| { | |
| "best_metric": 0.4288332760334015, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.38816108685104317, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0019408054342552159, | |
| "grad_norm": 1.7232019901275635, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5136, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0019408054342552159, | |
| "eval_loss": 1.874531865119934, | |
| "eval_runtime": 81.3442, | |
| "eval_samples_per_second": 10.671, | |
| "eval_steps_per_second": 2.668, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0038816108685104317, | |
| "grad_norm": 1.492817759513855, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4853, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.005822416302765648, | |
| "grad_norm": 1.3057711124420166, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4316, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0077632217370208634, | |
| "grad_norm": 1.655269980430603, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4361, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.009704027171276079, | |
| "grad_norm": 1.4750041961669922, | |
| "learning_rate": 5e-05, | |
| "loss": 0.4633, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011644832605531296, | |
| "grad_norm": 2.0782909393310547, | |
| "learning_rate": 6e-05, | |
| "loss": 0.4913, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.013585638039786511, | |
| "grad_norm": 1.9410984516143799, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4865, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.015526443474041727, | |
| "grad_norm": 1.9401772022247314, | |
| "learning_rate": 8e-05, | |
| "loss": 0.5406, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.017467248908296942, | |
| "grad_norm": 2.544038772583008, | |
| "learning_rate": 9e-05, | |
| "loss": 0.593, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.019408054342552158, | |
| "grad_norm": 2.2485432624816895, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5002, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.021348859776807377, | |
| "grad_norm": 3.3933422565460205, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.5997, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.023289665211062592, | |
| "grad_norm": 3.589137077331543, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.6316, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.025230470645317808, | |
| "grad_norm": 3.282288074493408, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.7404, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.027171276079573023, | |
| "grad_norm": 2.89556622505188, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.5094, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02911208151382824, | |
| "grad_norm": 4.256638050079346, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.6752, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.031052886948083454, | |
| "grad_norm": 4.366456508636475, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.7698, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03299369238233867, | |
| "grad_norm": 4.368771076202393, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.7455, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.034934497816593885, | |
| "grad_norm": 4.762805938720703, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.7844, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.036875303250849104, | |
| "grad_norm": 5.389572620391846, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.7462, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.038816108685104316, | |
| "grad_norm": 4.4080729484558105, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.7418, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.040756914119359534, | |
| "grad_norm": 4.218383312225342, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.6666, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04269771955361475, | |
| "grad_norm": 6.5666937828063965, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.743, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.044638524987869965, | |
| "grad_norm": 5.88185977935791, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.7389, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.046579330422125184, | |
| "grad_norm": 6.599873065948486, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.7555, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.048520135856380396, | |
| "grad_norm": 8.274136543273926, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.7586, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.050460941290635615, | |
| "grad_norm": 9.610648155212402, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.8773, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05240174672489083, | |
| "grad_norm": 10.665836334228516, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7605, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.054342552159146046, | |
| "grad_norm": 9.091414451599121, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.754, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.056283357593401265, | |
| "grad_norm": 11.075277328491211, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.7841, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05822416302765648, | |
| "grad_norm": 9.491761207580566, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.647, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.060164968461911696, | |
| "grad_norm": 8.139253616333008, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.6633, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.06210577389616691, | |
| "grad_norm": 7.97451639175415, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.6707, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06404657933042213, | |
| "grad_norm": 9.581368446350098, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.7825, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06598738476467735, | |
| "grad_norm": 63.50759506225586, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.8515, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06792819019893255, | |
| "grad_norm": 92.15950012207031, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.4107, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06986899563318777, | |
| "grad_norm": 44.622276306152344, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.0415, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07180980106744299, | |
| "grad_norm": 48.721153259277344, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.625, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07375060650169821, | |
| "grad_norm": 67.3081283569336, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.8686, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07569141193595343, | |
| "grad_norm": 26.684757232666016, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.5521, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07763221737020863, | |
| "grad_norm": 29.838045120239258, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.5138, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07957302280446385, | |
| "grad_norm": 18.04141616821289, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.3707, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08151382823871907, | |
| "grad_norm": 22.536243438720703, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.2001, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.08345463367297429, | |
| "grad_norm": 18.702436447143555, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.1195, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0853954391072295, | |
| "grad_norm": 17.435029983520508, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.0812, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08733624454148471, | |
| "grad_norm": 35.78749084472656, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0583, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08927704997573993, | |
| "grad_norm": 18.771339416503906, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.118, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.09121785540999515, | |
| "grad_norm": 16.208833694458008, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0683, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09315866084425037, | |
| "grad_norm": 12.883469581604004, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0457, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09509946627850557, | |
| "grad_norm": 7.016485691070557, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0485, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09704027171276079, | |
| "grad_norm": 14.671367645263672, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.126, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09704027171276079, | |
| "eval_loss": 4.280299663543701, | |
| "eval_runtime": 82.3393, | |
| "eval_samples_per_second": 10.542, | |
| "eval_steps_per_second": 2.635, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09898107714701601, | |
| "grad_norm": 74.30940246582031, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 6.8047, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.10092188258127123, | |
| "grad_norm": 57.95939636230469, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 4.6004, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10286268801552645, | |
| "grad_norm": 32.66792297363281, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.556, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.10480349344978165, | |
| "grad_norm": 19.1224308013916, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.5345, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.10674429888403687, | |
| "grad_norm": 7.216464042663574, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.94, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10868510431829209, | |
| "grad_norm": 3.6728262901306152, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.6137, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.11062590975254731, | |
| "grad_norm": 2.703936815261841, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.6146, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11256671518680253, | |
| "grad_norm": 1.8983687162399292, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.5714, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11450752062105773, | |
| "grad_norm": 1.801605224609375, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.5857, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11644832605531295, | |
| "grad_norm": 1.9191378355026245, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.6977, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11838913148956817, | |
| "grad_norm": 2.005643844604492, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.6146, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.12032993692382339, | |
| "grad_norm": 1.8634217977523804, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.6131, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1222707423580786, | |
| "grad_norm": 2.057691812515259, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.5836, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12421154779233382, | |
| "grad_norm": 2.572847604751587, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.6632, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12615235322658905, | |
| "grad_norm": 2.3367061614990234, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.5571, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12809315866084425, | |
| "grad_norm": 2.293715715408325, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.6468, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.13003396409509946, | |
| "grad_norm": 2.5277249813079834, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.6325, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1319747695293547, | |
| "grad_norm": 2.8323874473571777, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.5827, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1339155749636099, | |
| "grad_norm": 2.8978846073150635, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.6806, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.1358563803978651, | |
| "grad_norm": 4.2896270751953125, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.6588, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13779718583212033, | |
| "grad_norm": 3.737482786178589, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.7028, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.13973799126637554, | |
| "grad_norm": 2.8255558013916016, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.7127, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.14167879670063077, | |
| "grad_norm": 7.366374969482422, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.6581, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14361960213488598, | |
| "grad_norm": 2.2917308807373047, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.5928, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14556040756914118, | |
| "grad_norm": 3.0215542316436768, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8472, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14750121300339641, | |
| "grad_norm": 4.321956634521484, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.8806, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.14944201843765162, | |
| "grad_norm": 3.955066204071045, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.8262, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.15138282387190685, | |
| "grad_norm": 5.270688056945801, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.8037, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15332362930616206, | |
| "grad_norm": 3.847731828689575, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.7782, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.15526443474041726, | |
| "grad_norm": 4.927722454071045, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.6938, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1572052401746725, | |
| "grad_norm": 4.019804954528809, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.6472, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.1591460456089277, | |
| "grad_norm": 3.6839606761932373, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.6326, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.16108685104318293, | |
| "grad_norm": 3.269862174987793, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.625, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.16302765647743814, | |
| "grad_norm": 10.826176643371582, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.0978, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16496846191169334, | |
| "grad_norm": 55.355384826660156, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.2281, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.16690926734594858, | |
| "grad_norm": 33.33103942871094, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.9824, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.16885007278020378, | |
| "grad_norm": 37.37646484375, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.5676, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.170790878214459, | |
| "grad_norm": 36.94839859008789, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.4243, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.17273168364871422, | |
| "grad_norm": 26.722476959228516, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.2814, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.17467248908296942, | |
| "grad_norm": 47.58677673339844, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.9886, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17661329451722466, | |
| "grad_norm": 29.36099624633789, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.6394, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.17855409995147986, | |
| "grad_norm": 26.282276153564453, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.385, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.1804949053857351, | |
| "grad_norm": 28.603078842163086, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3408, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.1824357108199903, | |
| "grad_norm": 23.356945037841797, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2171, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1843765162542455, | |
| "grad_norm": 10.759159088134766, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.1167, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.18631732168850074, | |
| "grad_norm": 10.58999252319336, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.1587, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.18825812712275594, | |
| "grad_norm": 6.8319196701049805, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0881, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.19019893255701115, | |
| "grad_norm": 13.585020065307617, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0707, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.19213973799126638, | |
| "grad_norm": 14.443303108215332, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0682, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.19408054342552158, | |
| "grad_norm": 7.6578755378723145, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.1065, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19408054342552158, | |
| "eval_loss": 1.2820466756820679, | |
| "eval_runtime": 82.3895, | |
| "eval_samples_per_second": 10.535, | |
| "eval_steps_per_second": 2.634, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19602134885977682, | |
| "grad_norm": 27.670087814331055, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.7, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.19796215429403202, | |
| "grad_norm": 16.889015197753906, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.0854, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.19990295972828723, | |
| "grad_norm": 6.067689895629883, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.6199, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.20184376516254246, | |
| "grad_norm": 1.871160626411438, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.5381, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20378457059679767, | |
| "grad_norm": 1.4615622758865356, | |
| "learning_rate": 5e-05, | |
| "loss": 0.4677, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2057253760310529, | |
| "grad_norm": 1.1890963315963745, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.4587, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.2076661814653081, | |
| "grad_norm": 1.0661296844482422, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.4024, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2096069868995633, | |
| "grad_norm": 1.0485421419143677, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5118, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.21154779233381854, | |
| "grad_norm": 1.0046170949935913, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.4791, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.21348859776807375, | |
| "grad_norm": 1.1115124225616455, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.5694, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21542940320232898, | |
| "grad_norm": 0.8592437505722046, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.4441, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.21737020863658418, | |
| "grad_norm": 1.1286271810531616, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.5416, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2193110140708394, | |
| "grad_norm": 1.3943814039230347, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.532, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.22125181950509462, | |
| "grad_norm": 0.9902353882789612, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.5542, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.22319262493934983, | |
| "grad_norm": 0.9861344695091248, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.4915, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.22513343037360506, | |
| "grad_norm": 1.234898328781128, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.6066, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.22707423580786026, | |
| "grad_norm": 1.3861297369003296, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.64, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.22901504124211547, | |
| "grad_norm": 1.2729508876800537, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.5766, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2309558466763707, | |
| "grad_norm": 1.3564996719360352, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.5983, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2328966521106259, | |
| "grad_norm": 1.5852866172790527, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.6284, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2348374575448811, | |
| "grad_norm": 1.5961381196975708, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.6023, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.23677826297913634, | |
| "grad_norm": 3.5081632137298584, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.649, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.23871906841339155, | |
| "grad_norm": 1.6745890378952026, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.58, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.24065987384764678, | |
| "grad_norm": 2.6499383449554443, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.7133, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.242600679281902, | |
| "grad_norm": 2.544788122177124, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.6364, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2445414847161572, | |
| "grad_norm": 2.170379161834717, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.6807, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.24648229015041243, | |
| "grad_norm": 3.3199291229248047, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.837, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.24842309558466763, | |
| "grad_norm": 3.8902361392974854, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.737, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.25036390101892286, | |
| "grad_norm": 3.5638248920440674, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.6829, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.2523047064531781, | |
| "grad_norm": 4.286755561828613, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.6566, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2542455118874333, | |
| "grad_norm": 3.6197922229766846, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.5996, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2561863173216885, | |
| "grad_norm": 3.402855157852173, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.5818, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.25812712275594374, | |
| "grad_norm": 3.965913772583008, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.5624, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.2600679281901989, | |
| "grad_norm": 5.662968158721924, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.6588, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.26200873362445415, | |
| "grad_norm": 26.19328498840332, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.7597, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2639495390587094, | |
| "grad_norm": 22.501678466796875, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.0186, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.26589034449296456, | |
| "grad_norm": 15.385753631591797, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.9392, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2678311499272198, | |
| "grad_norm": 18.340078353881836, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.8334, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.269771955361475, | |
| "grad_norm": 22.590129852294922, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.5957, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2717127607957302, | |
| "grad_norm": 23.760828018188477, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.6336, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.27365356622998543, | |
| "grad_norm": 26.645437240600586, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.6011, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.27559437166424067, | |
| "grad_norm": 16.77243995666504, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3907, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2775351770984959, | |
| "grad_norm": 17.91073226928711, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.5283, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.2794759825327511, | |
| "grad_norm": 11.456660270690918, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3378, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.2814167879670063, | |
| "grad_norm": 11.108325958251953, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3111, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.28335759340126154, | |
| "grad_norm": 10.32477855682373, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3747, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.2852983988355167, | |
| "grad_norm": 11.42671012878418, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.202, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.28723920426977195, | |
| "grad_norm": 9.717866897583008, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.2681, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2891800097040272, | |
| "grad_norm": 20.730472564697266, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.2357, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.29112081513828236, | |
| "grad_norm": 12.306010246276855, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.1586, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.29112081513828236, | |
| "eval_loss": 0.43027618527412415, | |
| "eval_runtime": 82.3857, | |
| "eval_samples_per_second": 10.536, | |
| "eval_steps_per_second": 2.634, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2930616205725376, | |
| "grad_norm": 1.5048695802688599, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.37, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.29500242600679283, | |
| "grad_norm": 1.4473662376403809, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.3837, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.29694323144104806, | |
| "grad_norm": 1.4776946306228638, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.4211, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.29888403687530324, | |
| "grad_norm": 1.3038028478622437, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.3883, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.30082484230955847, | |
| "grad_norm": 1.5496819019317627, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.4149, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3027656477438137, | |
| "grad_norm": 1.6367863416671753, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.3448, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3047064531780689, | |
| "grad_norm": 1.4665528535842896, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.3953, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3066472586123241, | |
| "grad_norm": 1.6473298072814941, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.4127, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.30858806404657935, | |
| "grad_norm": 1.682902455329895, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.4146, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3105288694808345, | |
| "grad_norm": 1.5516167879104614, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.4092, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.31246967491508976, | |
| "grad_norm": 2.3393936157226562, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.5348, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.314410480349345, | |
| "grad_norm": 1.9260610342025757, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.557, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.31635128578360017, | |
| "grad_norm": 2.5166797637939453, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.511, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.3182920912178554, | |
| "grad_norm": 2.065892457962036, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.5195, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.32023289665211063, | |
| "grad_norm": 2.0890939235687256, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.526, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.32217370208636587, | |
| "grad_norm": 2.001455783843994, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.4953, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.32411450752062104, | |
| "grad_norm": 2.3791608810424805, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.5527, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3260553129548763, | |
| "grad_norm": 2.145820379257202, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.5662, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.3279961183891315, | |
| "grad_norm": 2.8020894527435303, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.6665, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3299369238233867, | |
| "grad_norm": 3.162172794342041, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.609, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3318777292576419, | |
| "grad_norm": 3.268869161605835, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.5982, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.33381853469189715, | |
| "grad_norm": 4.25205135345459, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.6915, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.33575934012615233, | |
| "grad_norm": 3.3226122856140137, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.6434, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.33770014556040756, | |
| "grad_norm": 2.6694557666778564, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.5814, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3396409509946628, | |
| "grad_norm": 3.830965518951416, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.6271, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.341581756428918, | |
| "grad_norm": 4.925380229949951, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.7041, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3435225618631732, | |
| "grad_norm": 4.1990156173706055, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.744, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.34546336729742844, | |
| "grad_norm": 5.960219860076904, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.7445, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.34740417273168367, | |
| "grad_norm": 4.452081203460693, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.6492, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.34934497816593885, | |
| "grad_norm": 4.32200813293457, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.6657, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3512857836001941, | |
| "grad_norm": 4.201397895812988, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.6161, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.3532265890344493, | |
| "grad_norm": 3.7781982421875, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.5944, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3551673944687045, | |
| "grad_norm": 5.680567741394043, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.7711, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.3571081999029597, | |
| "grad_norm": 11.954629898071289, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3028, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.35904900533721495, | |
| "grad_norm": 7.960069179534912, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.2086, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3609898107714702, | |
| "grad_norm": 9.26430606842041, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.2793, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.36293061620572536, | |
| "grad_norm": 11.756043434143066, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.213, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3648714216399806, | |
| "grad_norm": 20.615224838256836, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.2891, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.36681222707423583, | |
| "grad_norm": 16.722871780395508, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3722, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.368753032508491, | |
| "grad_norm": 10.050209999084473, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.1796, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.37069383794274624, | |
| "grad_norm": 7.875072479248047, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.1449, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.3726346433770015, | |
| "grad_norm": 9.175650596618652, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.1967, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.37457544881125665, | |
| "grad_norm": 6.732213973999023, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.2627, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3765162542455119, | |
| "grad_norm": 9.79932975769043, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3602, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.3784570596797671, | |
| "grad_norm": 11.805617332458496, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.1885, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3803978651140223, | |
| "grad_norm": 11.356443405151367, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.2876, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.3823386705482775, | |
| "grad_norm": 17.0898380279541, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.2711, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.38427947598253276, | |
| "grad_norm": 19.003767013549805, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.2415, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.386220281416788, | |
| "grad_norm": 16.432050704956055, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3056, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.38816108685104317, | |
| "grad_norm": 30.51390838623047, | |
| "learning_rate": 0.0, | |
| "loss": 0.1571, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.38816108685104317, | |
| "eval_loss": 0.4288332760334015, | |
| "eval_runtime": 82.4038, | |
| "eval_samples_per_second": 10.533, | |
| "eval_steps_per_second": 2.633, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.15717014142976e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |