| { | |
| "best_metric": 2.2500593662261963, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.010127607859023698, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010127607859023699, | |
| "grad_norm": 31.39533233642578, | |
| "learning_rate": 1e-05, | |
| "loss": 5.6035, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00010127607859023699, | |
| "eval_loss": 6.345576286315918, | |
| "eval_runtime": 1568.5781, | |
| "eval_samples_per_second": 10.602, | |
| "eval_steps_per_second": 2.651, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00020255215718047398, | |
| "grad_norm": 23.184526443481445, | |
| "learning_rate": 2e-05, | |
| "loss": 5.6016, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00030382823577071097, | |
| "grad_norm": 30.848772048950195, | |
| "learning_rate": 3e-05, | |
| "loss": 5.0115, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00040510431436094796, | |
| "grad_norm": 20.879568099975586, | |
| "learning_rate": 4e-05, | |
| "loss": 4.0997, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005063803929511849, | |
| "grad_norm": 56.598411560058594, | |
| "learning_rate": 5e-05, | |
| "loss": 3.3008, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006076564715414219, | |
| "grad_norm": 29.526737213134766, | |
| "learning_rate": 6e-05, | |
| "loss": 2.5314, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007089325501316589, | |
| "grad_norm": 12.449617385864258, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7549, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0008102086287218959, | |
| "grad_norm": 16.119483947753906, | |
| "learning_rate": 8e-05, | |
| "loss": 1.6059, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0009114847073121329, | |
| "grad_norm": 9.31202220916748, | |
| "learning_rate": 9e-05, | |
| "loss": 1.5359, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0010127607859023698, | |
| "grad_norm": 19.932828903198242, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7887, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001114036864492607, | |
| "grad_norm": 6.6560163497924805, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.1223, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0012153129430828439, | |
| "grad_norm": 7.101311206817627, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.3276, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0013165890216730808, | |
| "grad_norm": 5.894531726837158, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.4501, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0014178651002633178, | |
| "grad_norm": 6.54768180847168, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.4689, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0015191411788535547, | |
| "grad_norm": 6.256585597991943, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.5853, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0016204172574437918, | |
| "grad_norm": 8.225622177124023, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.6304, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0017216933360340288, | |
| "grad_norm": 5.209911346435547, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.3861, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0018229694146242657, | |
| "grad_norm": 7.358743667602539, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.4267, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0019242454932145027, | |
| "grad_norm": 5.794778823852539, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.4137, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0020255215718047396, | |
| "grad_norm": 5.440245151519775, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.7719, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0021267976503949767, | |
| "grad_norm": 9.398520469665527, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.8461, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002228073728985214, | |
| "grad_norm": 7.62516975402832, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.8011, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0023293498075754506, | |
| "grad_norm": 4.325904369354248, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.4682, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0024306258861656878, | |
| "grad_norm": 6.48047399520874, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.8256, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0025319019647559245, | |
| "grad_norm": 6.384871482849121, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.6956, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0026331780433461616, | |
| "grad_norm": 4.6459126472473145, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.6426, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002734454121936399, | |
| "grad_norm": 5.729737758636475, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.5419, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0028357302005266355, | |
| "grad_norm": 4.571890830993652, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.6178, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0029370062791168727, | |
| "grad_norm": 4.613413333892822, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.6403, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0030382823577071094, | |
| "grad_norm": 4.91149377822876, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.6804, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0031395584362973465, | |
| "grad_norm": 7.935057163238525, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.2117, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0032408345148875837, | |
| "grad_norm": 5.093746185302734, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.1726, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0033421105934778204, | |
| "grad_norm": 4.15211296081543, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.9019, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0034433866720680576, | |
| "grad_norm": 3.5006465911865234, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.7625, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0035446627506582947, | |
| "grad_norm": 3.709160089492798, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.7127, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0036459388292485314, | |
| "grad_norm": 3.0920586585998535, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.5979, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0037472149078387686, | |
| "grad_norm": 2.882312536239624, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.7251, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0038484909864290053, | |
| "grad_norm": 5.489690780639648, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.3286, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0039497670650192425, | |
| "grad_norm": 7.151604652404785, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.8412, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.004051043143609479, | |
| "grad_norm": 8.030261993408203, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 3.1621, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004152319222199717, | |
| "grad_norm": 7.285529136657715, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.7091, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0042535953007899535, | |
| "grad_norm": 5.098424911499023, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.3983, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.00435487137938019, | |
| "grad_norm": 5.375723838806152, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.3807, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004456147457970428, | |
| "grad_norm": 4.936964988708496, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.3285, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0045574235365606645, | |
| "grad_norm": 4.360125541687012, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.291, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004658699615150901, | |
| "grad_norm": 7.072688102722168, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.8103, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004759975693741138, | |
| "grad_norm": 6.983039379119873, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.6871, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0048612517723313755, | |
| "grad_norm": 8.898165702819824, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.8689, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.004962527850921612, | |
| "grad_norm": 5.214823246002197, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.5026, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005063803929511849, | |
| "grad_norm": 5.564656734466553, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.6417, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005063803929511849, | |
| "eval_loss": 2.4560940265655518, | |
| "eval_runtime": 1574.7033, | |
| "eval_samples_per_second": 10.561, | |
| "eval_steps_per_second": 2.64, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0051650800081020866, | |
| "grad_norm": 8.968050956726074, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.0716, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.005266356086692323, | |
| "grad_norm": 7.762667179107666, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.1966, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.00536763216528256, | |
| "grad_norm": 6.707168102264404, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.5668, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005468908243872798, | |
| "grad_norm": 5.985746383666992, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.343, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005570184322463034, | |
| "grad_norm": 4.789600372314453, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.4971, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.005671460401053271, | |
| "grad_norm": 3.0450186729431152, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.2251, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005772736479643509, | |
| "grad_norm": 3.3680098056793213, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.2603, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005874012558233745, | |
| "grad_norm": 3.324713945388794, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.5608, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.005975288636823982, | |
| "grad_norm": 3.841120481491089, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.8611, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006076564715414219, | |
| "grad_norm": 4.1734185218811035, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.2571, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.006177840794004456, | |
| "grad_norm": 3.5894811153411865, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.0733, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006279116872594693, | |
| "grad_norm": 5.320555210113525, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.1953, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.00638039295118493, | |
| "grad_norm": 5.043003082275391, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.1595, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006481669029775167, | |
| "grad_norm": 3.361215114593506, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.0175, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.006582945108365404, | |
| "grad_norm": 3.8890812397003174, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.1893, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006684221186955641, | |
| "grad_norm": 2.517195701599121, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.8345, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006785497265545878, | |
| "grad_norm": 6.489955425262451, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.1292, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006886773344136115, | |
| "grad_norm": 7.068796157836914, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.0766, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006988049422726352, | |
| "grad_norm": 4.267070293426514, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.3931, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007089325501316589, | |
| "grad_norm": 3.7103750705718994, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.3188, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.007190601579906826, | |
| "grad_norm": 5.24364709854126, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.616, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007291877658497063, | |
| "grad_norm": 5.233373641967773, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.3433, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0073931537370873, | |
| "grad_norm": 4.6338043212890625, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.5683, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.007494429815677537, | |
| "grad_norm": 8.574564933776855, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.4453, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.007595705894267774, | |
| "grad_norm": 4.522185802459717, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.5616, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007696981972858011, | |
| "grad_norm": 4.150303840637207, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.8283, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.007798258051448248, | |
| "grad_norm": 3.7501208782196045, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.3082, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007899534130038485, | |
| "grad_norm": 3.435241222381592, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.3736, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.008000810208628722, | |
| "grad_norm": 4.394081115722656, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.3067, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.008102086287218958, | |
| "grad_norm": 3.895310640335083, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.7659, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.008203362365809196, | |
| "grad_norm": 3.7235097885131836, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.9423, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008304638444399434, | |
| "grad_norm": 3.232382297515869, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.7102, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.00840591452298967, | |
| "grad_norm": 3.1203420162200928, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.5404, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008507190601579907, | |
| "grad_norm": 2.432556629180908, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.3299, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008608466680170145, | |
| "grad_norm": 2.7433419227600098, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.4803, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.00870974275876038, | |
| "grad_norm": 3.822559356689453, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.8033, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008811018837350618, | |
| "grad_norm": 2.104219436645508, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.4898, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.008912294915940856, | |
| "grad_norm": 2.9761338233947754, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.903, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.009013570994531091, | |
| "grad_norm": 4.711953163146973, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.3835, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.009114847073121329, | |
| "grad_norm": 5.131633758544922, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.4729, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.009216123151711565, | |
| "grad_norm": 5.405108451843262, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.5969, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.009317399230301802, | |
| "grad_norm": 4.3136467933654785, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.3014, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.00941867530889204, | |
| "grad_norm": 3.89194393157959, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.5294, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.009519951387482276, | |
| "grad_norm": 3.0902202129364014, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.9792, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.009621227466072513, | |
| "grad_norm": 3.7873964309692383, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.3636, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009722503544662751, | |
| "grad_norm": 3.423661708831787, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.1961, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009823779623252987, | |
| "grad_norm": 4.137724876403809, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.3676, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009925055701843224, | |
| "grad_norm": 3.5316386222839355, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.4104, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.010026331780433462, | |
| "grad_norm": 6.5205254554748535, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.7601, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.010127607859023698, | |
| "grad_norm": 4.362370014190674, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.3275, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.010127607859023698, | |
| "eval_loss": 2.2500593662261963, | |
| "eval_runtime": 1575.4029, | |
| "eval_samples_per_second": 10.556, | |
| "eval_steps_per_second": 2.639, | |
| "step": 100 | |
| } | |
| ], | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.727221903392768e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |