| { | |
| "best_metric": 2.3414769172668457, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8213552361396304, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004106776180698152, | |
| "grad_norm": 1.0899757146835327, | |
| "learning_rate": 1e-05, | |
| "loss": 2.5899, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004106776180698152, | |
| "eval_loss": 2.63228440284729, | |
| "eval_runtime": 44.3725, | |
| "eval_samples_per_second": 9.24, | |
| "eval_steps_per_second": 2.321, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008213552361396304, | |
| "grad_norm": 1.239064335823059, | |
| "learning_rate": 2e-05, | |
| "loss": 2.6126, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012320328542094456, | |
| "grad_norm": 1.2054294347763062, | |
| "learning_rate": 3e-05, | |
| "loss": 2.615, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01642710472279261, | |
| "grad_norm": 1.0069154500961304, | |
| "learning_rate": 4e-05, | |
| "loss": 2.6293, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02053388090349076, | |
| "grad_norm": 0.5620794892311096, | |
| "learning_rate": 5e-05, | |
| "loss": 2.4543, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.024640657084188913, | |
| "grad_norm": 1.1997545957565308, | |
| "learning_rate": 6e-05, | |
| "loss": 2.5298, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028747433264887063, | |
| "grad_norm": 1.3023887872695923, | |
| "learning_rate": 7e-05, | |
| "loss": 2.4727, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03285420944558522, | |
| "grad_norm": 0.7572681903839111, | |
| "learning_rate": 8e-05, | |
| "loss": 2.4752, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03696098562628337, | |
| "grad_norm": 0.512010931968689, | |
| "learning_rate": 9e-05, | |
| "loss": 2.4435, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04106776180698152, | |
| "grad_norm": 0.6867216229438782, | |
| "learning_rate": 0.0001, | |
| "loss": 2.4224, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.045174537987679675, | |
| "grad_norm": 0.5566503405570984, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.478, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.049281314168377825, | |
| "grad_norm": 0.31926611065864563, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.3247, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.053388090349075976, | |
| "grad_norm": 0.574629008769989, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.4465, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.057494866529774126, | |
| "grad_norm": 0.49882209300994873, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 2.3676, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.061601642710472276, | |
| "grad_norm": 0.3579407036304474, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 2.4046, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06570841889117043, | |
| "grad_norm": 0.2933938801288605, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.404, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06981519507186858, | |
| "grad_norm": 0.37158146500587463, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.3941, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07392197125256673, | |
| "grad_norm": 0.3912244141101837, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 2.4098, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07802874743326489, | |
| "grad_norm": 0.34771305322647095, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.4358, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08213552361396304, | |
| "grad_norm": 0.34832507371902466, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.3442, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08624229979466119, | |
| "grad_norm": 0.3815344572067261, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.4029, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09034907597535935, | |
| "grad_norm": 0.3517075777053833, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.4167, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0944558521560575, | |
| "grad_norm": 0.34069788455963135, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.3861, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09856262833675565, | |
| "grad_norm": 0.3173888921737671, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.4325, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1026694045174538, | |
| "grad_norm": 0.31365448236465454, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.3636, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10677618069815195, | |
| "grad_norm": 0.31624218821525574, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.3444, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11088295687885011, | |
| "grad_norm": 0.32835638523101807, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.3344, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11498973305954825, | |
| "grad_norm": 0.3243182897567749, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.4137, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11909650924024641, | |
| "grad_norm": 0.3379705846309662, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.3644, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12320328542094455, | |
| "grad_norm": 0.41938579082489014, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.4177, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1273100616016427, | |
| "grad_norm": 0.33362093567848206, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.4126, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13141683778234087, | |
| "grad_norm": 0.3650400638580322, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.3737, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13552361396303902, | |
| "grad_norm": 0.34693506360054016, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.4024, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13963039014373715, | |
| "grad_norm": 0.3428874909877777, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.3828, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1437371663244353, | |
| "grad_norm": 0.32840782403945923, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.3475, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14784394250513347, | |
| "grad_norm": 0.33618059754371643, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.4255, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15195071868583163, | |
| "grad_norm": 0.3571118414402008, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.4813, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15605749486652978, | |
| "grad_norm": 0.33389055728912354, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.3549, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1601642710472279, | |
| "grad_norm": 0.3478274345397949, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.4313, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16427104722792607, | |
| "grad_norm": 0.35416093468666077, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.4574, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16837782340862423, | |
| "grad_norm": 0.35133761167526245, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.4085, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17248459958932238, | |
| "grad_norm": 0.3783772587776184, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.412, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17659137577002054, | |
| "grad_norm": 0.37258121371269226, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.4296, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1806981519507187, | |
| "grad_norm": 0.35900548100471497, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.3735, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18480492813141683, | |
| "grad_norm": 0.3691897690296173, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.4797, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.188911704312115, | |
| "grad_norm": 0.364868700504303, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.372, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.19301848049281314, | |
| "grad_norm": 0.4236624836921692, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.3999, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1971252566735113, | |
| "grad_norm": 0.39137449860572815, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.4114, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.20123203285420946, | |
| "grad_norm": 0.40828824043273926, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.3196, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2053388090349076, | |
| "grad_norm": 0.43899235129356384, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.3784, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2053388090349076, | |
| "eval_loss": 2.397331476211548, | |
| "eval_runtime": 45.5288, | |
| "eval_samples_per_second": 9.005, | |
| "eval_steps_per_second": 2.262, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.20944558521560575, | |
| "grad_norm": 0.593197226524353, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.4285, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2135523613963039, | |
| "grad_norm": 0.38888007402420044, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.4083, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21765913757700206, | |
| "grad_norm": 0.3479970097541809, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.4482, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.22176591375770022, | |
| "grad_norm": 0.34518498182296753, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.2901, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22587268993839835, | |
| "grad_norm": 0.28937461972236633, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.3661, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.2299794661190965, | |
| "grad_norm": 0.2808569371700287, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.4229, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23408624229979466, | |
| "grad_norm": 0.25724881887435913, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.3183, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23819301848049282, | |
| "grad_norm": 0.24696066975593567, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 2.3377, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.24229979466119098, | |
| "grad_norm": 0.2605307996273041, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.3106, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2464065708418891, | |
| "grad_norm": 0.2634081542491913, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 2.4052, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.25051334702258726, | |
| "grad_norm": 0.2708021104335785, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 2.3389, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.2546201232032854, | |
| "grad_norm": 0.246720552444458, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 2.387, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2587268993839836, | |
| "grad_norm": 0.25167909264564514, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 2.2832, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.26283367556468173, | |
| "grad_norm": 0.24080559611320496, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 2.3747, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2669404517453799, | |
| "grad_norm": 0.2520730197429657, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 2.3256, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.27104722792607805, | |
| "grad_norm": 0.26964616775512695, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 2.282, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2751540041067762, | |
| "grad_norm": 0.2610137462615967, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 2.2795, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2792607802874743, | |
| "grad_norm": 0.256756067276001, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 2.2797, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.28336755646817247, | |
| "grad_norm": 0.2705448269844055, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 2.2721, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2874743326488706, | |
| "grad_norm": 0.2950783371925354, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 2.3952, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2915811088295688, | |
| "grad_norm": 0.2838421165943146, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 2.2959, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.29568788501026694, | |
| "grad_norm": 0.27813395857810974, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 2.3539, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2997946611909651, | |
| "grad_norm": 0.27417752146720886, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 2.3633, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.30390143737166325, | |
| "grad_norm": 0.2806124985218048, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 2.3968, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3080082135523614, | |
| "grad_norm": 0.29030802845954895, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 2.3808, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.31211498973305957, | |
| "grad_norm": 0.29869508743286133, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 2.3406, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3162217659137577, | |
| "grad_norm": 0.29406699538230896, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.3635, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3203285420944558, | |
| "grad_norm": 0.29778364300727844, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 2.3691, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.324435318275154, | |
| "grad_norm": 0.3259764313697815, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.3766, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.32854209445585214, | |
| "grad_norm": 0.30980968475341797, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.3656, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3326488706365503, | |
| "grad_norm": 0.3172374963760376, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.3187, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.33675564681724846, | |
| "grad_norm": 0.33012300729751587, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.37, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3408624229979466, | |
| "grad_norm": 0.3301931321620941, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.3215, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.34496919917864477, | |
| "grad_norm": 0.3144400417804718, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.2989, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3490759753593429, | |
| "grad_norm": 0.3052152991294861, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.3787, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3531827515400411, | |
| "grad_norm": 0.3368039131164551, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.4431, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35728952772073924, | |
| "grad_norm": 0.31848573684692383, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.4343, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3613963039014374, | |
| "grad_norm": 0.32096171379089355, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.2584, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3655030800821355, | |
| "grad_norm": 0.3516496419906616, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.3355, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36960985626283366, | |
| "grad_norm": 0.34917670488357544, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.3697, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3737166324435318, | |
| "grad_norm": 0.34420329332351685, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.2983, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37782340862423, | |
| "grad_norm": 0.3400464355945587, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.4064, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.38193018480492813, | |
| "grad_norm": 0.3461991548538208, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.2898, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3860369609856263, | |
| "grad_norm": 0.3436702489852905, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.347, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.39014373716632444, | |
| "grad_norm": 0.3600529730319977, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.339, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3942505133470226, | |
| "grad_norm": 0.3684062063694, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.4109, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.39835728952772076, | |
| "grad_norm": 0.39061105251312256, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.4639, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.4024640657084189, | |
| "grad_norm": 0.3650188744068146, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.3235, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.406570841889117, | |
| "grad_norm": 0.4232775866985321, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.3737, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4106776180698152, | |
| "grad_norm": 0.4435172975063324, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.3637, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4106776180698152, | |
| "eval_loss": 2.363128185272217, | |
| "eval_runtime": 45.3796, | |
| "eval_samples_per_second": 9.035, | |
| "eval_steps_per_second": 2.27, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.41478439425051333, | |
| "grad_norm": 0.37182050943374634, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.4067, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4188911704312115, | |
| "grad_norm": 0.3280104398727417, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.3489, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.42299794661190965, | |
| "grad_norm": 0.3283444344997406, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.4303, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4271047227926078, | |
| "grad_norm": 0.27891355752944946, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.2764, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.43121149897330596, | |
| "grad_norm": 0.2890203297138214, | |
| "learning_rate": 5e-05, | |
| "loss": 2.4135, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4353182751540041, | |
| "grad_norm": 0.2661040127277374, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 2.366, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4394250513347023, | |
| "grad_norm": 0.2523798644542694, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 2.3247, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.44353182751540043, | |
| "grad_norm": 0.24263252317905426, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 2.3438, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44763860369609854, | |
| "grad_norm": 0.2515197694301605, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 2.3421, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.4517453798767967, | |
| "grad_norm": 0.24143071472644806, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 2.3141, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.45585215605749485, | |
| "grad_norm": 0.2452765256166458, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 2.366, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.459958932238193, | |
| "grad_norm": 0.2681790888309479, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.3442, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.46406570841889117, | |
| "grad_norm": 0.24382120370864868, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 2.2685, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4681724845995893, | |
| "grad_norm": 0.24417120218276978, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 2.3444, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.4722792607802875, | |
| "grad_norm": 0.24526545405387878, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.3194, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.47638603696098564, | |
| "grad_norm": 0.24900095164775848, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 2.3091, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4804928131416838, | |
| "grad_norm": 0.2626597285270691, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 2.3742, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.48459958932238195, | |
| "grad_norm": 0.27201834321022034, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 2.2711, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4887063655030801, | |
| "grad_norm": 0.26552215218544006, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.3693, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4928131416837782, | |
| "grad_norm": 0.2703514099121094, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.3422, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.49691991786447637, | |
