| { | |
| "best_metric": 1.83381187915802, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.2658227848101267, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006329113924050633, | |
| "grad_norm": 3.68753719329834, | |
| "learning_rate": 1e-05, | |
| "loss": 5.2265, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006329113924050633, | |
| "eval_loss": 5.407243251800537, | |
| "eval_runtime": 7.8684, | |
| "eval_samples_per_second": 33.933, | |
| "eval_steps_per_second": 8.515, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.012658227848101266, | |
| "grad_norm": 3.5360615253448486, | |
| "learning_rate": 2e-05, | |
| "loss": 5.0398, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0189873417721519, | |
| "grad_norm": 3.8196494579315186, | |
| "learning_rate": 3e-05, | |
| "loss": 5.1714, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02531645569620253, | |
| "grad_norm": 3.5815281867980957, | |
| "learning_rate": 4e-05, | |
| "loss": 5.3216, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03164556962025317, | |
| "grad_norm": 3.5993247032165527, | |
| "learning_rate": 5e-05, | |
| "loss": 5.0941, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0379746835443038, | |
| "grad_norm": 3.444089651107788, | |
| "learning_rate": 6e-05, | |
| "loss": 4.8125, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04430379746835443, | |
| "grad_norm": 3.757082462310791, | |
| "learning_rate": 7e-05, | |
| "loss": 4.847, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.05063291139240506, | |
| "grad_norm": 2.973097562789917, | |
| "learning_rate": 8e-05, | |
| "loss": 4.4364, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.056962025316455694, | |
| "grad_norm": 2.9682071208953857, | |
| "learning_rate": 9e-05, | |
| "loss": 3.9288, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06329113924050633, | |
| "grad_norm": 3.043100118637085, | |
| "learning_rate": 0.0001, | |
| "loss": 3.6066, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06962025316455696, | |
| "grad_norm": 2.7873315811157227, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 3.5754, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0759493670886076, | |
| "grad_norm": 2.785930871963501, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 3.603, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.08227848101265822, | |
| "grad_norm": 3.0582737922668457, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 3.5674, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.08860759493670886, | |
| "grad_norm": 2.8924787044525146, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 3.382, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0949367088607595, | |
| "grad_norm": 2.608238697052002, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 3.1447, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 3.4497995376586914, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 3.3686, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.10759493670886076, | |
| "grad_norm": 2.796494722366333, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 3.2062, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.11392405063291139, | |
| "grad_norm": 3.322106122970581, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 3.1692, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.12025316455696203, | |
| "grad_norm": 2.33091402053833, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.7872, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.12658227848101267, | |
| "grad_norm": 2.7159955501556396, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.8737, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13291139240506328, | |
| "grad_norm": 2.792219400405884, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.7047, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.13924050632911392, | |
| "grad_norm": 2.682821750640869, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.9038, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.14556962025316456, | |
| "grad_norm": 2.6452414989471436, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 3.2227, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1518987341772152, | |
| "grad_norm": 2.768331527709961, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.9005, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.15822784810126583, | |
| "grad_norm": 3.05155348777771, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.8132, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.16455696202531644, | |
| "grad_norm": 2.6093695163726807, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.5272, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.17088607594936708, | |
| "grad_norm": 2.4727351665496826, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.6387, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.17721518987341772, | |
| "grad_norm": 3.1327879428863525, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.7982, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.18354430379746836, | |
| "grad_norm": 3.4934334754943848, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.4697, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.189873417721519, | |
| "grad_norm": 2.647977590560913, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.7398, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1962025316455696, | |
| "grad_norm": 2.445222854614258, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.1912, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 2.598055601119995, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.4392, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.2088607594936709, | |
| "grad_norm": 2.435616970062256, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.5867, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.21518987341772153, | |
| "grad_norm": 2.556702136993408, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.5121, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.22151898734177214, | |
| "grad_norm": 2.765544891357422, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.3471, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.22784810126582278, | |
| "grad_norm": 2.489150047302246, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.409, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.23417721518987342, | |
| "grad_norm": 2.6875150203704834, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.2156, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.24050632911392406, | |
| "grad_norm": 3.1702370643615723, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.366, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.2468354430379747, | |
| "grad_norm": 3.7439520359039307, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.1476, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 3.1675446033477783, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 3.7048, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.25949367088607594, | |
| "grad_norm": 2.9399471282958984, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 3.2541, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.26582278481012656, | |
| "grad_norm": 2.686516523361206, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 3.275, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2721518987341772, | |
| "grad_norm": 2.2668392658233643, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 3.0628, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.27848101265822783, | |
| "grad_norm": 2.3172688484191895, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.9878, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2848101265822785, | |
| "grad_norm": 2.148198127746582, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.9089, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.2911392405063291, | |
| "grad_norm": 2.2290523052215576, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.8187, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2974683544303797, | |
| "grad_norm": 1.8086093664169312, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.6843, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 1.6970168352127075, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.3247, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.310126582278481, | |
| "grad_norm": 1.5677433013916016, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.1573, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.31645569620253167, | |
| "grad_norm": 2.4579403400421143, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.9982, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.31645569620253167, | |
| "eval_loss": 2.3056161403656006, | |
| "eval_runtime": 7.8863, | |
| "eval_samples_per_second": 33.856, | |
| "eval_steps_per_second": 8.496, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3227848101265823, | |
| "grad_norm": 1.786394476890564, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.964, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.3291139240506329, | |
| "grad_norm": 1.8627042770385742, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.0741, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.33544303797468356, | |
| "grad_norm": 1.8683773279190063, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.9816, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.34177215189873417, | |
| "grad_norm": 1.6443946361541748, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.1488, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.34810126582278483, | |
| "grad_norm": 2.2271270751953125, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.2969, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.35443037974683544, | |
| "grad_norm": 2.036264419555664, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.2168, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.36075949367088606, | |
| "grad_norm": 1.8334553241729736, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.033, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3670886075949367, | |
| "grad_norm": 1.6956928968429565, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.9885, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.37341772151898733, | |
| "grad_norm": 2.1191928386688232, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.0429, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.379746835443038, | |
| "grad_norm": 1.7328978776931763, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.7138, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3860759493670886, | |
| "grad_norm": 1.8068017959594727, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.9531, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3924050632911392, | |
| "grad_norm": 1.7362067699432373, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 2.095, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.3987341772151899, | |
| "grad_norm": 1.7228310108184814, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 2.0825, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 1.9382199048995972, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.8795, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.41139240506329117, | |
| "grad_norm": 1.9357199668884277, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.795, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.4177215189873418, | |
| "grad_norm": 1.7983970642089844, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 2.1779, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.4240506329113924, | |
| "grad_norm": 2.0430989265441895, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 2.1682, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.43037974683544306, | |
| "grad_norm": 2.100715398788452, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 2.0459, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.43670886075949367, | |
| "grad_norm": 2.100064277648926, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.6107, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.4430379746835443, | |
| "grad_norm": 2.187488555908203, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.9462, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.44936708860759494, | |
| "grad_norm": 2.2040765285491943, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.922, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.45569620253164556, | |
| "grad_norm": 2.0525381565093994, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.8621, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.4620253164556962, | |
| "grad_norm": 2.097466468811035, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.7365, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.46835443037974683, | |
| "grad_norm": 2.297074794769287, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.9649, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.47468354430379744, | |
| "grad_norm": 2.3916752338409424, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.8918, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.4810126582278481, | |
| "grad_norm": 2.5592703819274902, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.9877, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.4873417721518987, | |
| "grad_norm": 2.3691787719726562, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.9568, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.4936708860759494, | |
| "grad_norm": 2.7181386947631836, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.9078, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 2.2957725524902344, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.8595, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 2.1818268299102783, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.9839, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5126582278481012, | |
| "grad_norm": 2.0579588413238525, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.8125, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.5189873417721519, | |
| "grad_norm": 1.861379861831665, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.4767, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5253164556962026, | |
| "grad_norm": 1.6677778959274292, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.4184, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5316455696202531, | |
| "grad_norm": 1.9435794353485107, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.3053, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.5379746835443038, | |
| "grad_norm": 1.7356312274932861, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.2776, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.5443037974683544, | |
| "grad_norm": 1.99918532371521, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.111, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5506329113924051, | |
| "grad_norm": 1.6710526943206787, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.7518, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5569620253164557, | |
| "grad_norm": 1.6463580131530762, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.1583, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5632911392405063, | |
| "grad_norm": 1.9608383178710938, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.7136, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.569620253164557, | |
| "grad_norm": 2.19707989692688, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.7322, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5759493670886076, | |
| "grad_norm": 2.0268800258636475, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8753, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.5822784810126582, | |
| "grad_norm": 1.7992669343948364, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.7356, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.5886075949367089, | |
| "grad_norm": 1.769127368927002, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.9601, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5949367088607594, | |
| "grad_norm": 1.8359941244125366, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.7752, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.6012658227848101, | |
| "grad_norm": 1.9174131155014038, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.8491, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 1.7310389280319214, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.6975, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.6139240506329114, | |
| "grad_norm": 1.9668468236923218, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.9186, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.620253164556962, | |
| "grad_norm": 1.953268051147461, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.0458, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.6265822784810127, | |
| "grad_norm": 1.7281947135925293, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.7306, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.6329113924050633, | |
| "grad_norm": 1.6690647602081299, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.6125, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6329113924050633, | |
| "eval_loss": 1.9758051633834839, | |
| "eval_runtime": 7.8851, | |
| "eval_samples_per_second": 33.861, | |
| "eval_steps_per_second": 8.497, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6392405063291139, | |
| "grad_norm": 1.9100043773651123, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.898, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.6455696202531646, | |
| "grad_norm": 2.0958266258239746, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.736, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.6518987341772152, | |
| "grad_norm": 2.159226179122925, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.886, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.6582278481012658, | |
| "grad_norm": 2.5595836639404297, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.0287, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6645569620253164, | |
| "grad_norm": 2.1910741329193115, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9264, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6708860759493671, | |
| "grad_norm": 2.020439863204956, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.6167, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6772151898734177, | |
| "grad_norm": 2.0922250747680664, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.8194, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.6835443037974683, | |
| "grad_norm": 2.085460901260376, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.7819, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.689873417721519, | |
| "grad_norm": 2.136093854904175, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.8623, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.6962025316455697, | |
| "grad_norm": 1.7850594520568848, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.8149, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.7025316455696202, | |
| "grad_norm": 2.205739736557007, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.68, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 2.071253776550293, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.7792, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.7151898734177216, | |
| "grad_norm": 2.2430472373962402, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.9237, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.7215189873417721, | |
| "grad_norm": 2.824376106262207, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 2.1701, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.7278481012658228, | |
| "grad_norm": 2.41506290435791, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.1441, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.7341772151898734, | |
| "grad_norm": 2.6198136806488037, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.8035, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.740506329113924, | |
| "grad_norm": 2.6884541511535645, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.9889, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.7468354430379747, | |
| "grad_norm": 2.3552210330963135, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 2.6257, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.7531645569620253, | |
| "grad_norm": 2.326866626739502, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.6255, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 2.1910130977630615, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.2653, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7658227848101266, | |
| "grad_norm": 1.8745567798614502, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.4079, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.7721518987341772, | |
| "grad_norm": 1.8924429416656494, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 2.4247, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.7784810126582279, | |
| "grad_norm": 1.793290615081787, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 2.165, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7848101265822784, | |
| "grad_norm": 2.060427665710449, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.8751, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7911392405063291, | |
| "grad_norm": 1.9701930284500122, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.8488, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.7974683544303798, | |
| "grad_norm": 1.7540119886398315, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.8847, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.8037974683544303, | |
| "grad_norm": 1.8235610723495483, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.717, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 1.8239582777023315, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.9106, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.8164556962025317, | |
| "grad_norm": 1.6905473470687866, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.7195, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.8227848101265823, | |
| "grad_norm": 1.8119703531265259, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.8231, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.8291139240506329, | |
| "grad_norm": 1.9453668594360352, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.5728, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.8354430379746836, | |
| "grad_norm": 1.88279128074646, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.7686, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.8417721518987342, | |
| "grad_norm": 1.6644700765609741, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.6289, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.8481012658227848, | |
| "grad_norm": 1.8488458395004272, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.0199, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.8544303797468354, | |
| "grad_norm": 1.7529346942901611, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.7583, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.8607594936708861, | |
| "grad_norm": 1.8415391445159912, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7896, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.8670886075949367, | |
| "grad_norm": 1.7497729063034058, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.738, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.8734177215189873, | |
| "grad_norm": 1.7767949104309082, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.6692, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.879746835443038, | |
| "grad_norm": 1.7927212715148926, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.7465, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.8860759493670886, | |
| "grad_norm": 2.12070894241333, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.013, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8924050632911392, | |
| "grad_norm": 1.8234907388687134, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.4714, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8987341772151899, | |
| "grad_norm": 2.12363600730896, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.7577, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.9050632911392406, | |
| "grad_norm": 2.5489907264709473, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.8927, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 2.0148303508758545, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.854, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.9177215189873418, | |
| "grad_norm": 1.8018898963928223, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.6669, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.9240506329113924, | |
| "grad_norm": 1.8505055904388428, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.6263, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.930379746835443, | |
| "grad_norm": 2.1651358604431152, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.8561, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.9367088607594937, | |
| "grad_norm": 2.619361639022827, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.9842, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.9430379746835443, | |
| "grad_norm": 1.9295697212219238, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.8995, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.9493670886075949, | |
| "grad_norm": 2.362227201461792, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.7865, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9493670886075949, | |
| "eval_loss": 1.8573416471481323, | |
| "eval_runtime": 7.8796, | |
| "eval_samples_per_second": 33.885, | |
| "eval_steps_per_second": 8.503, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9556962025316456, | |
| "grad_norm": 2.1369385719299316, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.6373, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.9620253164556962, | |
| "grad_norm": 2.5859482288360596, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.7759, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.9683544303797469, | |
| "grad_norm": 2.2409000396728516, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.8629, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.9746835443037974, | |
| "grad_norm": 2.356802225112915, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.7704, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.9810126582278481, | |
| "grad_norm": 2.301737070083618, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.7523, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.9873417721518988, | |
| "grad_norm": 2.5500082969665527, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.873, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.9936708860759493, | |
| "grad_norm": 1.890016794204712, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.9608, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.1641297340393066, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.4474, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.0063291139240507, | |
| "grad_norm": 1.7158830165863037, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.5425, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.0126582278481013, | |
| "grad_norm": 1.7165212631225586, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.4072, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.018987341772152, | |
| "grad_norm": 1.7016974687576294, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.2928, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.0253164556962024, | |
| "grad_norm": 1.6413666009902954, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.3178, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.0316455696202531, | |
| "grad_norm": 1.6856355667114258, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 2.1955, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.0379746835443038, | |
| "grad_norm": 1.5444437265396118, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.0975, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.0443037974683544, | |
| "grad_norm": 1.6907572746276855, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 2.1622, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 1.621337890625, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.7454, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.0569620253164558, | |
| "grad_norm": 1.605027437210083, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.7541, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.0632911392405062, | |
| "grad_norm": 1.285983681678772, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.4787, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.0696202531645569, | |
| "grad_norm": 1.549367070198059, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.3713, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.0759493670886076, | |
| "grad_norm": 1.3854743242263794, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.3681, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.0822784810126582, | |
| "grad_norm": 1.4103420972824097, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.343, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.0886075949367089, | |
| "grad_norm": 1.6694538593292236, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.518, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.0949367088607596, | |
| "grad_norm": 1.6379914283752441, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.6391, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.1012658227848102, | |
| "grad_norm": 1.6187005043029785, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.643, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.1075949367088607, | |
| "grad_norm": 1.604628086090088, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.5631, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.1139240506329113, | |
| "grad_norm": 1.6321837902069092, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.707, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.120253164556962, | |
| "grad_norm": 1.4925402402877808, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.5169, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.1265822784810127, | |
| "grad_norm": 1.5747805833816528, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.5427, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.1329113924050633, | |
| "grad_norm": 1.5690299272537231, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.3905, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.139240506329114, | |
| "grad_norm": 1.553519606590271, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.6417, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.1455696202531644, | |
| "grad_norm": 1.7300611734390259, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.6432, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 1.7533509731292725, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.8161, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.1582278481012658, | |
| "grad_norm": 1.8410526514053345, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.6242, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.1645569620253164, | |
| "grad_norm": 1.822778344154358, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.5752, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.1708860759493671, | |
| "grad_norm": 1.857142448425293, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.5192, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.1772151898734178, | |
| "grad_norm": 1.7670300006866455, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.5765, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.1835443037974684, | |
| "grad_norm": 1.9805082082748413, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.575, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.189873417721519, | |
| "grad_norm": 1.7736867666244507, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.615, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.1962025316455696, | |
| "grad_norm": 2.0030124187469482, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.6526, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.2025316455696202, | |
| "grad_norm": 1.8017233610153198, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.3771, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.2088607594936709, | |
| "grad_norm": 1.8634445667266846, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.5376, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.2151898734177216, | |
| "grad_norm": 2.1848528385162354, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.5987, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.2215189873417722, | |
| "grad_norm": 2.10420560836792, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.7893, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.2278481012658227, | |
| "grad_norm": 2.0422332286834717, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.7488, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.2341772151898733, | |
| "grad_norm": 2.021265745162964, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.602, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.240506329113924, | |
| "grad_norm": 2.130836248397827, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.4402, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.2468354430379747, | |
| "grad_norm": 2.325317621231079, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.8767, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 1.6750413179397583, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.3484, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.259493670886076, | |
| "grad_norm": 1.6284416913986206, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.262, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.2658227848101267, | |
| "grad_norm": 1.55083167552948, | |
| "learning_rate": 0.0, | |
| "loss": 2.0966, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.2658227848101267, | |
| "eval_loss": 1.83381187915802, | |
| "eval_runtime": 7.8847, | |
| "eval_samples_per_second": 33.863, | |
| "eval_steps_per_second": 8.498, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.441193709142016e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |