| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 36.00264, | |
| "global_step": 150000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 8.4652, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 6.089, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-05, | |
| "loss": 5.551, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4e-05, | |
| "loss": 4.8056, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5e-05, | |
| "loss": 4.0329, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6e-05, | |
| "loss": 2.3276, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7e-05, | |
| "loss": 1.7263, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8e-05, | |
| "loss": 1.6089, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9e-05, | |
| "loss": 1.5133, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2832, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.1773, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.2188, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.2291, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.1196, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.0393, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.0729, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.1187, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.0602, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.9615, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.9844, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.052, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.0116, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.9324, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.9189, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.0033, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.9903, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.9018, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.8627, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.9721, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.9775, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.8738, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.8174, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.9494, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.9505, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8665, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.7849, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.9324, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.9189, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.8631, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.7853, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8816, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.8999, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.861, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.7854, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.8388, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8872, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.8658, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.7754, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.806, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.8724, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_loss": 0.6730242967605591, | |
| "eval_runtime": 18.7856, | |
| "eval_samples_per_second": 266.161, | |
| "eval_steps_per_second": 2.129, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.8543, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.7843, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.7706, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.8663, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.8679, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.7686, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.7342, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.8567, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.86, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.777, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.7027, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.8554, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.8428, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.7819, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.6988, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.8313, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.823, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.7899, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.7085, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.7936, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.8175, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8032, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.7058, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.762, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8109, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.7983, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.7184, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.7324, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.8052, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.7942, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.728, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.7019, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.8063, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8064, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.7211, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.6696, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.804, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.7977, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.7326, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.6482, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.806, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.7806, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.7416, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.6584, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.7708, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.7716, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7505, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6682, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.7379, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.7713, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_loss": 0.5763441324234009, | |
| "eval_runtime": 12.9231, | |
| "eval_samples_per_second": 386.903, | |
| "eval_steps_per_second": 3.095, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.7622, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.6692, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.7129, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.7691, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 5e-05, | |
| "loss": 0.7535, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.6861, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.6842, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.7695, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.7641, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.6872, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.6529, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.7694, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.7678, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.6955, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.6236, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7744, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.7586, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7025, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.6197, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.764, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.7415, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.7161, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.6325, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.7278, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.7403, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.7252, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.6428, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.6984, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.7408, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.7326, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.6485, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.6746, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.7387, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.7322, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.6617, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.6487, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.7449, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.7443, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 33.01, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.6622, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 33.01, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.6161, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 33.02, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.7464, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.7394, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 34.01, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.6769, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 34.01, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.5946, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 34.02, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.7533, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.7271, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 35.01, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.6883, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 35.01, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.6052, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 35.02, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.7273, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.7186, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_loss": 0.5396238565444946, | |
| "eval_runtime": 12.6594, | |
| "eval_samples_per_second": 394.965, | |
| "eval_steps_per_second": 3.16, | |
| "step": 150000 | |
| } | |
| ], | |
| "max_steps": 200000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.5268097908736e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |