| { | |
| "best_metric": 1.4484697580337524, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.1049317943336831, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005246589716684155, | |
| "grad_norm": 1.1576203107833862, | |
| "learning_rate": 1e-05, | |
| "loss": 3.0341, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005246589716684155, | |
| "eval_loss": 5.070003986358643, | |
| "eval_runtime": 225.3527, | |
| "eval_samples_per_second": 14.244, | |
| "eval_steps_per_second": 3.563, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001049317943336831, | |
| "grad_norm": 1.779099702835083, | |
| "learning_rate": 2e-05, | |
| "loss": 3.6263, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0015739769150052466, | |
| "grad_norm": 1.921130657196045, | |
| "learning_rate": 3e-05, | |
| "loss": 3.8827, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002098635886673662, | |
| "grad_norm": 2.182387351989746, | |
| "learning_rate": 4e-05, | |
| "loss": 3.6948, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0026232948583420775, | |
| "grad_norm": 2.538600444793701, | |
| "learning_rate": 5e-05, | |
| "loss": 3.8578, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0031479538300104933, | |
| "grad_norm": 2.443502902984619, | |
| "learning_rate": 6e-05, | |
| "loss": 3.3757, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0036726128016789086, | |
| "grad_norm": 1.8905384540557861, | |
| "learning_rate": 7e-05, | |
| "loss": 3.4544, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004197271773347324, | |
| "grad_norm": 2.2534008026123047, | |
| "learning_rate": 8e-05, | |
| "loss": 3.1179, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00472193074501574, | |
| "grad_norm": 2.419541835784912, | |
| "learning_rate": 9e-05, | |
| "loss": 2.72, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.005246589716684155, | |
| "grad_norm": 2.5247607231140137, | |
| "learning_rate": 0.0001, | |
| "loss": 2.6931, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005771248688352571, | |
| "grad_norm": 4.06483268737793, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.2914, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0062959076600209865, | |
| "grad_norm": 2.091320753097534, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.8288, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.006820566631689402, | |
| "grad_norm": 2.039874315261841, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.7895, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.007345225603357817, | |
| "grad_norm": 1.7094388008117676, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.7866, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.007869884575026232, | |
| "grad_norm": 1.3983663320541382, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.6987, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.008394543546694649, | |
| "grad_norm": 2.4380130767822266, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.5988, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.008919202518363064, | |
| "grad_norm": 1.3538730144500732, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.7902, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.00944386149003148, | |
| "grad_norm": 1.3163968324661255, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.6041, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.009968520461699895, | |
| "grad_norm": 1.4796921014785767, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.6951, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01049317943336831, | |
| "grad_norm": 1.0245383977890015, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.5077, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011017838405036727, | |
| "grad_norm": 1.2236641645431519, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.8279, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.011542497376705142, | |
| "grad_norm": 1.142067313194275, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.642, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.012067156348373556, | |
| "grad_norm": 0.9749924540519714, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.7097, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.012591815320041973, | |
| "grad_norm": 1.0015535354614258, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.431, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.013116474291710388, | |
| "grad_norm": 0.9406030774116516, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.5726, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.013641133263378805, | |
| "grad_norm": 1.0338419675827026, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.5226, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01416579223504722, | |
| "grad_norm": 1.0772449970245361, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.7288, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.014690451206715634, | |
| "grad_norm": 1.0847026109695435, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.4969, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.015215110178384051, | |
| "grad_norm": 0.8713697195053101, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.3882, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.015739769150052464, | |
| "grad_norm": 1.143677830696106, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.53, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01626442812172088, | |
| "grad_norm": 1.400123119354248, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.6644, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.016789087093389297, | |
| "grad_norm": 0.9827703833580017, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.6692, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.017313746065057714, | |
| "grad_norm": 1.1392353773117065, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.6507, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.017838405036726127, | |
| "grad_norm": 1.22384512424469, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.7154, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.018363064008394544, | |
| "grad_norm": 0.9324412941932678, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.4078, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01888772298006296, | |
| "grad_norm": 1.0731570720672607, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.6842, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.019412381951731374, | |
| "grad_norm": 0.8245493769645691, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.1563, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.01993704092339979, | |
| "grad_norm": 0.8980649709701538, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.5008, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.020461699895068207, | |
| "grad_norm": 0.864634096622467, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.2187, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02098635886673662, | |
| "grad_norm": 0.9761171340942383, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.381, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.021511017838405037, | |
| "grad_norm": 1.0606969594955444, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.4, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.022035676810073453, | |
| "grad_norm": 1.2640045881271362, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.6077, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.022560335781741866, | |
| "grad_norm": 1.207768440246582, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.5555, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.023084994753410283, | |
| "grad_norm": 1.0950582027435303, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.2568, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0236096537250787, | |
| "grad_norm": 1.3536014556884766, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.2712, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.024134312696747113, | |
| "grad_norm": 1.1610876321792603, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.4622, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02465897166841553, | |
| "grad_norm": 0.9861035346984863, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.1073, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.025183630640083946, | |
| "grad_norm": 1.32306969165802, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.2821, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02570828961175236, | |
| "grad_norm": 1.0001379251480103, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.1953, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.026232948583420776, | |
| "grad_norm": 1.2933019399642944, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.1879, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.026232948583420776, | |
| "eval_loss": 1.6822460889816284, | |
| "eval_runtime": 226.9494, | |
| "eval_samples_per_second": 14.144, | |
| "eval_steps_per_second": 3.538, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.026757607555089193, | |
| "grad_norm": 4.4354071617126465, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.2723, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02728226652675761, | |
| "grad_norm": 2.7067105770111084, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.0828, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.027806925498426022, | |
| "grad_norm": 1.3014180660247803, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.6701, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02833158447009444, | |
| "grad_norm": 0.8862189054489136, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.5842, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.028856243441762856, | |
| "grad_norm": 1.0087246894836426, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.9389, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02938090241343127, | |
| "grad_norm": 1.0576592683792114, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.888, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.029905561385099685, | |
| "grad_norm": 0.9109050035476685, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.6875, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.030430220356768102, | |
| "grad_norm": 0.9095304608345032, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.6394, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.030954879328436515, | |
| "grad_norm": 0.8381404280662537, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.568, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03147953830010493, | |
| "grad_norm": 0.8554638624191284, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.7206, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.032004197271773345, | |
| "grad_norm": 0.8155702352523804, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.4357, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03252885624344176, | |
| "grad_norm": 0.8470095992088318, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.7245, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03305351521511018, | |
| "grad_norm": 0.9037229418754578, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.564, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.033578174186778595, | |
| "grad_norm": 1.0658760070800781, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.4438, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03410283315844701, | |
| "grad_norm": 1.0728156566619873, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.546, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03462749213011543, | |
| "grad_norm": 0.8077057600021362, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.6528, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03515215110178384, | |
| "grad_norm": 0.6561137437820435, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.5236, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.035676810073452254, | |
| "grad_norm": 0.8210721611976624, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.5082, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03620146904512067, | |
| "grad_norm": 0.8699805736541748, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.5826, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03672612801678909, | |
| "grad_norm": 0.8109676241874695, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.5513, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.037250786988457504, | |
| "grad_norm": 0.8149637579917908, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.4598, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03777544596012592, | |
| "grad_norm": 0.6916829347610474, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.5235, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03830010493179433, | |
| "grad_norm": 0.7610656023025513, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.6679, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03882476390346275, | |
| "grad_norm": 0.7549808621406555, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.3666, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.039349422875131164, | |
| "grad_norm": 0.9101117253303528, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.4156, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03987408184679958, | |
| "grad_norm": 0.8551775813102722, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.6248, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.040398740818468, | |
| "grad_norm": 0.6739151477813721, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.3195, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.040923399790136414, | |
| "grad_norm": 0.8481286764144897, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.4729, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.041448058761804824, | |
| "grad_norm": 0.7074611186981201, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.3106, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.04197271773347324, | |
| "grad_norm": 0.7907785773277283, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.6031, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04249737670514166, | |
| "grad_norm": 0.7411637902259827, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.3836, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04302203567681007, | |
| "grad_norm": 0.9525800347328186, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.6793, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04354669464847849, | |
| "grad_norm": 0.7306334972381592, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.2289, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04407135362014691, | |
| "grad_norm": 0.8034957051277161, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.2427, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04459601259181532, | |
| "grad_norm": 0.7753270268440247, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.4434, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04512067156348373, | |
| "grad_norm": 0.7781442999839783, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.3179, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04564533053515215, | |
| "grad_norm": 0.8402987122535706, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.3967, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.046169989506820566, | |
| "grad_norm": 1.0798988342285156, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.3569, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04669464847848898, | |
| "grad_norm": 0.9086529612541199, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.4316, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0472193074501574, | |
| "grad_norm": 0.9447148442268372, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.5306, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.047743966421825816, | |
| "grad_norm": 0.8512058258056641, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.3329, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.048268625393494226, | |
| "grad_norm": 0.8439379334449768, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.4061, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.04879328436516264, | |
| "grad_norm": 0.9092354774475098, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.4674, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04931794333683106, | |
| "grad_norm": 0.7518009543418884, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.2305, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.049842602308499476, | |
| "grad_norm": 0.9742012023925781, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.2046, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05036726128016789, | |
| "grad_norm": 0.8009738922119141, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.9592, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05089192025183631, | |
| "grad_norm": 1.0046333074569702, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.2396, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05141657922350472, | |
| "grad_norm": 1.0736814737319946, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.2732, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.051941238195173135, | |
| "grad_norm": 1.0017805099487305, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.0418, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.05246589716684155, | |
| "grad_norm": 1.1624499559402466, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.3055, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05246589716684155, | |
| "eval_loss": 1.5854398012161255, | |
| "eval_runtime": 226.8069, | |
| "eval_samples_per_second": 14.153, | |
| "eval_steps_per_second": 3.54, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05299055613850997, | |
| "grad_norm": 3.0951433181762695, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.053, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.053515215110178385, | |
| "grad_norm": 2.6890578269958496, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.9489, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0540398740818468, | |
| "grad_norm": 1.9763928651809692, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.072, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.05456453305351522, | |
| "grad_norm": 1.208033561706543, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.4995, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.05508919202518363, | |
| "grad_norm": 0.8731113076210022, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6535, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.055613850996852045, | |
| "grad_norm": 0.77850341796875, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.7701, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.05613850996852046, | |
| "grad_norm": 0.6573787331581116, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.4544, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.05666316894018888, | |
| "grad_norm": 0.8050903677940369, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.554, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.057187827911857295, | |
| "grad_norm": 0.7848634123802185, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.6746, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.05771248688352571, | |
| "grad_norm": 0.8570258021354675, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.5224, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05823714585519412, | |
| "grad_norm": 0.8733482360839844, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.505, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.05876180482686254, | |
| "grad_norm": 0.8403432369232178, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.5298, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.059286463798530954, | |
| "grad_norm": 0.9690472483634949, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.7143, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.05981112277019937, | |
| "grad_norm": 0.8156488537788391, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.4508, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06033578174186779, | |
| "grad_norm": 0.7401934266090393, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.5275, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.060860440713536204, | |
| "grad_norm": 0.7287182211875916, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.4969, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.061385099685204614, | |
| "grad_norm": 0.7605511546134949, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.4997, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06190975865687303, | |
| "grad_norm": 0.7882239818572998, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.646, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06243441762854145, | |
| "grad_norm": 0.7539576292037964, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.6236, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06295907660020986, | |
| "grad_norm": 0.7397094964981079, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.4905, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06348373557187828, | |
| "grad_norm": 0.6339027881622314, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.2408, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06400839454354669, | |
| "grad_norm": 0.8364681005477905, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.5084, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.06453305351521511, | |
| "grad_norm": 0.8842049837112427, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.5381, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.06505771248688352, | |
| "grad_norm": 0.766512393951416, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.546, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.06558237145855195, | |
| "grad_norm": 0.980896532535553, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.6172, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06610703043022036, | |
| "grad_norm": 0.8483807444572449, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.4834, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.06663168940188877, | |
| "grad_norm": 0.743341863155365, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.5558, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.06715634837355719, | |
| "grad_norm": 0.7407636642456055, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.4756, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0676810073452256, | |
| "grad_norm": 0.7860780954360962, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.5512, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.06820566631689402, | |
| "grad_norm": 0.752352237701416, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.376, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.06873032528856243, | |
| "grad_norm": 0.7883614301681519, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.533, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.06925498426023086, | |
| "grad_norm": 0.9203779101371765, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.4226, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.06977964323189927, | |
| "grad_norm": 0.873372495174408, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.6358, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07030430220356768, | |
| "grad_norm": 0.7549769282341003, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.43, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0708289611752361, | |
| "grad_norm": 0.7276805639266968, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.3683, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07135362014690451, | |
| "grad_norm": 0.7437852025032043, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.5458, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07187827911857293, | |
| "grad_norm": 0.8723979592323303, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.4498, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.07240293809024134, | |
| "grad_norm": 0.7789756655693054, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.5523, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.07292759706190975, | |
| "grad_norm": 0.7481866478919983, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.3149, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07345225603357818, | |
| "grad_norm": 0.8311874866485596, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.5739, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07397691500524659, | |
| "grad_norm": 0.7663030028343201, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.2044, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.07450157397691501, | |
| "grad_norm": 0.862217903137207, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.1892, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.07502623294858342, | |
| "grad_norm": 0.7785995006561279, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.3858, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.07555089192025184, | |
| "grad_norm": 0.7804452776908875, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.2495, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.07607555089192025, | |
| "grad_norm": 0.746898353099823, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.0519, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.07660020986358866, | |
| "grad_norm": 0.7804585099220276, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.2821, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.07712486883525708, | |
| "grad_norm": 0.9656455516815186, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.4167, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.0776495278069255, | |
| "grad_norm": 1.0540097951889038, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.2876, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.07817418677859392, | |
| "grad_norm": 1.1790372133255005, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.456, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.07869884575026233, | |
| "grad_norm": 1.4881292581558228, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.3363, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07869884575026233, | |
| "eval_loss": 1.4686623811721802, | |
| "eval_runtime": 226.8194, | |
| "eval_samples_per_second": 14.152, | |
| "eval_steps_per_second": 3.54, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07922350472193075, | |
| "grad_norm": 1.1341716051101685, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.663, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.07974816369359916, | |
| "grad_norm": 1.4673205614089966, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.8159, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08027282266526757, | |
| "grad_norm": 1.1706839799880981, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.7139, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.080797481636936, | |
| "grad_norm": 1.2127618789672852, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.5428, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0813221406086044, | |
| "grad_norm": 1.1465585231781006, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.7721, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08184679958027283, | |
| "grad_norm": 0.8733444213867188, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.6241, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08237145855194124, | |
| "grad_norm": 0.9330646395683289, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.6974, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08289611752360965, | |
| "grad_norm": 0.7993752360343933, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.5538, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08342077649527807, | |
| "grad_norm": 0.6594255566596985, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.408, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08394543546694648, | |
| "grad_norm": 0.6929203867912292, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.6504, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0844700944386149, | |
| "grad_norm": 0.76050865650177, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.7236, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.08499475341028331, | |
| "grad_norm": 0.6506959795951843, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.4091, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.08551941238195174, | |
| "grad_norm": 0.6918977499008179, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.557, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.08604407135362015, | |
| "grad_norm": 0.5875673294067383, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.3463, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.08656873032528856, | |
| "grad_norm": 0.7600943446159363, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.7546, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.08709338929695698, | |
| "grad_norm": 0.7193366289138794, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.5257, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.08761804826862539, | |
| "grad_norm": 0.6928470730781555, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.2352, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.08814270724029381, | |
| "grad_norm": 0.780381977558136, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.5033, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.08866736621196222, | |
| "grad_norm": 0.7388539910316467, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.516, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.08919202518363065, | |
| "grad_norm": 0.7183394432067871, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.3339, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.08971668415529906, | |
| "grad_norm": 0.8241344094276428, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.5293, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09024134312696747, | |
| "grad_norm": 0.7911620140075684, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.3069, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.09076600209863589, | |
| "grad_norm": 0.7571455240249634, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.5287, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.0912906610703043, | |
| "grad_norm": 0.9246066212654114, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.4598, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.09181532004197272, | |
| "grad_norm": 0.9424697160720825, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.6791, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09233997901364113, | |
| "grad_norm": 0.6656576991081238, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.2681, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.09286463798530954, | |
| "grad_norm": 0.8066405057907104, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.383, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.09338929695697797, | |
| "grad_norm": 0.9089919328689575, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.5856, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.09391395592864638, | |
| "grad_norm": 0.817192554473877, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.4945, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.0944386149003148, | |
| "grad_norm": 0.7961598634719849, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.4717, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09496327387198321, | |
| "grad_norm": 0.8352245688438416, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.4472, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.09548793284365163, | |
| "grad_norm": 0.900001585483551, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.4121, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.09601259181532004, | |
| "grad_norm": 0.8115286231040955, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.4419, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.09653725078698845, | |
| "grad_norm": 0.893621027469635, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.406, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.09706190975865688, | |
| "grad_norm": 0.8589741587638855, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.5105, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.09758656873032528, | |
| "grad_norm": 0.9568005204200745, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.3697, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.09811122770199371, | |
| "grad_norm": 1.0956547260284424, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.6071, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.09863588667366212, | |
| "grad_norm": 0.9156211018562317, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.4872, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.09916054564533054, | |
| "grad_norm": 0.9828962087631226, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.3937, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.09968520461699895, | |
| "grad_norm": 0.892189085483551, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.268, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10020986358866736, | |
| "grad_norm": 0.996194064617157, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.143, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.10073452256033578, | |
| "grad_norm": 1.1560606956481934, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.2767, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1012591815320042, | |
| "grad_norm": 1.2013170719146729, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.3289, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.10178384050367262, | |
| "grad_norm": 1.081543207168579, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.2053, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10230849947534103, | |
| "grad_norm": 1.294790506362915, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.484, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.10283315844700944, | |
| "grad_norm": 1.0419007539749146, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.219, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.10335781741867786, | |
| "grad_norm": 1.194427728652954, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.362, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.10388247639034627, | |
| "grad_norm": 1.2858350276947021, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.411, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.1044071353620147, | |
| "grad_norm": 1.369338870048523, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.1942, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1049317943336831, | |
| "grad_norm": 1.5181728601455688, | |
| "learning_rate": 0.0, | |
| "loss": 1.322, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1049317943336831, | |
| "eval_loss": 1.4484697580337524, | |
| "eval_runtime": 226.8848, | |
| "eval_samples_per_second": 14.148, | |
| "eval_steps_per_second": 3.539, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.684081292705792e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |