| { | |
| "best_metric": 1.3709410429000854, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.020255215718047396, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010127607859023699, | |
| "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 | |
| }, | |
| { | |
| "epoch": 0.010228883937613936, | |
| "grad_norm": 9.265037536621094, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.6577, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010330160016204173, | |
| "grad_norm": 8.436511039733887, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.749, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.010431436094794409, | |
| "grad_norm": 6.239524841308594, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.5332, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.010532712173384647, | |
| "grad_norm": 6.823162078857422, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.6346, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.010633988251974884, | |
| "grad_norm": 4.563138484954834, | |
| "learning_rate": 5e-05, | |
| "loss": 1.117, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.01073526433056512, | |
| "grad_norm": 4.2513203620910645, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.1082, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.010836540409155358, | |
| "grad_norm": 3.659548044204712, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.7426, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.010937816487745595, | |
| "grad_norm": 3.3887014389038086, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.9441, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.011039092566335831, | |
| "grad_norm": 5.242551803588867, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.1627, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.011140368644926069, | |
| "grad_norm": 5.448172092437744, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.9376, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.011241644723516306, | |
| "grad_norm": 4.152783393859863, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.9953, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.011342920802106542, | |
| "grad_norm": 2.2658233642578125, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.9487, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.01144419688069678, | |
| "grad_norm": 3.5433804988861084, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.1215, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.011545472959287017, | |
| "grad_norm": 3.221561908721924, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.0136, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.011646749037877253, | |
| "grad_norm": 3.5430595874786377, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.9319, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.01174802511646749, | |
| "grad_norm": 4.0749101638793945, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.9736, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.011849301195057726, | |
| "grad_norm": 3.0245511531829834, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.9719, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.011950577273647964, | |
| "grad_norm": 2.82936429977417, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.1158, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.012051853352238202, | |
| "grad_norm": 7.250178813934326, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0967, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.012153129430828438, | |
| "grad_norm": 6.545557975769043, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.8107, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.012254405509418675, | |
| "grad_norm": 4.1475043296813965, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.35, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.012355681588008913, | |
| "grad_norm": 5.050443649291992, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.4729, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.012456957666599149, | |
| "grad_norm": 3.237460136413574, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.4774, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.012558233745189386, | |
| "grad_norm": 4.389225006103516, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.5459, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.012659509823779624, | |
| "grad_norm": 3.2173123359680176, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.8471, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01276078590236986, | |
| "grad_norm": 3.876474380493164, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.3983, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.012862061980960097, | |
| "grad_norm": 4.244775772094727, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.6689, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012963338059550335, | |
| "grad_norm": 2.8168833255767822, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.0456, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.01306461413814057, | |
| "grad_norm": 2.9947640895843506, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.3599, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.013165890216730808, | |
| "grad_norm": 3.089524984359741, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.2929, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.013267166295321046, | |
| "grad_norm": 4.560943126678467, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.6056, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.013368442373911282, | |
| "grad_norm": 3.317545175552368, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.356, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.01346971845250152, | |
| "grad_norm": 2.595907688140869, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.4036, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.013570994531091757, | |
| "grad_norm": 2.746410369873047, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.5982, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.013672270609681993, | |
| "grad_norm": 2.330042839050293, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.4625, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.01377354668827223, | |
| "grad_norm": 2.7525479793548584, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.3657, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.013874822766862468, | |
| "grad_norm": 2.914900064468384, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.6761, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.013976098845452704, | |
| "grad_norm": 2.1483852863311768, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.4883, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.014077374924042941, | |
| "grad_norm": 3.72390079498291, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.1337, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.014178651002633179, | |
| "grad_norm": 4.195054531097412, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.4717, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.014279927081223415, | |
| "grad_norm": 3.6533384323120117, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.2632, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.014381203159813652, | |
| "grad_norm": 4.587025165557861, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.372, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.014482479238403888, | |
| "grad_norm": 3.2736153602600098, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.1278, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.014583755316994126, | |
| "grad_norm": 3.9641976356506348, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.3325, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.014685031395584363, | |
| "grad_norm": 3.7714686393737793, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.3855, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.0147863074741746, | |
| "grad_norm": 3.3997910022735596, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.8507, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.014887583552764837, | |
| "grad_norm": 3.5149173736572266, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.2801, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.014988859631355074, | |
| "grad_norm": 3.5408782958984375, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.1762, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01509013570994531, | |
| "grad_norm": 3.24328351020813, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.1446, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.015191411788535548, | |
| "grad_norm": 3.8849191665649414, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.2938, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015191411788535548, | |
| "eval_loss": 1.4395376443862915, | |
| "eval_runtime": 1577.1875, | |
| "eval_samples_per_second": 10.544, | |
| "eval_steps_per_second": 2.636, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.015292687867125785, | |
| "grad_norm": 3.577855348587036, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.9834, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.015393963945716021, | |
| "grad_norm": 3.4904537200927734, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.12, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.015495240024306259, | |
| "grad_norm": 2.1824731826782227, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.6947, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.015596516102896496, | |
| "grad_norm": 2.6002416610717773, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.9035, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.015697792181486734, | |
| "grad_norm": 2.642005205154419, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.9847, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.01579906826007697, | |
| "grad_norm": 2.108966588973999, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.0158, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.015900344338667206, | |
| "grad_norm": 2.7388312816619873, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.6122, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.016001620417257445, | |
| "grad_norm": 2.437753677368164, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.7889, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.01610289649584768, | |
| "grad_norm": 2.895228385925293, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.8055, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.016204172574437917, | |
| "grad_norm": 3.3598122596740723, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.6516, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.016305448653028156, | |
| "grad_norm": 5.975849628448486, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.2587, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.016406724731618392, | |
| "grad_norm": 2.5624125003814697, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.0677, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.016508000810208628, | |
| "grad_norm": 2.352781057357788, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.8555, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.016609276888798867, | |
| "grad_norm": 3.3142495155334473, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.9869, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.016710552967389103, | |
| "grad_norm": 2.431610107421875, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.7387, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.01681182904597934, | |
| "grad_norm": 2.484178304672241, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.9848, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.016913105124569578, | |
| "grad_norm": 4.131918907165527, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.927, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.017014381203159814, | |
| "grad_norm": 2.9562125205993652, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.2482, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01711565728175005, | |
| "grad_norm": 3.060600757598877, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.3198, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.01721693336034029, | |
| "grad_norm": 3.3929269313812256, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.3687, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.017318209438930525, | |
| "grad_norm": 3.2837131023406982, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.6001, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.01741948551752076, | |
| "grad_norm": 3.3161513805389404, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.2154, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.017520761596111, | |
| "grad_norm": 2.681596040725708, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.1706, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.017622037674701236, | |
| "grad_norm": 3.3178539276123047, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.5169, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.017723313753291472, | |
| "grad_norm": 3.5398411750793457, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.1858, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01782458983188171, | |
| "grad_norm": 3.443697214126587, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.5175, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.017925865910471947, | |
| "grad_norm": 2.9091107845306396, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.1557, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.018027141989062183, | |
| "grad_norm": 2.7438173294067383, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.1894, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.018128418067652422, | |
| "grad_norm": 3.149282693862915, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.133, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.018229694146242658, | |
| "grad_norm": 3.0966949462890625, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.4089, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.018330970224832894, | |
| "grad_norm": 2.108625888824463, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.2586, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.01843224630342313, | |
| "grad_norm": 1.9376676082611084, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.2564, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.01853352238201337, | |
| "grad_norm": 2.23464298248291, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.1772, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.018634798460603605, | |
| "grad_norm": 2.237391471862793, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.3342, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.01873607453919384, | |
| "grad_norm": 2.1162657737731934, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.1758, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01883735061778408, | |
| "grad_norm": 3.177123546600342, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.8013, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.018938626696374316, | |
| "grad_norm": 2.27571439743042, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.159, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.019039902774964552, | |
| "grad_norm": 2.396038293838501, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.4511, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01914117885355479, | |
| "grad_norm": 4.168811798095703, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.1186, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.019242454932145027, | |
| "grad_norm": 3.6993422508239746, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 2.1521, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.019343731010735263, | |
| "grad_norm": 3.8161509037017822, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.1989, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.019445007089325502, | |
| "grad_norm": 4.181246757507324, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.4176, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.019546283167915738, | |
| "grad_norm": 3.70147705078125, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.717, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.019647559246505974, | |
| "grad_norm": 3.016972541809082, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.8479, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.019748835325096213, | |
| "grad_norm": 3.836817502975464, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.9622, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.01985011140368645, | |
| "grad_norm": 3.564082622528076, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.1143, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.019951387482276685, | |
| "grad_norm": 4.26180362701416, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.1449, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.020052663560866924, | |
| "grad_norm": 3.6290478706359863, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.1276, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.02015393963945716, | |
| "grad_norm": 3.782374858856201, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.2057, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.020255215718047396, | |
| "grad_norm": 4.92399787902832, | |
| "learning_rate": 0.0, | |
| "loss": 2.3031, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.020255215718047396, | |
| "eval_loss": 1.3709410429000854, | |
| "eval_runtime": 1576.0099, | |
| "eval_samples_per_second": 10.552, | |
| "eval_steps_per_second": 2.638, | |
| "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": 3.454443806785536e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |