| { | |
| "best_metric": 1.0179662704467773, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.17656146546016332, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0008828073273008167, | |
| "grad_norm": 3.2828705310821533, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8697, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008828073273008167, | |
| "eval_loss": 2.3270158767700195, | |
| "eval_runtime": 38.3162, | |
| "eval_samples_per_second": 49.796, | |
| "eval_steps_per_second": 12.449, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0017656146546016333, | |
| "grad_norm": 3.5589444637298584, | |
| "learning_rate": 2e-05, | |
| "loss": 1.9468, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0026484219819024497, | |
| "grad_norm": 3.8853414058685303, | |
| "learning_rate": 3e-05, | |
| "loss": 1.895, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0035312293092032666, | |
| "grad_norm": 3.7132949829101562, | |
| "learning_rate": 4e-05, | |
| "loss": 1.758, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.004414036636504083, | |
| "grad_norm": 3.055154800415039, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8082, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0052968439638048995, | |
| "grad_norm": 3.010298490524292, | |
| "learning_rate": 6e-05, | |
| "loss": 1.5171, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.006179651291105716, | |
| "grad_norm": 2.4076344966888428, | |
| "learning_rate": 7e-05, | |
| "loss": 1.541, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.007062458618406533, | |
| "grad_norm": 2.4439666271209717, | |
| "learning_rate": 8e-05, | |
| "loss": 1.4547, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00794526594570735, | |
| "grad_norm": 2.055011510848999, | |
| "learning_rate": 9e-05, | |
| "loss": 1.3022, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.008828073273008165, | |
| "grad_norm": 1.6684356927871704, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3078, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.009710880600308983, | |
| "grad_norm": 1.7521171569824219, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.3426, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.010593687927609799, | |
| "grad_norm": 1.735939383506775, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.3363, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.011476495254910615, | |
| "grad_norm": 1.636270523071289, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.1968, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.012359302582211433, | |
| "grad_norm": 1.6050608158111572, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.2793, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.013242109909512249, | |
| "grad_norm": 1.4000378847122192, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.1612, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.014124917236813066, | |
| "grad_norm": 1.623569130897522, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.1665, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.015007724564113882, | |
| "grad_norm": 1.2394853830337524, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.125, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0158905318914147, | |
| "grad_norm": 1.233336091041565, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.2444, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.016773339218715516, | |
| "grad_norm": 1.110575556755066, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.1734, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01765614654601633, | |
| "grad_norm": 1.0778011083602905, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.1426, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.018538953873317148, | |
| "grad_norm": 1.0210111141204834, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.0438, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.019421761200617966, | |
| "grad_norm": 1.09498131275177, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.0852, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.02030456852791878, | |
| "grad_norm": 1.1424933671951294, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.145, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.021187375855219598, | |
| "grad_norm": 1.0362298488616943, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.0097, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.022070183182520416, | |
| "grad_norm": 1.0411286354064941, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.0837, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02295299050982123, | |
| "grad_norm": 1.1341434717178345, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.1176, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.023835797837122048, | |
| "grad_norm": 1.0587491989135742, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.948, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.024718605164422865, | |
| "grad_norm": 1.0277941226959229, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.0282, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.02560141249172368, | |
| "grad_norm": 1.275912880897522, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.3878, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.026484219819024497, | |
| "grad_norm": 1.2981607913970947, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.2537, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.027367027146325315, | |
| "grad_norm": 0.9924188256263733, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.0694, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.028249834473626133, | |
| "grad_norm": 1.3334228992462158, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.1636, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.029132641800926947, | |
| "grad_norm": 1.065547227859497, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.1861, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.030015449128227765, | |
| "grad_norm": 1.0047739744186401, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.1247, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.030898256455528583, | |
| "grad_norm": 1.1559410095214844, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.0303, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0317810637828294, | |
| "grad_norm": 1.1687241792678833, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.9959, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.032663871110130215, | |
| "grad_norm": 1.130322813987732, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.0834, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.03354667843743103, | |
| "grad_norm": 1.0708613395690918, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.1187, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.03442948576473185, | |
| "grad_norm": 1.0673141479492188, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.1363, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.03531229309203266, | |
| "grad_norm": 1.0976028442382812, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.1053, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03619510041933348, | |
| "grad_norm": 1.1204279661178589, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.985, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.037077907746634296, | |
| "grad_norm": 1.3454452753067017, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.4118, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.037960715073935114, | |
| "grad_norm": 1.149428129196167, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.0809, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.03884352240123593, | |
| "grad_norm": 1.0400094985961914, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.1268, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03972632972853675, | |
| "grad_norm": 1.1207729578018188, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.9689, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04060913705583756, | |
| "grad_norm": 1.3091357946395874, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.1772, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.04149194438313838, | |
| "grad_norm": 1.2577565908432007, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.2773, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.042374751710439196, | |
| "grad_norm": 1.2854396104812622, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.2758, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.043257559037740013, | |
| "grad_norm": 1.2923716306686401, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.1884, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.04414036636504083, | |
| "grad_norm": 1.361952781677246, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.1173, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04414036636504083, | |
| "eval_loss": 1.2455638647079468, | |
| "eval_runtime": 38.2858, | |
| "eval_samples_per_second": 49.836, | |
| "eval_steps_per_second": 12.459, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04502317369234165, | |
| "grad_norm": 3.5160176753997803, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.5233, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.04590598101964246, | |
| "grad_norm": 3.4013254642486572, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.5494, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.04678878834694328, | |
| "grad_norm": 2.6268138885498047, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.154, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.047671595674244095, | |
| "grad_norm": 1.5156739950180054, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.1239, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04855440300154491, | |
| "grad_norm": 1.1548527479171753, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.1214, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04943721032884573, | |
| "grad_norm": 1.4268921613693237, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.069, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.05032001765614655, | |
| "grad_norm": 1.5915234088897705, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.0367, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.05120282498344736, | |
| "grad_norm": 1.0153710842132568, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.9494, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.05208563231074818, | |
| "grad_norm": 0.9790281057357788, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.0383, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.052968439638048995, | |
| "grad_norm": 0.7598667144775391, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.9236, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05385124696534981, | |
| "grad_norm": 0.7081838846206665, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.0108, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.05473405429265063, | |
| "grad_norm": 0.7830983400344849, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.0278, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.05561686161995145, | |
| "grad_norm": 0.8189266920089722, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.1562, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.056499668947252266, | |
| "grad_norm": 0.7172884345054626, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.0595, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.057382476274553076, | |
| "grad_norm": 0.7670396566390991, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.0915, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.058265283601853894, | |
| "grad_norm": 0.7067275643348694, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.018, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.05914809092915471, | |
| "grad_norm": 0.6907898187637329, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.9909, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.06003089825645553, | |
| "grad_norm": 0.7496179938316345, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.0035, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.06091370558375635, | |
| "grad_norm": 0.6954419016838074, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.0178, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.061796512911057165, | |
| "grad_norm": 0.7761889696121216, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.0588, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06267932023835798, | |
| "grad_norm": 0.7610973715782166, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.9313, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0635621275656588, | |
| "grad_norm": 0.8635450005531311, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.0514, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.06444493489295962, | |
| "grad_norm": 0.9538561701774597, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.259, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.06532774222026043, | |
| "grad_norm": 0.7979761958122253, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.1682, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.06621054954756124, | |
| "grad_norm": 0.7394124269485474, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.0905, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06709335687486206, | |
| "grad_norm": 0.803189754486084, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.0583, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.06797616420216288, | |
| "grad_norm": 0.8538224101066589, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.04, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0688589715294637, | |
| "grad_norm": 0.7931703925132751, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.981, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.06974177885676451, | |
| "grad_norm": 0.8034219741821289, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.9427, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.07062458618406532, | |
| "grad_norm": 0.8855370283126831, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.0533, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07150739351136615, | |
| "grad_norm": 0.7561924457550049, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.0357, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.07239020083866696, | |
| "grad_norm": 0.8152874112129211, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.0248, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.07327300816596778, | |
| "grad_norm": 0.8663817644119263, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.9884, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.07415581549326859, | |
| "grad_norm": 0.9121115803718567, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.096, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.07503862282056942, | |
| "grad_norm": 0.8358010053634644, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.01, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.07592143014787023, | |
| "grad_norm": 0.9568318724632263, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.1565, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.07680423747517104, | |
| "grad_norm": 0.8760489821434021, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.0051, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.07768704480247186, | |
| "grad_norm": 0.910731852054596, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.0605, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.07856985212977267, | |
| "grad_norm": 0.8696299195289612, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.9884, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0794526594570735, | |
| "grad_norm": 0.9313849210739136, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.9193, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08033546678437431, | |
| "grad_norm": 0.8825989961624146, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.9621, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.08121827411167512, | |
| "grad_norm": 0.9456129670143127, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.9746, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.08210108143897595, | |
| "grad_norm": 1.000291109085083, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.2367, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.08298388876627676, | |
| "grad_norm": 1.0138276815414429, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.129, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.08386669609357758, | |
| "grad_norm": 1.0745936632156372, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.2886, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.08474950342087839, | |
| "grad_norm": 1.0831668376922607, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.2101, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.08563231074817922, | |
| "grad_norm": 1.239163875579834, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.3254, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.08651511807548003, | |
| "grad_norm": 1.0036643743515015, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.9825, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.08739792540278084, | |
| "grad_norm": 1.4951629638671875, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.4675, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.08828073273008166, | |
| "grad_norm": 1.3885345458984375, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.1356, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08828073273008166, | |
| "eval_loss": 1.1188737154006958, | |
| "eval_runtime": 38.3439, | |
| "eval_samples_per_second": 49.76, | |
| "eval_steps_per_second": 12.44, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08916354005738247, | |
| "grad_norm": 2.0964150428771973, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.3427, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0900463473846833, | |
| "grad_norm": 1.7808575630187988, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.1452, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.09092915471198411, | |
| "grad_norm": 1.8077600002288818, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.1804, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.09181196203928492, | |
| "grad_norm": 1.5133174657821655, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.0607, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.09269476936658574, | |
| "grad_norm": 1.092074990272522, | |
| "learning_rate": 5e-05, | |
| "loss": 0.9933, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.09357757669388655, | |
| "grad_norm": 0.9688529968261719, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.0794, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.09446038402118738, | |
| "grad_norm": 0.7483575344085693, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.0291, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.09534319134848819, | |
| "grad_norm": 0.8056850433349609, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.8974, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.09622599867578902, | |
| "grad_norm": 0.8146949410438538, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.0897, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.09710880600308983, | |
| "grad_norm": 0.8641289472579956, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.9679, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09799161333039064, | |
| "grad_norm": 0.8634604215621948, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.0662, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.09887442065769146, | |
| "grad_norm": 0.7673062682151794, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.8765, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.09975722798499227, | |
| "grad_norm": 0.8061811923980713, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.117, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1006400353122931, | |
| "grad_norm": 0.7722053527832031, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.0066, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.10152284263959391, | |
| "grad_norm": 0.7186280488967896, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.0151, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.10240564996689472, | |
| "grad_norm": 0.6878299117088318, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.9904, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.10328845729419554, | |
| "grad_norm": 0.7549110651016235, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.9945, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.10417126462149635, | |
| "grad_norm": 0.8615491390228271, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.1329, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.10505407194879718, | |
| "grad_norm": 0.7086193561553955, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.9388, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.10593687927609799, | |
| "grad_norm": 0.770940899848938, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.045, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10681968660339881, | |
| "grad_norm": 0.7125656604766846, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.9309, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.10770249393069962, | |
| "grad_norm": 0.7269070744514465, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.977, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.10858530125800044, | |
| "grad_norm": 0.7405799627304077, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.1038, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.10946810858530126, | |
| "grad_norm": 0.686782717704773, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.8536, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.11035091591260207, | |
| "grad_norm": 0.712872326374054, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.9517, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1112337232399029, | |
| "grad_norm": 0.7823908925056458, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.0083, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1121165305672037, | |
| "grad_norm": 0.6804335713386536, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.9927, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.11299933789450453, | |
| "grad_norm": 0.7232629656791687, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.935, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.11388214522180534, | |
| "grad_norm": 0.7596330046653748, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.8832, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.11476495254910615, | |
| "grad_norm": 0.7751919031143188, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.9471, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.11564775987640698, | |
| "grad_norm": 0.7972702980041504, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.0071, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.11653056720370779, | |
| "grad_norm": 0.8781986832618713, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.1418, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.11741337453100861, | |
| "grad_norm": 0.8987315893173218, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.0867, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.11829618185830942, | |
| "grad_norm": 0.8337807059288025, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.9745, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.11917898918561023, | |
| "grad_norm": 0.9067551493644714, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.0048, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.12006179651291106, | |
| "grad_norm": 0.8992189764976501, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.1514, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.12094460384021187, | |
| "grad_norm": 0.803499698638916, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.0236, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1218274111675127, | |
| "grad_norm": 0.7964784502983093, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.9221, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.1227102184948135, | |
| "grad_norm": 0.7766945958137512, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.9037, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.12359302582211433, | |
| "grad_norm": 0.8417460918426514, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.9649, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.12447583314941514, | |
| "grad_norm": 0.8503076434135437, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.0312, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.12535864047671597, | |
| "grad_norm": 0.978374183177948, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.0442, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.12624144780401678, | |
| "grad_norm": 0.9302412867546082, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.1717, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1271242551313176, | |
| "grad_norm": 1.0065759420394897, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.1741, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.1280070624586184, | |
| "grad_norm": 0.946594774723053, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.1837, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.12888986978591924, | |
| "grad_norm": 0.9386952519416809, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.0199, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.12977267711322005, | |
| "grad_norm": 0.9409152865409851, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.0907, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.13065548444052086, | |
| "grad_norm": 1.0664265155792236, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.0778, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.13153829176782167, | |
| "grad_norm": 1.0991122722625732, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.0545, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.13242109909512248, | |
| "grad_norm": 1.3308420181274414, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.4785, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.13242109909512248, | |
| "eval_loss": 1.0398972034454346, | |
| "eval_runtime": 38.3992, | |
| "eval_samples_per_second": 49.689, | |
| "eval_steps_per_second": 12.422, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.13330390642242332, | |
| "grad_norm": 1.113048791885376, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.0236, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.13418671374972413, | |
| "grad_norm": 0.9896067380905151, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.9432, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.13506952107702494, | |
| "grad_norm": 1.0587103366851807, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.9664, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.13595232840432575, | |
| "grad_norm": 1.043975830078125, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.9518, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.13683513573162656, | |
| "grad_norm": 0.9581245183944702, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.0899, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1377179430589274, | |
| "grad_norm": 0.8352423310279846, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.8906, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.1386007503862282, | |
| "grad_norm": 0.8045545220375061, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.9948, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.13948355771352902, | |
| "grad_norm": 0.7856590747833252, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.9166, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.14036636504082983, | |
| "grad_norm": 0.8514511585235596, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.0468, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.14124917236813064, | |
| "grad_norm": 0.7472115755081177, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.9245, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.14213197969543148, | |
| "grad_norm": 0.8417037725448608, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.0556, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1430147870227323, | |
| "grad_norm": 0.673015832901001, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.8544, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.1438975943500331, | |
| "grad_norm": 0.822077214717865, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.1305, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.14478040167733391, | |
| "grad_norm": 0.754014253616333, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.9398, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.14566320900463472, | |
| "grad_norm": 0.7298570275306702, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.0445, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.14654601633193556, | |
| "grad_norm": 0.7686694860458374, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.035, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.14742882365923637, | |
| "grad_norm": 0.7595144510269165, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.0837, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.14831163098653719, | |
| "grad_norm": 0.7648171186447144, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.9681, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.149194438313838, | |
| "grad_norm": 0.7730786800384521, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.0832, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.15007724564113883, | |
| "grad_norm": 0.7379984855651855, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.951, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.15096005296843965, | |
| "grad_norm": 0.6645638942718506, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.9498, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.15184286029574046, | |
| "grad_norm": 0.7368901968002319, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.9343, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.15272566762304127, | |
| "grad_norm": 0.6814242005348206, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.975, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.15360847495034208, | |
| "grad_norm": 0.7335926294326782, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.0102, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.15449128227764292, | |
| "grad_norm": 0.7502698302268982, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.9856, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.15537408960494373, | |
| "grad_norm": 0.7277138829231262, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.9038, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.15625689693224454, | |
| "grad_norm": 0.7511397004127502, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.9403, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.15713970425954535, | |
| "grad_norm": 0.756578803062439, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.9677, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.15802251158684616, | |
| "grad_norm": 0.8428077101707458, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.1702, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.158905318914147, | |
| "grad_norm": 0.8611013293266296, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.0389, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1597881262414478, | |
| "grad_norm": 0.7790641784667969, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.9727, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.16067093356874862, | |
| "grad_norm": 0.7951785922050476, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.9244, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.16155374089604943, | |
| "grad_norm": 0.8784627318382263, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.9776, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.16243654822335024, | |
| "grad_norm": 0.8529843091964722, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.0874, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.16331935555065108, | |
| "grad_norm": 0.8204442858695984, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.0073, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1642021628779519, | |
| "grad_norm": 0.7387585043907166, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.9589, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1650849702052527, | |
| "grad_norm": 0.9290441274642944, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.0061, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1659677775325535, | |
| "grad_norm": 0.7750152349472046, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.9998, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.16685058485985435, | |
| "grad_norm": 0.9057651162147522, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.0658, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.16773339218715516, | |
| "grad_norm": 0.8646574020385742, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.8308, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.16861619951445597, | |
| "grad_norm": 1.0296387672424316, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.2257, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.16949900684175678, | |
| "grad_norm": 0.8883101940155029, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.0171, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1703818141690576, | |
| "grad_norm": 0.9178954362869263, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.9304, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.17126462149635843, | |
| "grad_norm": 1.023403286933899, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.1031, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.17214742882365924, | |
| "grad_norm": 0.9290059804916382, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.9365, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.17303023615096005, | |
| "grad_norm": 1.048574447631836, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.1277, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.17391304347826086, | |
| "grad_norm": 1.1039568185806274, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.2084, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.17479585080556168, | |
| "grad_norm": 1.1607375144958496, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.1964, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.17567865813286251, | |
| "grad_norm": 1.340389370918274, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.2605, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.17656146546016332, | |
| "grad_norm": 1.6403038501739502, | |
| "learning_rate": 0.0, | |
| "loss": 1.2713, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.17656146546016332, | |
| "eval_loss": 1.0179662704467773, | |
| "eval_runtime": 38.396, | |
| "eval_samples_per_second": 49.693, | |
| "eval_steps_per_second": 12.423, | |
| "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": 4.26629570494464e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |