| { | |
| "best_metric": 1.118039608001709, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.044868199663488505, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0002243409983174425, | |
| "grad_norm": 1.2535195350646973, | |
| "learning_rate": 1e-05, | |
| "loss": 1.6978, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0002243409983174425, | |
| "eval_loss": 2.2106831073760986, | |
| "eval_runtime": 321.1125, | |
| "eval_samples_per_second": 23.381, | |
| "eval_steps_per_second": 5.845, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000448681996634885, | |
| "grad_norm": 1.4986181259155273, | |
| "learning_rate": 2e-05, | |
| "loss": 1.7227, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0006730229949523276, | |
| "grad_norm": 1.5583722591400146, | |
| "learning_rate": 3e-05, | |
| "loss": 1.7189, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00089736399326977, | |
| "grad_norm": 1.7785111665725708, | |
| "learning_rate": 4e-05, | |
| "loss": 1.9079, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0011217049915872126, | |
| "grad_norm": 1.4737913608551025, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7098, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0013460459899046551, | |
| "grad_norm": 1.5895984172821045, | |
| "learning_rate": 6e-05, | |
| "loss": 1.8817, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0015703869882220977, | |
| "grad_norm": 1.4774248600006104, | |
| "learning_rate": 7e-05, | |
| "loss": 1.5912, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00179472798653954, | |
| "grad_norm": 1.6116282939910889, | |
| "learning_rate": 8e-05, | |
| "loss": 1.7109, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0020190689848569826, | |
| "grad_norm": 1.568785309791565, | |
| "learning_rate": 9e-05, | |
| "loss": 1.6309, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.002243409983174425, | |
| "grad_norm": 1.7714145183563232, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6024, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0024677509814918677, | |
| "grad_norm": 1.6795310974121094, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.7471, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0026920919798093103, | |
| "grad_norm": 1.5556038618087769, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.6004, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.002916432978126753, | |
| "grad_norm": 1.5433200597763062, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.4154, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0031407739764441954, | |
| "grad_norm": 1.5615655183792114, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.4952, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0033651149747616375, | |
| "grad_norm": 1.5367473363876343, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.4814, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00358945597307908, | |
| "grad_norm": 1.503058671951294, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.5085, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0038137969713965226, | |
| "grad_norm": 1.7003638744354248, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.638, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.004038137969713965, | |
| "grad_norm": 1.4940476417541504, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.4553, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.004262478968031408, | |
| "grad_norm": 1.5899440050125122, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.3805, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00448681996634885, | |
| "grad_norm": 1.7690714597702026, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.599, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.004711160964666293, | |
| "grad_norm": 1.4902113676071167, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.4269, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.004935501962983735, | |
| "grad_norm": 1.8363754749298096, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.4929, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.005159842961301178, | |
| "grad_norm": 1.6948083639144897, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.3737, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0053841839596186205, | |
| "grad_norm": 1.4842345714569092, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.3563, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.005608524957936063, | |
| "grad_norm": 1.5867489576339722, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.3662, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.005832865956253506, | |
| "grad_norm": 1.4881802797317505, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.3966, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.006057206954570948, | |
| "grad_norm": 1.755458950996399, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.4032, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.006281547952888391, | |
| "grad_norm": 1.630266547203064, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.2872, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0065058889512058325, | |
| "grad_norm": 1.7784324884414673, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.4598, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.006730229949523275, | |
| "grad_norm": 1.650486707687378, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.4222, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.006954570947840718, | |
| "grad_norm": 1.9005061388015747, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.4273, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.00717891194615816, | |
| "grad_norm": 1.6536924839019775, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.3432, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.007403252944475603, | |
| "grad_norm": 1.6960432529449463, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.2885, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.007627593942793045, | |
| "grad_norm": 1.732762336730957, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.4664, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.007851934941110488, | |
| "grad_norm": 1.8983900547027588, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.5125, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.00807627593942793, | |
| "grad_norm": 1.6673097610473633, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.3431, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.008300616937745373, | |
| "grad_norm": 1.9295827150344849, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.5937, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.008524957936062815, | |
| "grad_norm": 2.07816743850708, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.3887, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.008749298934380258, | |
| "grad_norm": 2.1436219215393066, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.4596, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0089736399326977, | |
| "grad_norm": 2.2428765296936035, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.4262, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.009197980931015143, | |
| "grad_norm": 2.237614393234253, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.4507, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.009422321929332586, | |
| "grad_norm": 2.262399911880493, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.617, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.009646662927650028, | |
| "grad_norm": 2.208442211151123, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.3775, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.00987100392596747, | |
| "grad_norm": 2.1070003509521484, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.3109, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.010095344924284913, | |
| "grad_norm": 2.587320327758789, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.5739, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.010319685922602356, | |
| "grad_norm": 2.294077157974243, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.3254, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.010544026920919798, | |
| "grad_norm": 2.8567652702331543, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.6047, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.010768367919237241, | |
| "grad_norm": 3.087742567062378, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.336, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.010992708917554684, | |
| "grad_norm": 3.302703380584717, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.2902, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.011217049915872126, | |
| "grad_norm": 4.378942012786865, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.4073, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.011217049915872126, | |
| "eval_loss": 1.427445888519287, | |
| "eval_runtime": 323.2496, | |
| "eval_samples_per_second": 23.227, | |
| "eval_steps_per_second": 5.807, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.011441390914189569, | |
| "grad_norm": 3.126593589782715, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.7335, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.011665731912507011, | |
| "grad_norm": 2.6481401920318604, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.5729, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.011890072910824454, | |
| "grad_norm": 1.727514624595642, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.6804, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.012114413909141896, | |
| "grad_norm": 1.2480617761611938, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.4371, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.012338754907459339, | |
| "grad_norm": 1.2019025087356567, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.4222, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.012563095905776782, | |
| "grad_norm": 1.195434331893921, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.3863, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.012787436904094224, | |
| "grad_norm": 1.1391469240188599, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.2207, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.013011777902411665, | |
| "grad_norm": 1.1910752058029175, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.4788, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.013236118900729107, | |
| "grad_norm": 1.0737820863723755, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.3222, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.01346045989904655, | |
| "grad_norm": 1.0162631273269653, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.1773, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.013684800897363993, | |
| "grad_norm": 1.1553272008895874, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.3959, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.013909141895681435, | |
| "grad_norm": 1.0877430438995361, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.2248, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.014133482893998878, | |
| "grad_norm": 1.1033817529678345, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.3472, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.01435782389231632, | |
| "grad_norm": 1.2968347072601318, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.2236, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.014582164890633763, | |
| "grad_norm": 1.185168981552124, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.2695, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.014806505888951205, | |
| "grad_norm": 1.1989123821258545, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.2564, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.015030846887268648, | |
| "grad_norm": 1.1638737916946411, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.3146, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.01525518788558609, | |
| "grad_norm": 1.2309011220932007, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.229, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.015479528883903533, | |
| "grad_norm": 1.2377573251724243, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.2964, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.015703869882220976, | |
| "grad_norm": 1.3125661611557007, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.3079, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.015928210880538418, | |
| "grad_norm": 1.2609599828720093, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.3271, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.01615255187885586, | |
| "grad_norm": 1.3171006441116333, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.233, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.016376892877173303, | |
| "grad_norm": 1.1379790306091309, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.2051, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.016601233875490746, | |
| "grad_norm": 1.4332889318466187, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.3683, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.01682557487380819, | |
| "grad_norm": 1.2557628154754639, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.1368, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01704991587212563, | |
| "grad_norm": 1.4247772693634033, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.1477, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.017274256870443074, | |
| "grad_norm": 1.412876844406128, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.2012, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.017498597868760516, | |
| "grad_norm": 1.415099024772644, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.2443, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.01772293886707796, | |
| "grad_norm": 1.3193162679672241, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.1379, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0179472798653954, | |
| "grad_norm": 1.60520601272583, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.1652, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.018171620863712844, | |
| "grad_norm": 1.4677854776382446, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.1939, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.018395961862030286, | |
| "grad_norm": 1.654918909072876, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.2973, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.01862030286034773, | |
| "grad_norm": 1.6216435432434082, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.3242, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.01884464385866517, | |
| "grad_norm": 1.5336003303527832, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.1933, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.019068984856982614, | |
| "grad_norm": 1.4985709190368652, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.2003, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.019293325855300057, | |
| "grad_norm": 1.6437580585479736, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.253, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0195176668536175, | |
| "grad_norm": 1.7959896326065063, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.2056, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.01974200785193494, | |
| "grad_norm": 1.5936723947525024, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.19, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.019966348850252384, | |
| "grad_norm": 1.7353544235229492, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.1738, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.020190689848569827, | |
| "grad_norm": 1.9611711502075195, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.3455, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02041503084688727, | |
| "grad_norm": 1.6829428672790527, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.1503, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.020639371845204712, | |
| "grad_norm": 1.7608388662338257, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.0876, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.020863712843522154, | |
| "grad_norm": 2.420135736465454, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.3843, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.021088053841839597, | |
| "grad_norm": 2.545687675476074, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.5431, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.02131239484015704, | |
| "grad_norm": 2.2323973178863525, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.19, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.021536735838474482, | |
| "grad_norm": 2.3698513507843018, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.3178, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.021761076836791925, | |
| "grad_norm": 2.1286826133728027, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.0776, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.021985417835109367, | |
| "grad_norm": 2.6585943698883057, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.2897, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.02220975883342681, | |
| "grad_norm": 3.006467819213867, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.272, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.022434099831744252, | |
| "grad_norm": 5.164963722229004, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.3087, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.022434099831744252, | |
| "eval_loss": 1.2046418190002441, | |
| "eval_runtime": 323.5362, | |
| "eval_samples_per_second": 23.206, | |
| "eval_steps_per_second": 5.802, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.022658440830061695, | |
| "grad_norm": 0.9242492318153381, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.4462, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.022882781828379137, | |
| "grad_norm": 1.090623378753662, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.4612, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.02310712282669658, | |
| "grad_norm": 0.9142351746559143, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.2809, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.023331463825014023, | |
| "grad_norm": 0.9369277358055115, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.2703, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.023555804823331465, | |
| "grad_norm": 0.9184145331382751, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3298, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.023780145821648908, | |
| "grad_norm": 1.0177104473114014, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.238, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.02400448681996635, | |
| "grad_norm": 0.9094177484512329, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.1701, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.024228827818283793, | |
| "grad_norm": 0.938910961151123, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.2444, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.024453168816601235, | |
| "grad_norm": 1.0399653911590576, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.2409, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.024677509814918678, | |
| "grad_norm": 1.0955804586410522, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.2666, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.02490185081323612, | |
| "grad_norm": 1.0677424669265747, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.211, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.025126191811553563, | |
| "grad_norm": 1.0730018615722656, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.2708, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.025350532809871006, | |
| "grad_norm": 0.9894511103630066, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.1, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.025574873808188448, | |
| "grad_norm": 1.0025452375411987, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.0564, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.025799214806505887, | |
| "grad_norm": 1.0198055505752563, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.1394, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.02602355580482333, | |
| "grad_norm": 1.1933400630950928, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.2049, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.026247896803140772, | |
| "grad_norm": 1.259006142616272, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.1994, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.026472237801458215, | |
| "grad_norm": 1.6928879022598267, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.2408, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.026696578799775657, | |
| "grad_norm": 1.2442269325256348, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.2658, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0269209197980931, | |
| "grad_norm": 1.2106413841247559, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.1574, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.027145260796410543, | |
| "grad_norm": 1.247713327407837, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.1134, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.027369601794727985, | |
| "grad_norm": 1.0733723640441895, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.0592, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.027593942793045428, | |
| "grad_norm": 1.2934086322784424, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.0896, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.02781828379136287, | |
| "grad_norm": 1.1768771409988403, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.055, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.028042624789680313, | |
| "grad_norm": 1.508828043937683, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.1535, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.028266965787997755, | |
| "grad_norm": 1.1093461513519287, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.9791, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.028491306786315198, | |
| "grad_norm": 1.3151066303253174, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.0159, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.02871564778463264, | |
| "grad_norm": 1.287225365638733, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.1779, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.028939988782950083, | |
| "grad_norm": 1.366283655166626, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.0475, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.029164329781267526, | |
| "grad_norm": 1.4970052242279053, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.21, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.029388670779584968, | |
| "grad_norm": 1.6789995431900024, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.1716, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.02961301177790241, | |
| "grad_norm": 1.5438408851623535, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.2357, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.029837352776219853, | |
| "grad_norm": 1.4882869720458984, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.1639, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.030061693774537296, | |
| "grad_norm": 1.7601850032806396, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.1997, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.03028603477285474, | |
| "grad_norm": 1.7460747957229614, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.1608, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.03051037577117218, | |
| "grad_norm": 1.5243598222732544, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.2004, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.030734716769489624, | |
| "grad_norm": 1.4518287181854248, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.0194, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.030959057767807066, | |
| "grad_norm": 1.747273325920105, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.1875, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.03118339876612451, | |
| "grad_norm": 1.939194917678833, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.0184, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.03140773976444195, | |
| "grad_norm": 1.8796677589416504, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.1113, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0316320807627594, | |
| "grad_norm": 1.639658808708191, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.1186, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.031856421761076836, | |
| "grad_norm": 1.8173081874847412, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.123, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.03208076275939428, | |
| "grad_norm": 2.2189078330993652, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.1577, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.03230510375771172, | |
| "grad_norm": 2.105889320373535, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.089, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.03252944475602917, | |
| "grad_norm": 2.0868937969207764, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.0292, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.03275378575434661, | |
| "grad_norm": 2.0227229595184326, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.9974, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.03297812675266405, | |
| "grad_norm": 2.4205689430236816, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.0882, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.03320246775098149, | |
| "grad_norm": 2.5911386013031006, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.174, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.03342680874929894, | |
| "grad_norm": 2.834402322769165, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.9788, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.03365114974761638, | |
| "grad_norm": 3.1549575328826904, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.106, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03365114974761638, | |
| "eval_loss": 1.131042718887329, | |
| "eval_runtime": 323.1651, | |
| "eval_samples_per_second": 23.233, | |
| "eval_steps_per_second": 5.808, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03387549074593382, | |
| "grad_norm": 0.8614676594734192, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.3554, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.03409983174425126, | |
| "grad_norm": 0.9027763605117798, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.3378, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.0343241727425687, | |
| "grad_norm": 0.9020812511444092, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.2763, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.03454851374088615, | |
| "grad_norm": 0.9730469584465027, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.3594, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.034772854739203586, | |
| "grad_norm": 1.0548874139785767, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.2592, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.03499719573752103, | |
| "grad_norm": 1.1751259565353394, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.3561, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.03522153673583847, | |
| "grad_norm": 1.040268898010254, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.1385, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.03544587773415592, | |
| "grad_norm": 1.034965991973877, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.1944, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.035670218732473356, | |
| "grad_norm": 1.005588412284851, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.152, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0358945597307908, | |
| "grad_norm": 1.020017385482788, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.179, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.03611890072910824, | |
| "grad_norm": 1.0504347085952759, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.1554, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.03634324172742569, | |
| "grad_norm": 1.11807382106781, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.1324, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.03656758272574313, | |
| "grad_norm": 0.9994248747825623, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.1389, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.03679192372406057, | |
| "grad_norm": 0.9709587097167969, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.149, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.03701626472237801, | |
| "grad_norm": 1.1986273527145386, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.2284, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.03724060572069546, | |
| "grad_norm": 1.1978342533111572, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.1399, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0374649467190129, | |
| "grad_norm": 1.1640859842300415, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.2051, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.03768928771733034, | |
| "grad_norm": 0.9959132075309753, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.0146, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.03791362871564778, | |
| "grad_norm": 1.102310061454773, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.9959, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.03813796971396523, | |
| "grad_norm": 1.1588571071624756, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.1923, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03836231071228267, | |
| "grad_norm": 1.1732374429702759, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.1319, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.03858665171060011, | |
| "grad_norm": 1.260844349861145, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.0923, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.03881099270891755, | |
| "grad_norm": 1.2971606254577637, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.0556, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.039035333707235, | |
| "grad_norm": 1.2263072729110718, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.1448, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.03925967470555244, | |
| "grad_norm": 1.263629674911499, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.0225, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03948401570386988, | |
| "grad_norm": 1.3652485609054565, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.2126, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.03970835670218732, | |
| "grad_norm": 1.244150161743164, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.925, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.03993269770050477, | |
| "grad_norm": 1.1341102123260498, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.018, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.04015703869882221, | |
| "grad_norm": 1.3381551504135132, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.1376, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.040381379697139654, | |
| "grad_norm": 1.3847415447235107, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.1043, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.04060572069545709, | |
| "grad_norm": 1.2933632135391235, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.9109, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.04083006169377454, | |
| "grad_norm": 1.4092001914978027, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.0101, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.04105440269209198, | |
| "grad_norm": 1.4717921018600464, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.092, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.041278743690409424, | |
| "grad_norm": 1.4898322820663452, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.003, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.04150308468872686, | |
| "grad_norm": 1.434329867362976, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.0782, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.04172742568704431, | |
| "grad_norm": 1.4739409685134888, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.0491, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.04195176668536175, | |
| "grad_norm": 1.649688959121704, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.2685, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.042176107683679194, | |
| "grad_norm": 1.5006073713302612, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.0353, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.04240044868199663, | |
| "grad_norm": 1.5449039936065674, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.0905, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.04262478968031408, | |
| "grad_norm": 1.6924413442611694, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.0539, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.04284913067863152, | |
| "grad_norm": 2.0756149291992188, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.3964, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.043073471676948964, | |
| "grad_norm": 1.9471677541732788, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.1528, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0432978126752664, | |
| "grad_norm": 1.9654617309570312, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.1359, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.04352215367358385, | |
| "grad_norm": 2.175584554672241, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.0669, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.04374649467190129, | |
| "grad_norm": 1.6459085941314697, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.9547, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.043970835670218734, | |
| "grad_norm": 2.174954652786255, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.9913, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.044195176668536174, | |
| "grad_norm": 2.332949161529541, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.0769, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.04441951766685362, | |
| "grad_norm": 2.5872936248779297, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.8778, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.04464385866517106, | |
| "grad_norm": 2.374394655227661, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.8653, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.044868199663488505, | |
| "grad_norm": 3.0859851837158203, | |
| "learning_rate": 0.0, | |
| "loss": 0.9303, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.044868199663488505, | |
| "eval_loss": 1.118039608001709, | |
| "eval_runtime": 323.451, | |
| "eval_samples_per_second": 23.212, | |
| "eval_steps_per_second": 5.803, | |
| "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": 1.34707043893248e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |