323998, | |
| "grad_norm": 1.361155390739441, | |
| "learning_rate": 0.00014315960912052116, | |
| "loss": 0.9005, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.71334431630972, | |
| "grad_norm": 1.0135995149612427, | |
| "learning_rate": 0.00014291530944625404, | |
| "loss": 0.9508, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.7155409115870401, | |
| "grad_norm": 1.311626672744751, | |
| "learning_rate": 0.00014267100977198696, | |
| "loss": 0.9183, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.7177375068643603, | |
| "grad_norm": 1.8968180418014526, | |
| "learning_rate": 0.00014242671009771985, | |
| "loss": 0.8414, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.7199341021416803, | |
| "grad_norm": 1.1461299657821655, | |
| "learning_rate": 0.00014218241042345274, | |
| "loss": 0.9426, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.7221306974190007, | |
| "grad_norm": 0.8880915641784668, | |
| "learning_rate": 0.00014193811074918566, | |
| "loss": 0.7782, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.7243272926963207, | |
| "grad_norm": 1.6302744150161743, | |
| "learning_rate": 0.00014169381107491854, | |
| "loss": 1.147, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.7265238879736409, | |
| "grad_norm": 0.9286123514175415, | |
| "learning_rate": 0.00014144951140065146, | |
| "loss": 0.7536, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.728720483250961, | |
| "grad_norm": 1.5562793016433716, | |
| "learning_rate": 0.00014120521172638435, | |
| "loss": 0.8501, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.730917078528281, | |
| "grad_norm": 1.2620130777359009, | |
| "learning_rate": 0.00014096091205211724, | |
| "loss": 0.833, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.7331136738056014, | |
| "grad_norm": 1.2527226209640503, | |
| "learning_rate": 0.00014071661237785016, | |
| "loss": 0.9348, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.7353102690829214, | |
| "grad_norm": 1.2562363147735596, | |
| "learning_rate": 0.00014047231270358305, | |
| "loss": 1.0718, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.7375068643602416, | |
| "grad_norm": 1.8322001695632935, | |
| "learning_rate": 0.00014022801302931593, | |
| "loss": 0.8612, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.7397034596375618, | |
| "grad_norm": 0.9627997279167175, | |
| "learning_rate": 0.00013998371335504885, | |
| "loss": 0.8024, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.741900054914882, | |
| "grad_norm": 1.9179030656814575, | |
| "learning_rate": 0.00013973941368078174, | |
| "loss": 0.8768, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.7440966501922022, | |
| "grad_norm": 1.3875126838684082, | |
| "learning_rate": 0.00013949511400651463, | |
| "loss": 1.0192, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.7462932454695221, | |
| "grad_norm": 0.910798192024231, | |
| "learning_rate": 0.00013925081433224755, | |
| "loss": 0.7878, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.7484898407468425, | |
| "grad_norm": 1.455548644065857, | |
| "learning_rate": 0.00013900651465798043, | |
| "loss": 0.798, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.7506864360241625, | |
| "grad_norm": 1.2879135608673096, | |
| "learning_rate": 0.00013876221498371335, | |
| "loss": 1.0601, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.7528830313014827, | |
| "grad_norm": 1.5035905838012695, | |
| "learning_rate": 0.00013851791530944624, | |
| "loss": 0.8101, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.755079626578803, | |
| "grad_norm": 3.166656017303467, | |
| "learning_rate": 0.00013827361563517913, | |
| "loss": 0.9711, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.7572762218561229, | |
| "grad_norm": 1.075924277305603, | |
| "learning_rate": 0.00013802931596091205, | |
| "loss": 1.0037, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.7594728171334433, | |
| "grad_norm": 1.1928654909133911, | |
| "learning_rate": 0.00013778501628664493, | |
| "loss": 0.9749, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.7616694124107632, | |
| "grad_norm": 1.6487902402877808, | |
| "learning_rate": 0.00013754071661237782, | |
| "loss": 1.154, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.7638660076880834, | |
| "grad_norm": 1.7029637098312378, | |
| "learning_rate": 0.00013729641693811074, | |
| "loss": 1.1975, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.7660626029654036, | |
| "grad_norm": 1.1826850175857544, | |
| "learning_rate": 0.00013705211726384363, | |
| "loss": 1.171, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.7682591982427238, | |
| "grad_norm": 1.3277130126953125, | |
| "learning_rate": 0.00013680781758957652, | |
| "loss": 0.9323, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.770455793520044, | |
| "grad_norm": 1.1256794929504395, | |
| "learning_rate": 0.00013656351791530943, | |
| "loss": 0.83, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.772652388797364, | |
| "grad_norm": 1.1746410131454468, | |
| "learning_rate": 0.00013631921824104232, | |
| "loss": 0.8918, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.7748489840746844, | |
| "grad_norm": 1.9125680923461914, | |
| "learning_rate": 0.0001360749185667752, | |
| "loss": 1.0357, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.7770455793520044, | |
| "grad_norm": 0.9569590091705322, | |
| "learning_rate": 0.00013583061889250813, | |
| "loss": 0.9715, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.7792421746293245, | |
| "grad_norm": 1.3340052366256714, | |
| "learning_rate": 0.00013558631921824102, | |
| "loss": 0.651, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.7814387699066447, | |
| "grad_norm": 3.8345022201538086, | |
| "learning_rate": 0.00013534201954397394, | |
| "loss": 0.7422, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.7836353651839647, | |
| "grad_norm": 0.976948082447052, | |
| "learning_rate": 0.00013509771986970682, | |
| "loss": 0.8501, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.7858319604612851, | |
| "grad_norm": 1.1851125955581665, | |
| "learning_rate": 0.00013485342019543971, | |
| "loss": 1.0546, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.788028555738605, | |
| "grad_norm": 1.187682032585144, | |
| "learning_rate": 0.00013460912052117263, | |
| "loss": 0.9326, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.7902251510159253, | |
| "grad_norm": 1.0449515581130981, | |
| "learning_rate": 0.00013436482084690552, | |
| "loss": 0.9503, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.7924217462932455, | |
| "grad_norm": 1.541332483291626, | |
| "learning_rate": 0.0001341205211726384, | |
| "loss": 0.9396, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.7946183415705657, | |
| "grad_norm": 1.359718918800354, | |
| "learning_rate": 0.00013387622149837132, | |
| "loss": 0.795, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.7968149368478858, | |
| "grad_norm": 1.5417511463165283, | |
| "learning_rate": 0.00013363192182410421, | |
| "loss": 1.0083, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.7990115321252058, | |
| "grad_norm": 1.4850690364837646, | |
| "learning_rate": 0.0001333876221498371, | |
| "loss": 0.8988, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.8012081274025262, | |
| "grad_norm": 1.3120193481445312, | |
| "learning_rate": 0.00013314332247557002, | |
| "loss": 0.9497, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.8034047226798462, | |
| "grad_norm": 0.8782151341438293, | |
| "learning_rate": 0.0001328990228013029, | |
| "loss": 0.9036, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.8056013179571664, | |
| "grad_norm": 1.207207441329956, | |
| "learning_rate": 0.00013265472312703582, | |
| "loss": 0.9931, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.8077979132344866, | |
| "grad_norm": 1.3378241062164307, | |
| "learning_rate": 0.00013241042345276871, | |
| "loss": 1.0092, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.8099945085118065, | |
| "grad_norm": 1.6215455532073975, | |
| "learning_rate": 0.0001321661237785016, | |
| "loss": 0.8277, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.812191103789127, | |
| "grad_norm": 1.0759965181350708, | |
| "learning_rate": 0.00013192182410423452, | |
| "loss": 0.766, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.814387699066447, | |
| "grad_norm": 1.2844793796539307, | |
| "learning_rate": 0.0001316775244299674, | |
| "loss": 0.6559, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.8165842943437671, | |
| "grad_norm": 1.115490436553955, | |
| "learning_rate": 0.0001314332247557003, | |
| "loss": 0.9611, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.8187808896210873, | |
| "grad_norm": 1.7604005336761475, | |
| "learning_rate": 0.00013118892508143321, | |
| "loss": 0.8321, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.8209774848984075, | |
| "grad_norm": 1.3991718292236328, | |
| "learning_rate": 0.00013094462540716613, | |
| "loss": 0.7854, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.8231740801757277, | |
| "grad_norm": 1.0830546617507935, | |
| "learning_rate": 0.00013070032573289902, | |
| "loss": 0.6967, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.8253706754530477, | |
| "grad_norm": 0.8690131902694702, | |
| "learning_rate": 0.0001304560260586319, | |
| "loss": 0.6077, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.827567270730368, | |
| "grad_norm": 1.0183937549591064, | |
| "learning_rate": 0.00013021172638436483, | |
| "loss": 0.9258, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.829763866007688, | |
| "grad_norm": 1.7728548049926758, | |
| "learning_rate": 0.00012996742671009771, | |
| "loss": 0.9548, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.8319604612850082, | |
| "grad_norm": 1.1950143575668335, | |
| "learning_rate": 0.0001297231270358306, | |
| "loss": 1.0659, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.8341570565623284, | |
| "grad_norm": 1.6282366514205933, | |
| "learning_rate": 0.00012947882736156352, | |
| "loss": 0.9687, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.8363536518396484, | |
| "grad_norm": 1.2889693975448608, | |
| "learning_rate": 0.0001292345276872964, | |
| "loss": 0.8147, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.8385502471169688, | |
| "grad_norm": 1.084276556968689, | |
| "learning_rate": 0.00012899022801302933, | |
| "loss": 0.8125, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.8407468423942888, | |
| "grad_norm": 1.0461785793304443, | |
| "learning_rate": 0.00012874592833876221, | |
| "loss": 0.9188, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.842943437671609, | |
| "grad_norm": 1.2562017440795898, | |
| "learning_rate": 0.0001285016286644951, | |
| "loss": 0.7951, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.8451400329489291, | |
| "grad_norm": 2.189511299133301, | |
| "learning_rate": 0.00012825732899022802, | |
| "loss": 0.9201, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.8473366282262493, | |
| "grad_norm": 0.835127592086792, | |
| "learning_rate": 0.0001280130293159609, | |
| "loss": 0.8175, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.8495332235035695, | |
| "grad_norm": 2.7215588092803955, | |
| "learning_rate": 0.0001277687296416938, | |
| "loss": 0.8619, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.8517298187808895, | |
| "grad_norm": 1.1069096326828003, | |
| "learning_rate": 0.00012752442996742671, | |
| "loss": 1.1, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.85392641405821, | |
| "grad_norm": 2.0498270988464355, | |
| "learning_rate": 0.0001272801302931596, | |
| "loss": 0.9303, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.8561230093355299, | |
| "grad_norm": 1.3160767555236816, | |
| "learning_rate": 0.0001270358306188925, | |
| "loss": 0.8131, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.85831960461285, | |
| "grad_norm": 1.1988338232040405, | |
| "learning_rate": 0.0001267915309446254, | |
| "loss": 0.7993, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.8605161998901703, | |
| "grad_norm": 1.18996000289917, | |
| "learning_rate": 0.0001265472312703583, | |
| "loss": 1.0436, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.8627127951674904, | |
| "grad_norm": 1.085823893547058, | |
| "learning_rate": 0.0001263029315960912, | |
| "loss": 0.7747, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.8649093904448106, | |
| "grad_norm": 1.4252612590789795, | |
| "learning_rate": 0.0001260586319218241, | |
| "loss": 0.8054, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.8671059857221306, | |
| "grad_norm": 1.2093149423599243, | |
| "learning_rate": 0.000125814332247557, | |
| "loss": 0.801, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.869302580999451, | |
| "grad_norm": 1.007034420967102, | |
| "learning_rate": 0.0001255700325732899, | |
| "loss": 0.691, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.871499176276771, | |
| "grad_norm": 1.1088532209396362, | |
| "learning_rate": 0.0001253257328990228, | |
| "loss": 0.9028, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.8736957715540912, | |
| "grad_norm": 1.6077468395233154, | |
| "learning_rate": 0.0001250814332247557, | |
| "loss": 1.1585, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.8758923668314114, | |
| "grad_norm": 1.6345423460006714, | |
| "learning_rate": 0.0001248371335504886, | |
| "loss": 1.1166, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.8780889621087313, | |
| "grad_norm": 2.059034585952759, | |
| "learning_rate": 0.0001245928338762215, | |
| "loss": 0.7749, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.8802855573860517, | |
| "grad_norm": 1.4929571151733398, | |
| "learning_rate": 0.00012434853420195438, | |
| "loss": 1.0713, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.8824821526633717, | |
| "grad_norm": 2.498446464538574, | |
| "learning_rate": 0.0001241042345276873, | |
| "loss": 0.8104, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.884678747940692, | |
| "grad_norm": 0.9690324068069458, | |
| "learning_rate": 0.0001238599348534202, | |
| "loss": 0.9309, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.886875343218012, | |
| "grad_norm": 1.274396300315857, | |
| "learning_rate": 0.00012361563517915308, | |
| "loss": 0.7716, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.8890719384953323, | |
| "grad_norm": 0.9249063730239868, | |
| "learning_rate": 0.000123371335504886, | |
| "loss": 0.8114, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.8912685337726525, | |
| "grad_norm": 1.2150957584381104, | |
| "learning_rate": 0.00012312703583061888, | |
| "loss": 1.0017, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.8934651290499724, | |
| "grad_norm": 1.2546175718307495, | |
| "learning_rate": 0.00012288273615635177, | |
| "loss": 0.9169, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.8956617243272929, | |
| "grad_norm": 1.140419840812683, | |
| "learning_rate": 0.0001226384364820847, | |
| "loss": 1.1595, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.8978583196046128, | |
| "grad_norm": 2.788454055786133, | |
| "learning_rate": 0.00012239413680781758, | |
| "loss": 0.773, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.900054914881933, | |
| "grad_norm": 1.1368335485458374, | |
| "learning_rate": 0.0001221498371335505, | |
| "loss": 0.9305, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.9022515101592532, | |
| "grad_norm": 1.1291612386703491, | |
| "learning_rate": 0.00012190553745928338, | |
| "loss": 0.842, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.9044481054365732, | |
| "grad_norm": 0.9890234470367432, | |
| "learning_rate": 0.00012166123778501629, | |
| "loss": 0.8236, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.9066447007138936, | |
| "grad_norm": 1.678726077079773, | |
| "learning_rate": 0.00012141693811074918, | |
| "loss": 1.1744, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.9088412959912135, | |
| "grad_norm": 1.3313941955566406, | |
| "learning_rate": 0.00012117263843648208, | |
| "loss": 1.1439, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.9110378912685337, | |
| "grad_norm": 1.555583119392395, | |
| "learning_rate": 0.00012092833876221498, | |
| "loss": 0.8426, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.913234486545854, | |
| "grad_norm": 1.408030390739441, | |
| "learning_rate": 0.00012068403908794787, | |
| "loss": 0.8473, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.9154310818231741, | |
| "grad_norm": 1.2433571815490723, | |
| "learning_rate": 0.00012043973941368077, | |
| "loss": 1.054, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.9176276771004943, | |
| "grad_norm": 1.2735772132873535, | |
| "learning_rate": 0.00012019543973941368, | |
| "loss": 1.0058, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.9198242723778143, | |
| "grad_norm": 0.8596783876419067, | |
| "learning_rate": 0.00011995114006514658, | |
| "loss": 1.1413, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.9220208676551347, | |
| "grad_norm": 1.3398058414459229, | |
| "learning_rate": 0.00011970684039087947, | |
| "loss": 1.0245, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.9242174629324547, | |
| "grad_norm": 2.7426981925964355, | |
| "learning_rate": 0.00011946254071661237, | |
| "loss": 1.0204, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.9264140582097748, | |
| "grad_norm": 1.1783690452575684, | |
| "learning_rate": 0.00011921824104234527, | |
| "loss": 0.7817, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.928610653487095, | |
| "grad_norm": 1.8914209604263306, | |
| "learning_rate": 0.00011897394136807816, | |
| "loss": 0.9806, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.930807248764415, | |
| "grad_norm": 0.8673095703125, | |
| "learning_rate": 0.00011872964169381106, | |
| "loss": 0.8618, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.9330038440417354, | |
| "grad_norm": 1.635268211364746, | |
| "learning_rate": 0.00011848534201954397, | |
| "loss": 0.8492, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.9352004393190554, | |
| "grad_norm": 1.011803150177002, | |
| "learning_rate": 0.00011824104234527687, | |
| "loss": 0.9225, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.9373970345963756, | |
| "grad_norm": 0.8291391134262085, | |
| "learning_rate": 0.00011799674267100976, | |
| "loss": 0.7335, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.9395936298736958, | |
| "grad_norm": 1.0854843854904175, | |
| "learning_rate": 0.00011775244299674266, | |
| "loss": 0.9847, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.941790225151016, | |
| "grad_norm": 1.5232948064804077, | |
| "learning_rate": 0.00011750814332247557, | |
| "loss": 0.9486, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.9439868204283361, | |
| "grad_norm": 0.9641236066818237, | |
| "learning_rate": 0.00011726384364820845, | |
| "loss": 1.3126, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.9461834157056561, | |
| "grad_norm": 1.0800381898880005, | |
| "learning_rate": 0.00011701954397394136, | |
| "loss": 0.8741, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.9483800109829765, | |
| "grad_norm": 0.8853541612625122, | |
| "learning_rate": 0.00011677524429967426, | |
| "loss": 0.8182, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.9505766062602965, | |
| "grad_norm": 1.125565528869629, | |
| "learning_rate": 0.00011653094462540716, | |
| "loss": 0.8039, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.9527732015376167, | |
| "grad_norm": 1.5178683996200562, | |
| "learning_rate": 0.00011628664495114005, | |
| "loss": 0.8788, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.9549697968149369, | |
| "grad_norm": 1.0295542478561401, | |
| "learning_rate": 0.00011604234527687295, | |
| "loss": 0.9124, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.9571663920922568, | |
| "grad_norm": 0.7818289399147034, | |
| "learning_rate": 0.00011579804560260586, | |
| "loss": 1.002, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.9593629873695773, | |
| "grad_norm": 1.2037432193756104, | |
| "learning_rate": 0.00011555374592833876, | |
| "loss": 0.8485, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.9615595826468972, | |
| "grad_norm": 1.2548919916152954, | |
| "learning_rate": 0.00011530944625407165, | |
| "loss": 0.8572, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.9637561779242174, | |
| "grad_norm": 1.3632498979568481, | |
| "learning_rate": 0.00011506514657980455, | |
| "loss": 0.9463, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.9659527732015376, | |
| "grad_norm": 1.0141692161560059, | |
| "learning_rate": 0.00011482084690553745, | |
| "loss": 0.7037, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.9681493684788578, | |
| "grad_norm": 1.1087349653244019, | |
| "learning_rate": 0.00011457654723127034, | |
| "loss": 1.065, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.970345963756178, | |
| "grad_norm": 0.9798489809036255, | |
| "learning_rate": 0.00011433224755700325, | |
| "loss": 0.7692, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.972542559033498, | |
| "grad_norm": 1.509859323501587, | |
| "learning_rate": 0.00011408794788273615, | |
| "loss": 1.0534, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.9747391543108184, | |
| "grad_norm": 1.6241490840911865, | |
| "learning_rate": 0.00011384364820846905, | |
| "loss": 1.0387, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 1.9769357495881383, | |
| "grad_norm": 0.8627262115478516, | |
| "learning_rate": 0.00011359934853420194, | |
| "loss": 0.798, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.9791323448654585, | |
| "grad_norm": 0.9451064467430115, | |
| "learning_rate": 0.00011335504885993484, | |
| "loss": 0.9857, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 1.9813289401427787, | |
| "grad_norm": 1.2718652486801147, | |
| "learning_rate": 0.00011311074918566775, | |
| "loss": 0.9738, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 1.9835255354200987, | |
| "grad_norm": 1.1287384033203125, | |
| "learning_rate": 0.00011286644951140064, | |
| "loss": 0.9321, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 1.985722130697419, | |
| "grad_norm": 2.24863338470459, | |
| "learning_rate": 0.00011262214983713354, | |
| "loss": 0.7237, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.987918725974739, | |
| "grad_norm": 1.4581634998321533, | |
| "learning_rate": 0.00011237785016286644, | |
| "loss": 0.7733, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.9901153212520593, | |
| "grad_norm": 1.1087408065795898, | |
| "learning_rate": 0.00011213355048859934, | |
| "loss": 1.0261, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 1.9923119165293794, | |
| "grad_norm": 1.469053864479065, | |
| "learning_rate": 0.00011188925081433223, | |
| "loss": 0.7781, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 1.9945085118066996, | |
| "grad_norm": 0.8977324962615967, | |
| "learning_rate": 0.00011164495114006514, | |
| "loss": 0.754, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 1.9967051070840198, | |
| "grad_norm": 1.6588226556777954, | |
| "learning_rate": 0.00011140065146579804, | |
| "loss": 0.6791, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 1.9989017023613398, | |
| "grad_norm": 2.111530065536499, | |
| "learning_rate": 0.00011115635179153093, | |
| "loss": 0.9089, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.00109829763866, | |
| "grad_norm": 1.1633527278900146, | |
| "learning_rate": 0.00011091205211726383, | |
| "loss": 1.0454, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.00329489291598, | |
| "grad_norm": 1.1369563341140747, | |
| "learning_rate": 0.00011066775244299673, | |
| "loss": 0.8206, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.0054914881933006, | |
| "grad_norm": 0.995354950428009, | |
| "learning_rate": 0.00011042345276872964, | |
| "loss": 0.6817, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.0076880834706206, | |
| "grad_norm": 1.3472247123718262, | |
| "learning_rate": 0.00011017915309446253, | |
| "loss": 0.8746, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.0098846787479405, | |
| "grad_norm": 1.5700223445892334, | |
| "learning_rate": 0.00010993485342019543, | |
| "loss": 0.8236, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.012081274025261, | |
| "grad_norm": 0.8956323266029358, | |
| "learning_rate": 0.00010969055374592833, | |
| "loss": 1.0753, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.014277869302581, | |
| "grad_norm": 1.6048280000686646, | |
| "learning_rate": 0.00010944625407166122, | |
| "loss": 1.0931, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.0164744645799013, | |
| "grad_norm": 1.0637688636779785, | |
| "learning_rate": 0.00010920195439739412, | |
| "loss": 1.0596, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.0186710598572213, | |
| "grad_norm": 1.286287784576416, | |
| "learning_rate": 0.00010895765472312703, | |
| "loss": 0.8256, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.0208676551345413, | |
| "grad_norm": 1.4362157583236694, | |
| "learning_rate": 0.00010871335504885993, | |
| "loss": 0.8228, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.0230642504118617, | |
| "grad_norm": 1.3563312292099, | |
| "learning_rate": 0.00010846905537459282, | |
| "loss": 0.7615, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.0252608456891816, | |
| "grad_norm": 0.9290498495101929, | |
| "learning_rate": 0.00010822475570032572, | |
| "loss": 0.8972, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.027457440966502, | |
| "grad_norm": 0.9655089378356934, | |
| "learning_rate": 0.00010798045602605862, | |
| "loss": 0.925, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.029654036243822, | |
| "grad_norm": 1.7136750221252441, | |
| "learning_rate": 0.00010773615635179153, | |
| "loss": 1.0087, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.0318506315211424, | |
| "grad_norm": 1.2151069641113281, | |
| "learning_rate": 0.00010749185667752442, | |
| "loss": 0.8218, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.0340472267984624, | |
| "grad_norm": 0.849277138710022, | |
| "learning_rate": 0.00010724755700325732, | |
| "loss": 0.8666, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.0362438220757824, | |
| "grad_norm": 1.1864643096923828, | |
| "learning_rate": 0.00010700325732899022, | |
| "loss": 0.9731, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.0384404173531028, | |
| "grad_norm": 0.8906531929969788, | |
| "learning_rate": 0.00010675895765472311, | |
| "loss": 0.8102, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.0406370126304227, | |
| "grad_norm": 0.8965123891830444, | |
| "learning_rate": 0.00010651465798045601, | |
| "loss": 0.741, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.042833607907743, | |
| "grad_norm": 1.1413078308105469, | |
| "learning_rate": 0.00010627035830618892, | |
| "loss": 0.722, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.045030203185063, | |
| "grad_norm": 0.9153503179550171, | |
| "learning_rate": 0.00010602605863192182, | |
| "loss": 0.8273, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.047226798462383, | |
| "grad_norm": 1.157155156135559, | |
| "learning_rate": 0.00010578175895765471, | |
| "loss": 0.8128, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.0494233937397035, | |
| "grad_norm": 1.1913141012191772, | |
| "learning_rate": 0.00010553745928338761, | |
| "loss": 0.8199, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.0516199890170235, | |
| "grad_norm": 1.1848474740982056, | |
| "learning_rate": 0.00010529315960912051, | |
| "loss": 0.806, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.053816584294344, | |
| "grad_norm": 1.0001102685928345, | |
| "learning_rate": 0.0001050488599348534, | |
| "loss": 0.8142, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.056013179571664, | |
| "grad_norm": 1.0069324970245361, | |
| "learning_rate": 0.0001048045602605863, | |
| "loss": 1.0328, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.0582097748489843, | |
| "grad_norm": 1.4018255472183228, | |
| "learning_rate": 0.00010456026058631921, | |
| "loss": 0.9502, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.0604063701263042, | |
| "grad_norm": 1.194178581237793, | |
| "learning_rate": 0.00010431596091205211, | |
| "loss": 0.675, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.062602965403624, | |
| "grad_norm": 1.252380132675171, | |
| "learning_rate": 0.000104071661237785, | |
| "loss": 1.0248, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.0647995606809446, | |
| "grad_norm": 0.827702522277832, | |
| "learning_rate": 0.0001038273615635179, | |
| "loss": 0.7724, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.0669961559582646, | |
| "grad_norm": 0.8210299611091614, | |
| "learning_rate": 0.0001035830618892508, | |
| "loss": 0.8821, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.069192751235585, | |
| "grad_norm": 1.618456244468689, | |
| "learning_rate": 0.0001033387622149837, | |
| "loss": 0.6774, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.071389346512905, | |
| "grad_norm": 2.8957290649414062, | |
| "learning_rate": 0.0001030944625407166, | |
| "loss": 0.8639, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.073585941790225, | |
| "grad_norm": 1.407447099685669, | |
| "learning_rate": 0.0001028501628664495, | |
| "loss": 0.7323, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.0757825370675453, | |
| "grad_norm": 1.1115045547485352, | |
| "learning_rate": 0.0001026058631921824, | |
| "loss": 0.8532, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.0779791323448653, | |
| "grad_norm": 1.036726713180542, | |
| "learning_rate": 0.00010236156351791529, | |
| "loss": 0.8251, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.0801757276221857, | |
| "grad_norm": 1.275976538658142, | |
| "learning_rate": 0.0001021172638436482, | |
| "loss": 0.6882, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.0823723228995057, | |
| "grad_norm": 1.1043846607208252, | |
| "learning_rate": 0.0001018729641693811, | |
| "loss": 0.7632, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.084568918176826, | |
| "grad_norm": 1.3968265056610107, | |
| "learning_rate": 0.00010162866449511399, | |
| "loss": 1.0236, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.086765513454146, | |
| "grad_norm": 1.0273393392562866, | |
| "learning_rate": 0.00010138436482084689, | |
| "loss": 0.9206, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.088962108731466, | |
| "grad_norm": 1.180243730545044, | |
| "learning_rate": 0.00010114006514657979, | |
| "loss": 0.9142, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.0911587040087865, | |
| "grad_norm": 1.5757319927215576, | |
| "learning_rate": 0.0001008957654723127, | |
| "loss": 0.9064, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.0933552992861064, | |
| "grad_norm": 1.2129454612731934, | |
| "learning_rate": 0.00010065146579804558, | |
| "loss": 0.9792, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.095551894563427, | |
| "grad_norm": 0.7025288343429565, | |
| "learning_rate": 0.00010040716612377849, | |
| "loss": 0.7824, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.097748489840747, | |
| "grad_norm": 1.0241849422454834, | |
| "learning_rate": 0.00010016286644951139, | |
| "loss": 0.9198, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.099945085118067, | |
| "grad_norm": 0.9621145725250244, | |
| "learning_rate": 9.991856677524428e-05, | |
| "loss": 0.8266, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.102141680395387, | |
| "grad_norm": 1.557124376296997, | |
| "learning_rate": 9.967426710097718e-05, | |
| "loss": 0.9184, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.104338275672707, | |
| "grad_norm": 0.8569768667221069, | |
| "learning_rate": 9.942996742671008e-05, | |
| "loss": 0.7586, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.1065348709500276, | |
| "grad_norm": 1.0364352464675903, | |
| "learning_rate": 9.918566775244299e-05, | |
| "loss": 0.8661, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.1087314662273475, | |
| "grad_norm": 0.8486453890800476, | |
| "learning_rate": 9.894136807817588e-05, | |
| "loss": 0.8093, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.110928061504668, | |
| "grad_norm": 1.7316734790802002, | |
| "learning_rate": 9.869706840390878e-05, | |
| "loss": 1.0561, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.113124656781988, | |
| "grad_norm": 12.940186500549316, | |
| "learning_rate": 9.845276872964168e-05, | |
| "loss": 0.8509, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.115321252059308, | |
| "grad_norm": 2.602274179458618, | |
| "learning_rate": 9.820846905537458e-05, | |
| "loss": 0.7027, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.1175178473366283, | |
| "grad_norm": 1.218983769416809, | |
| "learning_rate": 9.796416938110747e-05, | |
| "loss": 0.8518, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.1197144426139483, | |
| "grad_norm": 1.2135183811187744, | |
| "learning_rate": 9.771986970684038e-05, | |
| "loss": 0.8, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.1219110378912687, | |
| "grad_norm": 2.395167827606201, | |
| "learning_rate": 9.747557003257328e-05, | |
| "loss": 0.8689, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.1241076331685886, | |
| "grad_norm": 1.162548542022705, | |
| "learning_rate": 9.723127035830617e-05, | |
| "loss": 0.7942, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.1263042284459086, | |
| "grad_norm": 0.9186055064201355, | |
| "learning_rate": 9.698697068403907e-05, | |
| "loss": 0.8278, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.128500823723229, | |
| "grad_norm": 1.3132836818695068, | |
| "learning_rate": 9.674267100977197e-05, | |
| "loss": 0.7537, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.130697419000549, | |
| "grad_norm": 0.9659914374351501, | |
| "learning_rate": 9.649837133550488e-05, | |
| "loss": 0.9606, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.1328940142778694, | |
| "grad_norm": 3.9964468479156494, | |
| "learning_rate": 9.625407166123777e-05, | |
| "loss": 0.8398, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.1350906095551894, | |
| "grad_norm": 1.4134726524353027, | |
| "learning_rate": 9.600977198697067e-05, | |
| "loss": 1.0316, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.13728720483251, | |
| "grad_norm": 1.150215744972229, | |
| "learning_rate": 9.576547231270357e-05, | |
| "loss": 0.8058, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.1394838001098297, | |
| "grad_norm": 1.152235507965088, | |
| "learning_rate": 9.552117263843646e-05, | |
| "loss": 0.9237, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.1416803953871497, | |
| "grad_norm": 1.1499593257904053, | |
| "learning_rate": 9.527687296416936e-05, | |
| "loss": 0.9465, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.14387699066447, | |
| "grad_norm": 1.9223628044128418, | |
| "learning_rate": 9.503257328990227e-05, | |
| "loss": 0.9064, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.14607358594179, | |
| "grad_norm": 0.9833508729934692, | |
| "learning_rate": 9.478827361563517e-05, | |
| "loss": 0.8701, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.1482701812191105, | |
| "grad_norm": 1.2317014932632446, | |
| "learning_rate": 9.454397394136806e-05, | |
| "loss": 0.8028, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.1504667764964305, | |
| "grad_norm": 2.2281908988952637, | |
| "learning_rate": 9.429967426710096e-05, | |
| "loss": 0.8844, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.152663371773751, | |
| "grad_norm": 1.4272838830947876, | |
| "learning_rate": 9.405537459283386e-05, | |
| "loss": 0.9874, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.154859967051071, | |
| "grad_norm": 1.1508535146713257, | |
| "learning_rate": 9.381107491856675e-05, | |
| "loss": 0.7099, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.157056562328391, | |
| "grad_norm": 1.0197069644927979, | |
| "learning_rate": 9.356677524429967e-05, | |
| "loss": 0.7358, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.1592531576057112, | |
| "grad_norm": 1.6753896474838257, | |
| "learning_rate": 9.332247557003257e-05, | |
| "loss": 1.1028, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.161449752883031, | |
| "grad_norm": 1.0567740201950073, | |
| "learning_rate": 9.307817589576547e-05, | |
| "loss": 0.7508, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.1636463481603516, | |
| "grad_norm": 1.1761212348937988, | |
| "learning_rate": 9.283387622149838e-05, | |
| "loss": 0.7395, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.1658429434376716, | |
| "grad_norm": 1.691198706626892, | |
| "learning_rate": 9.258957654723127e-05, | |
| "loss": 0.8154, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.168039538714992, | |
| "grad_norm": 1.4455115795135498, | |
| "learning_rate": 9.234527687296417e-05, | |
| "loss": 1.0294, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.170236133992312, | |
| "grad_norm": 0.8547191619873047, | |
| "learning_rate": 9.210097719869707e-05, | |
| "loss": 0.7019, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.172432729269632, | |
| "grad_norm": 0.7419626116752625, | |
| "learning_rate": 9.185667752442996e-05, | |
| "loss": 0.8297, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.1746293245469523, | |
| "grad_norm": 1.1484277248382568, | |
| "learning_rate": 9.161237785016286e-05, | |
| "loss": 1.0279, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.1768259198242723, | |
| "grad_norm": 1.7566430568695068, | |
| "learning_rate": 9.136807817589577e-05, | |
| "loss": 0.8116, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.1790225151015927, | |
| "grad_norm": 1.0973962545394897, | |
| "learning_rate": 9.112377850162867e-05, | |
| "loss": 0.7568, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.1812191103789127, | |
| "grad_norm": 1.1992583274841309, | |
| "learning_rate": 9.087947882736156e-05, | |
| "loss": 0.9238, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.1834157056562327, | |
| "grad_norm": 1.7046031951904297, | |
| "learning_rate": 9.063517915309446e-05, | |
| "loss": 0.7861, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.185612300933553, | |
| "grad_norm": 1.3837920427322388, | |
| "learning_rate": 9.039087947882736e-05, | |
| "loss": 0.8473, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.187808896210873, | |
| "grad_norm": 1.1505504846572876, | |
| "learning_rate": 9.014657980456025e-05, | |
| "loss": 1.0042, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.1900054914881935, | |
| "grad_norm": 1.048252820968628, | |
| "learning_rate": 8.990228013029316e-05, | |
| "loss": 0.9184, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.1922020867655134, | |
| "grad_norm": 1.2969130277633667, | |
| "learning_rate": 8.965798045602606e-05, | |
| "loss": 0.8576, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.1943986820428334, | |
| "grad_norm": 1.209368348121643, | |
| "learning_rate": 8.941368078175896e-05, | |
| "loss": 1.0012, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.196595277320154, | |
| "grad_norm": 0.8410612940788269, | |
| "learning_rate": 8.916938110749185e-05, | |
| "loss": 0.775, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.1987918725974738, | |
| "grad_norm": 1.0059013366699219, | |
| "learning_rate": 8.892508143322475e-05, | |
| "loss": 0.5974, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.200988467874794, | |
| "grad_norm": 1.314413070678711, | |
| "learning_rate": 8.868078175895766e-05, | |
| "loss": 0.766, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.203185063152114, | |
| "grad_norm": 1.7778496742248535, | |
| "learning_rate": 8.843648208469055e-05, | |
| "loss": 0.8514, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.2053816584294346, | |
| "grad_norm": 1.0296496152877808, | |
| "learning_rate": 8.819218241042345e-05, | |
| "loss": 0.7581, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.2075782537067545, | |
| "grad_norm": 0.7333378195762634, | |
| "learning_rate": 8.794788273615635e-05, | |
| "loss": 0.762, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.2097748489840745, | |
| "grad_norm": 1.3215105533599854, | |
| "learning_rate": 8.770358306188925e-05, | |
| "loss": 0.8474, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.211971444261395, | |
| "grad_norm": 0.9052922129631042, | |
| "learning_rate": 8.745928338762214e-05, | |
| "loss": 0.6966, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.214168039538715, | |
| "grad_norm": 1.2188451290130615, | |
| "learning_rate": 8.721498371335505e-05, | |
| "loss": 1.1153, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.2163646348160353, | |
| "grad_norm": 0.9124431014060974, | |
| "learning_rate": 8.697068403908795e-05, | |
| "loss": 0.8864, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.2185612300933553, | |
| "grad_norm": 0.8480375409126282, | |
| "learning_rate": 8.672638436482085e-05, | |
| "loss": 0.7897, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.2207578253706757, | |
| "grad_norm": 0.8407557606697083, | |
| "learning_rate": 8.648208469055374e-05, | |
| "loss": 0.8304, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.2229544206479956, | |
| "grad_norm": 1.1393015384674072, | |
| "learning_rate": 8.623778501628664e-05, | |
| "loss": 1.1172, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.2251510159253156, | |
| "grad_norm": 1.7269963026046753, | |
| "learning_rate": 8.599348534201955e-05, | |
| "loss": 0.8489, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 2.227347611202636, | |
| "grad_norm": 1.364884614944458, | |
| "learning_rate": 8.574918566775243e-05, | |
| "loss": 1.0099, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.229544206479956, | |
| "grad_norm": 2.5326216220855713, | |
| "learning_rate": 8.550488599348534e-05, | |
| "loss": 1.0842, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.2317408017572764, | |
| "grad_norm": 0.9562110304832458, | |
| "learning_rate": 8.526058631921824e-05, | |
| "loss": 0.9105, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.2339373970345964, | |
| "grad_norm": 1.571959137916565, | |
| "learning_rate": 8.501628664495114e-05, | |
| "loss": 1.0383, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 2.2361339923119163, | |
| "grad_norm": 1.2061654329299927, | |
| "learning_rate": 8.477198697068403e-05, | |
| "loss": 1.0225, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.2383305875892368, | |
| "grad_norm": 0.9113017916679382, | |
| "learning_rate": 8.452768729641694e-05, | |
| "loss": 0.9426, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 2.2405271828665567, | |
| "grad_norm": 3.3736088275909424, | |
| "learning_rate": 8.428338762214984e-05, | |
| "loss": 0.8176, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.242723778143877, | |
| "grad_norm": 0.8913333415985107, | |
| "learning_rate": 8.403908794788273e-05, | |
| "loss": 0.8625, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 2.244920373421197, | |
| "grad_norm": 1.267503261566162, | |
| "learning_rate": 8.379478827361563e-05, | |
| "loss": 0.6476, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.247116968698517, | |
| "grad_norm": 1.1047229766845703, | |
| "learning_rate": 8.355048859934853e-05, | |
| "loss": 0.9819, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 2.2493135639758375, | |
| "grad_norm": 1.089673638343811, | |
| "learning_rate": 8.330618892508144e-05, | |
| "loss": 0.8298, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.2515101592531575, | |
| "grad_norm": 0.829242467880249, | |
| "learning_rate": 8.306188925081432e-05, | |
| "loss": 0.7925, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.253706754530478, | |
| "grad_norm": 0.9690037965774536, | |
| "learning_rate": 8.281758957654723e-05, | |
| "loss": 0.766, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.255903349807798, | |
| "grad_norm": 1.8586704730987549, | |
| "learning_rate": 8.257328990228013e-05, | |
| "loss": 0.846, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 2.2580999450851182, | |
| "grad_norm": 1.0639582872390747, | |
| "learning_rate": 8.232899022801302e-05, | |
| "loss": 0.9403, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.260296540362438, | |
| "grad_norm": 1.0850247144699097, | |
| "learning_rate": 8.208469055374592e-05, | |
| "loss": 0.906, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 2.262493135639758, | |
| "grad_norm": 1.5845433473587036, | |
| "learning_rate": 8.184039087947882e-05, | |
| "loss": 0.7753, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.2646897309170786, | |
| "grad_norm": 1.117396354675293, | |
| "learning_rate": 8.159609120521173e-05, | |
| "loss": 0.7598, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 2.2668863261943986, | |
| "grad_norm": 1.0254619121551514, | |
| "learning_rate": 8.135179153094462e-05, | |
| "loss": 0.7259, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.269082921471719, | |
| "grad_norm": 0.7966324090957642, | |
| "learning_rate": 8.110749185667752e-05, | |
| "loss": 1.0262, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 2.271279516749039, | |
| "grad_norm": 1.3414199352264404, | |
| "learning_rate": 8.086319218241042e-05, | |
| "loss": 0.7843, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.2734761120263594, | |
| "grad_norm": 1.0072284936904907, | |
| "learning_rate": 8.061889250814331e-05, | |
| "loss": 0.7707, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.2756727073036793, | |
| "grad_norm": 0.8906221389770508, | |
| "learning_rate": 8.037459283387621e-05, | |
| "loss": 0.7972, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.2778693025809993, | |
| "grad_norm": 1.3249026536941528, | |
| "learning_rate": 8.013029315960912e-05, | |
| "loss": 0.7648, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 2.2800658978583197, | |
| "grad_norm": 1.1036349534988403, | |
| "learning_rate": 7.988599348534202e-05, | |
| "loss": 0.7828, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.2822624931356397, | |
| "grad_norm": 1.4306989908218384, | |
| "learning_rate": 7.964169381107491e-05, | |
| "loss": 0.8, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 2.28445908841296, | |
| "grad_norm": 1.1071574687957764, | |
| "learning_rate": 7.939739413680781e-05, | |
| "loss": 0.8418, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.28665568369028, | |
| "grad_norm": 1.484405279159546, | |
| "learning_rate": 7.915309446254071e-05, | |
| "loss": 0.7931, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 2.2888522789676005, | |
| "grad_norm": 1.2407186031341553, | |
| "learning_rate": 7.890879478827362e-05, | |
| "loss": 0.9376, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.2910488742449204, | |
| "grad_norm": 1.0445892810821533, | |
| "learning_rate": 7.86644951140065e-05, | |
| "loss": 0.6961, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 2.2932454695222404, | |
| "grad_norm": 4.345700263977051, | |
| "learning_rate": 7.842019543973941e-05, | |
| "loss": 0.8988, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.295442064799561, | |
| "grad_norm": 1.0117210149765015, | |
| "learning_rate": 7.817589576547231e-05, | |
| "loss": 0.9598, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.297638660076881, | |
| "grad_norm": 0.826719343662262, | |
| "learning_rate": 7.79315960912052e-05, | |
| "loss": 0.7618, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.2998352553542007, | |
| "grad_norm": 0.9766603112220764, | |
| "learning_rate": 7.76872964169381e-05, | |
| "loss": 0.618, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 2.302031850631521, | |
| "grad_norm": 1.4053963422775269, | |
| "learning_rate": 7.7442996742671e-05, | |
| "loss": 0.8552, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.304228445908841, | |
| "grad_norm": 1.1446764469146729, | |
| "learning_rate": 7.719869706840391e-05, | |
| "loss": 0.8592, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 2.3064250411861615, | |
| "grad_norm": 0.7592353224754333, | |
| "learning_rate": 7.69543973941368e-05, | |
| "loss": 0.8853, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.3086216364634815, | |
| "grad_norm": 1.5565050840377808, | |
| "learning_rate": 7.67100977198697e-05, | |
| "loss": 0.6694, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.310818231740802, | |
| "grad_norm": 0.9414224028587341, | |
| "learning_rate": 7.64657980456026e-05, | |
| "loss": 0.7777, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.313014827018122, | |
| "grad_norm": 0.9869343042373657, | |
| "learning_rate": 7.622149837133549e-05, | |
| "loss": 0.8936, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.315211422295442, | |
| "grad_norm": 0.9009634852409363, | |
| "learning_rate": 7.59771986970684e-05, | |
| "loss": 0.8041, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.3174080175727623, | |
| "grad_norm": 1.1592154502868652, | |
| "learning_rate": 7.57328990228013e-05, | |
| "loss": 1.0348, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.3196046128500822, | |
| "grad_norm": 1.4870555400848389, | |
| "learning_rate": 7.54885993485342e-05, | |
| "loss": 0.8373, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.3218012081274026, | |
| "grad_norm": 1.32427978515625, | |
| "learning_rate": 7.524429967426709e-05, | |
| "loss": 1.0835, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 2.3239978034047226, | |
| "grad_norm": 0.78367018699646, | |
| "learning_rate": 7.5e-05, | |
| "loss": 0.603, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.326194398682043, | |
| "grad_norm": 1.3375253677368164, | |
| "learning_rate": 7.47557003257329e-05, | |
| "loss": 0.7714, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 2.328390993959363, | |
| "grad_norm": 0.8829704523086548, | |
| "learning_rate": 7.451140065146579e-05, | |
| "loss": 0.793, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.330587589236683, | |
| "grad_norm": 1.0981749296188354, | |
| "learning_rate": 7.426710097719869e-05, | |
| "loss": 0.8559, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 2.3327841845140034, | |
| "grad_norm": 0.8963515162467957, | |
| "learning_rate": 7.402280130293159e-05, | |
| "loss": 0.8233, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.3349807797913233, | |
| "grad_norm": 1.379052996635437, | |
| "learning_rate": 7.37785016286645e-05, | |
| "loss": 1.0092, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 2.3371773750686438, | |
| "grad_norm": 1.0106465816497803, | |
| "learning_rate": 7.353420195439738e-05, | |
| "loss": 0.9123, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.3393739703459637, | |
| "grad_norm": 0.9500569701194763, | |
| "learning_rate": 7.328990228013029e-05, | |
| "loss": 0.8799, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.341570565623284, | |
| "grad_norm": 0.926685631275177, | |
| "learning_rate": 7.304560260586319e-05, | |
| "loss": 0.971, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.343767160900604, | |
| "grad_norm": 0.9143648147583008, | |
| "learning_rate": 7.280130293159608e-05, | |
| "loss": 0.6812, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 2.345963756177924, | |
| "grad_norm": 1.1228523254394531, | |
| "learning_rate": 7.255700325732898e-05, | |
| "loss": 0.9212, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.3481603514552445, | |
| "grad_norm": 1.8963797092437744, | |
| "learning_rate": 7.231270358306188e-05, | |
| "loss": 1.0675, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 2.3503569467325645, | |
| "grad_norm": 1.213624358177185, | |
| "learning_rate": 7.206840390879479e-05, | |
| "loss": 0.7898, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.352553542009885, | |
| "grad_norm": 0.985992968082428, | |
| "learning_rate": 7.182410423452768e-05, | |
| "loss": 0.9013, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 2.354750137287205, | |
| "grad_norm": 1.101683259010315, | |
| "learning_rate": 7.157980456026058e-05, | |
| "loss": 0.6841, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.356946732564525, | |
| "grad_norm": 0.8792346715927124, | |
| "learning_rate": 7.133550488599348e-05, | |
| "loss": 0.7187, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 2.359143327841845, | |
| "grad_norm": 2.0248584747314453, | |
| "learning_rate": 7.109120521172637e-05, | |
| "loss": 0.9014, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.361339923119165, | |
| "grad_norm": 1.3683643341064453, | |
| "learning_rate": 7.084690553745927e-05, | |
| "loss": 0.9346, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.3635365183964856, | |
| "grad_norm": 1.316840648651123, | |
| "learning_rate": 7.060260586319218e-05, | |
| "loss": 0.8979, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.3657331136738056, | |
| "grad_norm": 1.21073317527771, | |
| "learning_rate": 7.035830618892508e-05, | |
| "loss": 0.8275, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 2.3679297089511255, | |
| "grad_norm": 1.5338951349258423, | |
| "learning_rate": 7.011400651465797e-05, | |
| "loss": 0.9043, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.370126304228446, | |
| "grad_norm": 1.6079583168029785, | |
| "learning_rate": 6.986970684039087e-05, | |
| "loss": 0.6892, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 2.372322899505766, | |
| "grad_norm": 1.3436959981918335, | |
| "learning_rate": 6.962540716612377e-05, | |
| "loss": 0.8897, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.3745194947830863, | |
| "grad_norm": 1.4026193618774414, | |
| "learning_rate": 6.938110749185668e-05, | |
| "loss": 1.1274, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 2.3767160900604063, | |
| "grad_norm": 1.1094086170196533, | |
| "learning_rate": 6.913680781758956e-05, | |
| "loss": 0.8223, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.3789126853377267, | |
| "grad_norm": 1.4587323665618896, | |
| "learning_rate": 6.889250814332247e-05, | |
| "loss": 0.8555, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 2.3811092806150467, | |
| "grad_norm": 1.388769507408142, | |
| "learning_rate": 6.864820846905537e-05, | |
| "loss": 1.0072, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.3833058758923666, | |
| "grad_norm": 1.1615240573883057, | |
| "learning_rate": 6.840390879478826e-05, | |
| "loss": 1.0839, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.385502471169687, | |
| "grad_norm": 1.2834014892578125, | |
| "learning_rate": 6.815960912052116e-05, | |
| "loss": 0.7851, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.387699066447007, | |
| "grad_norm": 1.6230034828186035, | |
| "learning_rate": 6.791530944625406e-05, | |
| "loss": 1.1561, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 2.3898956617243274, | |
| "grad_norm": 1.105741262435913, | |
| "learning_rate": 6.767100977198697e-05, | |
| "loss": 0.7021, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.3920922570016474, | |
| "grad_norm": 1.7024222612380981, | |
| "learning_rate": 6.742671009771986e-05, | |
| "loss": 1.0061, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 2.394288852278968, | |
| "grad_norm": 1.3809547424316406, | |
| "learning_rate": 6.718241042345276e-05, | |
| "loss": 0.826, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.396485447556288, | |
| "grad_norm": 1.0459635257720947, | |
| "learning_rate": 6.693811074918566e-05, | |
| "loss": 0.8135, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 2.3986820428336078, | |
| "grad_norm": 1.4703642129898071, | |
| "learning_rate": 6.669381107491855e-05, | |
| "loss": 0.8355, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.400878638110928, | |
| "grad_norm": 1.2681678533554077, | |
| "learning_rate": 6.644951140065145e-05, | |
| "loss": 0.8863, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 2.403075233388248, | |
| "grad_norm": 0.9711754322052002, | |
| "learning_rate": 6.620521172638436e-05, | |
| "loss": 0.9471, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.4052718286655685, | |
| "grad_norm": 1.3253278732299805, | |
| "learning_rate": 6.596091205211726e-05, | |
| "loss": 0.737, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.4074684239428885, | |
| "grad_norm": 1.1183706521987915, | |
| "learning_rate": 6.571661237785015e-05, | |
| "loss": 0.7369, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.409665019220209, | |
| "grad_norm": 1.3476358652114868, | |
| "learning_rate": 6.547231270358307e-05, | |
| "loss": 0.9299, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 2.411861614497529, | |
| "grad_norm": 1.1968330144882202, | |
| "learning_rate": 6.522801302931595e-05, | |
| "loss": 0.9321, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.414058209774849, | |
| "grad_norm": 1.8310567140579224, | |
| "learning_rate": 6.498371335504886e-05, | |
| "loss": 0.9395, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 2.4162548050521693, | |
| "grad_norm": 1.2409788370132446, | |
| "learning_rate": 6.473941368078176e-05, | |
| "loss": 0.9209, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.4184514003294892, | |
| "grad_norm": 1.065801739692688, | |
| "learning_rate": 6.449511400651466e-05, | |
| "loss": 0.8192, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 2.420647995606809, | |
| "grad_norm": 1.2083854675292969, | |
| "learning_rate": 6.425081433224755e-05, | |
| "loss": 0.933, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.4228445908841296, | |
| "grad_norm": 1.0434120893478394, | |
| "learning_rate": 6.400651465798045e-05, | |
| "loss": 0.8065, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.4250411861614496, | |
| "grad_norm": 2.861186981201172, | |
| "learning_rate": 6.376221498371336e-05, | |
| "loss": 0.939, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.42723778143877, | |
| "grad_norm": 1.0325102806091309, | |
| "learning_rate": 6.351791530944625e-05, | |
| "loss": 0.8962, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.42943437671609, | |
| "grad_norm": 1.3520573377609253, | |
| "learning_rate": 6.327361563517915e-05, | |
| "loss": 0.7714, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.4316309719934104, | |
| "grad_norm": 1.4123116731643677, | |
| "learning_rate": 6.302931596091205e-05, | |
| "loss": 0.9877, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 2.4338275672707304, | |
| "grad_norm": 1.2660409212112427, | |
| "learning_rate": 6.278501628664495e-05, | |
| "loss": 1.0343, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.4360241625480503, | |
| "grad_norm": 1.3444467782974243, | |
| "learning_rate": 6.254071661237784e-05, | |
| "loss": 0.8782, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 2.4382207578253707, | |
| "grad_norm": 1.0368796586990356, | |
| "learning_rate": 6.229641693811075e-05, | |
| "loss": 0.8901, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.4404173531026907, | |
| "grad_norm": 1.1158747673034668, | |
| "learning_rate": 6.205211726384365e-05, | |
| "loss": 0.8605, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 2.442613948380011, | |
| "grad_norm": 0.8776071667671204, | |
| "learning_rate": 6.180781758957654e-05, | |
| "loss": 0.6398, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.444810543657331, | |
| "grad_norm": 1.1517044305801392, | |
| "learning_rate": 6.156351791530944e-05, | |
| "loss": 1.1219, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 2.4470071389346515, | |
| "grad_norm": 1.0181858539581299, | |
| "learning_rate": 6.131921824104234e-05, | |
| "loss": 1.0752, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.4492037342119715, | |
| "grad_norm": 0.8813405632972717, | |
| "learning_rate": 6.107491856677525e-05, | |
| "loss": 0.8201, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.4514003294892914, | |
| "grad_norm": 0.9770997166633606, | |
| "learning_rate": 6.083061889250814e-05, | |
| "loss": 0.7619, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.453596924766612, | |
| "grad_norm": 1.0907503366470337, | |
| "learning_rate": 6.058631921824104e-05, | |
| "loss": 0.8062, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 2.455793520043932, | |
| "grad_norm": 1.7084773778915405, | |
| "learning_rate": 6.0342019543973935e-05, | |
| "loss": 0.9198, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.4579901153212522, | |
| "grad_norm": 1.1684274673461914, | |
| "learning_rate": 6.009771986970684e-05, | |
| "loss": 0.8897, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 2.460186710598572, | |
| "grad_norm": 1.1164532899856567, | |
| "learning_rate": 5.9853420195439734e-05, | |
| "loss": 0.8091, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.4623833058758926, | |
| "grad_norm": 2.3498594760894775, | |
| "learning_rate": 5.9609120521172636e-05, | |
| "loss": 0.9212, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 2.4645799011532126, | |
| "grad_norm": 1.512941837310791, | |
| "learning_rate": 5.936482084690553e-05, | |
| "loss": 1.2465, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.4667764964305325, | |
| "grad_norm": 1.8065310716629028, | |
| "learning_rate": 5.9120521172638435e-05, | |
| "loss": 0.9141, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 2.468973091707853, | |
| "grad_norm": 1.0092695951461792, | |
| "learning_rate": 5.887622149837133e-05, | |
| "loss": 0.9218, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.471169686985173, | |
| "grad_norm": 0.8607140183448792, | |
| "learning_rate": 5.863192182410423e-05, | |
| "loss": 0.7359, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.473366282262493, | |
| "grad_norm": 1.5194811820983887, | |
| "learning_rate": 5.838762214983713e-05, | |
| "loss": 0.982, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.4755628775398133, | |
| "grad_norm": 1.4082212448120117, | |
| "learning_rate": 5.8143322475570026e-05, | |
| "loss": 0.8133, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 2.4777594728171333, | |
| "grad_norm": 1.0593407154083252, | |
| "learning_rate": 5.789902280130293e-05, | |
| "loss": 0.8832, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.4799560680944537, | |
| "grad_norm": 1.3378733396530151, | |
| "learning_rate": 5.7654723127035825e-05, | |
| "loss": 1.0643, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 2.4821526633717736, | |
| "grad_norm": 2.411097288131714, | |
| "learning_rate": 5.741042345276873e-05, | |
| "loss": 0.8441, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.484349258649094, | |
| "grad_norm": 1.1326396465301514, | |
| "learning_rate": 5.716612377850162e-05, | |
| "loss": 0.8668, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 2.486545853926414, | |
| "grad_norm": 1.0524181127548218, | |
| "learning_rate": 5.6921824104234526e-05, | |
| "loss": 0.8589, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.488742449203734, | |
| "grad_norm": 1.0454916954040527, | |
| "learning_rate": 5.667752442996742e-05, | |
| "loss": 0.7247, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 2.4909390444810544, | |
| "grad_norm": 1.6732035875320435, | |
| "learning_rate": 5.643322475570032e-05, | |
| "loss": 0.8747, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.4931356397583744, | |
| "grad_norm": 3.5578784942626953, | |
| "learning_rate": 5.618892508143322e-05, | |
| "loss": 0.8685, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.495332235035695, | |
| "grad_norm": 1.0309659242630005, | |
| "learning_rate": 5.594462540716612e-05, | |
| "loss": 0.8147, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.4975288303130148, | |
| "grad_norm": 1.073958396911621, | |
| "learning_rate": 5.570032573289902e-05, | |
| "loss": 0.6074, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 2.499725425590335, | |
| "grad_norm": 1.9705623388290405, | |
| "learning_rate": 5.5456026058631915e-05, | |
| "loss": 0.9174, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.501922020867655, | |
| "grad_norm": 1.3620948791503906, | |
| "learning_rate": 5.521172638436482e-05, | |
| "loss": 1.0521, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.504118616144975, | |
| "grad_norm": 1.1516716480255127, | |
| "learning_rate": 5.4967426710097714e-05, | |
| "loss": 0.7685, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.5063152114222955, | |
| "grad_norm": 0.8479107022285461, | |
| "learning_rate": 5.472312703583061e-05, | |
| "loss": 0.754, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 2.5085118066996155, | |
| "grad_norm": 1.369097352027893, | |
| "learning_rate": 5.447882736156351e-05, | |
| "loss": 0.9374, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 2.510708401976936, | |
| "grad_norm": 0.9345399737358093, | |
| "learning_rate": 5.423452768729641e-05, | |
| "loss": 0.9754, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 2.512904997254256, | |
| "grad_norm": 0.9029057621955872, | |
| "learning_rate": 5.399022801302931e-05, | |
| "loss": 0.853, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 2.5151015925315763, | |
| "grad_norm": 1.0849556922912598, | |
| "learning_rate": 5.374592833876221e-05, | |
| "loss": 0.8151, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.5172981878088962, | |
| "grad_norm": 1.7522032260894775, | |
| "learning_rate": 5.350162866449511e-05, | |
| "loss": 0.8365, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 2.519494783086216, | |
| "grad_norm": 1.3333768844604492, | |
| "learning_rate": 5.3257328990228006e-05, | |
| "loss": 0.7697, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 2.5216913783635366, | |
| "grad_norm": 1.9205724000930786, | |
| "learning_rate": 5.301302931596091e-05, | |
| "loss": 0.737, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 2.5238879736408566, | |
| "grad_norm": 1.00103759765625, | |
| "learning_rate": 5.2768729641693805e-05, | |
| "loss": 0.948, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 2.5260845689181766, | |
| "grad_norm": 1.011674165725708, | |
| "learning_rate": 5.25244299674267e-05, | |
| "loss": 1.1004, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.528281164195497, | |
| "grad_norm": 0.9842965006828308, | |
| "learning_rate": 5.2280130293159604e-05, | |
| "loss": 0.8981, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 2.5304777594728174, | |
| "grad_norm": 1.2793430089950562, | |
| "learning_rate": 5.20358306188925e-05, | |
| "loss": 0.8596, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 2.5326743547501374, | |
| "grad_norm": 0.8217779397964478, | |
| "learning_rate": 5.17915309446254e-05, | |
| "loss": 0.8776, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 2.5348709500274573, | |
| "grad_norm": 0.9789122939109802, | |
| "learning_rate": 5.15472312703583e-05, | |
| "loss": 0.8407, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 2.5370675453047777, | |
| "grad_norm": 0.9248374104499817, | |
| "learning_rate": 5.13029315960912e-05, | |
| "loss": 0.8402, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.5392641405820977, | |
| "grad_norm": 1.5190105438232422, | |
| "learning_rate": 5.10586319218241e-05, | |
| "loss": 0.9892, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.5414607358594177, | |
| "grad_norm": 1.2966643571853638, | |
| "learning_rate": 5.081433224755699e-05, | |
| "loss": 0.9808, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 2.543657331136738, | |
| "grad_norm": 0.855202317237854, | |
| "learning_rate": 5.0570032573289896e-05, | |
| "loss": 0.7486, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 2.545853926414058, | |
| "grad_norm": 1.2291526794433594, | |
| "learning_rate": 5.032573289902279e-05, | |
| "loss": 0.7575, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 2.5480505216913785, | |
| "grad_norm": 0.7679459452629089, | |
| "learning_rate": 5.0081433224755695e-05, | |
| "loss": 0.8006, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.5502471169686984, | |
| "grad_norm": 0.9735711812973022, | |
| "learning_rate": 4.983713355048859e-05, | |
| "loss": 0.6879, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 2.552443712246019, | |
| "grad_norm": 1.0061559677124023, | |
| "learning_rate": 4.9592833876221493e-05, | |
| "loss": 0.8422, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 2.554640307523339, | |
| "grad_norm": 1.277942419052124, | |
| "learning_rate": 4.934853420195439e-05, | |
| "loss": 0.6927, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 2.556836902800659, | |
| "grad_norm": 1.1331262588500977, | |
| "learning_rate": 4.910423452768729e-05, | |
| "loss": 0.9251, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 2.559033498077979, | |
| "grad_norm": 1.0753247737884521, | |
| "learning_rate": 4.885993485342019e-05, | |
| "loss": 0.9339, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.561230093355299, | |
| "grad_norm": 0.8370097875595093, | |
| "learning_rate": 4.8615635179153084e-05, | |
| "loss": 0.7791, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 2.5634266886326196, | |
| "grad_norm": 0.8946124911308289, | |
| "learning_rate": 4.837133550488599e-05, | |
| "loss": 0.7178, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 2.5656232839099395, | |
| "grad_norm": 2.765690803527832, | |
| "learning_rate": 4.812703583061888e-05, | |
| "loss": 0.9205, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.56781987918726, | |
| "grad_norm": 1.1060473918914795, | |
| "learning_rate": 4.7882736156351786e-05, | |
| "loss": 0.809, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 2.57001647446458, | |
| "grad_norm": 1.0659475326538086, | |
| "learning_rate": 4.763843648208468e-05, | |
| "loss": 0.7603, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.5722130697419, | |
| "grad_norm": 0.9563616514205933, | |
| "learning_rate": 4.7394136807817584e-05, | |
| "loss": 0.7433, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 2.5744096650192203, | |
| "grad_norm": 1.6382946968078613, | |
| "learning_rate": 4.714983713355048e-05, | |
| "loss": 0.7462, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 2.5766062602965403, | |
| "grad_norm": 1.1870077848434448, | |
| "learning_rate": 4.6905537459283376e-05, | |
| "loss": 0.8028, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 2.5788028555738602, | |
| "grad_norm": 1.0493857860565186, | |
| "learning_rate": 4.6661237785016286e-05, | |
| "loss": 0.8993, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 2.5809994508511807, | |
| "grad_norm": 0.9559311270713806, | |
| "learning_rate": 4.641693811074919e-05, | |
| "loss": 0.9137, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.583196046128501, | |
| "grad_norm": 1.0354852676391602, | |
| "learning_rate": 4.6172638436482085e-05, | |
| "loss": 0.798, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 2.585392641405821, | |
| "grad_norm": 0.8822775483131409, | |
| "learning_rate": 4.592833876221498e-05, | |
| "loss": 0.9833, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 2.587589236683141, | |
| "grad_norm": 2.044999837875366, | |
| "learning_rate": 4.568403908794788e-05, | |
| "loss": 0.7995, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 2.5897858319604614, | |
| "grad_norm": 1.6066820621490479, | |
| "learning_rate": 4.543973941368078e-05, | |
| "loss": 0.9919, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 2.5919824272377814, | |
| "grad_norm": 1.0398755073547363, | |
| "learning_rate": 4.519543973941368e-05, | |
| "loss": 0.7157, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.5941790225151014, | |
| "grad_norm": 1.6578835248947144, | |
| "learning_rate": 4.495114006514658e-05, | |
| "loss": 1.0445, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 2.5963756177924218, | |
| "grad_norm": 1.2569780349731445, | |
| "learning_rate": 4.470684039087948e-05, | |
| "loss": 0.8845, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 2.598572213069742, | |
| "grad_norm": 0.7564569711685181, | |
| "learning_rate": 4.446254071661238e-05, | |
| "loss": 0.8126, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 2.600768808347062, | |
| "grad_norm": 1.0271086692810059, | |
| "learning_rate": 4.421824104234527e-05, | |
| "loss": 0.8989, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.602965403624382, | |
| "grad_norm": 1.1719675064086914, | |
| "learning_rate": 4.3973941368078175e-05, | |
| "loss": 0.934, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.6051619989017025, | |
| "grad_norm": 0.8848729729652405, | |
| "learning_rate": 4.372964169381107e-05, | |
| "loss": 0.8541, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 2.6073585941790225, | |
| "grad_norm": 0.9240156412124634, | |
| "learning_rate": 4.3485342019543974e-05, | |
| "loss": 0.7308, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 2.6095551894563425, | |
| "grad_norm": 1.3479830026626587, | |
| "learning_rate": 4.324104234527687e-05, | |
| "loss": 0.8604, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 2.611751784733663, | |
| "grad_norm": 1.5111393928527832, | |
| "learning_rate": 4.299674267100977e-05, | |
| "loss": 0.8487, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 2.613948380010983, | |
| "grad_norm": 1.0924351215362549, | |
| "learning_rate": 4.275244299674267e-05, | |
| "loss": 0.8793, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.6161449752883033, | |
| "grad_norm": 0.9148683547973633, | |
| "learning_rate": 4.250814332247557e-05, | |
| "loss": 0.7591, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 2.618341570565623, | |
| "grad_norm": 1.1730152368545532, | |
| "learning_rate": 4.226384364820847e-05, | |
| "loss": 0.6494, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 2.6205381658429436, | |
| "grad_norm": 1.2757980823516846, | |
| "learning_rate": 4.2019543973941364e-05, | |
| "loss": 1.3453, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 2.6227347611202636, | |
| "grad_norm": 1.2076345682144165, | |
| "learning_rate": 4.1775244299674266e-05, | |
| "loss": 0.9615, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 2.6249313563975836, | |
| "grad_norm": 1.3630986213684082, | |
| "learning_rate": 4.153094462540716e-05, | |
| "loss": 0.5929, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.627127951674904, | |
| "grad_norm": 1.2658156156539917, | |
| "learning_rate": 4.1286644951140065e-05, | |
| "loss": 0.8333, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 2.629324546952224, | |
| "grad_norm": 1.5235031843185425, | |
| "learning_rate": 4.104234527687296e-05, | |
| "loss": 0.8443, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 2.6315211422295444, | |
| "grad_norm": 0.7697120308876038, | |
| "learning_rate": 4.0798045602605864e-05, | |
| "loss": 0.8747, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 2.6337177375068643, | |
| "grad_norm": 0.899372935295105, | |
| "learning_rate": 4.055374592833876e-05, | |
| "loss": 0.6747, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 2.6359143327841847, | |
| "grad_norm": 1.242138385772705, | |
| "learning_rate": 4.0309446254071656e-05, | |
| "loss": 0.8163, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.6381109280615047, | |
| "grad_norm": 0.952166736125946, | |
| "learning_rate": 4.006514657980456e-05, | |
| "loss": 0.6725, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 2.6403075233388247, | |
| "grad_norm": 1.2235968112945557, | |
| "learning_rate": 3.9820846905537454e-05, | |
| "loss": 0.6599, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 2.642504118616145, | |
| "grad_norm": 1.394832968711853, | |
| "learning_rate": 3.957654723127036e-05, | |
| "loss": 1.089, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 2.644700713893465, | |
| "grad_norm": 1.0789695978164673, | |
| "learning_rate": 3.933224755700325e-05, | |
| "loss": 0.7631, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 2.646897309170785, | |
| "grad_norm": 1.1938402652740479, | |
| "learning_rate": 3.9087947882736156e-05, | |
| "loss": 0.8656, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.6490939044481054, | |
| "grad_norm": 0.9868903160095215, | |
| "learning_rate": 3.884364820846905e-05, | |
| "loss": 0.9899, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 2.651290499725426, | |
| "grad_norm": 1.3135132789611816, | |
| "learning_rate": 3.8599348534201955e-05, | |
| "loss": 0.9245, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 2.653487095002746, | |
| "grad_norm": 1.2097454071044922, | |
| "learning_rate": 3.835504885993485e-05, | |
| "loss": 0.8725, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 2.655683690280066, | |
| "grad_norm": 1.2300525903701782, | |
| "learning_rate": 3.8110749185667747e-05, | |
| "loss": 0.7239, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 2.657880285557386, | |
| "grad_norm": 1.6410201787948608, | |
| "learning_rate": 3.786644951140065e-05, | |
| "loss": 1.0612, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.660076880834706, | |
| "grad_norm": 5.413272380828857, | |
| "learning_rate": 3.7622149837133545e-05, | |
| "loss": 0.8339, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 2.662273476112026, | |
| "grad_norm": 1.8787386417388916, | |
| "learning_rate": 3.737785016286645e-05, | |
| "loss": 0.8176, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 2.6644700713893466, | |
| "grad_norm": 1.0435289144515991, | |
| "learning_rate": 3.7133550488599344e-05, | |
| "loss": 0.8885, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 1.3319135904312134, | |
| "learning_rate": 3.688925081433225e-05, | |
| "loss": 0.7998, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 2.668863261943987, | |
| "grad_norm": 0.9780111312866211, | |
| "learning_rate": 3.664495114006514e-05, | |
| "loss": 0.7838, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.671059857221307, | |
| "grad_norm": 1.0613712072372437, | |
| "learning_rate": 3.640065146579804e-05, | |
| "loss": 0.913, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 2.6732564524986273, | |
| "grad_norm": 1.2312910556793213, | |
| "learning_rate": 3.615635179153094e-05, | |
| "loss": 0.9319, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 2.6754530477759473, | |
| "grad_norm": 1.5348732471466064, | |
| "learning_rate": 3.591205211726384e-05, | |
| "loss": 0.8286, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 2.6776496430532672, | |
| "grad_norm": 1.3830292224884033, | |
| "learning_rate": 3.566775244299674e-05, | |
| "loss": 0.778, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 2.6798462383305877, | |
| "grad_norm": 1.391552448272705, | |
| "learning_rate": 3.5423452768729636e-05, | |
| "loss": 0.8366, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.6820428336079076, | |
| "grad_norm": 1.8366199731826782, | |
| "learning_rate": 3.517915309446254e-05, | |
| "loss": 1.0616, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 2.684239428885228, | |
| "grad_norm": 1.1593966484069824, | |
| "learning_rate": 3.4934853420195435e-05, | |
| "loss": 0.6405, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 2.686436024162548, | |
| "grad_norm": 1.3455157279968262, | |
| "learning_rate": 3.469055374592834e-05, | |
| "loss": 0.944, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 2.6886326194398684, | |
| "grad_norm": 1.5045685768127441, | |
| "learning_rate": 3.4446254071661234e-05, | |
| "loss": 0.8224, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 2.6908292147171884, | |
| "grad_norm": 0.8928258419036865, | |
| "learning_rate": 3.420195439739413e-05, | |
| "loss": 0.9118, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.6930258099945084, | |
| "grad_norm": 1.0776011943817139, | |
| "learning_rate": 3.395765472312703e-05, | |
| "loss": 0.6991, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 2.6952224052718288, | |
| "grad_norm": 0.9437106847763062, | |
| "learning_rate": 3.371335504885993e-05, | |
| "loss": 0.8943, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 2.6974190005491487, | |
| "grad_norm": 1.0122783184051514, | |
| "learning_rate": 3.346905537459283e-05, | |
| "loss": 0.8268, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 2.6996155958264687, | |
| "grad_norm": 0.9164775013923645, | |
| "learning_rate": 3.322475570032573e-05, | |
| "loss": 0.915, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 2.701812191103789, | |
| "grad_norm": 1.0424001216888428, | |
| "learning_rate": 3.298045602605863e-05, | |
| "loss": 0.7043, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.7040087863811095, | |
| "grad_norm": 1.1365594863891602, | |
| "learning_rate": 3.273615635179153e-05, | |
| "loss": 0.7721, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 2.7062053816584295, | |
| "grad_norm": 1.3189200162887573, | |
| "learning_rate": 3.249185667752443e-05, | |
| "loss": 0.9579, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 2.7084019769357495, | |
| "grad_norm": 1.2020214796066284, | |
| "learning_rate": 3.224755700325733e-05, | |
| "loss": 0.7895, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 2.71059857221307, | |
| "grad_norm": 1.0074046850204468, | |
| "learning_rate": 3.200325732899023e-05, | |
| "loss": 0.8517, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 2.71279516749039, | |
| "grad_norm": 0.8530951738357544, | |
| "learning_rate": 3.175895765472312e-05, | |
| "loss": 0.9096, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.71499176276771, | |
| "grad_norm": 1.581216812133789, | |
| "learning_rate": 3.1514657980456026e-05, | |
| "loss": 1.096, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 2.7171883580450302, | |
| "grad_norm": 1.045326590538025, | |
| "learning_rate": 3.127035830618892e-05, | |
| "loss": 0.88, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 2.71938495332235, | |
| "grad_norm": 0.9708707928657532, | |
| "learning_rate": 3.1026058631921825e-05, | |
| "loss": 0.8234, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 2.7215815485996706, | |
| "grad_norm": 1.4384783506393433, | |
| "learning_rate": 3.078175895765472e-05, | |
| "loss": 0.8896, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 2.7237781438769906, | |
| "grad_norm": 1.037429690361023, | |
| "learning_rate": 3.0537459283387624e-05, | |
| "loss": 0.8639, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.725974739154311, | |
| "grad_norm": 1.0188984870910645, | |
| "learning_rate": 3.029315960912052e-05, | |
| "loss": 0.6669, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 2.728171334431631, | |
| "grad_norm": 1.037124514579773, | |
| "learning_rate": 3.004885993485342e-05, | |
| "loss": 0.8323, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 2.730367929708951, | |
| "grad_norm": 0.9664590954780579, | |
| "learning_rate": 2.9804560260586318e-05, | |
| "loss": 0.9373, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 2.7325645249862713, | |
| "grad_norm": 1.217652678489685, | |
| "learning_rate": 2.9560260586319218e-05, | |
| "loss": 0.9453, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 2.7347611202635913, | |
| "grad_norm": 1.0391770601272583, | |
| "learning_rate": 2.9315960912052114e-05, | |
| "loss": 0.8426, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.7369577155409117, | |
| "grad_norm": 2.063138008117676, | |
| "learning_rate": 2.9071661237785013e-05, | |
| "loss": 0.8813, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 2.7391543108182317, | |
| "grad_norm": 1.309981107711792, | |
| "learning_rate": 2.8827361563517912e-05, | |
| "loss": 1.3591, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 2.741350906095552, | |
| "grad_norm": 1.037191390991211, | |
| "learning_rate": 2.858306188925081e-05, | |
| "loss": 0.8844, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 2.743547501372872, | |
| "grad_norm": 1.0931987762451172, | |
| "learning_rate": 2.833876221498371e-05, | |
| "loss": 0.9707, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 2.745744096650192, | |
| "grad_norm": 2.16721248626709, | |
| "learning_rate": 2.809446254071661e-05, | |
| "loss": 0.9061, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.7479406919275124, | |
| "grad_norm": 0.9501640200614929, | |
| "learning_rate": 2.785016286644951e-05, | |
| "loss": 0.9927, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 2.7501372872048324, | |
| "grad_norm": 1.8190885782241821, | |
| "learning_rate": 2.760586319218241e-05, | |
| "loss": 0.943, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 2.752333882482153, | |
| "grad_norm": 1.047094702720642, | |
| "learning_rate": 2.7361563517915305e-05, | |
| "loss": 0.7121, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 2.754530477759473, | |
| "grad_norm": 0.8964688181877136, | |
| "learning_rate": 2.7117263843648204e-05, | |
| "loss": 0.7554, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 2.756727073036793, | |
| "grad_norm": 1.3040779829025269, | |
| "learning_rate": 2.6872964169381104e-05, | |
| "loss": 1.013, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.758923668314113, | |
| "grad_norm": 1.169255018234253, | |
| "learning_rate": 2.6628664495114003e-05, | |
| "loss": 0.6653, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 2.761120263591433, | |
| "grad_norm": 1.123046875, | |
| "learning_rate": 2.6384364820846903e-05, | |
| "loss": 0.8708, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 2.7633168588687536, | |
| "grad_norm": 0.8622208833694458, | |
| "learning_rate": 2.6140065146579802e-05, | |
| "loss": 0.6874, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 2.7655134541460735, | |
| "grad_norm": 1.216880440711975, | |
| "learning_rate": 2.58957654723127e-05, | |
| "loss": 1.015, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 2.7677100494233935, | |
| "grad_norm": 1.2886595726013184, | |
| "learning_rate": 2.56514657980456e-05, | |
| "loss": 0.9048, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.769906644700714, | |
| "grad_norm": 1.3192064762115479, | |
| "learning_rate": 2.5407166123778497e-05, | |
| "loss": 0.8396, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 2.7721032399780343, | |
| "grad_norm": 1.3086817264556885, | |
| "learning_rate": 2.5162866449511396e-05, | |
| "loss": 0.8741, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 2.7742998352553543, | |
| "grad_norm": 1.2222366333007812, | |
| "learning_rate": 2.4918566775244295e-05, | |
| "loss": 0.9113, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 2.7764964305326743, | |
| "grad_norm": 1.302921175956726, | |
| "learning_rate": 2.4674267100977195e-05, | |
| "loss": 1.0418, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 2.7786930258099947, | |
| "grad_norm": 0.9135136604309082, | |
| "learning_rate": 2.4429967426710094e-05, | |
| "loss": 0.9731, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.7808896210873146, | |
| "grad_norm": 0.7827017307281494, | |
| "learning_rate": 2.4185667752442993e-05, | |
| "loss": 0.7232, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 2.7830862163646346, | |
| "grad_norm": 1.0402188301086426, | |
| "learning_rate": 2.3941368078175893e-05, | |
| "loss": 0.7635, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 2.785282811641955, | |
| "grad_norm": 1.075492262840271, | |
| "learning_rate": 2.3697068403908792e-05, | |
| "loss": 0.8346, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 2.787479406919275, | |
| "grad_norm": 1.4890602827072144, | |
| "learning_rate": 2.3452768729641688e-05, | |
| "loss": 0.7228, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 2.7896760021965954, | |
| "grad_norm": 2.4160094261169434, | |
| "learning_rate": 2.3208469055374594e-05, | |
| "loss": 0.7726, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.7918725974739154, | |
| "grad_norm": 0.985327959060669, | |
| "learning_rate": 2.296416938110749e-05, | |
| "loss": 0.733, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 2.7940691927512358, | |
| "grad_norm": 1.300789475440979, | |
| "learning_rate": 2.271986970684039e-05, | |
| "loss": 0.8913, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 2.7962657880285557, | |
| "grad_norm": 1.114559292793274, | |
| "learning_rate": 2.247557003257329e-05, | |
| "loss": 0.878, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 2.7984623833058757, | |
| "grad_norm": 1.0144329071044922, | |
| "learning_rate": 2.223127035830619e-05, | |
| "loss": 1.064, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 2.800658978583196, | |
| "grad_norm": 1.7015892267227173, | |
| "learning_rate": 2.1986970684039088e-05, | |
| "loss": 0.8405, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.802855573860516, | |
| "grad_norm": 1.109307050704956, | |
| "learning_rate": 2.1742671009771987e-05, | |
| "loss": 0.9851, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 2.8050521691378365, | |
| "grad_norm": 1.7427302598953247, | |
| "learning_rate": 2.1498371335504886e-05, | |
| "loss": 1.1001, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 2.8072487644151565, | |
| "grad_norm": 1.247683048248291, | |
| "learning_rate": 2.1254071661237786e-05, | |
| "loss": 0.9303, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 2.809445359692477, | |
| "grad_norm": 1.0203958749771118, | |
| "learning_rate": 2.1009771986970682e-05, | |
| "loss": 0.7621, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 2.811641954969797, | |
| "grad_norm": 1.2967689037322998, | |
| "learning_rate": 2.076547231270358e-05, | |
| "loss": 0.8049, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.813838550247117, | |
| "grad_norm": 1.1701385974884033, | |
| "learning_rate": 2.052117263843648e-05, | |
| "loss": 0.9015, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 2.8160351455244372, | |
| "grad_norm": 1.0702723264694214, | |
| "learning_rate": 2.027687296416938e-05, | |
| "loss": 0.8226, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 2.818231740801757, | |
| "grad_norm": 1.3522272109985352, | |
| "learning_rate": 2.003257328990228e-05, | |
| "loss": 1.0621, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 2.820428336079077, | |
| "grad_norm": 1.0732183456420898, | |
| "learning_rate": 1.978827361563518e-05, | |
| "loss": 0.97, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 2.8226249313563976, | |
| "grad_norm": 1.1556763648986816, | |
| "learning_rate": 1.9543973941368078e-05, | |
| "loss": 0.9431, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.824821526633718, | |
| "grad_norm": 1.019116997718811, | |
| "learning_rate": 1.9299674267100977e-05, | |
| "loss": 1.0621, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 2.827018121911038, | |
| "grad_norm": 1.1273375749588013, | |
| "learning_rate": 1.9055374592833873e-05, | |
| "loss": 0.7635, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 2.829214717188358, | |
| "grad_norm": 1.4081268310546875, | |
| "learning_rate": 1.8811074918566773e-05, | |
| "loss": 0.9682, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 2.8314113124656783, | |
| "grad_norm": 1.1118184328079224, | |
| "learning_rate": 1.8566775244299672e-05, | |
| "loss": 0.7911, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 2.8336079077429983, | |
| "grad_norm": 1.4093362092971802, | |
| "learning_rate": 1.832247557003257e-05, | |
| "loss": 0.963, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.8358045030203183, | |
| "grad_norm": 1.334221363067627, | |
| "learning_rate": 1.807817589576547e-05, | |
| "loss": 0.7566, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 2.8380010982976387, | |
| "grad_norm": 1.2348610162734985, | |
| "learning_rate": 1.783387622149837e-05, | |
| "loss": 0.7868, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 2.8401976935749587, | |
| "grad_norm": 1.260190486907959, | |
| "learning_rate": 1.758957654723127e-05, | |
| "loss": 0.6996, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 2.842394288852279, | |
| "grad_norm": 1.1248668432235718, | |
| "learning_rate": 1.734527687296417e-05, | |
| "loss": 0.9437, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 2.844590884129599, | |
| "grad_norm": 1.7150044441223145, | |
| "learning_rate": 1.7100977198697065e-05, | |
| "loss": 0.9285, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.8467874794069195, | |
| "grad_norm": 1.5910953283309937, | |
| "learning_rate": 1.6856677524429964e-05, | |
| "loss": 0.994, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 2.8489840746842394, | |
| "grad_norm": 0.8757630586624146, | |
| "learning_rate": 1.6612377850162864e-05, | |
| "loss": 0.7491, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 2.8511806699615594, | |
| "grad_norm": 1.3299140930175781, | |
| "learning_rate": 1.6368078175895766e-05, | |
| "loss": 1.1164, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 2.85337726523888, | |
| "grad_norm": 1.3504143953323364, | |
| "learning_rate": 1.6123778501628666e-05, | |
| "loss": 0.9145, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 2.8555738605161998, | |
| "grad_norm": 1.729332447052002, | |
| "learning_rate": 1.587947882736156e-05, | |
| "loss": 1.0842, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.85777045579352, | |
| "grad_norm": 0.9344820976257324, | |
| "learning_rate": 1.563517915309446e-05, | |
| "loss": 0.8679, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 2.85996705107084, | |
| "grad_norm": 1.1068594455718994, | |
| "learning_rate": 1.539087947882736e-05, | |
| "loss": 0.7283, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 2.8621636463481606, | |
| "grad_norm": 0.9639398455619812, | |
| "learning_rate": 1.514657980456026e-05, | |
| "loss": 1.1114, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 2.8643602416254805, | |
| "grad_norm": 1.1765731573104858, | |
| "learning_rate": 1.4902280130293159e-05, | |
| "loss": 0.6964, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 2.8665568369028005, | |
| "grad_norm": 1.215322494506836, | |
| "learning_rate": 1.4657980456026057e-05, | |
| "loss": 0.8301, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.868753432180121, | |
| "grad_norm": 0.8946697115898132, | |
| "learning_rate": 1.4413680781758956e-05, | |
| "loss": 0.8393, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 2.870950027457441, | |
| "grad_norm": 1.343819499015808, | |
| "learning_rate": 1.4169381107491856e-05, | |
| "loss": 0.9394, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 2.873146622734761, | |
| "grad_norm": 1.0153502225875854, | |
| "learning_rate": 1.3925081433224755e-05, | |
| "loss": 0.7729, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 2.8753432180120813, | |
| "grad_norm": 0.943464994430542, | |
| "learning_rate": 1.3680781758957653e-05, | |
| "loss": 0.8021, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 2.8775398132894017, | |
| "grad_norm": 1.4244797229766846, | |
| "learning_rate": 1.3436482084690552e-05, | |
| "loss": 1.0023, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.8797364085667216, | |
| "grad_norm": 1.106939435005188, | |
| "learning_rate": 1.3192182410423451e-05, | |
| "loss": 0.8714, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 2.8819330038440416, | |
| "grad_norm": 1.0381802320480347, | |
| "learning_rate": 1.294788273615635e-05, | |
| "loss": 0.8543, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 2.884129599121362, | |
| "grad_norm": 1.3506184816360474, | |
| "learning_rate": 1.2703583061889248e-05, | |
| "loss": 0.859, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 2.886326194398682, | |
| "grad_norm": 0.9586111903190613, | |
| "learning_rate": 1.2459283387622148e-05, | |
| "loss": 0.864, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 2.888522789676002, | |
| "grad_norm": 0.9992175102233887, | |
| "learning_rate": 1.2214983713355047e-05, | |
| "loss": 0.7907, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.8907193849533224, | |
| "grad_norm": 1.261605143547058, | |
| "learning_rate": 1.1970684039087946e-05, | |
| "loss": 0.8529, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 2.892915980230643, | |
| "grad_norm": 0.9609043002128601, | |
| "learning_rate": 1.1726384364820844e-05, | |
| "loss": 0.7055, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 2.8951125755079627, | |
| "grad_norm": 1.319348692893982, | |
| "learning_rate": 1.1482084690553745e-05, | |
| "loss": 0.7441, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 2.8973091707852827, | |
| "grad_norm": 3.7486751079559326, | |
| "learning_rate": 1.1237785016286644e-05, | |
| "loss": 0.6907, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 2.899505766062603, | |
| "grad_norm": 1.3148154020309448, | |
| "learning_rate": 1.0993485342019544e-05, | |
| "loss": 1.0001, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.901702361339923, | |
| "grad_norm": 1.0113205909729004, | |
| "learning_rate": 1.0749185667752443e-05, | |
| "loss": 0.6387, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 2.903898956617243, | |
| "grad_norm": 1.1947811841964722, | |
| "learning_rate": 1.0504885993485341e-05, | |
| "loss": 0.8781, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 2.9060955518945635, | |
| "grad_norm": 1.2437448501586914, | |
| "learning_rate": 1.026058631921824e-05, | |
| "loss": 0.7778, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 2.9082921471718834, | |
| "grad_norm": 1.320144534111023, | |
| "learning_rate": 1.001628664495114e-05, | |
| "loss": 0.8277, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 2.910488742449204, | |
| "grad_norm": 0.9644711017608643, | |
| "learning_rate": 9.771986970684039e-06, | |
| "loss": 0.7722, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.912685337726524, | |
| "grad_norm": 5.376433849334717, | |
| "learning_rate": 9.527687296416937e-06, | |
| "loss": 0.866, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 2.9148819330038442, | |
| "grad_norm": 1.0609482526779175, | |
| "learning_rate": 9.283387622149836e-06, | |
| "loss": 0.8958, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 2.917078528281164, | |
| "grad_norm": 1.0237983465194702, | |
| "learning_rate": 9.039087947882735e-06, | |
| "loss": 0.7229, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 2.919275123558484, | |
| "grad_norm": 0.7842565774917603, | |
| "learning_rate": 8.794788273615635e-06, | |
| "loss": 0.6121, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 2.9214717188358046, | |
| "grad_norm": 1.414908528327942, | |
| "learning_rate": 8.550488599348532e-06, | |
| "loss": 0.8335, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.9236683141131246, | |
| "grad_norm": 0.9051879048347473, | |
| "learning_rate": 8.306188925081432e-06, | |
| "loss": 0.9257, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 2.925864909390445, | |
| "grad_norm": 0.9372172951698303, | |
| "learning_rate": 8.061889250814333e-06, | |
| "loss": 0.5061, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 2.928061504667765, | |
| "grad_norm": 1.2651898860931396, | |
| "learning_rate": 7.81758957654723e-06, | |
| "loss": 0.772, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 2.9302580999450853, | |
| "grad_norm": 0.87308669090271, | |
| "learning_rate": 7.57328990228013e-06, | |
| "loss": 0.7354, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 2.9324546952224053, | |
| "grad_norm": 1.1068834066390991, | |
| "learning_rate": 7.328990228013028e-06, | |
| "loss": 0.7072, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.9346512904997253, | |
| "grad_norm": 1.157633662223816, | |
| "learning_rate": 7.084690553745928e-06, | |
| "loss": 0.7362, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 2.9368478857770457, | |
| "grad_norm": 1.1332664489746094, | |
| "learning_rate": 6.840390879478826e-06, | |
| "loss": 0.7427, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 2.9390444810543657, | |
| "grad_norm": 1.5844314098358154, | |
| "learning_rate": 6.596091205211726e-06, | |
| "loss": 0.8045, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 2.9412410763316856, | |
| "grad_norm": 18.48008918762207, | |
| "learning_rate": 6.351791530944624e-06, | |
| "loss": 0.7844, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 2.943437671609006, | |
| "grad_norm": 1.0144950151443481, | |
| "learning_rate": 6.1074918566775235e-06, | |
| "loss": 0.5907, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.9456342668863265, | |
| "grad_norm": 1.2525478601455688, | |
| "learning_rate": 5.863192182410422e-06, | |
| "loss": 0.8114, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 2.9478308621636464, | |
| "grad_norm": 0.811509370803833, | |
| "learning_rate": 5.618892508143322e-06, | |
| "loss": 0.8491, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 2.9500274574409664, | |
| "grad_norm": 0.8506098985671997, | |
| "learning_rate": 5.374592833876222e-06, | |
| "loss": 0.9222, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 2.952224052718287, | |
| "grad_norm": 1.1636351346969604, | |
| "learning_rate": 5.13029315960912e-06, | |
| "loss": 1.0773, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 2.9544206479956068, | |
| "grad_norm": 1.531467318534851, | |
| "learning_rate": 4.8859934853420195e-06, | |
| "loss": 0.9243, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.9566172432729267, | |
| "grad_norm": 1.4680067300796509, | |
| "learning_rate": 4.641693811074918e-06, | |
| "loss": 0.7544, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 2.958813838550247, | |
| "grad_norm": 1.066221833229065, | |
| "learning_rate": 4.397394136807817e-06, | |
| "loss": 0.7252, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 2.961010433827567, | |
| "grad_norm": 1.2599577903747559, | |
| "learning_rate": 4.153094462540716e-06, | |
| "loss": 0.73, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 2.9632070291048875, | |
| "grad_norm": 1.5725163221359253, | |
| "learning_rate": 3.908794788273615e-06, | |
| "loss": 1.077, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 2.9654036243822075, | |
| "grad_norm": 1.251369595527649, | |
| "learning_rate": 3.664495114006514e-06, | |
| "loss": 0.762, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.967600219659528, | |
| "grad_norm": 1.0914812088012695, | |
| "learning_rate": 3.420195439739413e-06, | |
| "loss": 1.003, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 2.969796814936848, | |
| "grad_norm": 1.1189730167388916, | |
| "learning_rate": 3.175895765472312e-06, | |
| "loss": 0.798, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 2.971993410214168, | |
| "grad_norm": 0.9660980105400085, | |
| "learning_rate": 2.931596091205211e-06, | |
| "loss": 0.8536, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 2.9741900054914883, | |
| "grad_norm": 1.1298117637634277, | |
| "learning_rate": 2.687296416938111e-06, | |
| "loss": 1.0098, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 2.9763866007688082, | |
| "grad_norm": 0.9191001057624817, | |
| "learning_rate": 2.4429967426710097e-06, | |
| "loss": 0.7406, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 2.9785831960461286, | |
| "grad_norm": 2.6848766803741455, | |
| "learning_rate": 2.1986970684039087e-06, | |
| "loss": 0.8959, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 2.9807797913234486, | |
| "grad_norm": 1.144676685333252, | |
| "learning_rate": 1.9543973941368076e-06, | |
| "loss": 0.9057, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 2.982976386600769, | |
| "grad_norm": 1.20539128780365, | |
| "learning_rate": 1.7100977198697066e-06, | |
| "loss": 0.9064, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 2.985172981878089, | |
| "grad_norm": 0.8768987655639648, | |
| "learning_rate": 1.4657980456026055e-06, | |
| "loss": 1.0208, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 2.987369577155409, | |
| "grad_norm": 1.3136026859283447, | |
| "learning_rate": 1.2214983713355049e-06, | |
| "loss": 0.7389, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.9895661724327294, | |
| "grad_norm": 1.249769926071167, | |
| "learning_rate": 9.771986970684038e-07, | |
| "loss": 0.8536, | |
| "step": 1361 | |
| }, | |
| { | |
| "epoch": 2.9917627677100493, | |
| "grad_norm": 1.6826989650726318, | |
| "learning_rate": 7.328990228013028e-07, | |
| "loss": 1.0375, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 2.9939593629873693, | |
| "grad_norm": 1.329368233680725, | |
| "learning_rate": 4.885993485342019e-07, | |
| "loss": 0.8677, | |
| "step": 1363 | |
| }, | |
| { | |
| "epoch": 2.9961559582646897, | |
| "grad_norm": 1.3290849924087524, | |
| "learning_rate": 2.4429967426710095e-07, | |
| "loss": 0.928, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 2.99835255354201, | |
| "grad_norm": 0.9792918562889099, | |
| "learning_rate": 0.0, | |
| "loss": 0.9456, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 2.99835255354201, | |
| "step": 1365, | |
| "total_flos": 1.9088629453081805e+17, | |
| "train_loss": 1.0000693833434975, | |
| "train_runtime": 2797.1733, | |
| "train_samples_per_second": 7.811, | |
| "train_steps_per_second": 0.488 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1365, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 1.9088629453081805e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |