| { | |
| "best_metric": 11.87588882446289, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.05779511631267158, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00028897558156335787, | |
| "grad_norm": 0.1662612408399582, | |
| "learning_rate": 1e-05, | |
| "loss": 11.9392, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00028897558156335787, | |
| "eval_loss": 11.937902450561523, | |
| "eval_runtime": 57.3039, | |
| "eval_samples_per_second": 101.721, | |
| "eval_steps_per_second": 25.443, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0005779511631267157, | |
| "grad_norm": 0.1898190826177597, | |
| "learning_rate": 2e-05, | |
| "loss": 11.9385, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0008669267446900737, | |
| "grad_norm": 0.23333746194839478, | |
| "learning_rate": 3e-05, | |
| "loss": 11.936, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0011559023262534315, | |
| "grad_norm": 0.22852833569049835, | |
| "learning_rate": 4e-05, | |
| "loss": 11.9374, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0014448779078167894, | |
| "grad_norm": 0.20863816142082214, | |
| "learning_rate": 5e-05, | |
| "loss": 11.9391, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0017338534893801473, | |
| "grad_norm": 0.21721790730953217, | |
| "learning_rate": 6e-05, | |
| "loss": 11.9362, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0020228290709435053, | |
| "grad_norm": 0.21195803582668304, | |
| "learning_rate": 7e-05, | |
| "loss": 11.9369, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.002311804652506863, | |
| "grad_norm": 0.20923803746700287, | |
| "learning_rate": 8e-05, | |
| "loss": 11.9357, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.002600780234070221, | |
| "grad_norm": 0.2240084558725357, | |
| "learning_rate": 9e-05, | |
| "loss": 11.9315, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.002889755815633579, | |
| "grad_norm": 0.24352015554904938, | |
| "learning_rate": 0.0001, | |
| "loss": 11.9296, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.003178731397196937, | |
| "grad_norm": 0.22750435769557953, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 11.9317, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0034677069787602947, | |
| "grad_norm": 0.27325478196144104, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 11.9264, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.003756682560323653, | |
| "grad_norm": 0.2954610586166382, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 11.9238, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0040456581418870105, | |
| "grad_norm": 0.3273448348045349, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 11.9213, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.004334633723450368, | |
| "grad_norm": 0.3554380536079407, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 11.9201, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.004623609305013726, | |
| "grad_norm": 0.3640565574169159, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 11.9179, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0049125848865770845, | |
| "grad_norm": 0.39382296800613403, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 11.9134, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.005201560468140442, | |
| "grad_norm": 0.38839638233184814, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 11.9102, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0054905360497038, | |
| "grad_norm": 0.43728920817375183, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 11.9076, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.005779511631267158, | |
| "grad_norm": 0.40401872992515564, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 11.9071, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.006068487212830516, | |
| "grad_norm": 0.39821857213974, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 11.9039, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.006357462794393874, | |
| "grad_norm": 0.35098686814308167, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 11.8988, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.006646438375957232, | |
| "grad_norm": 0.36843863129615784, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 11.8981, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.006935413957520589, | |
| "grad_norm": 0.3306281864643097, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 11.8983, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.007224389539083947, | |
| "grad_norm": 0.29317593574523926, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 11.8959, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.007513365120647306, | |
| "grad_norm": 0.30842310190200806, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 11.8918, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.007802340702210663, | |
| "grad_norm": 0.27182668447494507, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 11.8946, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.008091316283774021, | |
| "grad_norm": 0.2669576108455658, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 11.8926, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.00838029186533738, | |
| "grad_norm": 0.2471419870853424, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 11.8912, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.008669267446900737, | |
| "grad_norm": 0.23154519498348236, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 11.8833, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.008958243028464095, | |
| "grad_norm": 0.20965711772441864, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 11.8854, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.009247218610027452, | |
| "grad_norm": 0.21031919121742249, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 11.8807, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.00953619419159081, | |
| "grad_norm": 0.17498861253261566, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 11.8851, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.009825169773154169, | |
| "grad_norm": 0.19231873750686646, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 11.8814, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.010114145354717526, | |
| "grad_norm": 0.1473800390958786, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 11.8776, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.010403120936280884, | |
| "grad_norm": 0.20245665311813354, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 11.8738, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.010692096517844241, | |
| "grad_norm": 0.16064031422138214, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 11.88, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0109810720994076, | |
| "grad_norm": 0.14997169375419617, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 11.8783, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.011270047680970958, | |
| "grad_norm": 0.14303217828273773, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 11.8837, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.011559023262534315, | |
| "grad_norm": 0.18485713005065918, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 11.8801, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.011847998844097674, | |
| "grad_norm": 0.1774420440196991, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 11.8742, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.012136974425661032, | |
| "grad_norm": 0.16609381139278412, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 11.8773, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01242595000722439, | |
| "grad_norm": 0.1143420934677124, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 11.8739, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.012714925588787748, | |
| "grad_norm": 0.15700672566890717, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 11.8843, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.013003901170351105, | |
| "grad_norm": 0.1422463357448578, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 11.8775, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.013292876751914463, | |
| "grad_norm": 0.14469079673290253, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 11.8845, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.013581852333477822, | |
| "grad_norm": 0.12706555426120758, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 11.889, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.013870827915041179, | |
| "grad_norm": 0.1611371487379074, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 11.8882, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.014159803496604537, | |
| "grad_norm": 0.16547800600528717, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 11.8953, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.014448779078167894, | |
| "grad_norm": 0.21278278529644012, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 11.8898, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.014448779078167894, | |
| "eval_loss": 11.878070831298828, | |
| "eval_runtime": 55.8059, | |
| "eval_samples_per_second": 104.451, | |
| "eval_steps_per_second": 26.126, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.014737754659731253, | |
| "grad_norm": 0.16262826323509216, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 11.8881, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.015026730241294611, | |
| "grad_norm": 0.1444682627916336, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 11.8799, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.015315705822857968, | |
| "grad_norm": 0.13952292501926422, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 11.8751, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.015604681404421327, | |
| "grad_norm": 0.14246442914009094, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 11.8753, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.015893656985984685, | |
| "grad_norm": 0.13902343809604645, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 11.875, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.016182632567548042, | |
| "grad_norm": 0.12585100531578064, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 11.8762, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0164716081491114, | |
| "grad_norm": 0.11996205896139145, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 11.876, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.01676058373067476, | |
| "grad_norm": 0.12303932756185532, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 11.8751, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.017049559312238116, | |
| "grad_norm": 0.12264929711818695, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 11.8785, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.017338534893801473, | |
| "grad_norm": 0.26482313871383667, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 11.874, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.017627510475364833, | |
| "grad_norm": 0.11297342926263809, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 11.8701, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.01791648605692819, | |
| "grad_norm": 0.10358939319849014, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 11.8711, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.018205461638491547, | |
| "grad_norm": 0.09097111225128174, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 11.871, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.018494437220054904, | |
| "grad_norm": 0.10291942954063416, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 11.8736, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.018783412801618264, | |
| "grad_norm": 0.07753278315067291, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 11.8751, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.01907238838318162, | |
| "grad_norm": 0.07438040524721146, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 11.8739, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.019361363964744978, | |
| "grad_norm": 0.10090917348861694, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 11.8732, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.019650339546308338, | |
| "grad_norm": 0.07704046368598938, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 11.8775, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.019939315127871695, | |
| "grad_norm": 0.07107193768024445, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 11.8719, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.020228290709435052, | |
| "grad_norm": 0.06724188476800919, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 11.8785, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.020517266290998412, | |
| "grad_norm": 0.07781937718391418, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 11.8789, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.02080624187256177, | |
| "grad_norm": 0.08545064181089401, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 11.8764, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.021095217454125126, | |
| "grad_norm": 0.08945167064666748, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 11.8754, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.021384193035688483, | |
| "grad_norm": 0.06463414430618286, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 11.8723, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.021673168617251843, | |
| "grad_norm": 0.07582932710647583, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 11.8774, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0219621441988152, | |
| "grad_norm": 0.07794112712144852, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 11.8806, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.022251119780378557, | |
| "grad_norm": 0.08979761600494385, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 11.8799, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.022540095361941917, | |
| "grad_norm": 0.0700855702161789, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 11.8774, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.022829070943505274, | |
| "grad_norm": 0.21359409391880035, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 11.878, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.02311804652506863, | |
| "grad_norm": 0.0810750424861908, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 11.8812, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02340702210663199, | |
| "grad_norm": 0.07715233415365219, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 11.8752, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.023695997688195348, | |
| "grad_norm": 0.07944002002477646, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 11.8716, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.023984973269758705, | |
| "grad_norm": 0.06868164241313934, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 11.873, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.024273948851322065, | |
| "grad_norm": 0.09088148176670074, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 11.8782, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.024562924432885422, | |
| "grad_norm": 0.08253742009401321, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 11.8689, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.02485190001444878, | |
| "grad_norm": 0.07526249438524246, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 11.8718, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.025140875596012136, | |
| "grad_norm": 0.10941426455974579, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 11.8805, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.025429851177575496, | |
| "grad_norm": 0.10719692707061768, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 11.8822, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.025718826759138853, | |
| "grad_norm": 0.10798517614603043, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 11.8768, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.02600780234070221, | |
| "grad_norm": 0.1495065987110138, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 11.8726, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02629677792226557, | |
| "grad_norm": 0.1519833654165268, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 11.8749, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.026585753503828927, | |
| "grad_norm": 0.10712479054927826, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 11.8784, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.026874729085392284, | |
| "grad_norm": 0.13148878514766693, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 11.8772, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.027163704666955644, | |
| "grad_norm": 0.10946288704872131, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 11.8812, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.027452680248519, | |
| "grad_norm": 0.11115029454231262, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 11.8842, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.027741655830082357, | |
| "grad_norm": 0.10704438388347626, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 11.8809, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.028030631411645714, | |
| "grad_norm": 0.12639576196670532, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 11.8794, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.028319606993209075, | |
| "grad_norm": 0.10369209945201874, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 11.8835, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.02860858257477243, | |
| "grad_norm": 0.1431816667318344, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 11.8797, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.02889755815633579, | |
| "grad_norm": 0.15824422240257263, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 11.879, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02889755815633579, | |
| "eval_loss": 11.876375198364258, | |
| "eval_runtime": 55.1639, | |
| "eval_samples_per_second": 105.667, | |
| "eval_steps_per_second": 26.43, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02918653373789915, | |
| "grad_norm": 0.09448030591011047, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 11.8769, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.029475509319462505, | |
| "grad_norm": 0.09203047305345535, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 11.8809, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.029764484901025862, | |
| "grad_norm": 0.10378605127334595, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 11.8723, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.030053460482589223, | |
| "grad_norm": 0.07438343018293381, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 11.8738, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.03034243606415258, | |
| "grad_norm": 0.08363966643810272, | |
| "learning_rate": 5e-05, | |
| "loss": 11.8704, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.030631411645715936, | |
| "grad_norm": 0.07607440650463104, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 11.8709, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.030920387227279293, | |
| "grad_norm": 0.07229042798280716, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 11.8734, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.031209362808842653, | |
| "grad_norm": 0.0885797068476677, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 11.8713, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.031498338390406014, | |
| "grad_norm": 0.08899901807308197, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 11.8746, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.03178731397196937, | |
| "grad_norm": 0.0725225955247879, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 11.8724, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.03207628955353273, | |
| "grad_norm": 0.07378460466861725, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 11.868, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.032365265135096084, | |
| "grad_norm": 0.0828651487827301, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 11.8712, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.03265424071665944, | |
| "grad_norm": 0.0730227530002594, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 11.8706, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0329432162982228, | |
| "grad_norm": 0.07580568641424179, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 11.8719, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.033232191879786155, | |
| "grad_norm": 0.06368804723024368, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 11.8718, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.03352116746134952, | |
| "grad_norm": 0.08921920508146286, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 11.8736, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.033810143042912875, | |
| "grad_norm": 0.0802396759390831, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 11.8763, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.03409911862447623, | |
| "grad_norm": 0.06945715099573135, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 11.8732, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.03438809420603959, | |
| "grad_norm": 0.07443312555551529, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 11.871, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.034677069787602946, | |
| "grad_norm": 0.07826009392738342, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 11.8703, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0349660453691663, | |
| "grad_norm": 0.06413071602582932, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 11.8765, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.03525502095072967, | |
| "grad_norm": 0.06165166571736336, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 11.8742, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.035543996532293023, | |
| "grad_norm": 0.06768231838941574, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 11.8728, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.03583297211385638, | |
| "grad_norm": 0.08700777590274811, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 11.8732, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.03612194769541974, | |
| "grad_norm": 0.05918419361114502, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 11.8749, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.036410923276983094, | |
| "grad_norm": 0.05309590324759483, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 11.8737, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.03669989885854645, | |
| "grad_norm": 0.07097380608320236, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 11.8802, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.03698887444010981, | |
| "grad_norm": 0.0766899362206459, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 11.8779, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.03727785002167317, | |
| "grad_norm": 0.06368804723024368, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 11.8867, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.03756682560323653, | |
| "grad_norm": 0.06508305668830872, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 11.882, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.037855801184799885, | |
| "grad_norm": 0.08121135830879211, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 11.8766, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.03814477676636324, | |
| "grad_norm": 0.07969960570335388, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 11.8755, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0384337523479266, | |
| "grad_norm": 0.08134438842535019, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 11.8782, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.038722727929489956, | |
| "grad_norm": 0.07282125949859619, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 11.8758, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.03901170351105332, | |
| "grad_norm": 0.0943041443824768, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 11.8676, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.039300679092616676, | |
| "grad_norm": 0.09704722464084625, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 11.8776, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.03958965467418003, | |
| "grad_norm": 0.08427353948354721, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 11.8774, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.03987863025574339, | |
| "grad_norm": 0.07390493154525757, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 11.8738, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.04016760583730675, | |
| "grad_norm": 0.11108248680830002, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 11.8792, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.040456581418870104, | |
| "grad_norm": 0.117488794028759, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 11.8702, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04074555700043346, | |
| "grad_norm": 0.112480029463768, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 11.8731, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.041034532581996824, | |
| "grad_norm": 0.11315959692001343, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 11.8755, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.04132350816356018, | |
| "grad_norm": 0.1187959685921669, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 11.8769, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.04161248374512354, | |
| "grad_norm": 0.08921589702367783, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 11.8715, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.041901459326686895, | |
| "grad_norm": 0.10372520238161087, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 11.8759, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.04219043490825025, | |
| "grad_norm": 0.10234490782022476, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 11.8821, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.04247941048981361, | |
| "grad_norm": 0.10286494344472885, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 11.8825, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.042768386071376965, | |
| "grad_norm": 0.0980636477470398, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 11.8819, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.04305736165294033, | |
| "grad_norm": 0.1364598423242569, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 11.8917, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.043346337234503686, | |
| "grad_norm": 0.11411885917186737, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 11.8883, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.043346337234503686, | |
| "eval_loss": 11.875956535339355, | |
| "eval_runtime": 54.9073, | |
| "eval_samples_per_second": 106.161, | |
| "eval_steps_per_second": 26.554, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04363531281606704, | |
| "grad_norm": 0.08824671804904938, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 11.8799, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0439242883976304, | |
| "grad_norm": 0.08539780229330063, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 11.88, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.044213263979193757, | |
| "grad_norm": 0.05796075984835625, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 11.8723, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.04450223956075711, | |
| "grad_norm": 0.08143189549446106, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 11.8752, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.04479121514232048, | |
| "grad_norm": 0.06231541931629181, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 11.8708, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.045080190723883834, | |
| "grad_norm": 0.06014825403690338, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 11.8727, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.04536916630544719, | |
| "grad_norm": 0.07167522609233856, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 11.8758, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.04565814188701055, | |
| "grad_norm": 0.11054030060768127, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 11.8743, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.045947117468573904, | |
| "grad_norm": 0.07149700075387955, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 11.8698, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.04623609305013726, | |
| "grad_norm": 0.05823131650686264, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 11.8709, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.04652506863170062, | |
| "grad_norm": 0.05063368007540703, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 11.8702, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.04681404421326398, | |
| "grad_norm": 0.07420355081558228, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 11.8714, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.04710301979482734, | |
| "grad_norm": 0.054763611406087875, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 11.8719, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.047391995376390696, | |
| "grad_norm": 0.06387555599212646, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 11.8699, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.04768097095795405, | |
| "grad_norm": 0.06474991887807846, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 11.8763, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.04796994653951741, | |
| "grad_norm": 0.06635675579309464, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 11.8682, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.048258922121080766, | |
| "grad_norm": 0.05499062314629555, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 11.8744, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.04854789770264413, | |
| "grad_norm": 0.05452093109488487, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 11.8758, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.04883687328420749, | |
| "grad_norm": 0.0677514374256134, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 11.8707, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.049125848865770844, | |
| "grad_norm": 0.0678141862154007, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 11.8767, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0494148244473342, | |
| "grad_norm": 0.0642886683344841, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 11.8783, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.04970380002889756, | |
| "grad_norm": 0.08009558171033859, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 11.8754, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.049992775610460914, | |
| "grad_norm": 0.05672631412744522, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 11.872, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.05028175119202427, | |
| "grad_norm": 0.06327534466981888, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 11.8754, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.050570726773587635, | |
| "grad_norm": 0.06763085722923279, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 11.8807, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.05085970235515099, | |
| "grad_norm": 0.05437658354640007, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 11.8829, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.05114867793671435, | |
| "grad_norm": 0.0749303475022316, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 11.8779, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.051437653518277705, | |
| "grad_norm": 0.06602714955806732, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 11.8802, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.05172662909984106, | |
| "grad_norm": 0.05587778240442276, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 11.8799, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05201560468140442, | |
| "grad_norm": 0.06512124091386795, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 11.8785, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.052304580262967776, | |
| "grad_norm": 0.07296937704086304, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 11.8717, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.05259355584453114, | |
| "grad_norm": 0.05819704011082649, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 11.883, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.052882531426094496, | |
| "grad_norm": 0.05296211689710617, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 11.8713, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.05317150700765785, | |
| "grad_norm": 0.08113344013690948, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 11.8739, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.05346048258922121, | |
| "grad_norm": 0.07535281032323837, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 11.8735, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.05374945817078457, | |
| "grad_norm": 0.07368480414152145, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 11.869, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.054038433752347924, | |
| "grad_norm": 0.0869714617729187, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 11.8765, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.05432740933391129, | |
| "grad_norm": 0.07330923527479172, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 11.8766, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.054616384915474644, | |
| "grad_norm": 0.07783111184835434, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 11.8743, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.054905360497038, | |
| "grad_norm": 0.09879664331674576, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 11.8771, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.05519433607860136, | |
| "grad_norm": 0.10569307208061218, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 11.8759, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.055483311660164715, | |
| "grad_norm": 0.0905223861336708, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 11.8817, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.05577228724172807, | |
| "grad_norm": 0.09453146159648895, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 11.8735, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.05606126282329143, | |
| "grad_norm": 0.10134062170982361, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 11.8776, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.05635023840485479, | |
| "grad_norm": 0.08888080716133118, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 11.8807, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.05663921398641815, | |
| "grad_norm": 0.11437094956636429, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 11.8804, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.056928189567981506, | |
| "grad_norm": 0.11957822740077972, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 11.881, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.05721716514954486, | |
| "grad_norm": 0.15185602009296417, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 11.8869, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.05750614073110822, | |
| "grad_norm": 0.1162104457616806, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 11.8869, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.05779511631267158, | |
| "grad_norm": 0.16675637662410736, | |
| "learning_rate": 0.0, | |
| "loss": 11.8887, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05779511631267158, | |
| "eval_loss": 11.87588882446289, | |
| "eval_runtime": 54.5453, | |
| "eval_samples_per_second": 106.865, | |
| "eval_steps_per_second": 26.73, | |
| "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": 220633694208000.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |