| { | |
| "best_metric": 1.6457138061523438, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.13003901170351106, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006501950585175553, | |
| "grad_norm": 3.2897896766662598, | |
| "learning_rate": 1e-05, | |
| "loss": 2.0283, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006501950585175553, | |
| "eval_loss": 3.44756817817688, | |
| "eval_runtime": 90.0523, | |
| "eval_samples_per_second": 28.772, | |
| "eval_steps_per_second": 7.196, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0013003901170351106, | |
| "grad_norm": 3.1339213848114014, | |
| "learning_rate": 2e-05, | |
| "loss": 1.9172, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0019505851755526658, | |
| "grad_norm": 3.7369251251220703, | |
| "learning_rate": 3e-05, | |
| "loss": 2.146, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002600780234070221, | |
| "grad_norm": 3.616701126098633, | |
| "learning_rate": 4e-05, | |
| "loss": 2.1511, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003250975292587776, | |
| "grad_norm": 3.113723039627075, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9094, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0039011703511053317, | |
| "grad_norm": 2.2736775875091553, | |
| "learning_rate": 6e-05, | |
| "loss": 1.8222, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004551365409622887, | |
| "grad_norm": 1.570701003074646, | |
| "learning_rate": 7e-05, | |
| "loss": 1.898, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005201560468140442, | |
| "grad_norm": 1.7534939050674438, | |
| "learning_rate": 8e-05, | |
| "loss": 1.7586, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005851755526657998, | |
| "grad_norm": 1.9343774318695068, | |
| "learning_rate": 9e-05, | |
| "loss": 1.7696, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.006501950585175552, | |
| "grad_norm": 1.8362518548965454, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8114, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007152145643693108, | |
| "grad_norm": 1.7032122611999512, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.8475, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.007802340702210663, | |
| "grad_norm": 1.3912725448608398, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.5605, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.008452535760728219, | |
| "grad_norm": 1.5769200325012207, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.697, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.009102730819245773, | |
| "grad_norm": 1.658319354057312, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.6825, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00975292587776333, | |
| "grad_norm": 1.5307115316390991, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.5677, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.010403120936280884, | |
| "grad_norm": 1.5279511213302612, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.5374, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.011053315994798439, | |
| "grad_norm": 1.6088758707046509, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.8303, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.011703511053315995, | |
| "grad_norm": 1.585939884185791, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.6869, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01235370611183355, | |
| "grad_norm": 1.4861834049224854, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.4943, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.013003901170351105, | |
| "grad_norm": 1.5812008380889893, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.5888, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.013654096228868661, | |
| "grad_norm": 1.6460590362548828, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.586, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.014304291287386216, | |
| "grad_norm": 1.539414405822754, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.5051, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01495448634590377, | |
| "grad_norm": 1.6561790704727173, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.4821, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.015604681404421327, | |
| "grad_norm": 1.9184123277664185, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.711, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.016254876462938883, | |
| "grad_norm": 1.8694337606430054, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.5594, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.016905071521456438, | |
| "grad_norm": 1.8018032312393188, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.4768, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.017555266579973992, | |
| "grad_norm": 2.0624754428863525, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.7323, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.018205461638491547, | |
| "grad_norm": 2.1648051738739014, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.7729, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0188556566970091, | |
| "grad_norm": 2.3907740116119385, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.3934, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01950585175552666, | |
| "grad_norm": 2.478781223297119, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.5685, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.020156046814044214, | |
| "grad_norm": 2.4775261878967285, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.5554, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02080624187256177, | |
| "grad_norm": 2.721911907196045, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.5665, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.021456436931079324, | |
| "grad_norm": 2.8904712200164795, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.5612, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.022106631989596878, | |
| "grad_norm": 3.3987462520599365, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.386, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.022756827048114433, | |
| "grad_norm": 3.020430564880371, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.5512, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02340702210663199, | |
| "grad_norm": 3.224257230758667, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.0187, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.024057217165149546, | |
| "grad_norm": 3.259340286254883, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.4873, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0247074122236671, | |
| "grad_norm": 3.7515547275543213, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.089, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.025357607282184655, | |
| "grad_norm": 4.221808433532715, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.1419, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02600780234070221, | |
| "grad_norm": 3.694406270980835, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.6599, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.026657997399219768, | |
| "grad_norm": 3.2924211025238037, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.3823, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.027308192457737322, | |
| "grad_norm": 5.164480686187744, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.8636, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.027958387516254877, | |
| "grad_norm": 4.975708484649658, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.8434, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02860858257477243, | |
| "grad_norm": 4.463617324829102, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.6642, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.029258777633289986, | |
| "grad_norm": 4.744454860687256, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.1711, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02990897269180754, | |
| "grad_norm": 6.025666236877441, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.3716, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0305591677503251, | |
| "grad_norm": 6.508204460144043, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.1102, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.031209362808842653, | |
| "grad_norm": 5.709312915802002, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.1194, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03185955786736021, | |
| "grad_norm": 9.023933410644531, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.5849, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.032509752925877766, | |
| "grad_norm": 9.542643547058105, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.4991, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.032509752925877766, | |
| "eval_loss": 1.9646611213684082, | |
| "eval_runtime": 90.7958, | |
| "eval_samples_per_second": 28.537, | |
| "eval_steps_per_second": 7.137, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03315994798439532, | |
| "grad_norm": 3.2122533321380615, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.9224, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.033810143042912875, | |
| "grad_norm": 3.290236234664917, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.8479, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03446033810143043, | |
| "grad_norm": 2.634038209915161, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.7921, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.035110533159947985, | |
| "grad_norm": 1.9074913263320923, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.7204, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03576072821846554, | |
| "grad_norm": 1.7936071157455444, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.6915, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.036410923276983094, | |
| "grad_norm": 1.543684720993042, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.6863, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03706111833550065, | |
| "grad_norm": 1.4650025367736816, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.7075, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0377113133940182, | |
| "grad_norm": 1.251019835472107, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.6222, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03836150845253576, | |
| "grad_norm": 1.2934004068374634, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.7139, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03901170351105332, | |
| "grad_norm": 1.4878422021865845, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.6775, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03966189856957087, | |
| "grad_norm": 1.612789511680603, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.5977, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.04031209362808843, | |
| "grad_norm": 1.2940391302108765, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.5164, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04096228868660598, | |
| "grad_norm": 1.391506552696228, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.652, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.04161248374512354, | |
| "grad_norm": 1.2718766927719116, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.4272, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04226267880364109, | |
| "grad_norm": 1.259207010269165, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.4945, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04291287386215865, | |
| "grad_norm": 1.2823281288146973, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.4277, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.043563068920676205, | |
| "grad_norm": 1.4021849632263184, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.6008, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.044213263979193757, | |
| "grad_norm": 1.3278549909591675, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.4096, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.044863459037711315, | |
| "grad_norm": 1.4589539766311646, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.5087, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.045513654096228866, | |
| "grad_norm": 1.462446928024292, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.4616, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.046163849154746424, | |
| "grad_norm": 1.577277660369873, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.5672, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04681404421326398, | |
| "grad_norm": 1.610364317893982, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.4999, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04746423927178153, | |
| "grad_norm": 1.5940923690795898, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.5778, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04811443433029909, | |
| "grad_norm": 1.5773005485534668, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.4325, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04876462938881664, | |
| "grad_norm": 1.785773754119873, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.6098, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0494148244473342, | |
| "grad_norm": 1.9534682035446167, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.7123, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05006501950585176, | |
| "grad_norm": 2.0544018745422363, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.5178, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.05071521456436931, | |
| "grad_norm": 2.129631519317627, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.505, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05136540962288687, | |
| "grad_norm": 2.116121292114258, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.4793, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.05201560468140442, | |
| "grad_norm": 2.2213428020477295, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.4858, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05266579973992198, | |
| "grad_norm": 2.378239154815674, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.4172, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.053315994798439535, | |
| "grad_norm": 2.4855735301971436, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.4769, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.053966189856957086, | |
| "grad_norm": 2.3395750522613525, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.1569, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.054616384915474644, | |
| "grad_norm": 2.9003002643585205, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.7208, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.055266579973992196, | |
| "grad_norm": 2.8214237689971924, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.6679, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.055916775032509754, | |
| "grad_norm": 3.1319596767425537, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.4479, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.056566970091027305, | |
| "grad_norm": 3.692051887512207, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.6194, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.05721716514954486, | |
| "grad_norm": 3.415761709213257, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.6039, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.05786736020806242, | |
| "grad_norm": 3.417778968811035, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.6117, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.05851755526657997, | |
| "grad_norm": 3.6709556579589844, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.6009, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05916775032509753, | |
| "grad_norm": 3.5706937313079834, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8983, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.05981794538361508, | |
| "grad_norm": 3.9006922245025635, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.6683, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06046814044213264, | |
| "grad_norm": 4.110403537750244, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.8464, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0611183355006502, | |
| "grad_norm": 4.5136213302612305, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.1915, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.06176853055916775, | |
| "grad_norm": 6.277181625366211, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.8335, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06241872561768531, | |
| "grad_norm": 4.6575775146484375, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.2307, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.06306892067620286, | |
| "grad_norm": 5.150646686553955, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.0925, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.06371911573472042, | |
| "grad_norm": 7.0254902839660645, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.677, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.06436931079323797, | |
| "grad_norm": 7.098849773406982, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.2844, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.06501950585175553, | |
| "grad_norm": 11.306396484375, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 3.0443, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06501950585175553, | |
| "eval_loss": 1.8381491899490356, | |
| "eval_runtime": 90.7409, | |
| "eval_samples_per_second": 28.554, | |
| "eval_steps_per_second": 7.141, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06566970091027308, | |
| "grad_norm": 2.2785446643829346, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 1.7277, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.06631989596879063, | |
| "grad_norm": 2.329958200454712, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.7344, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.06697009102730819, | |
| "grad_norm": 2.0368316173553467, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.7262, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.06762028608582575, | |
| "grad_norm": 2.069119453430176, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.5644, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.06827048114434331, | |
| "grad_norm": 1.8223001956939697, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7701, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06892067620286085, | |
| "grad_norm": 1.5952881574630737, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.5158, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06957087126137841, | |
| "grad_norm": 1.389872431755066, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.6127, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.07022106631989597, | |
| "grad_norm": 1.283721685409546, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.5289, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07087126137841353, | |
| "grad_norm": 1.1220057010650635, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.5817, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.07152145643693109, | |
| "grad_norm": 1.1918190717697144, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.5914, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07217165149544863, | |
| "grad_norm": 1.160514235496521, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.5442, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.07282184655396619, | |
| "grad_norm": 1.1976782083511353, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.4824, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.07347204161248375, | |
| "grad_norm": 1.2484989166259766, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.5711, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0741222366710013, | |
| "grad_norm": 1.3540258407592773, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.6888, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.07477243172951886, | |
| "grad_norm": 1.3416101932525635, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.53, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0754226267880364, | |
| "grad_norm": 1.3969745635986328, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.5536, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.07607282184655396, | |
| "grad_norm": 1.4391306638717651, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.614, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.07672301690507152, | |
| "grad_norm": 1.460862398147583, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.4015, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.07737321196358908, | |
| "grad_norm": 1.5547221899032593, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.5811, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.07802340702210664, | |
| "grad_norm": 1.6407545804977417, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.7403, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07867360208062418, | |
| "grad_norm": 1.6028615236282349, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.5175, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.07932379713914174, | |
| "grad_norm": 1.759552240371704, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.5523, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0799739921976593, | |
| "grad_norm": 1.862209677696228, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.5584, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.08062418725617686, | |
| "grad_norm": 1.8480825424194336, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.5117, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0812743823146944, | |
| "grad_norm": 1.8010337352752686, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.5137, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08192457737321196, | |
| "grad_norm": 1.8761563301086426, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.4736, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.08257477243172952, | |
| "grad_norm": 1.9912925958633423, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.6898, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.08322496749024708, | |
| "grad_norm": 2.0787229537963867, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.3937, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.08387516254876463, | |
| "grad_norm": 2.2982161045074463, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.3034, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.08452535760728218, | |
| "grad_norm": 2.3474977016448975, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.4216, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08517555266579974, | |
| "grad_norm": 2.3278427124023438, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.1997, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0858257477243173, | |
| "grad_norm": 2.549748659133911, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.8476, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.08647594278283485, | |
| "grad_norm": 2.46454119682312, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.4477, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.08712613784135241, | |
| "grad_norm": 2.5464067459106445, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.6201, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.08777633289986995, | |
| "grad_norm": 2.5115482807159424, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.439, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.08842652795838751, | |
| "grad_norm": 3.1942481994628906, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7011, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.08907672301690507, | |
| "grad_norm": 2.8990478515625, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.5849, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.08972691807542263, | |
| "grad_norm": 3.035853385925293, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.6251, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.09037711313394019, | |
| "grad_norm": 3.2503368854522705, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.7843, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.09102730819245773, | |
| "grad_norm": 3.748659610748291, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.0034, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09167750325097529, | |
| "grad_norm": 3.4162096977233887, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.8084, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.09232769830949285, | |
| "grad_norm": 4.435839653015137, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.3281, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.0929778933680104, | |
| "grad_norm": 4.22627067565918, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.9283, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.09362808842652796, | |
| "grad_norm": 3.963848352432251, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.0417, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.09427828348504551, | |
| "grad_norm": 4.55359411239624, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.0615, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09492847854356307, | |
| "grad_norm": 5.180711269378662, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.1736, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.09557867360208062, | |
| "grad_norm": 5.014293670654297, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.9006, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.09622886866059818, | |
| "grad_norm": 5.979249477386475, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.8095, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.09687906371911574, | |
| "grad_norm": 9.18662166595459, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.8576, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.09752925877763328, | |
| "grad_norm": 11.332324028015137, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 3.1411, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09752925877763328, | |
| "eval_loss": 1.671252727508545, | |
| "eval_runtime": 90.7068, | |
| "eval_samples_per_second": 28.565, | |
| "eval_steps_per_second": 7.144, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09817945383615084, | |
| "grad_norm": 0.9760773181915283, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.4542, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0988296488946684, | |
| "grad_norm": 1.084282636642456, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.5224, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.09947984395318596, | |
| "grad_norm": 1.3246732950210571, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.659, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.10013003901170352, | |
| "grad_norm": 1.384027361869812, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.6052, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.10078023407022106, | |
| "grad_norm": 1.380460500717163, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.7433, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.10143042912873862, | |
| "grad_norm": 1.416570782661438, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.5922, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.10208062418725618, | |
| "grad_norm": 1.3455933332443237, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.5548, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.10273081924577374, | |
| "grad_norm": 1.3731634616851807, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.5529, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.1033810143042913, | |
| "grad_norm": 1.4273340702056885, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.5362, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.10403120936280884, | |
| "grad_norm": 1.322783350944519, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.5484, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1046814044213264, | |
| "grad_norm": 1.3814400434494019, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.4714, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.10533159947984395, | |
| "grad_norm": 1.4020730257034302, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.6761, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.10598179453836151, | |
| "grad_norm": 1.3990076780319214, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.5292, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.10663198959687907, | |
| "grad_norm": 1.346995234489441, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.5009, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.10728218465539661, | |
| "grad_norm": 1.4257160425186157, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.4682, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.10793237971391417, | |
| "grad_norm": 1.37079918384552, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.4034, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.10858257477243173, | |
| "grad_norm": 1.4135793447494507, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.5825, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.10923276983094929, | |
| "grad_norm": 1.381396770477295, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.403, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.10988296488946683, | |
| "grad_norm": 1.4749950170516968, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.6121, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.11053315994798439, | |
| "grad_norm": 1.462877869606018, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.607, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11118335500650195, | |
| "grad_norm": 1.5816662311553955, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.6085, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.11183355006501951, | |
| "grad_norm": 1.8168708086013794, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.6652, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.11248374512353707, | |
| "grad_norm": 1.5931594371795654, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.4476, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.11313394018205461, | |
| "grad_norm": 1.8875209093093872, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.6712, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.11378413524057217, | |
| "grad_norm": 1.7895739078521729, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.4214, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11443433029908973, | |
| "grad_norm": 1.8276786804199219, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.5169, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.11508452535760728, | |
| "grad_norm": 2.088595390319824, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.4725, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.11573472041612484, | |
| "grad_norm": 2.0828213691711426, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.5394, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.11638491547464239, | |
| "grad_norm": 2.3032801151275635, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.5125, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.11703511053315994, | |
| "grad_norm": 2.4829626083374023, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.4702, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1176853055916775, | |
| "grad_norm": 3.2075982093811035, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.5139, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.11833550065019506, | |
| "grad_norm": 2.7486572265625, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.4814, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.11898569570871262, | |
| "grad_norm": 2.388666868209839, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.4523, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.11963589076723016, | |
| "grad_norm": 3.4033751487731934, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.5523, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.12028608582574772, | |
| "grad_norm": 3.1661531925201416, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.4877, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12093628088426528, | |
| "grad_norm": 2.6967453956604004, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.2437, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.12158647594278284, | |
| "grad_norm": 3.191657543182373, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.6093, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1222366710013004, | |
| "grad_norm": 3.4261112213134766, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.7214, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.12288686605981794, | |
| "grad_norm": 3.1970436573028564, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.5344, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1235370611183355, | |
| "grad_norm": 3.5080649852752686, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.5922, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.12418725617685306, | |
| "grad_norm": 3.912050247192383, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.5595, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.12483745123537061, | |
| "grad_norm": 3.4891788959503174, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.7351, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.12548764629388817, | |
| "grad_norm": 3.7075014114379883, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.8078, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.12613784135240572, | |
| "grad_norm": 4.0296149253845215, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.0384, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1267880364109233, | |
| "grad_norm": 4.49355411529541, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.331, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.12743823146944083, | |
| "grad_norm": 4.66786003112793, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.1507, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.12808842652795838, | |
| "grad_norm": 5.419894695281982, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.0951, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.12873862158647595, | |
| "grad_norm": 6.154303550720215, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.4956, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.1293888166449935, | |
| "grad_norm": 7.455812454223633, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.7904, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.13003901170351106, | |
| "grad_norm": 11.858441352844238, | |
| "learning_rate": 0.0, | |
| "loss": 2.7366, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13003901170351106, | |
| "eval_loss": 1.6457138061523438, | |
| "eval_runtime": 90.6627, | |
| "eval_samples_per_second": 28.578, | |
| "eval_steps_per_second": 7.147, | |
| "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": 8.111678115387802e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |