| { | |
| "best_metric": 0.48696738481521606, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.3401360544217687, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0017006802721088435, | |
| "grad_norm": 0.16959799826145172, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5694, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0017006802721088435, | |
| "eval_loss": 0.7016671299934387, | |
| "eval_runtime": 71.4118, | |
| "eval_samples_per_second": 13.863, | |
| "eval_steps_per_second": 3.473, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003401360544217687, | |
| "grad_norm": 0.18853634595870972, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6119, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00510204081632653, | |
| "grad_norm": 0.18308532238006592, | |
| "learning_rate": 3e-05, | |
| "loss": 0.6166, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.006802721088435374, | |
| "grad_norm": 0.21027779579162598, | |
| "learning_rate": 4e-05, | |
| "loss": 0.6103, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.008503401360544218, | |
| "grad_norm": 0.19290000200271606, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6287, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01020408163265306, | |
| "grad_norm": 0.18025891482830048, | |
| "learning_rate": 6e-05, | |
| "loss": 0.6208, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.011904761904761904, | |
| "grad_norm": 0.18125878274440765, | |
| "learning_rate": 7e-05, | |
| "loss": 0.6344, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.013605442176870748, | |
| "grad_norm": 0.17453987896442413, | |
| "learning_rate": 8e-05, | |
| "loss": 0.6131, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.015306122448979591, | |
| "grad_norm": 0.22241806983947754, | |
| "learning_rate": 9e-05, | |
| "loss": 0.6083, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.017006802721088437, | |
| "grad_norm": 0.22235193848609924, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6033, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01870748299319728, | |
| "grad_norm": 0.22243435680866241, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.6101, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02040816326530612, | |
| "grad_norm": 0.212838277220726, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.6282, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.022108843537414966, | |
| "grad_norm": 0.18950781226158142, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.587, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.023809523809523808, | |
| "grad_norm": 0.18078520894050598, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.566, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.025510204081632654, | |
| "grad_norm": 0.19135352969169617, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.606, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.027210884353741496, | |
| "grad_norm": 0.17477868497371674, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.5983, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02891156462585034, | |
| "grad_norm": 0.17936213314533234, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.5803, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.030612244897959183, | |
| "grad_norm": 0.171107217669487, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.6004, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03231292517006803, | |
| "grad_norm": 0.17990447580814362, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.5798, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.034013605442176874, | |
| "grad_norm": 0.17869217693805695, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.556, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03571428571428571, | |
| "grad_norm": 0.18821021914482117, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.579, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.03741496598639456, | |
| "grad_norm": 0.1777324676513672, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.5772, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0391156462585034, | |
| "grad_norm": 0.17420151829719543, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.5682, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04081632653061224, | |
| "grad_norm": 0.1808144897222519, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.5921, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04251700680272109, | |
| "grad_norm": 0.18310721218585968, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.5789, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04421768707482993, | |
| "grad_norm": 0.18186233937740326, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.5858, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.04591836734693878, | |
| "grad_norm": 0.1901922971010208, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.5872, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.047619047619047616, | |
| "grad_norm": 0.18851330876350403, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.609, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.04931972789115646, | |
| "grad_norm": 0.17681556940078735, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.5502, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05102040816326531, | |
| "grad_norm": 0.19169022142887115, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.5866, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05272108843537415, | |
| "grad_norm": 0.1974894106388092, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.5671, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.05442176870748299, | |
| "grad_norm": 0.19867075979709625, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.5513, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.05612244897959184, | |
| "grad_norm": 0.212591752409935, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.5431, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.05782312925170068, | |
| "grad_norm": 0.20537696778774261, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.5254, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.05952380952380952, | |
| "grad_norm": 0.19645805656909943, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.54, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.061224489795918366, | |
| "grad_norm": 0.21156303584575653, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.5601, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.06292517006802721, | |
| "grad_norm": 0.21462957561016083, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.5519, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.06462585034013606, | |
| "grad_norm": 0.2154107540845871, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.5123, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0663265306122449, | |
| "grad_norm": 0.2208368331193924, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.5221, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.06802721088435375, | |
| "grad_norm": 0.2187241166830063, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.5106, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06972789115646258, | |
| "grad_norm": 0.22782184183597565, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.5342, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07142857142857142, | |
| "grad_norm": 0.23117218911647797, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5013, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.07312925170068027, | |
| "grad_norm": 0.25789204239845276, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5135, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.07482993197278912, | |
| "grad_norm": 0.27129191160202026, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.5327, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.07653061224489796, | |
| "grad_norm": 0.27148500084877014, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.4719, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0782312925170068, | |
| "grad_norm": 0.2911697030067444, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.4779, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.07993197278911565, | |
| "grad_norm": 0.318223774433136, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.472, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.08163265306122448, | |
| "grad_norm": 0.36965641379356384, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.506, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.08333333333333333, | |
| "grad_norm": 0.5182129144668579, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.5451, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.08503401360544217, | |
| "grad_norm": 0.9564126133918762, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.4366, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08503401360544217, | |
| "eval_loss": 0.5648638010025024, | |
| "eval_runtime": 72.6373, | |
| "eval_samples_per_second": 13.629, | |
| "eval_steps_per_second": 3.414, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08673469387755102, | |
| "grad_norm": 0.22063733637332916, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.505, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.08843537414965986, | |
| "grad_norm": 0.2718883454799652, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.5658, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.09013605442176871, | |
| "grad_norm": 0.23576916754245758, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.5559, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.09183673469387756, | |
| "grad_norm": 0.2260821908712387, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.5725, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0935374149659864, | |
| "grad_norm": 0.2174433320760727, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.5552, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.09523809523809523, | |
| "grad_norm": 0.18729886412620544, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.5591, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.09693877551020408, | |
| "grad_norm": 0.18305058777332306, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.5186, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.09863945578231292, | |
| "grad_norm": 0.18157657980918884, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.5212, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.10034013605442177, | |
| "grad_norm": 0.2037811428308487, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.5548, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 0.19771002233028412, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.5718, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.10374149659863946, | |
| "grad_norm": 0.1701708436012268, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.5323, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1054421768707483, | |
| "grad_norm": 0.16258856654167175, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.5529, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.10714285714285714, | |
| "grad_norm": 0.15664634108543396, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.5361, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.10884353741496598, | |
| "grad_norm": 0.1598876416683197, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.5361, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.11054421768707483, | |
| "grad_norm": 0.15704718232154846, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.5496, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.11224489795918367, | |
| "grad_norm": 0.15310242772102356, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.5342, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.11394557823129252, | |
| "grad_norm": 0.16328680515289307, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.5541, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.11564625850340136, | |
| "grad_norm": 0.1636633723974228, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.5231, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.11734693877551021, | |
| "grad_norm": 0.17201963067054749, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.5296, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.11904761904761904, | |
| "grad_norm": 0.17560014128684998, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.5538, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12074829931972789, | |
| "grad_norm": 0.17567718029022217, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.525, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.12244897959183673, | |
| "grad_norm": 0.17158040404319763, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.5249, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.12414965986394558, | |
| "grad_norm": 0.17191287875175476, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.5435, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.12585034013605442, | |
| "grad_norm": 0.1712241917848587, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.5402, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.12755102040816327, | |
| "grad_norm": 0.17726421356201172, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.5372, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1292517006802721, | |
| "grad_norm": 0.17581725120544434, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.5363, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.13095238095238096, | |
| "grad_norm": 0.17728029191493988, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.5372, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1326530612244898, | |
| "grad_norm": 0.17805948853492737, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.4983, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.13435374149659865, | |
| "grad_norm": 0.18020778894424438, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.5372, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1360544217687075, | |
| "grad_norm": 0.1803048700094223, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.5289, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1377551020408163, | |
| "grad_norm": 0.18663981556892395, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.5464, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.13945578231292516, | |
| "grad_norm": 0.18048711121082306, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.5241, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.141156462585034, | |
| "grad_norm": 0.1833612471818924, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.5321, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.14285714285714285, | |
| "grad_norm": 0.18568553030490875, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.5172, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1445578231292517, | |
| "grad_norm": 0.18531528115272522, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.4875, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.14625850340136054, | |
| "grad_norm": 0.19250451028347015, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.5063, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.14795918367346939, | |
| "grad_norm": 0.20437608659267426, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.5094, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.14965986394557823, | |
| "grad_norm": 0.2079511433839798, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.5031, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.15136054421768708, | |
| "grad_norm": 0.2023497223854065, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.4824, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.15306122448979592, | |
| "grad_norm": 0.2186407893896103, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.4727, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15476190476190477, | |
| "grad_norm": 0.2154996246099472, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.4293, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1564625850340136, | |
| "grad_norm": 0.22082054615020752, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.4432, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.15816326530612246, | |
| "grad_norm": 0.3104734420776367, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.4886, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.1598639455782313, | |
| "grad_norm": 0.2640933692455292, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.4422, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.16156462585034015, | |
| "grad_norm": 0.2691980004310608, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.484, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.16326530612244897, | |
| "grad_norm": 0.29049983620643616, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.4608, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.1649659863945578, | |
| "grad_norm": 0.32303106784820557, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.4776, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.38841065764427185, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.4841, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1683673469387755, | |
| "grad_norm": 0.591713011264801, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.4225, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.17006802721088435, | |
| "grad_norm": 1.1758267879486084, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.245, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17006802721088435, | |
| "eval_loss": 0.5157535672187805, | |
| "eval_runtime": 72.3745, | |
| "eval_samples_per_second": 13.679, | |
| "eval_steps_per_second": 3.427, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1717687074829932, | |
| "grad_norm": 0.18069548904895782, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.5215, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.17346938775510204, | |
| "grad_norm": 0.19141064584255219, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.5354, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.17517006802721088, | |
| "grad_norm": 0.19714467227458954, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.5284, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.17687074829931973, | |
| "grad_norm": 0.18312111496925354, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.5473, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.17857142857142858, | |
| "grad_norm": 0.3003970980644226, | |
| "learning_rate": 5e-05, | |
| "loss": 0.5312, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.18027210884353742, | |
| "grad_norm": 0.17633511126041412, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.5273, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.18197278911564627, | |
| "grad_norm": 0.17157384753227234, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.5433, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1836734693877551, | |
| "grad_norm": 0.17225749790668488, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5396, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.18537414965986396, | |
| "grad_norm": 0.16851207613945007, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.5155, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.1870748299319728, | |
| "grad_norm": 0.18357497453689575, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.519, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.18877551020408162, | |
| "grad_norm": 0.17682017385959625, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.4996, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 0.1814984381198883, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.5354, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.1921768707482993, | |
| "grad_norm": 0.18621817231178284, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.5382, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.19387755102040816, | |
| "grad_norm": 0.1769561767578125, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.5186, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.195578231292517, | |
| "grad_norm": 0.1833718717098236, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.5163, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.19727891156462585, | |
| "grad_norm": 0.1793185919523239, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.5354, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.1989795918367347, | |
| "grad_norm": 0.17421963810920715, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.5188, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.20068027210884354, | |
| "grad_norm": 0.1665407121181488, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.522, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.20238095238095238, | |
| "grad_norm": 0.1696515828371048, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.5068, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 0.1721905916929245, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.5385, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.20578231292517007, | |
| "grad_norm": 0.17645381391048431, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.5407, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.20748299319727892, | |
| "grad_norm": 0.17688636481761932, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.5304, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.20918367346938777, | |
| "grad_norm": 0.1737985759973526, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.4851, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.2108843537414966, | |
| "grad_norm": 0.1777268350124359, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.5331, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.21258503401360543, | |
| "grad_norm": 0.18043720722198486, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.5253, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21428571428571427, | |
| "grad_norm": 0.18539422750473022, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.5603, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.21598639455782312, | |
| "grad_norm": 0.1766299158334732, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.5205, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.21768707482993196, | |
| "grad_norm": 0.19391000270843506, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.5358, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2193877551020408, | |
| "grad_norm": 0.18630413711071014, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.5198, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.22108843537414966, | |
| "grad_norm": 0.1979658007621765, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.5403, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2227891156462585, | |
| "grad_norm": 0.19212418794631958, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.5007, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.22448979591836735, | |
| "grad_norm": 0.18956515192985535, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.5016, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.2261904761904762, | |
| "grad_norm": 0.19607217609882355, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.507, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.22789115646258504, | |
| "grad_norm": 0.20650868117809296, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.4962, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.22959183673469388, | |
| "grad_norm": 0.20123378932476044, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.4667, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.23129251700680273, | |
| "grad_norm": 0.2093091607093811, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.4657, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.23299319727891157, | |
| "grad_norm": 0.1981552243232727, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.444, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.23469387755102042, | |
| "grad_norm": 0.2081432193517685, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.4453, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.23639455782312926, | |
| "grad_norm": 0.2073698192834854, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.489, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.23809523809523808, | |
| "grad_norm": 0.21433527767658234, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.4287, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.23979591836734693, | |
| "grad_norm": 0.23464985191822052, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.4375, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.24149659863945577, | |
| "grad_norm": 0.22605937719345093, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.4237, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.24319727891156462, | |
| "grad_norm": 0.25338131189346313, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.4475, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.24489795918367346, | |
| "grad_norm": 0.26331523060798645, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3964, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.2465986394557823, | |
| "grad_norm": 0.2676081657409668, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3821, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.24829931972789115, | |
| "grad_norm": 0.2934207320213318, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.4465, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.3128584027290344, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.4523, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.25170068027210885, | |
| "grad_norm": 0.3836762309074402, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.4168, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2534013605442177, | |
| "grad_norm": 0.5722853541374207, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.4489, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.25510204081632654, | |
| "grad_norm": 0.7652419209480286, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.3521, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25510204081632654, | |
| "eval_loss": 0.4918357729911804, | |
| "eval_runtime": 72.3893, | |
| "eval_samples_per_second": 13.676, | |
| "eval_steps_per_second": 3.426, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2568027210884354, | |
| "grad_norm": 0.14883185923099518, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.4939, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.2585034013605442, | |
| "grad_norm": 0.1577376425266266, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.5104, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2602040816326531, | |
| "grad_norm": 0.16027680039405823, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.5332, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.2619047619047619, | |
| "grad_norm": 0.15759849548339844, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.4854, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.26360544217687076, | |
| "grad_norm": 0.15275824069976807, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.5032, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2653061224489796, | |
| "grad_norm": 0.15402601659297943, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.4779, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.26700680272108845, | |
| "grad_norm": 0.15565964579582214, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.5175, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.2687074829931973, | |
| "grad_norm": 0.16247078776359558, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.5132, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.27040816326530615, | |
| "grad_norm": 0.16090863943099976, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.5286, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.272108843537415, | |
| "grad_norm": 0.16414028406143188, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.5058, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.27380952380952384, | |
| "grad_norm": 0.16387270390987396, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.5209, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.2755102040816326, | |
| "grad_norm": 0.16053496301174164, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.5274, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.27721088435374147, | |
| "grad_norm": 0.16361047327518463, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.514, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.2789115646258503, | |
| "grad_norm": 0.16737475991249084, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.5087, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.28061224489795916, | |
| "grad_norm": 0.16898231208324432, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.5013, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.282312925170068, | |
| "grad_norm": 0.1683025360107422, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.5108, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.28401360544217685, | |
| "grad_norm": 0.16656532883644104, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.5238, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 0.1694728136062622, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.505, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.28741496598639454, | |
| "grad_norm": 0.17515990138053894, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.5324, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.2891156462585034, | |
| "grad_norm": 0.17463481426239014, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.5323, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.29081632653061223, | |
| "grad_norm": 0.17755144834518433, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.5186, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2925170068027211, | |
| "grad_norm": 0.17801311612129211, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.5073, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.2942176870748299, | |
| "grad_norm": 0.17718392610549927, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.5107, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.29591836734693877, | |
| "grad_norm": 0.1812407672405243, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.52, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.2976190476190476, | |
| "grad_norm": 0.1797058880329132, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.5134, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.29931972789115646, | |
| "grad_norm": 0.1781056523323059, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.5176, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3010204081632653, | |
| "grad_norm": 0.1884399950504303, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.5133, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.30272108843537415, | |
| "grad_norm": 0.18392810225486755, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.4936, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.304421768707483, | |
| "grad_norm": 0.18322356045246124, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.4612, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 0.19549725949764252, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.5156, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3078231292517007, | |
| "grad_norm": 0.19811411201953888, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.4863, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.30952380952380953, | |
| "grad_norm": 0.19696243107318878, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.5145, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3112244897959184, | |
| "grad_norm": 0.18621185421943665, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.4403, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.3129251700680272, | |
| "grad_norm": 0.1986522227525711, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.4632, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.31462585034013607, | |
| "grad_norm": 0.19403479993343353, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.528, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3163265306122449, | |
| "grad_norm": 0.20525862276554108, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.5147, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.31802721088435376, | |
| "grad_norm": 0.20824183523654938, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.4832, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3197278911564626, | |
| "grad_norm": 0.20856603980064392, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.4359, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.32142857142857145, | |
| "grad_norm": 0.21307384967803955, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.4599, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.3231292517006803, | |
| "grad_norm": 0.22089087963104248, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.465, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.32482993197278914, | |
| "grad_norm": 0.21890276670455933, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.4784, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.32653061224489793, | |
| "grad_norm": 0.2296353578567505, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.4732, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.3282312925170068, | |
| "grad_norm": 0.23591068387031555, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.4127, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3299319727891156, | |
| "grad_norm": 0.24621009826660156, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.4257, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.33163265306122447, | |
| "grad_norm": 0.26626601815223694, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.4173, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.27958881855010986, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.4883, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.33503401360544216, | |
| "grad_norm": 0.28960496187210083, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.4141, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.336734693877551, | |
| "grad_norm": 0.3632633090019226, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4301, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.33843537414965985, | |
| "grad_norm": 0.3938828110694885, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.4803, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 0.6212309002876282, | |
| "learning_rate": 0.0, | |
| "loss": 0.325, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3401360544217687, | |
| "eval_loss": 0.48696738481521606, | |
| "eval_runtime": 72.3824, | |
| "eval_samples_per_second": 13.677, | |
| "eval_steps_per_second": 3.426, | |
| "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": 2.690927703443374e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |