|
{ |
|
"best_metric": 1.3934108018875122, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.045493318168893945, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0002274665908444697, |
|
"grad_norm": 34.041351318359375, |
|
"learning_rate": 1e-05, |
|
"loss": 8.9316, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0002274665908444697, |
|
"eval_loss": 2.5400564670562744, |
|
"eval_runtime": 568.3619, |
|
"eval_samples_per_second": 13.027, |
|
"eval_steps_per_second": 3.257, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0004549331816889394, |
|
"grad_norm": 44.586666107177734, |
|
"learning_rate": 2e-05, |
|
"loss": 10.7919, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.0006823997725334092, |
|
"grad_norm": 32.4596061706543, |
|
"learning_rate": 3e-05, |
|
"loss": 7.4133, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.0009098663633778788, |
|
"grad_norm": 23.579296112060547, |
|
"learning_rate": 4e-05, |
|
"loss": 6.9578, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.0011373329542223485, |
|
"grad_norm": 21.836469650268555, |
|
"learning_rate": 5e-05, |
|
"loss": 5.6818, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.0013647995450668184, |
|
"grad_norm": 19.26651954650879, |
|
"learning_rate": 6e-05, |
|
"loss": 5.5704, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.001592266135911288, |
|
"grad_norm": 20.165502548217773, |
|
"learning_rate": 7e-05, |
|
"loss": 5.208, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.0018197327267557577, |
|
"grad_norm": 18.88631820678711, |
|
"learning_rate": 8e-05, |
|
"loss": 5.2784, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.0020471993176002273, |
|
"grad_norm": 18.446062088012695, |
|
"learning_rate": 9e-05, |
|
"loss": 5.0863, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.002274665908444697, |
|
"grad_norm": 18.037721633911133, |
|
"learning_rate": 0.0001, |
|
"loss": 4.797, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.002502132499289167, |
|
"grad_norm": 14.005064010620117, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 4.8758, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.0027295990901336368, |
|
"grad_norm": 14.213648796081543, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 5.0906, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.0029570656809781064, |
|
"grad_norm": 14.11625862121582, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 5.1718, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.003184532271822576, |
|
"grad_norm": 15.320588111877441, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 4.8488, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.0034119988626670457, |
|
"grad_norm": 14.174674987792969, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 4.8103, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.0036394654535115154, |
|
"grad_norm": 15.885099411010742, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 4.6694, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.003866932044355985, |
|
"grad_norm": 14.515787124633789, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 4.6005, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.004094398635200455, |
|
"grad_norm": 18.154613494873047, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 5.0799, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.004321865226044924, |
|
"grad_norm": 17.530113220214844, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 5.9696, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.004549331816889394, |
|
"grad_norm": 18.2506160736084, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 5.6355, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.004776798407733864, |
|
"grad_norm": 15.541585922241211, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 5.4084, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.005004264998578334, |
|
"grad_norm": 15.226705551147461, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 5.8269, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.005231731589422804, |
|
"grad_norm": 14.854185104370117, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 5.6565, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.0054591981802672735, |
|
"grad_norm": 15.64132308959961, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 6.2298, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.005686664771111743, |
|
"grad_norm": 14.779892921447754, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 5.6871, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.005914131361956213, |
|
"grad_norm": 14.125879287719727, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 5.8338, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.0061415979528006825, |
|
"grad_norm": 18.950382232666016, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 5.8387, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.006369064543645152, |
|
"grad_norm": 16.46758270263672, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 6.8297, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.006596531134489622, |
|
"grad_norm": 17.215564727783203, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 7.088, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.0068239977253340914, |
|
"grad_norm": 14.567110061645508, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 6.1848, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.007051464316178561, |
|
"grad_norm": 17.58001708984375, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 7.4265, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.007278930907023031, |
|
"grad_norm": 14.948777198791504, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 7.1257, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.0075063974978675, |
|
"grad_norm": 15.770895004272461, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 7.5041, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.00773386408871197, |
|
"grad_norm": 16.993030548095703, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 8.2977, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.00796133067955644, |
|
"grad_norm": 18.309011459350586, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 8.6731, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.00818879727040091, |
|
"grad_norm": 16.03449249267578, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 8.7035, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.008416263861245379, |
|
"grad_norm": 17.210514068603516, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 7.5915, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.008643730452089849, |
|
"grad_norm": 17.723649978637695, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 7.882, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.008871197042934318, |
|
"grad_norm": 15.294744491577148, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 8.3729, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.009098663633778788, |
|
"grad_norm": 16.79600715637207, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 9.3703, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.009326130224623258, |
|
"grad_norm": 17.177452087402344, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 8.1529, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.009553596815467727, |
|
"grad_norm": 18.984376907348633, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 9.1076, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.009781063406312199, |
|
"grad_norm": 17.823272705078125, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 8.743, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.010008529997156668, |
|
"grad_norm": 22.286880493164062, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 9.4842, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.010235996588001138, |
|
"grad_norm": 18.42691993713379, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 7.6027, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.010463463178845608, |
|
"grad_norm": 20.371795654296875, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 9.2501, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.010690929769690077, |
|
"grad_norm": 21.78159523010254, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 9.0573, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.010918396360534547, |
|
"grad_norm": 26.65410804748535, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 10.2407, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.011145862951379017, |
|
"grad_norm": 31.913434982299805, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 10.8729, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.011373329542223486, |
|
"grad_norm": 26.928504943847656, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 10.6478, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.011373329542223486, |
|
"eval_loss": 1.745985507965088, |
|
"eval_runtime": 572.8926, |
|
"eval_samples_per_second": 12.924, |
|
"eval_steps_per_second": 3.231, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.011600796133067956, |
|
"grad_norm": 19.977048873901367, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 6.9754, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.011828262723912426, |
|
"grad_norm": 22.3455867767334, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 7.0525, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.012055729314756895, |
|
"grad_norm": 21.204179763793945, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 6.0618, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.012283195905601365, |
|
"grad_norm": 17.540246963500977, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 5.0722, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.012510662496445835, |
|
"grad_norm": 11.772398948669434, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 4.6221, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.012738129087290304, |
|
"grad_norm": 14.863117218017578, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 5.0881, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.012965595678134774, |
|
"grad_norm": 14.302623748779297, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 4.8498, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.013193062268979244, |
|
"grad_norm": 12.283204078674316, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 4.1844, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.013420528859823713, |
|
"grad_norm": 12.973774909973145, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 4.5245, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.013647995450668183, |
|
"grad_norm": 11.702144622802734, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 4.0904, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.013875462041512653, |
|
"grad_norm": 12.34423542022705, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 3.8874, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.014102928632357122, |
|
"grad_norm": 10.625223159790039, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 3.7295, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.014330395223201592, |
|
"grad_norm": 9.918624877929688, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 3.4615, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.014557861814046062, |
|
"grad_norm": 9.898490905761719, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 3.6777, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.014785328404890531, |
|
"grad_norm": 10.252726554870605, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 3.9065, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.015012794995735, |
|
"grad_norm": 13.20054817199707, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 4.5182, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.01524026158657947, |
|
"grad_norm": 12.68508529663086, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 4.7883, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.01546772817742394, |
|
"grad_norm": 13.369020462036133, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 5.698, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.01569519476826841, |
|
"grad_norm": 12.121949195861816, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 4.6271, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.01592266135911288, |
|
"grad_norm": 14.011739730834961, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 6.1048, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.01615012794995735, |
|
"grad_norm": 11.493091583251953, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 4.8471, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.01637759454080182, |
|
"grad_norm": 11.066706657409668, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 5.2317, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.01660506113164629, |
|
"grad_norm": 12.930093765258789, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 4.6176, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.016832527722490758, |
|
"grad_norm": 12.51268482208252, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 6.0917, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.017059994313335228, |
|
"grad_norm": 14.068593978881836, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 6.0103, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.017287460904179697, |
|
"grad_norm": 14.852651596069336, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 7.102, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.017514927495024167, |
|
"grad_norm": 13.94018840789795, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 6.6789, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.017742394085868637, |
|
"grad_norm": 13.79462718963623, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 6.241, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.017969860676713106, |
|
"grad_norm": 15.39090347290039, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 6.6139, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.018197327267557576, |
|
"grad_norm": 13.380969047546387, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 6.5099, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.018424793858402046, |
|
"grad_norm": 14.356781005859375, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 6.6345, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.018652260449246515, |
|
"grad_norm": 14.944310188293457, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 7.1505, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.018879727040090985, |
|
"grad_norm": 16.056324005126953, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 7.0958, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.019107193630935455, |
|
"grad_norm": 12.605941772460938, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 6.4196, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.019334660221779928, |
|
"grad_norm": 12.993471145629883, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 6.6034, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.019562126812624397, |
|
"grad_norm": 14.480472564697266, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 8.1206, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.019789593403468867, |
|
"grad_norm": 13.966653823852539, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 7.2278, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.020017059994313337, |
|
"grad_norm": 15.118941307067871, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 7.9223, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.020244526585157806, |
|
"grad_norm": 13.609482765197754, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 8.2732, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.020471993176002276, |
|
"grad_norm": 14.514567375183105, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 7.9276, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.020699459766846746, |
|
"grad_norm": 13.03682804107666, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 8.0556, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.020926926357691215, |
|
"grad_norm": 16.205564498901367, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 8.0025, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.021154392948535685, |
|
"grad_norm": 14.616352081298828, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 8.5207, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.021381859539380155, |
|
"grad_norm": 14.196993827819824, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 8.028, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.021609326130224624, |
|
"grad_norm": 13.343052864074707, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 7.2808, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.021836792721069094, |
|
"grad_norm": 13.771421432495117, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 6.7428, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.022064259311913564, |
|
"grad_norm": 18.806198120117188, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 8.0513, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.022291725902758033, |
|
"grad_norm": 18.11672592163086, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 7.6447, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.022519192493602503, |
|
"grad_norm": 23.170137405395508, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 8.4192, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.022746659084446973, |
|
"grad_norm": 33.44895935058594, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 11.1173, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.022746659084446973, |
|
"eval_loss": 1.5323618650436401, |
|
"eval_runtime": 571.5375, |
|
"eval_samples_per_second": 12.955, |
|
"eval_steps_per_second": 3.239, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.022974125675291442, |
|
"grad_norm": 14.957679748535156, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 5.8546, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.023201592266135912, |
|
"grad_norm": 19.272850036621094, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 6.6302, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.02342905885698038, |
|
"grad_norm": 19.960105895996094, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 5.8994, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.02365652544782485, |
|
"grad_norm": 13.906604766845703, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 4.9998, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.02388399203866932, |
|
"grad_norm": 11.011517524719238, |
|
"learning_rate": 5e-05, |
|
"loss": 4.657, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.02411145862951379, |
|
"grad_norm": 12.705352783203125, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 4.7983, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.02433892522035826, |
|
"grad_norm": 11.65358829498291, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 4.7326, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.02456639181120273, |
|
"grad_norm": 9.971208572387695, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 3.8575, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.0247938584020472, |
|
"grad_norm": 9.415610313415527, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 3.45, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.02502132499289167, |
|
"grad_norm": 12.18667984008789, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 4.3955, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.02524879158373614, |
|
"grad_norm": 9.86085319519043, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 3.5571, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.02547625817458061, |
|
"grad_norm": 10.178373336791992, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 3.8681, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.025703724765425078, |
|
"grad_norm": 10.37630558013916, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 3.7967, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.025931191356269548, |
|
"grad_norm": 9.967632293701172, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 3.7198, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.026158657947114018, |
|
"grad_norm": 10.731016159057617, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 4.0524, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.026386124537958487, |
|
"grad_norm": 10.638973236083984, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 4.0039, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.026613591128802957, |
|
"grad_norm": 10.42179012298584, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 3.6658, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.026841057719647426, |
|
"grad_norm": 10.291654586791992, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 3.8606, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.027068524310491896, |
|
"grad_norm": 11.214519500732422, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 4.3043, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.027295990901336366, |
|
"grad_norm": 13.954524040222168, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 4.7875, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.027523457492180835, |
|
"grad_norm": 12.026573181152344, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 4.7223, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.027750924083025305, |
|
"grad_norm": 11.638748168945312, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 4.7275, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.027978390673869775, |
|
"grad_norm": 12.984579086303711, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 4.4276, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.028205857264714244, |
|
"grad_norm": 11.86833667755127, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 4.9858, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.028433323855558714, |
|
"grad_norm": 12.810986518859863, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 6.52, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.028660790446403184, |
|
"grad_norm": 11.11644172668457, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 5.1678, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.028888257037247653, |
|
"grad_norm": 13.12958812713623, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 6.5639, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.029115723628092123, |
|
"grad_norm": 12.201437950134277, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 6.0811, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.029343190218936593, |
|
"grad_norm": 12.989922523498535, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 5.3872, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.029570656809781062, |
|
"grad_norm": 13.592166900634766, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 6.0792, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.029798123400625532, |
|
"grad_norm": 12.639175415039062, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 6.5032, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.03002558999147, |
|
"grad_norm": 12.010916709899902, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 6.4719, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.03025305658231447, |
|
"grad_norm": 11.63107967376709, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 6.0408, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.03048052317315894, |
|
"grad_norm": 13.046253204345703, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 6.1969, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.03070798976400341, |
|
"grad_norm": 14.268304824829102, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 6.7242, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.03093545635484788, |
|
"grad_norm": 13.387679100036621, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 6.7124, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.03116292294569235, |
|
"grad_norm": 12.30424976348877, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 6.2506, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.03139038953653682, |
|
"grad_norm": 13.062273025512695, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 7.0967, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.03161785612738129, |
|
"grad_norm": 14.494675636291504, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 6.9732, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.03184532271822576, |
|
"grad_norm": 15.215642929077148, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 6.973, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.03207278930907023, |
|
"grad_norm": 15.166794776916504, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 7.6123, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.0323002558999147, |
|
"grad_norm": 17.052576065063477, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 8.5165, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.03252772249075917, |
|
"grad_norm": 14.346969604492188, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 7.4776, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.03275518908160364, |
|
"grad_norm": 20.73611831665039, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 7.3791, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.03298265567244811, |
|
"grad_norm": 29.446413040161133, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 5.9631, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.03321012226329258, |
|
"grad_norm": 14.22492504119873, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 7.5374, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.03343758885413705, |
|
"grad_norm": 17.104326248168945, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 8.2033, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.033665055444981516, |
|
"grad_norm": 17.587560653686523, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 7.3793, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.03389252203582599, |
|
"grad_norm": 16.36390495300293, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 7.5105, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.034119988626670456, |
|
"grad_norm": 21.527830123901367, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 7.9748, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.034119988626670456, |
|
"eval_loss": 1.422861933708191, |
|
"eval_runtime": 572.3031, |
|
"eval_samples_per_second": 12.937, |
|
"eval_steps_per_second": 3.234, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.03434745521751493, |
|
"grad_norm": 13.004639625549316, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 4.9985, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.034574921808359395, |
|
"grad_norm": 17.859960556030273, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 5.6094, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.03480238839920387, |
|
"grad_norm": 13.085841178894043, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 4.4637, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.035029854990048334, |
|
"grad_norm": 11.08560562133789, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 3.7434, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.03525732158089281, |
|
"grad_norm": 12.804967880249023, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 4.4985, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.03548478817173727, |
|
"grad_norm": 10.449297904968262, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 3.9085, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.03571225476258175, |
|
"grad_norm": 11.833451271057129, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 3.3967, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.03593972135342621, |
|
"grad_norm": 9.631769180297852, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 3.4055, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.036167187944270686, |
|
"grad_norm": 9.996049880981445, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 3.4965, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.03639465453511515, |
|
"grad_norm": 10.321172714233398, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 3.6611, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.036622121125959625, |
|
"grad_norm": 8.969747543334961, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 3.4414, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.03684958771680409, |
|
"grad_norm": 9.434122085571289, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 3.912, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.037077054307648565, |
|
"grad_norm": 9.716062545776367, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 3.8831, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.03730452089849303, |
|
"grad_norm": 9.628499031066895, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 3.5424, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.037531987489337504, |
|
"grad_norm": 10.095179557800293, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 3.8517, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.03775945408018197, |
|
"grad_norm": 9.444663047790527, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 4.1422, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.03798692067102644, |
|
"grad_norm": 8.723136901855469, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 3.6762, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.03821438726187091, |
|
"grad_norm": 10.722376823425293, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 3.6847, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.03844185385271538, |
|
"grad_norm": 10.953653335571289, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 4.2501, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.038669320443559856, |
|
"grad_norm": 10.628883361816406, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 4.3282, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.03889678703440432, |
|
"grad_norm": 10.546460151672363, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 4.8413, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.039124253625248795, |
|
"grad_norm": 11.666546821594238, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 4.876, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.03935172021609326, |
|
"grad_norm": 12.602836608886719, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 5.9223, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.039579186806937734, |
|
"grad_norm": 13.231545448303223, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 5.8793, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.0398066533977822, |
|
"grad_norm": 10.380090713500977, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 4.4184, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.040034119988626674, |
|
"grad_norm": 11.95571517944336, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 5.6347, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.04026158657947114, |
|
"grad_norm": 12.159829139709473, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 6.0605, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.04048905317031561, |
|
"grad_norm": 11.637267112731934, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 5.5362, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.04071651976116008, |
|
"grad_norm": 12.93325138092041, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 6.2703, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.04094398635200455, |
|
"grad_norm": 11.345155715942383, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 5.8386, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.04117145294284902, |
|
"grad_norm": 13.935602188110352, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 6.5126, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.04139891953369349, |
|
"grad_norm": 15.000650405883789, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 7.1408, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.04162638612453796, |
|
"grad_norm": 12.500697135925293, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 6.5353, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.04185385271538243, |
|
"grad_norm": 13.510952949523926, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 6.9013, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.0420813193062269, |
|
"grad_norm": 12.489456176757812, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 6.4148, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.04230878589707137, |
|
"grad_norm": 13.093245506286621, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 6.2208, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.042536252487915836, |
|
"grad_norm": 14.346376419067383, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 7.715, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.04276371907876031, |
|
"grad_norm": 15.307150840759277, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 6.9647, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.042991185669604776, |
|
"grad_norm": 13.886529922485352, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 6.8288, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.04321865226044925, |
|
"grad_norm": 15.29570198059082, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 7.7172, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.043446118851293715, |
|
"grad_norm": 14.093865394592285, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 7.1941, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.04367358544213819, |
|
"grad_norm": 14.034331321716309, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 7.332, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.043901052032982654, |
|
"grad_norm": 14.35676097869873, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 7.7477, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.04412851862382713, |
|
"grad_norm": 21.333402633666992, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 5.9969, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.044355985214671594, |
|
"grad_norm": 11.25492000579834, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 4.8391, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.04458345180551607, |
|
"grad_norm": 16.54047203063965, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 8.5932, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.04481091839636053, |
|
"grad_norm": 14.195585250854492, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 6.2815, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.045038384987205006, |
|
"grad_norm": 17.41058349609375, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 7.3235, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.04526585157804947, |
|
"grad_norm": 15.436453819274902, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 7.2891, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.045493318168893945, |
|
"grad_norm": 23.008148193359375, |
|
"learning_rate": 0.0, |
|
"loss": 9.4996, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.045493318168893945, |
|
"eval_loss": 1.3934108018875122, |
|
"eval_runtime": 572.5451, |
|
"eval_samples_per_second": 12.932, |
|
"eval_steps_per_second": 3.233, |
|
"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": 2.995657493352284e+17, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|