| { | |
| "best_metric": 0.3198044002056122, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.007874325760856726, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.937162880428363e-05, | |
| "grad_norm": 0.1493980437517166, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4842, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.937162880428363e-05, | |
| "eval_loss": 0.6386224031448364, | |
| "eval_runtime": 2000.5635, | |
| "eval_samples_per_second": 21.383, | |
| "eval_steps_per_second": 5.346, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 7.874325760856726e-05, | |
| "grad_norm": 0.15225277841091156, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4723, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0001181148864128509, | |
| "grad_norm": 0.15469801425933838, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4648, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00015748651521713453, | |
| "grad_norm": 0.1572115123271942, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4614, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00019685814402141817, | |
| "grad_norm": 0.17931561172008514, | |
| "learning_rate": 5e-05, | |
| "loss": 0.4572, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0002362297728257018, | |
| "grad_norm": 0.17100964486598969, | |
| "learning_rate": 6e-05, | |
| "loss": 0.452, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00027560140162998545, | |
| "grad_norm": 0.17576375603675842, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4686, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00031497303043426906, | |
| "grad_norm": 0.19541922211647034, | |
| "learning_rate": 8e-05, | |
| "loss": 0.4868, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0003543446592385527, | |
| "grad_norm": 0.2201804667711258, | |
| "learning_rate": 9e-05, | |
| "loss": 0.4394, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00039371628804283634, | |
| "grad_norm": 0.20416098833084106, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4674, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00043308791684711995, | |
| "grad_norm": 0.190375417470932, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.4404, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0004724595456514036, | |
| "grad_norm": 0.18620383739471436, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.4312, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0005118311744556872, | |
| "grad_norm": 0.19505620002746582, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.4261, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0005512028032599709, | |
| "grad_norm": 0.18445885181427002, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.4201, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0005905744320642545, | |
| "grad_norm": 0.1769520491361618, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.4373, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0006299460608685381, | |
| "grad_norm": 0.17001008987426758, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.4088, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0006693176896728218, | |
| "grad_norm": 0.1816432774066925, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.4117, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0007086893184771055, | |
| "grad_norm": 0.1911289542913437, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.4113, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.000748060947281389, | |
| "grad_norm": 0.17927183210849762, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.4, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0007874325760856727, | |
| "grad_norm": 0.19351114332675934, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.4031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0008268042048899563, | |
| "grad_norm": 0.20432829856872559, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.3982, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0008661758336942399, | |
| "grad_norm": 0.2100755125284195, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.4125, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0009055474624985236, | |
| "grad_norm": 0.2107066661119461, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.393, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0009449190913028072, | |
| "grad_norm": 0.20557498931884766, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.3947, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0009842907201070908, | |
| "grad_norm": 0.23331749439239502, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.4078, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0010236623489113745, | |
| "grad_norm": 0.21170148253440857, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3773, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0010630339777156581, | |
| "grad_norm": 0.23207199573516846, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.3934, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0011024056065199418, | |
| "grad_norm": 0.2046716958284378, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.3622, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0011417772353242255, | |
| "grad_norm": 0.22365449368953705, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3728, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.001181148864128509, | |
| "grad_norm": 0.2250414490699768, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3824, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0012205204929327926, | |
| "grad_norm": 0.22676680982112885, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3724, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0012598921217370762, | |
| "grad_norm": 0.24163582921028137, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.3865, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.00129926375054136, | |
| "grad_norm": 0.2499234825372696, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3775, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0013386353793456436, | |
| "grad_norm": 0.2501504719257355, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.3667, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0013780070081499272, | |
| "grad_norm": 0.28612300753593445, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3857, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.001417378636954211, | |
| "grad_norm": 0.2889863848686218, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.3975, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0014567502657584943, | |
| "grad_norm": 0.30571281909942627, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.4242, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.001496121894562778, | |
| "grad_norm": 0.32020753622055054, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.3906, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0015354935233670617, | |
| "grad_norm": 0.35397595167160034, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.3875, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0015748651521713453, | |
| "grad_norm": 0.38060539960861206, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4096, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.001614236780975629, | |
| "grad_norm": 0.3947359621524811, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.4432, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0016536084097799127, | |
| "grad_norm": 0.43858495354652405, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.4264, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0016929800385841961, | |
| "grad_norm": 0.49851053953170776, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.4603, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0017323516673884798, | |
| "grad_norm": 0.4782646894454956, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.4383, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0017717232961927635, | |
| "grad_norm": 0.5415741801261902, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.508, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0018110949249970471, | |
| "grad_norm": 0.6462609171867371, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.4773, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0018504665538013308, | |
| "grad_norm": 0.7398020029067993, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.4995, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0018898381826056145, | |
| "grad_norm": 0.8590586185455322, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4858, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0019292098114098981, | |
| "grad_norm": 0.9482952952384949, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.5083, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0019685814402141816, | |
| "grad_norm": 1.7293684482574463, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.7293, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0019685814402141816, | |
| "eval_loss": 0.4911499321460724, | |
| "eval_runtime": 2011.4927, | |
| "eval_samples_per_second": 21.267, | |
| "eval_steps_per_second": 5.317, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0020079530690184652, | |
| "grad_norm": 1.1663459539413452, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.5298, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.002047324697822749, | |
| "grad_norm": 0.5759194493293762, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.4563, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0020866963266270326, | |
| "grad_norm": 0.29835495352745056, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.4226, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0021260679554313162, | |
| "grad_norm": 0.21884822845458984, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3939, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0021654395842356, | |
| "grad_norm": 0.20164261758327484, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.3646, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0022048112130398836, | |
| "grad_norm": 0.2158244103193283, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.3915, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0022441828418441672, | |
| "grad_norm": 0.21263156831264496, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.3837, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.002283554470648451, | |
| "grad_norm": 0.2191099375486374, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.3949, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0023229260994527346, | |
| "grad_norm": 0.2002435326576233, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3933, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.002362297728257018, | |
| "grad_norm": 0.18316560983657837, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.3923, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0024016693570613015, | |
| "grad_norm": 0.17700187861919403, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.3724, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.002441040985865585, | |
| "grad_norm": 0.166967511177063, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.3523, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.002480412614669869, | |
| "grad_norm": 0.15993095934391022, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.3531, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0025197842434741525, | |
| "grad_norm": 0.17065750062465668, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3646, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.002559155872278436, | |
| "grad_norm": 0.16862858831882477, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3406, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.00259852750108272, | |
| "grad_norm": 0.17622439563274384, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.3591, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0026378991298870035, | |
| "grad_norm": 0.17540805041790009, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.3625, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.002677270758691287, | |
| "grad_norm": 0.17689818143844604, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.3394, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.002716642387495571, | |
| "grad_norm": 0.19289594888687134, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.3324, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0027560140162998545, | |
| "grad_norm": 0.1848318725824356, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3633, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.002795385645104138, | |
| "grad_norm": 0.18961237370967865, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.3688, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.002834757273908422, | |
| "grad_norm": 0.18003550171852112, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3136, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.002874128902712705, | |
| "grad_norm": 0.20321786403656006, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3678, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0029135005315169887, | |
| "grad_norm": 0.17587269842624664, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.336, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0029528721603212724, | |
| "grad_norm": 0.18733522295951843, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.3299, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.002992243789125556, | |
| "grad_norm": 0.18234075605869293, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.334, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0030316154179298397, | |
| "grad_norm": 0.19873753190040588, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.3323, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0030709870467341234, | |
| "grad_norm": 0.19613344967365265, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.3248, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.003110358675538407, | |
| "grad_norm": 0.21492932736873627, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.3604, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0031497303043426907, | |
| "grad_norm": 0.22016383707523346, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3544, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0031891019331469744, | |
| "grad_norm": 0.21575337648391724, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.3244, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.003228473561951258, | |
| "grad_norm": 0.23265521228313446, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3495, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0032678451907555417, | |
| "grad_norm": 0.23755455017089844, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.3396, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0033072168195598254, | |
| "grad_norm": 0.27804362773895264, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.3647, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.003346588448364109, | |
| "grad_norm": 0.2547140121459961, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.3365, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0033859600771683923, | |
| "grad_norm": 0.2836421728134155, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.3552, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.003425331705972676, | |
| "grad_norm": 0.31407999992370605, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.3426, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0034647033347769596, | |
| "grad_norm": 0.2865597903728485, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3242, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0035040749635812433, | |
| "grad_norm": 0.3280423879623413, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.3735, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.003543446592385527, | |
| "grad_norm": 0.3588223457336426, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.385, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0035828182211898106, | |
| "grad_norm": 0.38680195808410645, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.3801, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0036221898499940943, | |
| "grad_norm": 0.39277011156082153, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.3565, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.003661561478798378, | |
| "grad_norm": 0.44393858313560486, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3805, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0037009331076026616, | |
| "grad_norm": 0.5041418671607971, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.3686, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0037403047364069453, | |
| "grad_norm": 0.6213234663009644, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.4352, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.003779676365211229, | |
| "grad_norm": 0.5778347849845886, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3747, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0038190479940155126, | |
| "grad_norm": 0.8748276233673096, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.4385, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0038584196228197963, | |
| "grad_norm": 1.0055272579193115, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5019, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0038977912516240795, | |
| "grad_norm": 0.9788726568222046, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.5448, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.003937162880428363, | |
| "grad_norm": 1.643742322921753, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.5097, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003937162880428363, | |
| "eval_loss": 0.37032032012939453, | |
| "eval_runtime": 2015.0366, | |
| "eval_samples_per_second": 21.229, | |
| "eval_steps_per_second": 5.308, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003976534509232647, | |
| "grad_norm": 0.2651832103729248, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.3912, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0040159061380369305, | |
| "grad_norm": 0.24760182201862335, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.3616, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.004055277766841214, | |
| "grad_norm": 0.24754416942596436, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.3743, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.004094649395645498, | |
| "grad_norm": 0.3310948610305786, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.3546, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0041340210244497815, | |
| "grad_norm": 0.28900614380836487, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3344, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.004173392653254065, | |
| "grad_norm": 0.2792387008666992, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.3684, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.004212764282058349, | |
| "grad_norm": 0.20933227241039276, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.3505, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.0042521359108626325, | |
| "grad_norm": 0.19124023616313934, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.3413, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.004291507539666916, | |
| "grad_norm": 0.18814119696617126, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.3342, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0043308791684712, | |
| "grad_norm": 0.1860683709383011, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.3307, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0043702507972754835, | |
| "grad_norm": 0.17513607442378998, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.3409, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.004409622426079767, | |
| "grad_norm": 0.18182946741580963, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3275, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.004448994054884051, | |
| "grad_norm": 0.1895298808813095, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.3334, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0044883656836883345, | |
| "grad_norm": 0.18013550341129303, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.3241, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.004527737312492618, | |
| "grad_norm": 0.1788962334394455, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.3054, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.004567108941296902, | |
| "grad_norm": 0.19194889068603516, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.3425, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.0046064805701011855, | |
| "grad_norm": 0.18574917316436768, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.3075, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.004645852198905469, | |
| "grad_norm": 0.21845579147338867, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.3473, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.004685223827709752, | |
| "grad_norm": 0.18899574875831604, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.3215, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.004724595456514036, | |
| "grad_norm": 0.17867511510849, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.3118, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.004763967085318319, | |
| "grad_norm": 0.20106135308742523, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.3161, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.004803338714122603, | |
| "grad_norm": 0.19418220221996307, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.3344, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.004842710342926887, | |
| "grad_norm": 0.19221185147762299, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.3276, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.00488208197173117, | |
| "grad_norm": 0.19667880237102509, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.3091, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.004921453600535454, | |
| "grad_norm": 0.20428292453289032, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.3172, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.004960825229339738, | |
| "grad_norm": 0.19734974205493927, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.3109, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.005000196858144021, | |
| "grad_norm": 0.20616655051708221, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.3265, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.005039568486948305, | |
| "grad_norm": 0.20463624596595764, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.293, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.005078940115752589, | |
| "grad_norm": 0.21404369175434113, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.2928, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.005118311744556872, | |
| "grad_norm": 0.22799019515514374, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.304, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.005157683373361156, | |
| "grad_norm": 0.24096277356147766, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.3149, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.00519705500216544, | |
| "grad_norm": 0.22695647180080414, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.2959, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.005236426630969723, | |
| "grad_norm": 0.25506260991096497, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.3338, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.005275798259774007, | |
| "grad_norm": 0.2578275799751282, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3004, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.005315169888578291, | |
| "grad_norm": 0.2474546581506729, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.3029, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.005354541517382574, | |
| "grad_norm": 0.2818843722343445, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.312, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.005393913146186858, | |
| "grad_norm": 0.27845391631126404, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.2965, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.005433284774991142, | |
| "grad_norm": 0.2858252227306366, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3258, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.005472656403795425, | |
| "grad_norm": 0.32283511757850647, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.3297, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.005512028032599709, | |
| "grad_norm": 0.3427778482437134, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3106, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.005551399661403993, | |
| "grad_norm": 0.3494415879249573, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.3011, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.005590771290208276, | |
| "grad_norm": 0.4077412486076355, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3639, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.00563014291901256, | |
| "grad_norm": 0.47255849838256836, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3215, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.005669514547816844, | |
| "grad_norm": 0.5124167799949646, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.34, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.005708886176621126, | |
| "grad_norm": 0.542047381401062, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3923, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.00574825780542541, | |
| "grad_norm": 0.5466054081916809, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3414, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.005787629434229694, | |
| "grad_norm": 0.5823593139648438, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.3369, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.005827001063033977, | |
| "grad_norm": 0.7636677026748657, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.4091, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.005866372691838261, | |
| "grad_norm": 0.8901990056037903, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.4089, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.005905744320642545, | |
| "grad_norm": 1.376933217048645, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.4716, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005905744320642545, | |
| "eval_loss": 0.3277217447757721, | |
| "eval_runtime": 2013.3176, | |
| "eval_samples_per_second": 21.248, | |
| "eval_steps_per_second": 5.312, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005945115949446828, | |
| "grad_norm": 0.19697712361812592, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.3445, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.005984487578251112, | |
| "grad_norm": 0.2107386291027069, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.3313, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.006023859207055396, | |
| "grad_norm": 0.18995290994644165, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.3538, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.006063230835859679, | |
| "grad_norm": 0.19262909889221191, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.3283, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.006102602464663963, | |
| "grad_norm": 0.21958303451538086, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.3146, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.006141974093468247, | |
| "grad_norm": 0.1917515993118286, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.3043, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.00618134572227253, | |
| "grad_norm": 0.20367880165576935, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.3172, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.006220717351076814, | |
| "grad_norm": 0.17750227451324463, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.2909, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.006260088979881098, | |
| "grad_norm": 0.184042826294899, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3052, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.006299460608685381, | |
| "grad_norm": 0.19311654567718506, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.3004, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.006338832237489665, | |
| "grad_norm": 0.1915796399116516, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3013, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.006378203866293949, | |
| "grad_norm": 0.18793466687202454, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.3288, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.006417575495098232, | |
| "grad_norm": 0.19275987148284912, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2952, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.006456947123902516, | |
| "grad_norm": 0.19124749302864075, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.3002, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0064963187527068, | |
| "grad_norm": 0.1956319510936737, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3071, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.006535690381511083, | |
| "grad_norm": 0.19477754831314087, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.3049, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.006575062010315367, | |
| "grad_norm": 0.1876872181892395, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.2937, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.006614433639119651, | |
| "grad_norm": 0.20833444595336914, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3005, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.006653805267923934, | |
| "grad_norm": 0.19107308983802795, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.2899, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.006693176896728218, | |
| "grad_norm": 0.19436660408973694, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.2859, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.006732548525532501, | |
| "grad_norm": 0.20439302921295166, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3018, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0067719201543367845, | |
| "grad_norm": 0.19564414024353027, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.2819, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.006811291783141068, | |
| "grad_norm": 0.206856831908226, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.3041, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.006850663411945352, | |
| "grad_norm": 0.212832972407341, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.308, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0068900350407496355, | |
| "grad_norm": 0.21448437869548798, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.2916, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.006929406669553919, | |
| "grad_norm": 0.21251370012760162, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.3065, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.006968778298358203, | |
| "grad_norm": 0.2179366797208786, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.2969, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0070081499271624865, | |
| "grad_norm": 0.2313450574874878, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.3149, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.00704752155596677, | |
| "grad_norm": 0.22925609350204468, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.3087, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.007086893184771054, | |
| "grad_norm": 0.23077863454818726, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.2997, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0071262648135753375, | |
| "grad_norm": 0.23666110634803772, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.3021, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.007165636442379621, | |
| "grad_norm": 0.25907155871391296, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.2875, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.007205008071183905, | |
| "grad_norm": 0.2466953694820404, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.2999, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0072443796999881885, | |
| "grad_norm": 0.2594219446182251, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3204, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.007283751328792472, | |
| "grad_norm": 0.27320852875709534, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3077, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.007323122957596756, | |
| "grad_norm": 0.2888332009315491, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.3224, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.0073624945864010395, | |
| "grad_norm": 0.2894538938999176, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3128, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.007401866215205323, | |
| "grad_norm": 0.3097641170024872, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.3308, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.007441237844009607, | |
| "grad_norm": 0.32769468426704407, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3056, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0074806094728138905, | |
| "grad_norm": 0.3616294264793396, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3317, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.007519981101618174, | |
| "grad_norm": 0.3736540973186493, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.3382, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.007559352730422458, | |
| "grad_norm": 0.3888842761516571, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.3011, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0075987243592267415, | |
| "grad_norm": 0.41784223914146423, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3424, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.007638095988031025, | |
| "grad_norm": 0.46386027336120605, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3893, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.007677467616835309, | |
| "grad_norm": 0.5194088816642761, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.3598, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0077168392456395925, | |
| "grad_norm": 0.5213820934295654, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.345, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.007756210874443875, | |
| "grad_norm": 0.6496748924255371, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3255, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.007795582503248159, | |
| "grad_norm": 0.9071391224861145, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.4531, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.007834954132052444, | |
| "grad_norm": 1.0409427881240845, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.4495, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.007874325760856726, | |
| "grad_norm": 1.514607548713684, | |
| "learning_rate": 0.0, | |
| "loss": 0.5581, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.007874325760856726, | |
| "eval_loss": 0.3198044002056122, | |
| "eval_runtime": 2013.9681, | |
| "eval_samples_per_second": 21.241, | |
| "eval_steps_per_second": 5.31, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.439681531609088e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |