| { | |
| "best_metric": 0.8103886842727661, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.5706134094151213, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0028530670470756064, | |
| "grad_norm": 2.7992610931396484, | |
| "learning_rate": 1e-05, | |
| "loss": 2.7379, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0028530670470756064, | |
| "eval_loss": 4.149313926696777, | |
| "eval_runtime": 43.2092, | |
| "eval_samples_per_second": 13.654, | |
| "eval_steps_per_second": 3.425, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005706134094151213, | |
| "grad_norm": 3.6649270057678223, | |
| "learning_rate": 2e-05, | |
| "loss": 3.3691, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.008559201141226819, | |
| "grad_norm": 3.85835862159729, | |
| "learning_rate": 3e-05, | |
| "loss": 3.6013, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011412268188302425, | |
| "grad_norm": 3.976522207260132, | |
| "learning_rate": 4e-05, | |
| "loss": 3.6466, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.014265335235378032, | |
| "grad_norm": 3.449662923812866, | |
| "learning_rate": 5e-05, | |
| "loss": 3.3128, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.017118402282453638, | |
| "grad_norm": 3.3031234741210938, | |
| "learning_rate": 6e-05, | |
| "loss": 2.9783, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.019971469329529243, | |
| "grad_norm": 2.935482978820801, | |
| "learning_rate": 7e-05, | |
| "loss": 2.6386, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02282453637660485, | |
| "grad_norm": 2.6723663806915283, | |
| "learning_rate": 8e-05, | |
| "loss": 2.2127, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.025677603423680456, | |
| "grad_norm": 2.8913800716400146, | |
| "learning_rate": 9e-05, | |
| "loss": 1.8363, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.028530670470756064, | |
| "grad_norm": 3.16005802154541, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4834, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03138373751783167, | |
| "grad_norm": 3.2690377235412598, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.4439, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.034236804564907276, | |
| "grad_norm": 2.2833311557769775, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.3592, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.037089871611982884, | |
| "grad_norm": 1.751080870628357, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.2227, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.039942938659058486, | |
| "grad_norm": 1.6022005081176758, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.148, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.042796005706134094, | |
| "grad_norm": 1.2305362224578857, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.1343, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0456490727532097, | |
| "grad_norm": 1.2019984722137451, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.0619, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04850213980028531, | |
| "grad_norm": 1.369314193725586, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9922, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05135520684736091, | |
| "grad_norm": 1.1308224201202393, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9969, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05420827389443652, | |
| "grad_norm": 1.3146789073944092, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.1297, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05706134094151213, | |
| "grad_norm": 1.0999287366867065, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.9673, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05991440798858773, | |
| "grad_norm": 1.1856595277786255, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.986, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06276747503566334, | |
| "grad_norm": 1.0935935974121094, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.0622, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06562054208273894, | |
| "grad_norm": 1.1153169870376587, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.9862, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06847360912981455, | |
| "grad_norm": 1.0085804462432861, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.9354, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07132667617689016, | |
| "grad_norm": 0.8231331706047058, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.8846, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07417974322396577, | |
| "grad_norm": 0.8864841461181641, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.9359, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07703281027104136, | |
| "grad_norm": 0.968312680721283, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.9645, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07988587731811697, | |
| "grad_norm": 0.8740638494491577, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.9106, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08273894436519258, | |
| "grad_norm": 0.8972720503807068, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.9534, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08559201141226819, | |
| "grad_norm": 0.8092889189720154, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.8828, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0884450784593438, | |
| "grad_norm": 0.8146161437034607, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.846, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0912981455064194, | |
| "grad_norm": 1.073032021522522, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9755, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09415121255349501, | |
| "grad_norm": 0.834766149520874, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.9749, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09700427960057062, | |
| "grad_norm": 0.9171404242515564, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.9198, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09985734664764621, | |
| "grad_norm": 0.7169038653373718, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8649, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.10271041369472182, | |
| "grad_norm": 0.8518374562263489, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.8701, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.10556348074179743, | |
| "grad_norm": 0.8100901246070862, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.8672, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.10841654778887304, | |
| "grad_norm": 0.8432615995407104, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.0118, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11126961483594865, | |
| "grad_norm": 0.8186846375465393, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.9473, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11412268188302425, | |
| "grad_norm": 0.7907865047454834, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.9723, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11697574893009986, | |
| "grad_norm": 0.8026462197303772, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8334, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11982881597717546, | |
| "grad_norm": 0.7831972241401672, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.9834, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12268188302425106, | |
| "grad_norm": 0.7319669723510742, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.8226, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12553495007132667, | |
| "grad_norm": 0.7991037368774414, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.9327, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.12838801711840228, | |
| "grad_norm": 0.7444028854370117, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.8459, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1312410841654779, | |
| "grad_norm": 0.7931735515594482, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.9871, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1340941512125535, | |
| "grad_norm": 0.6636390686035156, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.8724, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1369472182596291, | |
| "grad_norm": 0.7103052735328674, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.9285, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1398002853067047, | |
| "grad_norm": 0.8651701211929321, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.9808, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14265335235378032, | |
| "grad_norm": 0.92668217420578, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.9602, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14265335235378032, | |
| "eval_loss": 0.9245217442512512, | |
| "eval_runtime": 44.2152, | |
| "eval_samples_per_second": 13.344, | |
| "eval_steps_per_second": 3.347, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14550641940085593, | |
| "grad_norm": 1.163582682609558, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.0826, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14835948644793154, | |
| "grad_norm": 0.860844075679779, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.8401, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.15121255349500715, | |
| "grad_norm": 0.7308872938156128, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8534, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.15406562054208273, | |
| "grad_norm": 0.7423079013824463, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.8807, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.15691868758915833, | |
| "grad_norm": 0.6839932799339294, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.8896, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15977175463623394, | |
| "grad_norm": 0.7370685935020447, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.7753, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16262482168330955, | |
| "grad_norm": 0.6701346635818481, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.8892, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.16547788873038516, | |
| "grad_norm": 0.6396673321723938, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.8337, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.16833095577746077, | |
| "grad_norm": 0.6428176164627075, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.9081, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.17118402282453637, | |
| "grad_norm": 0.6929706335067749, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.7898, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17403708987161198, | |
| "grad_norm": 0.6850852966308594, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.9256, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1768901569186876, | |
| "grad_norm": 0.6981192827224731, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.8607, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1797432239657632, | |
| "grad_norm": 0.6592012047767639, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.854, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.1825962910128388, | |
| "grad_norm": 0.6796256899833679, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.8215, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.18544935805991442, | |
| "grad_norm": 0.6764621138572693, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.9007, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18830242510699002, | |
| "grad_norm": 0.8056644201278687, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.8839, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.19115549215406563, | |
| "grad_norm": 0.7033549547195435, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.8739, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.19400855920114124, | |
| "grad_norm": 0.5774713754653931, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.803, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.19686162624821682, | |
| "grad_norm": 0.5844163298606873, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8469, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.19971469329529243, | |
| "grad_norm": 0.6252447366714478, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.8198, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.20256776034236804, | |
| "grad_norm": 0.8064114451408386, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.8879, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.20542082738944364, | |
| "grad_norm": 0.6135920286178589, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.7782, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.20827389443651925, | |
| "grad_norm": 0.6385617256164551, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.8332, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.21112696148359486, | |
| "grad_norm": 0.552251398563385, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8453, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.21398002853067047, | |
| "grad_norm": 0.6212217807769775, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.8323, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.21683309557774608, | |
| "grad_norm": 0.7010154724121094, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.8174, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.21968616262482168, | |
| "grad_norm": 0.639336884021759, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.845, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2225392296718973, | |
| "grad_norm": 0.604095458984375, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.8682, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2253922967189729, | |
| "grad_norm": 0.5614564418792725, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.8213, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2282453637660485, | |
| "grad_norm": 0.6672795414924622, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.8927, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23109843081312412, | |
| "grad_norm": 0.6684776544570923, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.9162, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.23395149786019973, | |
| "grad_norm": 0.5671095848083496, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.8638, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.23680456490727533, | |
| "grad_norm": 0.6492483615875244, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.9219, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.2396576319543509, | |
| "grad_norm": 0.6074286103248596, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8927, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.24251069900142652, | |
| "grad_norm": 0.6289499402046204, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.8405, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.24536376604850213, | |
| "grad_norm": 0.6300058960914612, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.8794, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.24821683309557774, | |
| "grad_norm": 0.5989180207252502, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.8615, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.25106990014265335, | |
| "grad_norm": 0.6682757139205933, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.911, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.25392296718972895, | |
| "grad_norm": 0.7386969327926636, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.8821, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.25677603423680456, | |
| "grad_norm": 0.5840036273002625, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.8406, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25962910128388017, | |
| "grad_norm": 0.5873855948448181, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.847, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2624821683309558, | |
| "grad_norm": 0.6577649712562561, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.8779, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2653352353780314, | |
| "grad_norm": 0.633472740650177, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.858, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.268188302425107, | |
| "grad_norm": 0.6111188530921936, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.9075, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2710413694721826, | |
| "grad_norm": 0.6667155027389526, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.9141, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2738944365192582, | |
| "grad_norm": 0.5913331508636475, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.8175, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2767475035663338, | |
| "grad_norm": 0.5792120695114136, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.9016, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2796005706134094, | |
| "grad_norm": 0.7187466025352478, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.8577, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.28245363766048504, | |
| "grad_norm": 0.6599798798561096, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.8689, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.28530670470756064, | |
| "grad_norm": 0.7733741998672485, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.9826, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28530670470756064, | |
| "eval_loss": 0.8544186353683472, | |
| "eval_runtime": 44.2802, | |
| "eval_samples_per_second": 13.324, | |
| "eval_steps_per_second": 3.342, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28815977175463625, | |
| "grad_norm": 0.6371700763702393, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.9248, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.29101283880171186, | |
| "grad_norm": 0.582548201084137, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.8146, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.29386590584878747, | |
| "grad_norm": 0.5589428544044495, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.8198, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2967189728958631, | |
| "grad_norm": 0.5687887072563171, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8444, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2995720399429387, | |
| "grad_norm": 0.5136210322380066, | |
| "learning_rate": 5e-05, | |
| "loss": 0.7763, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3024251069900143, | |
| "grad_norm": 0.5174034833908081, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.7564, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.30527817403708984, | |
| "grad_norm": 0.4906408488750458, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.7724, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.30813124108416545, | |
| "grad_norm": 0.5513845682144165, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.7418, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.31098430813124106, | |
| "grad_norm": 0.5535486936569214, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.7917, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.31383737517831667, | |
| "grad_norm": 0.5535765290260315, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.7924, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3166904422253923, | |
| "grad_norm": 0.49853044748306274, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.7095, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3195435092724679, | |
| "grad_norm": 0.5564809441566467, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.7611, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3223965763195435, | |
| "grad_norm": 0.5671737790107727, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.8143, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3252496433666191, | |
| "grad_norm": 0.5639700293540955, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.7561, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3281027104136947, | |
| "grad_norm": 0.5924969911575317, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7904, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3309557774607703, | |
| "grad_norm": 0.5341849327087402, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7737, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.3338088445078459, | |
| "grad_norm": 0.5442417860031128, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.7571, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.33666191155492153, | |
| "grad_norm": 0.5302444100379944, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7788, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.33951497860199714, | |
| "grad_norm": 0.5593897104263306, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.7699, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.34236804564907275, | |
| "grad_norm": 0.5808508992195129, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.8509, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.34522111269614836, | |
| "grad_norm": 0.5292861461639404, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.718, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.34807417974322397, | |
| "grad_norm": 0.5898811221122742, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.7856, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3509272467902996, | |
| "grad_norm": 0.5789220929145813, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.8545, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.3537803138373752, | |
| "grad_norm": 0.5804517865180969, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.7609, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3566333808844508, | |
| "grad_norm": 0.5174581408500671, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.7816, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3594864479315264, | |
| "grad_norm": 0.5828277468681335, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.7722, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.362339514978602, | |
| "grad_norm": 0.5580853819847107, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.8082, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.3651925820256776, | |
| "grad_norm": 0.5669066905975342, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.8631, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3680456490727532, | |
| "grad_norm": 0.5581154823303223, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.8243, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.37089871611982883, | |
| "grad_norm": 0.5976192951202393, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.9098, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.37375178316690444, | |
| "grad_norm": 0.5638813972473145, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.8379, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.37660485021398005, | |
| "grad_norm": 0.5447173714637756, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.7449, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.37945791726105566, | |
| "grad_norm": 0.5775190591812134, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.8201, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.38231098430813126, | |
| "grad_norm": 0.5627070069313049, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.8006, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.38516405135520687, | |
| "grad_norm": 0.6440720558166504, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.8581, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3880171184022825, | |
| "grad_norm": 0.5219485759735107, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.8322, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3908701854493581, | |
| "grad_norm": 0.54684978723526, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.8127, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.39372325249643364, | |
| "grad_norm": 0.5484306812286377, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.765, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.39657631954350925, | |
| "grad_norm": 0.612368106842041, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.9168, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.39942938659058486, | |
| "grad_norm": 0.5576868653297424, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.8076, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.40228245363766046, | |
| "grad_norm": 0.5026583075523376, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.7803, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4051355206847361, | |
| "grad_norm": 0.5895950198173523, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.8483, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.4079885877318117, | |
| "grad_norm": 0.5736966133117676, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.8441, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4108416547788873, | |
| "grad_norm": 0.5163689255714417, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.7666, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4136947218259629, | |
| "grad_norm": 0.5051657557487488, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.7557, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4165477888730385, | |
| "grad_norm": 0.5903991460800171, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.8585, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4194008559201141, | |
| "grad_norm": 0.6192286610603333, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.9259, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4222539229671897, | |
| "grad_norm": 0.653674840927124, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.8998, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.42510699001426533, | |
| "grad_norm": 0.6450168490409851, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.7758, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.42796005706134094, | |
| "grad_norm": 0.754431426525116, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.0083, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.42796005706134094, | |
| "eval_loss": 0.8174534440040588, | |
| "eval_runtime": 44.1274, | |
| "eval_samples_per_second": 13.37, | |
| "eval_steps_per_second": 3.354, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.43081312410841655, | |
| "grad_norm": 0.5710943341255188, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.9289, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.43366619115549215, | |
| "grad_norm": 0.5718706250190735, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.7838, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.43651925820256776, | |
| "grad_norm": 0.5101996064186096, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.7348, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.43937232524964337, | |
| "grad_norm": 0.5495441555976868, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.7565, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.442225392296719, | |
| "grad_norm": 0.5264116525650024, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.7803, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4450784593437946, | |
| "grad_norm": 0.4827347695827484, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.7598, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4479315263908702, | |
| "grad_norm": 0.4847021698951721, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.7653, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.4507845934379458, | |
| "grad_norm": 0.46831297874450684, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.7432, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.4536376604850214, | |
| "grad_norm": 0.5538355708122253, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.7706, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.456490727532097, | |
| "grad_norm": 0.5456767082214355, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.7238, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4593437945791726, | |
| "grad_norm": 0.510344922542572, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.7961, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.46219686162624823, | |
| "grad_norm": 0.5119059085845947, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.8135, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.46504992867332384, | |
| "grad_norm": 0.537604570388794, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.8445, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.46790299572039945, | |
| "grad_norm": 0.5314122438430786, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.7474, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.47075606276747506, | |
| "grad_norm": 0.4871026277542114, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.7464, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.47360912981455067, | |
| "grad_norm": 0.4640917181968689, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.7164, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.4764621968616263, | |
| "grad_norm": 0.5186988711357117, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.7444, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.4793152639087018, | |
| "grad_norm": 0.48805567622184753, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.7467, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.48216833095577744, | |
| "grad_norm": 0.5131738781929016, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.768, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.48502139800285304, | |
| "grad_norm": 0.5057012438774109, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.7508, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.48787446504992865, | |
| "grad_norm": 0.4922831952571869, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.7759, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.49072753209700426, | |
| "grad_norm": 0.5268054604530334, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.7905, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.49358059914407987, | |
| "grad_norm": 0.4818932116031647, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.7277, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.4964336661911555, | |
| "grad_norm": 0.5188502669334412, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.7417, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.4992867332382311, | |
| "grad_norm": 0.5614742636680603, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.8516, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5021398002853067, | |
| "grad_norm": 0.5036123991012573, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.7537, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5049928673323824, | |
| "grad_norm": 0.5600584745407104, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.8352, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5078459343794579, | |
| "grad_norm": 0.504410445690155, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.7724, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5106990014265336, | |
| "grad_norm": 0.5405497550964355, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.8254, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5135520684736091, | |
| "grad_norm": 0.5208389163017273, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.7788, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5164051355206848, | |
| "grad_norm": 0.5008552074432373, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.7517, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5192582025677603, | |
| "grad_norm": 0.5555158257484436, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.7226, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5221112696148359, | |
| "grad_norm": 0.5501844882965088, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.7685, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5249643366619116, | |
| "grad_norm": 0.5697614550590515, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.7982, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5278174037089871, | |
| "grad_norm": 0.6065782904624939, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.8692, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5306704707560628, | |
| "grad_norm": 0.5230533480644226, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.7677, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5335235378031383, | |
| "grad_norm": 0.5687628984451294, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.8118, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.536376604850214, | |
| "grad_norm": 0.5855110287666321, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.8243, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5392296718972895, | |
| "grad_norm": 0.5566447973251343, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.8549, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5420827389443652, | |
| "grad_norm": 0.593215823173523, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.8222, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5449358059914408, | |
| "grad_norm": 0.5914384126663208, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.8683, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5477888730385164, | |
| "grad_norm": 0.5680501461029053, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.8363, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.550641940085592, | |
| "grad_norm": 0.5938475131988525, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.8068, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5534950071326676, | |
| "grad_norm": 0.6767430901527405, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.8217, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5563480741797432, | |
| "grad_norm": 0.6291959285736084, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.8215, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.5592011412268189, | |
| "grad_norm": 0.5889286398887634, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.8574, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.5620542082738944, | |
| "grad_norm": 0.6157364249229431, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.814, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.5649072753209701, | |
| "grad_norm": 0.7902777194976807, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.9372, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5677603423680456, | |
| "grad_norm": 0.6634673476219177, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.9016, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.5706134094151213, | |
| "grad_norm": 0.8335981965065002, | |
| "learning_rate": 0.0, | |
| "loss": 0.9725, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5706134094151213, | |
| "eval_loss": 0.8103886842727661, | |
| "eval_runtime": 44.1543, | |
| "eval_samples_per_second": 13.362, | |
| "eval_steps_per_second": 3.352, | |
| "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": 3.017027157491712e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |