| { | |
| "best_metric": 1.2555583715438843, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.21580547112462, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0060790273556231, | |
| "grad_norm": 1.7836625576019287, | |
| "learning_rate": 1e-05, | |
| "loss": 1.9276, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0060790273556231, | |
| "eval_loss": 2.0459976196289062, | |
| "eval_runtime": 30.8801, | |
| "eval_samples_per_second": 8.97, | |
| "eval_steps_per_second": 2.267, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0121580547112462, | |
| "grad_norm": 1.9655598402023315, | |
| "learning_rate": 2e-05, | |
| "loss": 1.9162, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0182370820668693, | |
| "grad_norm": 1.9556907415390015, | |
| "learning_rate": 3e-05, | |
| "loss": 2.0912, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0243161094224924, | |
| "grad_norm": 1.7137624025344849, | |
| "learning_rate": 4e-05, | |
| "loss": 2.1189, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.030395136778115502, | |
| "grad_norm": 1.6796306371688843, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0105, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0364741641337386, | |
| "grad_norm": 1.5738074779510498, | |
| "learning_rate": 6e-05, | |
| "loss": 1.864, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0425531914893617, | |
| "grad_norm": 1.609481930732727, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7974, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0486322188449848, | |
| "grad_norm": 1.9356203079223633, | |
| "learning_rate": 8e-05, | |
| "loss": 1.65, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0547112462006079, | |
| "grad_norm": 2.287429094314575, | |
| "learning_rate": 9e-05, | |
| "loss": 1.7972, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.060790273556231005, | |
| "grad_norm": 1.8350834846496582, | |
| "learning_rate": 0.0001, | |
| "loss": 1.661, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0668693009118541, | |
| "grad_norm": 1.586941123008728, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.5518, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0729483282674772, | |
| "grad_norm": 1.7649606466293335, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.627, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0790273556231003, | |
| "grad_norm": 1.5410637855529785, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.4936, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0851063829787234, | |
| "grad_norm": 1.5190150737762451, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.5412, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0911854103343465, | |
| "grad_norm": 1.5291023254394531, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.4773, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0972644376899696, | |
| "grad_norm": 1.5268183946609497, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.6176, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1033434650455927, | |
| "grad_norm": 1.3530791997909546, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.6504, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1094224924012158, | |
| "grad_norm": 1.4574755430221558, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.6461, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.11550151975683891, | |
| "grad_norm": 1.6941369771957397, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.8071, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.12158054711246201, | |
| "grad_norm": 1.4535752534866333, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.5129, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1276595744680851, | |
| "grad_norm": 1.5512946844100952, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.6627, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1337386018237082, | |
| "grad_norm": 1.5676143169403076, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.747, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.1398176291793313, | |
| "grad_norm": 1.3140560388565063, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.5061, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1458966565349544, | |
| "grad_norm": 1.2685281038284302, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.2963, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1519756838905775, | |
| "grad_norm": 1.1902039051055908, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.3054, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1580547112462006, | |
| "grad_norm": 1.386483907699585, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.3126, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1641337386018237, | |
| "grad_norm": 1.163313388824463, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.2615, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1702127659574468, | |
| "grad_norm": 1.1470832824707031, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.2127, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1762917933130699, | |
| "grad_norm": 1.2081419229507446, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.2246, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.182370820668693, | |
| "grad_norm": 1.1016731262207031, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.201, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1884498480243161, | |
| "grad_norm": 1.1176942586898804, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.2514, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.1945288753799392, | |
| "grad_norm": 1.2162182331085205, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.2817, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.2006079027355623, | |
| "grad_norm": 1.2303563356399536, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.2787, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.2066869300911854, | |
| "grad_norm": 1.1269229650497437, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.1797, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.2127659574468085, | |
| "grad_norm": 1.1142123937606812, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.2678, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.2188449848024316, | |
| "grad_norm": 1.098849892616272, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.1874, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.22492401215805471, | |
| "grad_norm": 1.128641963005066, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.1545, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.23100303951367782, | |
| "grad_norm": 1.1264187097549438, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.3152, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.23708206686930092, | |
| "grad_norm": 1.2174651622772217, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.3264, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.24316109422492402, | |
| "grad_norm": 1.1203888654708862, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.2648, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.24924012158054712, | |
| "grad_norm": 1.2735846042633057, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.3305, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.2553191489361702, | |
| "grad_norm": 0.92659592628479, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.5845, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2613981762917933, | |
| "grad_norm": 0.928512692451477, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.4695, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2674772036474164, | |
| "grad_norm": 1.0293956995010376, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.5067, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2735562310030395, | |
| "grad_norm": 1.198373794555664, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.5005, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.2796352583586626, | |
| "grad_norm": 1.1086431741714478, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.6231, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 1.0705828666687012, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.4303, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.2917933130699088, | |
| "grad_norm": 1.0497395992279053, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.3332, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2978723404255319, | |
| "grad_norm": 1.237426996231079, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.393, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.303951367781155, | |
| "grad_norm": 1.14468514919281, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.4106, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.303951367781155, | |
| "eval_loss": 1.3674739599227905, | |
| "eval_runtime": 31.284, | |
| "eval_samples_per_second": 8.854, | |
| "eval_steps_per_second": 2.238, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3100303951367781, | |
| "grad_norm": 1.1763962507247925, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.4482, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.3161094224924012, | |
| "grad_norm": 1.0938806533813477, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.2576, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3221884498480243, | |
| "grad_norm": 1.1484322547912598, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.4166, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3282674772036474, | |
| "grad_norm": 1.2330974340438843, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.4815, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.3343465045592705, | |
| "grad_norm": 1.5895949602127075, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.4773, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.3404255319148936, | |
| "grad_norm": 1.3721816539764404, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.5345, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3465045592705167, | |
| "grad_norm": 1.2874829769134521, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.5054, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3525835866261398, | |
| "grad_norm": 1.477946162223816, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.711, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.3586626139817629, | |
| "grad_norm": 1.6561298370361328, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.5596, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.364741641337386, | |
| "grad_norm": 1.4493569135665894, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.6255, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3708206686930091, | |
| "grad_norm": 1.5102639198303223, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.4314, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3768996960486322, | |
| "grad_norm": 1.2875674962997437, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.1642, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.3829787234042553, | |
| "grad_norm": 1.536087155342102, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.1629, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.3890577507598784, | |
| "grad_norm": 1.0621429681777954, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.1772, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3951367781155015, | |
| "grad_norm": 1.1528043746948242, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.2588, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.4012158054711246, | |
| "grad_norm": 1.2233235836029053, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.4096, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.4072948328267477, | |
| "grad_norm": 1.1939024925231934, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.1564, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.4133738601823708, | |
| "grad_norm": 1.1092005968093872, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.1575, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.4194528875379939, | |
| "grad_norm": 1.131093144416809, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.3206, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.425531914893617, | |
| "grad_norm": 1.012694001197815, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.2485, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.4316109422492401, | |
| "grad_norm": 1.2496697902679443, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.2219, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.4376899696048632, | |
| "grad_norm": 1.052462100982666, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.12, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.44376899696048633, | |
| "grad_norm": 1.2121566534042358, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.2185, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.44984802431610943, | |
| "grad_norm": 1.045432209968567, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.1954, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.45592705167173253, | |
| "grad_norm": 1.0931679010391235, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.1167, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.46200607902735563, | |
| "grad_norm": 1.0873976945877075, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.2239, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.46808510638297873, | |
| "grad_norm": 1.2558058500289917, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.2211, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.47416413373860183, | |
| "grad_norm": 1.164841890335083, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.2813, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.48024316109422494, | |
| "grad_norm": 1.3772122859954834, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.1648, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.48632218844984804, | |
| "grad_norm": 1.0816856622695923, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.2033, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.49240121580547114, | |
| "grad_norm": 1.166603922843933, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.209, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.49848024316109424, | |
| "grad_norm": 1.2450183629989624, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.314, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5045592705167173, | |
| "grad_norm": 0.9764375686645508, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.394, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5106382978723404, | |
| "grad_norm": 1.041074514389038, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.5352, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.5167173252279635, | |
| "grad_norm": 0.9813753366470337, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.4155, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.5227963525835866, | |
| "grad_norm": 1.0731186866760254, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.3959, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5288753799392097, | |
| "grad_norm": 1.0571465492248535, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.4873, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5349544072948328, | |
| "grad_norm": 1.2374310493469238, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.3827, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.541033434650456, | |
| "grad_norm": 1.1906167268753052, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.2807, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.547112462006079, | |
| "grad_norm": 1.1364439725875854, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.2452, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5531914893617021, | |
| "grad_norm": 1.1996861696243286, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.2793, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5592705167173252, | |
| "grad_norm": 1.119796633720398, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.2084, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.5653495440729484, | |
| "grad_norm": 1.1832455396652222, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.3079, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 1.2578562498092651, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.2807, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.5775075987841946, | |
| "grad_norm": 1.5265213251113892, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.3349, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.5835866261398176, | |
| "grad_norm": 1.4674025774002075, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.3907, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.5896656534954408, | |
| "grad_norm": 1.2287399768829346, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.2637, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.5957446808510638, | |
| "grad_norm": 1.4680312871932983, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.5601, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.601823708206687, | |
| "grad_norm": 1.3695200681686401, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.2223, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.60790273556231, | |
| "grad_norm": 1.4715473651885986, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.4056, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.60790273556231, | |
| "eval_loss": 1.292844295501709, | |
| "eval_runtime": 31.3094, | |
| "eval_samples_per_second": 8.847, | |
| "eval_steps_per_second": 2.236, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6139817629179332, | |
| "grad_norm": 1.4016162157058716, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.4079, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.6200607902735562, | |
| "grad_norm": 2.0157923698425293, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.481, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.6261398176291794, | |
| "grad_norm": 1.2046968936920166, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.2524, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.6322188449848024, | |
| "grad_norm": 1.0978970527648926, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.1627, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6382978723404256, | |
| "grad_norm": 1.0364453792572021, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2258, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6443768996960486, | |
| "grad_norm": 1.1342463493347168, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.2528, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6504559270516718, | |
| "grad_norm": 1.000739574432373, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.1265, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.6565349544072948, | |
| "grad_norm": 1.0854004621505737, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.1973, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.662613981762918, | |
| "grad_norm": 1.032932162284851, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.1833, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.668693009118541, | |
| "grad_norm": 0.9852817058563232, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.1426, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6747720364741642, | |
| "grad_norm": 1.092152714729309, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.2261, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.6808510638297872, | |
| "grad_norm": 1.0326870679855347, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.1596, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.6869300911854104, | |
| "grad_norm": 1.0733891725540161, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.2205, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.6930091185410334, | |
| "grad_norm": 0.9702630639076233, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.1775, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.6990881458966566, | |
| "grad_norm": 1.062697410583496, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.0724, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.7051671732522796, | |
| "grad_norm": 1.03281831741333, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.1681, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.7112462006079028, | |
| "grad_norm": 1.1306233406066895, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.1371, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.7173252279635258, | |
| "grad_norm": 1.0361230373382568, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.1704, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.723404255319149, | |
| "grad_norm": 1.0658708810806274, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0982, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.729483282674772, | |
| "grad_norm": 1.0389710664749146, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.2011, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7355623100303952, | |
| "grad_norm": 1.2019336223602295, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.2158, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.7416413373860182, | |
| "grad_norm": 1.113996982574463, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.2178, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.7477203647416414, | |
| "grad_norm": 1.206320881843567, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.278, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7537993920972644, | |
| "grad_norm": 0.8871518969535828, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.3995, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7598784194528876, | |
| "grad_norm": 1.10869562625885, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.3967, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.7659574468085106, | |
| "grad_norm": 1.0009129047393799, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.4259, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.7720364741641338, | |
| "grad_norm": 1.0479925870895386, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.4265, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.7781155015197568, | |
| "grad_norm": 0.9976812601089478, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.257, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.78419452887538, | |
| "grad_norm": 1.355043649673462, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.4366, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.790273556231003, | |
| "grad_norm": 1.089213490486145, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.2735, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.7963525835866262, | |
| "grad_norm": 2.232626438140869, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.5396, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.8024316109422492, | |
| "grad_norm": 1.3629820346832275, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.3665, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.8085106382978723, | |
| "grad_norm": 1.199716329574585, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.4773, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.8145896656534954, | |
| "grad_norm": 1.2396886348724365, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.3163, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.8206686930091185, | |
| "grad_norm": 1.2613160610198975, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.1356, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.8267477203647416, | |
| "grad_norm": 1.2099186182022095, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.4594, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.8328267477203647, | |
| "grad_norm": 1.326059341430664, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.2732, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.8389057750759878, | |
| "grad_norm": 1.2106437683105469, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.244, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.8449848024316109, | |
| "grad_norm": 1.2956374883651733, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.2384, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.851063829787234, | |
| "grad_norm": 1.666294813156128, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.2934, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8571428571428571, | |
| "grad_norm": 1.3977957963943481, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.3895, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8632218844984803, | |
| "grad_norm": 1.3760126829147339, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.3124, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.8693009118541033, | |
| "grad_norm": 1.3664624691009521, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.3252, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.8753799392097265, | |
| "grad_norm": 1.1247613430023193, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.2056, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.8814589665653495, | |
| "grad_norm": 1.2257062196731567, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.0573, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.8875379939209727, | |
| "grad_norm": 1.0265038013458252, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.1623, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.8936170212765957, | |
| "grad_norm": 1.020497441291809, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.1581, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.8996960486322189, | |
| "grad_norm": 0.9764475226402283, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.0843, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.9057750759878419, | |
| "grad_norm": 1.0857138633728027, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.2159, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.9118541033434651, | |
| "grad_norm": 1.0098153352737427, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.2268, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9118541033434651, | |
| "eval_loss": 1.258838176727295, | |
| "eval_runtime": 31.056, | |
| "eval_samples_per_second": 8.919, | |
| "eval_steps_per_second": 2.254, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9179331306990881, | |
| "grad_norm": 1.0707805156707764, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.2064, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.9240121580547113, | |
| "grad_norm": 0.9760774970054626, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.0687, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.9300911854103343, | |
| "grad_norm": 0.9699292778968811, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.0737, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.9361702127659575, | |
| "grad_norm": 0.9776856303215027, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.1459, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.9422492401215805, | |
| "grad_norm": 1.2479808330535889, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.2061, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.9483282674772037, | |
| "grad_norm": 0.9862396121025085, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.0816, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.9544072948328267, | |
| "grad_norm": 1.002426028251648, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.2156, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.9604863221884499, | |
| "grad_norm": 1.0932667255401611, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.1449, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.9665653495440729, | |
| "grad_norm": 0.9802281260490417, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.0786, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.9726443768996961, | |
| "grad_norm": 1.0398657321929932, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.1442, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9787234042553191, | |
| "grad_norm": 1.0206019878387451, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.1332, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.9848024316109423, | |
| "grad_norm": 1.0815765857696533, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.2285, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.9908814589665653, | |
| "grad_norm": 1.1101170778274536, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.0357, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.9969604863221885, | |
| "grad_norm": 1.3740544319152832, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.2763, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.0030395136778116, | |
| "grad_norm": 1.6474155187606812, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.8005, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.0091185410334347, | |
| "grad_norm": 0.8269138336181641, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.1268, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.0151975683890577, | |
| "grad_norm": 0.9584388136863708, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.1263, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.0212765957446808, | |
| "grad_norm": 0.9003273248672485, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.1763, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.027355623100304, | |
| "grad_norm": 0.9121745824813843, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.2929, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.033434650455927, | |
| "grad_norm": 0.8949875831604004, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.0734, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.0395136778115501, | |
| "grad_norm": 0.9615802764892578, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.0933, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.0455927051671732, | |
| "grad_norm": 0.9501930475234985, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.0371, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.0516717325227964, | |
| "grad_norm": 0.9965758323669434, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.0524, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.0577507598784195, | |
| "grad_norm": 0.9461009502410889, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.0621, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.0638297872340425, | |
| "grad_norm": 0.8981134295463562, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.9009, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.0699088145896656, | |
| "grad_norm": 0.9697230458259583, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.9235, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.0759878419452888, | |
| "grad_norm": 0.9853894710540771, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.9314, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.082066869300912, | |
| "grad_norm": 1.1321412324905396, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.0284, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.088145896656535, | |
| "grad_norm": 1.121160626411438, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.9478, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.094224924012158, | |
| "grad_norm": 1.3388609886169434, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.0532, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.1003039513677813, | |
| "grad_norm": 1.1614882946014404, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.0538, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.1063829787234043, | |
| "grad_norm": 1.2139602899551392, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.1108, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.1124620060790273, | |
| "grad_norm": 1.5596733093261719, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.9484, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.1185410334346504, | |
| "grad_norm": 1.1934853792190552, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.0519, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.1246200607902737, | |
| "grad_norm": 1.2559603452682495, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.2086, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.1306990881458967, | |
| "grad_norm": 1.0586079359054565, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.0069, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.1367781155015197, | |
| "grad_norm": 0.9022791981697083, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.8929, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 0.8893226385116577, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.9946, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.148936170212766, | |
| "grad_norm": 0.8484130501747131, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.8628, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.155015197568389, | |
| "grad_norm": 0.9051937460899353, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.8818, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.1610942249240122, | |
| "grad_norm": 1.166377305984497, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.9814, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.1671732522796352, | |
| "grad_norm": 0.8817613124847412, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.87, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.1732522796352582, | |
| "grad_norm": 0.8438370227813721, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.987, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.1793313069908815, | |
| "grad_norm": 0.7979356050491333, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.8422, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.1854103343465046, | |
| "grad_norm": 0.8640711903572083, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.8591, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.1914893617021276, | |
| "grad_norm": 1.0389556884765625, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.0762, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.1975683890577509, | |
| "grad_norm": 0.9843372106552124, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.9142, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.203647416413374, | |
| "grad_norm": 0.9549326300621033, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.8878, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.209726443768997, | |
| "grad_norm": 0.8847540020942688, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.936, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.21580547112462, | |
| "grad_norm": 0.8382249474525452, | |
| "learning_rate": 0.0, | |
| "loss": 0.9145, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.21580547112462, | |
| "eval_loss": 1.2555583715438843, | |
| "eval_runtime": 31.3501, | |
| "eval_samples_per_second": 8.836, | |
| "eval_steps_per_second": 2.233, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 4.603794878135009e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |