|
{ |
|
"best_metric": 0.7724684476852417, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.46893317702227433, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0023446658851113715, |
|
"grad_norm": 23.405542373657227, |
|
"learning_rate": 1e-05, |
|
"loss": 5.4807, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0023446658851113715, |
|
"eval_loss": 1.620888352394104, |
|
"eval_runtime": 52.3146, |
|
"eval_samples_per_second": 13.725, |
|
"eval_steps_per_second": 3.441, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.004689331770222743, |
|
"grad_norm": 30.190401077270508, |
|
"learning_rate": 2e-05, |
|
"loss": 5.8935, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.007033997655334115, |
|
"grad_norm": 26.080381393432617, |
|
"learning_rate": 3e-05, |
|
"loss": 5.84, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.009378663540445486, |
|
"grad_norm": 23.042579650878906, |
|
"learning_rate": 4e-05, |
|
"loss": 5.3222, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.011723329425556858, |
|
"grad_norm": 23.48046112060547, |
|
"learning_rate": 5e-05, |
|
"loss": 5.391, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01406799531066823, |
|
"grad_norm": 16.22538185119629, |
|
"learning_rate": 6e-05, |
|
"loss": 4.4877, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.016412661195779603, |
|
"grad_norm": 16.708181381225586, |
|
"learning_rate": 7e-05, |
|
"loss": 4.2436, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.01875732708089097, |
|
"grad_norm": 13.979721069335938, |
|
"learning_rate": 8e-05, |
|
"loss": 4.0923, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.021101992966002344, |
|
"grad_norm": 12.381362915039062, |
|
"learning_rate": 9e-05, |
|
"loss": 3.9154, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.023446658851113716, |
|
"grad_norm": 11.103169441223145, |
|
"learning_rate": 0.0001, |
|
"loss": 4.0311, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02579132473622509, |
|
"grad_norm": 11.883439064025879, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 3.9611, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.02813599062133646, |
|
"grad_norm": 10.65888500213623, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 3.92, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.03048065650644783, |
|
"grad_norm": 14.587371826171875, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 4.176, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.032825322391559206, |
|
"grad_norm": 9.647926330566406, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 3.3739, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.035169988276670575, |
|
"grad_norm": 12.973896980285645, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 3.777, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.03751465416178194, |
|
"grad_norm": 10.148499488830566, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 3.6683, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.03985932004689332, |
|
"grad_norm": 10.17501163482666, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 3.4995, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.04220398593200469, |
|
"grad_norm": 10.771862030029297, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 3.8145, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.044548651817116064, |
|
"grad_norm": 7.992787837982178, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 3.294, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.04689331770222743, |
|
"grad_norm": 8.624823570251465, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 3.1573, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0492379835873388, |
|
"grad_norm": 9.035672187805176, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 3.5599, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.05158264947245018, |
|
"grad_norm": 9.139631271362305, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 3.5685, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.053927315357561546, |
|
"grad_norm": 9.682453155517578, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 3.4256, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.05627198124267292, |
|
"grad_norm": 8.52701473236084, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 3.0577, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.05861664712778429, |
|
"grad_norm": 13.335333824157715, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 3.9149, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.06096131301289566, |
|
"grad_norm": 8.877936363220215, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 3.4293, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.06330597889800703, |
|
"grad_norm": 11.6506929397583, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 3.3192, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.06565064478311841, |
|
"grad_norm": 11.447749137878418, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 3.2359, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.06799531066822978, |
|
"grad_norm": 8.89906120300293, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 2.9944, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.07033997655334115, |
|
"grad_norm": 8.394125938415527, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 2.9144, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.07268464243845252, |
|
"grad_norm": 9.476370811462402, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 3.0997, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.07502930832356389, |
|
"grad_norm": 8.897541046142578, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 3.0782, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.07737397420867527, |
|
"grad_norm": 9.168450355529785, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 3.2234, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.07971864009378664, |
|
"grad_norm": 8.860148429870605, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 2.9236, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.08206330597889801, |
|
"grad_norm": 8.4860258102417, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 3.3617, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.08440797186400938, |
|
"grad_norm": 8.206006050109863, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 3.0589, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.08675263774912075, |
|
"grad_norm": 9.936936378479004, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 3.3066, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.08909730363423213, |
|
"grad_norm": 8.540318489074707, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 3.3055, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.0914419695193435, |
|
"grad_norm": 7.713901519775391, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 3.2, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.09378663540445487, |
|
"grad_norm": 7.591506004333496, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 3.2786, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.09613130128956623, |
|
"grad_norm": 7.995643615722656, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 3.1609, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.0984759671746776, |
|
"grad_norm": 7.518246173858643, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 2.7857, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.10082063305978899, |
|
"grad_norm": 9.68978214263916, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 2.7311, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.10316529894490035, |
|
"grad_norm": 11.168111801147461, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 3.0827, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.10550996483001172, |
|
"grad_norm": 9.186760902404785, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 2.5625, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.10785463071512309, |
|
"grad_norm": 8.938870429992676, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 2.5811, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.11019929660023446, |
|
"grad_norm": 9.78221321105957, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 2.5146, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.11254396248534584, |
|
"grad_norm": 9.555784225463867, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 2.4811, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.11488862837045721, |
|
"grad_norm": 9.54387092590332, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 2.1377, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.11723329425556858, |
|
"grad_norm": 11.764355659484863, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 2.9033, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.11723329425556858, |
|
"eval_loss": 1.0513248443603516, |
|
"eval_runtime": 53.327, |
|
"eval_samples_per_second": 13.464, |
|
"eval_steps_per_second": 3.375, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.11957796014067995, |
|
"grad_norm": 24.633373260498047, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 6.7308, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.12192262602579132, |
|
"grad_norm": 17.573421478271484, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 5.7586, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.1242672919109027, |
|
"grad_norm": 8.933952331542969, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 4.2271, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.12661195779601406, |
|
"grad_norm": 9.926009178161621, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 4.4827, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.12895662368112543, |
|
"grad_norm": 10.70360279083252, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 3.7757, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.13130128956623682, |
|
"grad_norm": 8.511175155639648, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 3.934, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.1336459554513482, |
|
"grad_norm": 7.956448078155518, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 3.9996, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.13599062133645956, |
|
"grad_norm": 9.405022621154785, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 3.8999, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.13833528722157093, |
|
"grad_norm": 7.946819305419922, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 3.9496, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.1406799531066823, |
|
"grad_norm": 8.405508041381836, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 3.579, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.14302461899179367, |
|
"grad_norm": 8.906330108642578, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 3.6143, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.14536928487690504, |
|
"grad_norm": 9.379605293273926, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 3.7798, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.1477139507620164, |
|
"grad_norm": 8.477218627929688, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 3.588, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.15005861664712777, |
|
"grad_norm": 7.908034801483154, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 3.589, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.15240328253223914, |
|
"grad_norm": 8.690335273742676, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 3.4561, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.15474794841735054, |
|
"grad_norm": 8.06505298614502, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 3.8101, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.1570926143024619, |
|
"grad_norm": 6.990350246429443, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 3.4838, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.15943728018757328, |
|
"grad_norm": 7.266732692718506, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 3.2637, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.16178194607268465, |
|
"grad_norm": 8.722139358520508, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 3.803, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.16412661195779601, |
|
"grad_norm": 6.49424409866333, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 3.2726, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.16647127784290738, |
|
"grad_norm": 8.221322059631348, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 4.0207, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.16881594372801875, |
|
"grad_norm": 7.128183841705322, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 3.0073, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.17116060961313012, |
|
"grad_norm": 7.4666619300842285, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 3.0837, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.1735052754982415, |
|
"grad_norm": 6.74752950668335, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 3.2209, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.17584994138335286, |
|
"grad_norm": 6.79481315612793, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 3.1971, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.17819460726846426, |
|
"grad_norm": 7.2510528564453125, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 3.0597, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.18053927315357562, |
|
"grad_norm": 7.681369781494141, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 2.7634, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.182883939038687, |
|
"grad_norm": 8.846420288085938, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 3.0827, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.18522860492379836, |
|
"grad_norm": 8.721245765686035, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 3.1422, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.18757327080890973, |
|
"grad_norm": 7.962874412536621, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 3.2284, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.1899179366940211, |
|
"grad_norm": 7.202088832855225, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 2.6309, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.19226260257913247, |
|
"grad_norm": 7.756674289703369, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 3.0028, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.19460726846424384, |
|
"grad_norm": 7.410325527191162, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 2.7875, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.1969519343493552, |
|
"grad_norm": 7.595869064331055, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 2.9473, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.19929660023446658, |
|
"grad_norm": 7.9515557289123535, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 3.1198, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.20164126611957797, |
|
"grad_norm": 8.794443130493164, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 3.2781, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.20398593200468934, |
|
"grad_norm": 9.458645820617676, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 3.2948, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.2063305978898007, |
|
"grad_norm": 7.8468146324157715, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 3.0067, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.20867526377491208, |
|
"grad_norm": 9.829216003417969, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 2.8751, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.21101992966002345, |
|
"grad_norm": 7.089529991149902, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 2.7765, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.21336459554513482, |
|
"grad_norm": 8.202539443969727, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 3.2057, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.21570926143024619, |
|
"grad_norm": 8.448012351989746, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 3.3169, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.21805392731535755, |
|
"grad_norm": 7.2261576652526855, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 2.8668, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.22039859320046892, |
|
"grad_norm": 7.020049095153809, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 2.8429, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.2227432590855803, |
|
"grad_norm": 7.415134906768799, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 2.5114, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.2250879249706917, |
|
"grad_norm": 7.972723484039307, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 2.8023, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.22743259085580306, |
|
"grad_norm": 8.125452995300293, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 2.9841, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.22977725674091443, |
|
"grad_norm": 6.471831798553467, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 2.3228, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.2321219226260258, |
|
"grad_norm": 6.832740783691406, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 2.2556, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.23446658851113716, |
|
"grad_norm": 7.426825523376465, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 2.4321, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.23446658851113716, |
|
"eval_loss": 0.9207891225814819, |
|
"eval_runtime": 53.3707, |
|
"eval_samples_per_second": 13.453, |
|
"eval_steps_per_second": 3.373, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.23681125439624853, |
|
"grad_norm": 13.413337707519531, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 5.202, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.2391559202813599, |
|
"grad_norm": 12.255110740661621, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 4.7768, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.24150058616647127, |
|
"grad_norm": 8.698344230651855, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 3.8756, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.24384525205158264, |
|
"grad_norm": 7.275907516479492, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 4.2124, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.246189917936694, |
|
"grad_norm": 5.887223720550537, |
|
"learning_rate": 5e-05, |
|
"loss": 3.471, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.2485345838218054, |
|
"grad_norm": 7.182607650756836, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 4.0454, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.2508792497069168, |
|
"grad_norm": 7.129477500915527, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 3.672, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.2532239155920281, |
|
"grad_norm": 6.107865810394287, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 3.3338, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.2555685814771395, |
|
"grad_norm": 6.434869289398193, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 3.6837, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.25791324736225085, |
|
"grad_norm": 6.784361839294434, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 3.4953, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.26025791324736225, |
|
"grad_norm": 6.471665859222412, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 3.4507, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.26260257913247365, |
|
"grad_norm": 6.709517955780029, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 3.5703, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.264947245017585, |
|
"grad_norm": 7.378337860107422, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 4.0464, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.2672919109026964, |
|
"grad_norm": 6.960837364196777, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 3.6552, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.2696365767878077, |
|
"grad_norm": 6.859847545623779, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 3.1426, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.2719812426729191, |
|
"grad_norm": 6.819009780883789, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 3.3986, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.27432590855803046, |
|
"grad_norm": 6.8491010665893555, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 3.2384, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.27667057444314186, |
|
"grad_norm": 6.187265872955322, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 3.1369, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.2790152403282532, |
|
"grad_norm": 6.026250839233398, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 3.2787, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.2813599062133646, |
|
"grad_norm": 6.356221675872803, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 3.0797, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.283704572098476, |
|
"grad_norm": 6.473607540130615, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 3.156, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.28604923798358733, |
|
"grad_norm": 7.452845096588135, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 3.6072, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.28839390386869873, |
|
"grad_norm": 6.59464168548584, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 3.3267, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.29073856975381007, |
|
"grad_norm": 6.750889301300049, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 2.996, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.29308323563892147, |
|
"grad_norm": 6.107902526855469, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 2.8911, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.2954279015240328, |
|
"grad_norm": 6.825888633728027, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 2.9248, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.2977725674091442, |
|
"grad_norm": 5.761615753173828, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 2.6024, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.30011723329425555, |
|
"grad_norm": 7.221162796020508, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 2.8481, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.30246189917936694, |
|
"grad_norm": 6.933948516845703, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 3.3723, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.3048065650644783, |
|
"grad_norm": 6.539292812347412, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 2.8156, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.3071512309495897, |
|
"grad_norm": 6.422267913818359, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 3.1214, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.3094958968347011, |
|
"grad_norm": 6.599564075469971, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 3.1743, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.3118405627198124, |
|
"grad_norm": 6.437052249908447, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 2.6728, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.3141852286049238, |
|
"grad_norm": 6.3518500328063965, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 2.4424, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.31652989449003516, |
|
"grad_norm": 7.512455463409424, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 2.9071, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.31887456037514655, |
|
"grad_norm": 6.249546527862549, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 2.7378, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.3212192262602579, |
|
"grad_norm": 6.139853000640869, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 2.8643, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.3235638921453693, |
|
"grad_norm": 6.650813102722168, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 2.7508, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.32590855803048063, |
|
"grad_norm": 7.451976776123047, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 3.0694, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.32825322391559203, |
|
"grad_norm": 7.3050456047058105, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 2.8099, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.3305978898007034, |
|
"grad_norm": 7.736148834228516, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 3.2716, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.33294255568581477, |
|
"grad_norm": 7.5445756912231445, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 2.7682, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.33528722157092616, |
|
"grad_norm": 8.870253562927246, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 2.8229, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.3376318874560375, |
|
"grad_norm": 7.864132404327393, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 2.6235, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.3399765533411489, |
|
"grad_norm": 6.567162990570068, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 2.3168, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.34232121922626024, |
|
"grad_norm": 8.08400821685791, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 2.8536, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.34466588511137164, |
|
"grad_norm": 6.88629674911499, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 2.2925, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.347010550996483, |
|
"grad_norm": 6.940112113952637, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 2.3887, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.3493552168815944, |
|
"grad_norm": 7.423598766326904, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 2.4429, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.3516998827667057, |
|
"grad_norm": 8.786375045776367, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 2.5743, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.3516998827667057, |
|
"eval_loss": 0.8272587656974792, |
|
"eval_runtime": 53.3907, |
|
"eval_samples_per_second": 13.448, |
|
"eval_steps_per_second": 3.371, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.3540445486518171, |
|
"grad_norm": 12.490707397460938, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 4.7622, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.3563892145369285, |
|
"grad_norm": 10.970746994018555, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 4.2459, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.35873388042203985, |
|
"grad_norm": 10.433035850524902, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 3.9727, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.36107854630715125, |
|
"grad_norm": 8.777443885803223, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 3.6843, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.3634232121922626, |
|
"grad_norm": 7.379918098449707, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 3.5217, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.365767878077374, |
|
"grad_norm": 6.822660446166992, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 3.7399, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.36811254396248533, |
|
"grad_norm": 6.766895771026611, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 3.4668, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.3704572098475967, |
|
"grad_norm": 6.219024658203125, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 3.325, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.37280187573270807, |
|
"grad_norm": 6.84614896774292, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 3.4695, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.37514654161781946, |
|
"grad_norm": 6.653427600860596, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 3.2945, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.37749120750293086, |
|
"grad_norm": 6.658251762390137, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 3.8596, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.3798358733880422, |
|
"grad_norm": 7.170784950256348, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 3.6789, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.3821805392731536, |
|
"grad_norm": 6.30677604675293, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 3.1439, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.38452520515826494, |
|
"grad_norm": 7.19639778137207, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 3.4967, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.38686987104337633, |
|
"grad_norm": 6.956872463226318, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 3.1998, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.3892145369284877, |
|
"grad_norm": 6.73064661026001, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 3.3462, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.39155920281359907, |
|
"grad_norm": 6.21381950378418, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 3.306, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.3939038686987104, |
|
"grad_norm": 6.727031230926514, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 3.1663, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.3962485345838218, |
|
"grad_norm": 7.383633136749268, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 3.4129, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.39859320046893315, |
|
"grad_norm": 6.748716354370117, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 3.3349, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.40093786635404455, |
|
"grad_norm": 6.933330535888672, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 3.4379, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.40328253223915594, |
|
"grad_norm": 5.849376678466797, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 3.0163, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.4056271981242673, |
|
"grad_norm": 6.345935344696045, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 2.971, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.4079718640093787, |
|
"grad_norm": 7.133333206176758, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 3.1947, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.41031652989449, |
|
"grad_norm": 6.377161502838135, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 2.9808, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.4126611957796014, |
|
"grad_norm": 6.192864418029785, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 2.9739, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.41500586166471276, |
|
"grad_norm": 5.448167324066162, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 2.485, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.41735052754982416, |
|
"grad_norm": 5.94972562789917, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 2.6537, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.4196951934349355, |
|
"grad_norm": 6.432535171508789, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 2.7788, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.4220398593200469, |
|
"grad_norm": 6.361620903015137, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 2.8393, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.4243845252051583, |
|
"grad_norm": 6.261866569519043, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 3.1024, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.42672919109026963, |
|
"grad_norm": 6.204875469207764, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 2.5939, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.42907385697538103, |
|
"grad_norm": 5.875705242156982, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 2.6644, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.43141852286049237, |
|
"grad_norm": 6.217267990112305, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 3.0126, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.43376318874560377, |
|
"grad_norm": 6.96459436416626, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 2.7384, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.4361078546307151, |
|
"grad_norm": 6.288431644439697, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 2.7716, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.4384525205158265, |
|
"grad_norm": 6.644599914550781, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 2.5964, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.44079718640093785, |
|
"grad_norm": 5.972239971160889, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 2.5109, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.44314185228604924, |
|
"grad_norm": 6.647230625152588, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 2.9682, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.4454865181711606, |
|
"grad_norm": 6.839677333831787, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 3.084, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.447831184056272, |
|
"grad_norm": 7.2198662757873535, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 2.701, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.4501758499413834, |
|
"grad_norm": 7.756752014160156, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 3.0758, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.4525205158264947, |
|
"grad_norm": 7.570158958435059, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 2.9631, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.4548651817116061, |
|
"grad_norm": 5.9812116622924805, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 2.0871, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.45720984759671746, |
|
"grad_norm": 6.631890296936035, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 2.1681, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.45955451348182885, |
|
"grad_norm": 7.978712558746338, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 2.5791, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.4618991793669402, |
|
"grad_norm": 6.869262218475342, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 2.8627, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.4642438452520516, |
|
"grad_norm": 8.06986141204834, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 2.6656, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.46658851113716293, |
|
"grad_norm": 8.685197830200195, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 2.4931, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.46893317702227433, |
|
"grad_norm": 9.024889945983887, |
|
"learning_rate": 0.0, |
|
"loss": 2.4721, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.46893317702227433, |
|
"eval_loss": 0.7724684476852417, |
|
"eval_runtime": 53.3719, |
|
"eval_samples_per_second": 13.453, |
|
"eval_steps_per_second": 3.373, |
|
"step": 200 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 200, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 50, |
|
"stateful_callbacks": { |
|
"EarlyStoppingCallback": { |
|
"args": { |
|
"early_stopping_patience": 5, |
|
"early_stopping_threshold": 0.0 |
|
}, |
|
"attributes": { |
|
"early_stopping_patience_counter": 0 |
|
} |
|
}, |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 2.876309235396772e+17, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|