| "grad_norm": 0.26164478063583374, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.348, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5010266940451745, | |
| "grad_norm": 0.2846229076385498, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 2.3089, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5051334702258727, | |
| "grad_norm": 0.275920033454895, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 2.3168, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5092402464065708, | |
| "grad_norm": 0.28674623370170593, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.2891, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5133470225872689, | |
| "grad_norm": 0.2734324336051941, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 2.3072, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5174537987679672, | |
| "grad_norm": 0.2848536968231201, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 2.3871, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5215605749486653, | |
| "grad_norm": 0.30089619755744934, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 2.3321, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5256673511293635, | |
| "grad_norm": 0.2885209023952484, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 2.3473, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5297741273100616, | |
| "grad_norm": 0.2830330729484558, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 2.3278, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5338809034907598, | |
| "grad_norm": 0.2817167639732361, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 2.3005, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5379876796714579, | |
| "grad_norm": 0.2884472608566284, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.3133, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5420944558521561, | |
| "grad_norm": 0.2849859595298767, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 2.3397, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5462012320328542, | |
| "grad_norm": 0.3301396369934082, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 2.3441, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5503080082135524, | |
| "grad_norm": 0.3178142011165619, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.316, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5544147843942505, | |
| "grad_norm": 0.30877718329429626, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.2844, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5585215605749486, | |
| "grad_norm": 0.3154371678829193, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.3523, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5626283367556468, | |
| "grad_norm": 0.3181307017803192, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.3735, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5667351129363449, | |
| "grad_norm": 0.32128021121025085, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.3183, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5708418891170431, | |
| "grad_norm": 0.327861487865448, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.4364, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5749486652977412, | |
| "grad_norm": 0.32442885637283325, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.4305, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5790554414784395, | |
| "grad_norm": 0.3366475999355316, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.3708, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5831622176591376, | |
| "grad_norm": 0.3531894385814667, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.4358, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5872689938398358, | |
| "grad_norm": 0.3379497826099396, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.3567, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5913757700205339, | |
| "grad_norm": 0.35603824257850647, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.3427, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5954825462012321, | |
| "grad_norm": 0.35133254528045654, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.3505, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5995893223819302, | |
| "grad_norm": 0.3662244379520416, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.4185, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6036960985626283, | |
| "grad_norm": 0.37666887044906616, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.365, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6078028747433265, | |
| "grad_norm": 0.3985544443130493, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.4046, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6119096509240246, | |
| "grad_norm": 0.4093400835990906, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.3763, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6160164271047228, | |
| "grad_norm": 0.43639683723449707, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.3459, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6160164271047228, | |
| "eval_loss": 2.344456672668457, | |
| "eval_runtime": 45.5018, | |
| "eval_samples_per_second": 9.011, | |
| "eval_steps_per_second": 2.264, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6201232032854209, | |
| "grad_norm": 0.2805425524711609, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.3709, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6242299794661191, | |
| "grad_norm": 0.309853732585907, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.429, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6283367556468172, | |
| "grad_norm": 0.33582112193107605, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.3477, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6324435318275154, | |
| "grad_norm": 0.29468798637390137, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.3359, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6365503080082136, | |
| "grad_norm": 0.28757351636886597, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.3597, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6406570841889117, | |
| "grad_norm": 0.27962955832481384, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.3355, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6447638603696099, | |
| "grad_norm": 0.267606645822525, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 2.3099, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.648870636550308, | |
| "grad_norm": 0.257451593875885, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 2.3082, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6529774127310062, | |
| "grad_norm": 0.26463115215301514, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.3496, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6570841889117043, | |
| "grad_norm": 0.2509690225124359, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.2506, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6611909650924025, | |
| "grad_norm": 0.2563090920448303, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.3512, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6652977412731006, | |
| "grad_norm": 0.24937409162521362, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.3201, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6694045174537988, | |
| "grad_norm": 0.24155984818935394, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 2.3305, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6735112936344969, | |
| "grad_norm": 0.24877284467220306, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.2932, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6776180698151951, | |
| "grad_norm": 0.2494582235813141, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 2.2714, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6817248459958932, | |
| "grad_norm": 0.25818485021591187, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 2.2696, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6858316221765913, | |
| "grad_norm": 0.2597648501396179, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 2.3384, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6899383983572895, | |
| "grad_norm": 0.2583504617214203, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 2.2874, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6940451745379876, | |
| "grad_norm": 0.25682592391967773, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 2.3395, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6981519507186859, | |
| "grad_norm": 0.27451130747795105, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 2.2966, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.702258726899384, | |
| "grad_norm": 0.269361287355423, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 2.325, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7063655030800822, | |
| "grad_norm": 0.2751413583755493, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 2.3112, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7104722792607803, | |
| "grad_norm": 0.27977946400642395, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 2.2973, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7145790554414785, | |
| "grad_norm": 0.27026399970054626, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 2.381, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7186858316221766, | |
| "grad_norm": 0.28504830598831177, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 2.2967, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7227926078028748, | |
| "grad_norm": 0.30635881423950195, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 2.276, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7268993839835729, | |
| "grad_norm": 0.2944948077201843, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 2.2743, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.731006160164271, | |
| "grad_norm": 0.2819386422634125, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.2921, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7351129363449692, | |
| "grad_norm": 0.296308696269989, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 2.3994, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7392197125256673, | |
| "grad_norm": 0.304075688123703, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 2.271, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7433264887063655, | |
| "grad_norm": 0.30602318048477173, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 2.3283, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7474332648870636, | |
| "grad_norm": 0.303476482629776, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.3518, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7515400410677618, | |
| "grad_norm": 0.30692949891090393, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 2.3687, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.75564681724846, | |
| "grad_norm": 0.32387521862983704, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.2876, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7597535934291582, | |
| "grad_norm": 0.3372074365615845, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.3528, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7638603696098563, | |
| "grad_norm": 0.3233869671821594, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.3374, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7679671457905544, | |
| "grad_norm": 0.3280622661113739, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.3641, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7720739219712526, | |
| "grad_norm": 0.32876065373420715, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 2.3269, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7761806981519507, | |
| "grad_norm": 0.33632636070251465, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.2996, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7802874743326489, | |
| "grad_norm": 0.3384513556957245, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 2.3976, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.784394250513347, | |
| "grad_norm": 0.3339103162288666, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.2832, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7885010266940452, | |
| "grad_norm": 0.34259554743766785, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.3546, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7926078028747433, | |
| "grad_norm": 0.3554518520832062, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.3662, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7967145790554415, | |
| "grad_norm": 0.365813672542572, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.2609, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8008213552361396, | |
| "grad_norm": 0.38189005851745605, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.4067, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.8049281314168378, | |
| "grad_norm": 0.3823890686035156, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.3932, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8090349075975359, | |
| "grad_norm": 0.39428114891052246, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.3534, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.813141683778234, | |
| "grad_norm": 0.4303341805934906, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.39, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8172484599589322, | |
| "grad_norm": 0.4249311089515686, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.4358, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8213552361396304, | |
| "grad_norm": 0.5224833488464355, | |
| "learning_rate": 0.0, | |
| "loss": 2.3921, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8213552361396304, | |
| "eval_loss": 2.3414769172668457, | |
| "eval_runtime": 45.2628, | |
| "eval_samples_per_second": 9.058, | |
| "eval_steps_per_second": 2.276, | |
| "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": 4.2577795760101786e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |