| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.03374957813027337, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00016874789065136686, | |
| "grad_norm": 1.204833279189188e-05, | |
| "learning_rate": 1.007e-05, | |
| "loss": 23.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00016874789065136686, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 31.0185, | |
| "eval_samples_per_second": 80.436, | |
| "eval_steps_per_second": 20.117, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0003374957813027337, | |
| "grad_norm": 1.1431009625084698e-05, | |
| "learning_rate": 2.014e-05, | |
| "loss": 23.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0005062436719541006, | |
| "grad_norm": 1.5091093700903002e-05, | |
| "learning_rate": 3.0209999999999997e-05, | |
| "loss": 23.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0006749915626054674, | |
| "grad_norm": 1.993294426938519e-05, | |
| "learning_rate": 4.028e-05, | |
| "loss": 23.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0008437394532568343, | |
| "grad_norm": 1.4757441022084095e-05, | |
| "learning_rate": 5.035e-05, | |
| "loss": 23.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0010124873439082012, | |
| "grad_norm": 1.3547244634537492e-05, | |
| "learning_rate": 6.0419999999999994e-05, | |
| "loss": 23.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.001181235234559568, | |
| "grad_norm": 1.6674213838996366e-05, | |
| "learning_rate": 7.049e-05, | |
| "loss": 23.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0013499831252109348, | |
| "grad_norm": 1.7742673662723973e-05, | |
| "learning_rate": 8.056e-05, | |
| "loss": 23.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0015187310158623017, | |
| "grad_norm": 1.2505891390901525e-05, | |
| "learning_rate": 9.062999999999999e-05, | |
| "loss": 23.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0016874789065136687, | |
| "grad_norm": 1.3835131539963186e-05, | |
| "learning_rate": 0.0001007, | |
| "loss": 23.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0018562267971650355, | |
| "grad_norm": 2.4900897187762894e-05, | |
| "learning_rate": 0.00010017, | |
| "loss": 23.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0020249746878164025, | |
| "grad_norm": 1.7070025933207944e-05, | |
| "learning_rate": 9.963999999999999e-05, | |
| "loss": 23.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0021937225784677693, | |
| "grad_norm": 2.7361658794688992e-05, | |
| "learning_rate": 9.910999999999999e-05, | |
| "loss": 23.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.002362470469119136, | |
| "grad_norm": 1.2697471902356483e-05, | |
| "learning_rate": 9.858e-05, | |
| "loss": 23.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.002531218359770503, | |
| "grad_norm": 1.9702922145370394e-05, | |
| "learning_rate": 9.805e-05, | |
| "loss": 23.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0026999662504218697, | |
| "grad_norm": 2.4902374207158573e-05, | |
| "learning_rate": 9.752e-05, | |
| "loss": 23.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0028687141410732365, | |
| "grad_norm": 2.721859527810011e-05, | |
| "learning_rate": 9.698999999999999e-05, | |
| "loss": 23.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0030374620317246033, | |
| "grad_norm": 1.8586168152978644e-05, | |
| "learning_rate": 9.646e-05, | |
| "loss": 23.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.00320620992237597, | |
| "grad_norm": 1.872859866125509e-05, | |
| "learning_rate": 9.593e-05, | |
| "loss": 23.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0033749578130273373, | |
| "grad_norm": 3.305126665509306e-05, | |
| "learning_rate": 9.539999999999999e-05, | |
| "loss": 23.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.003543705703678704, | |
| "grad_norm": 2.0460800442378968e-05, | |
| "learning_rate": 9.487e-05, | |
| "loss": 23.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.003712453594330071, | |
| "grad_norm": 2.6154682927881368e-05, | |
| "learning_rate": 9.434e-05, | |
| "loss": 23.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0038812014849814377, | |
| "grad_norm": 2.1470910724019632e-05, | |
| "learning_rate": 9.381e-05, | |
| "loss": 23.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.004049949375632805, | |
| "grad_norm": 2.203008989454247e-05, | |
| "learning_rate": 9.327999999999999e-05, | |
| "loss": 23.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.004218697266284171, | |
| "grad_norm": 2.6265855922247283e-05, | |
| "learning_rate": 9.274999999999999e-05, | |
| "loss": 23.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.004387445156935539, | |
| "grad_norm": 3.531806214596145e-05, | |
| "learning_rate": 9.222e-05, | |
| "loss": 23.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.004556193047586905, | |
| "grad_norm": 3.6674504372058436e-05, | |
| "learning_rate": 9.169e-05, | |
| "loss": 23.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.004724940938238272, | |
| "grad_norm": 3.1309191399486735e-05, | |
| "learning_rate": 9.116e-05, | |
| "loss": 23.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0048936888288896386, | |
| "grad_norm": 2.5531946448609233e-05, | |
| "learning_rate": 9.062999999999999e-05, | |
| "loss": 23.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.005062436719541006, | |
| "grad_norm": 3.693735197884962e-05, | |
| "learning_rate": 9.01e-05, | |
| "loss": 23.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.005231184610192373, | |
| "grad_norm": 3.671792001114227e-05, | |
| "learning_rate": 8.957e-05, | |
| "loss": 23.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.005399932500843739, | |
| "grad_norm": 2.8840735467383638e-05, | |
| "learning_rate": 8.903999999999999e-05, | |
| "loss": 23.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.005568680391495107, | |
| "grad_norm": 4.73731524834875e-05, | |
| "learning_rate": 8.850999999999999e-05, | |
| "loss": 23.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.005737428282146473, | |
| "grad_norm": 4.9730784667190164e-05, | |
| "learning_rate": 8.798e-05, | |
| "loss": 23.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.00590617617279784, | |
| "grad_norm": 4.867700408794917e-05, | |
| "learning_rate": 8.745e-05, | |
| "loss": 23.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.006074924063449207, | |
| "grad_norm": 4.8206897190539166e-05, | |
| "learning_rate": 8.692e-05, | |
| "loss": 23.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.006243671954100574, | |
| "grad_norm": 5.011281609768048e-05, | |
| "learning_rate": 8.638999999999999e-05, | |
| "loss": 23.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.00641241984475194, | |
| "grad_norm": 8.074403740465641e-05, | |
| "learning_rate": 8.586e-05, | |
| "loss": 23.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0065811677354033074, | |
| "grad_norm": 6.544839561684057e-05, | |
| "learning_rate": 8.533e-05, | |
| "loss": 23.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.006749915626054675, | |
| "grad_norm": 7.167661533458158e-05, | |
| "learning_rate": 8.479999999999999e-05, | |
| "loss": 23.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.006918663516706041, | |
| "grad_norm": 6.072892210795544e-05, | |
| "learning_rate": 8.427e-05, | |
| "loss": 23.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.007087411407357408, | |
| "grad_norm": 4.846288720727898e-05, | |
| "learning_rate": 8.374e-05, | |
| "loss": 23.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.007256159298008775, | |
| "grad_norm": 7.074383756844327e-05, | |
| "learning_rate": 8.321e-05, | |
| "loss": 23.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.007424907188660142, | |
| "grad_norm": 4.74240405310411e-05, | |
| "learning_rate": 8.268e-05, | |
| "loss": 23.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.007593655079311508, | |
| "grad_norm": 0.00011488777818158269, | |
| "learning_rate": 8.214999999999999e-05, | |
| "loss": 23.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0077624029699628755, | |
| "grad_norm": 9.618495823815465e-05, | |
| "learning_rate": 8.162e-05, | |
| "loss": 23.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.007931150860614243, | |
| "grad_norm": 7.153845945140347e-05, | |
| "learning_rate": 8.108999999999998e-05, | |
| "loss": 23.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.00809989875126561, | |
| "grad_norm": 0.00012329178571235389, | |
| "learning_rate": 8.056e-05, | |
| "loss": 23.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.008268646641916975, | |
| "grad_norm": 9.25509084481746e-05, | |
| "learning_rate": 8.003e-05, | |
| "loss": 23.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.008437394532568343, | |
| "grad_norm": 0.00014846926205791533, | |
| "learning_rate": 7.95e-05, | |
| "loss": 23.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.008437394532568343, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 31.0281, | |
| "eval_samples_per_second": 80.411, | |
| "eval_steps_per_second": 20.111, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.00860614242321971, | |
| "grad_norm": 3.109794488409534e-05, | |
| "learning_rate": 7.897e-05, | |
| "loss": 23.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.008774890313871077, | |
| "grad_norm": 4.0770904888631776e-05, | |
| "learning_rate": 7.843999999999999e-05, | |
| "loss": 23.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.008943638204522443, | |
| "grad_norm": 4.711969813797623e-05, | |
| "learning_rate": 7.790999999999999e-05, | |
| "loss": 23.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.00911238609517381, | |
| "grad_norm": 5.1224196795374155e-05, | |
| "learning_rate": 7.738e-05, | |
| "loss": 23.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.009281133985825177, | |
| "grad_norm": 3.525058491504751e-05, | |
| "learning_rate": 7.685e-05, | |
| "loss": 23.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.009449881876476544, | |
| "grad_norm": 4.9862577725434676e-05, | |
| "learning_rate": 7.632e-05, | |
| "loss": 23.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.009618629767127912, | |
| "grad_norm": 9.993968706112355e-05, | |
| "learning_rate": 7.578999999999999e-05, | |
| "loss": 23.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.009787377657779277, | |
| "grad_norm": 4.630363400792703e-05, | |
| "learning_rate": 7.526e-05, | |
| "loss": 23.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.009956125548430644, | |
| "grad_norm": 3.37816818500869e-05, | |
| "learning_rate": 7.473e-05, | |
| "loss": 23.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.010124873439082012, | |
| "grad_norm": 7.735610415693372e-05, | |
| "learning_rate": 7.419999999999999e-05, | |
| "loss": 23.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.010293621329733379, | |
| "grad_norm": 3.4660544770304114e-05, | |
| "learning_rate": 7.367e-05, | |
| "loss": 23.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.010462369220384746, | |
| "grad_norm": 7.357754657277837e-05, | |
| "learning_rate": 7.314e-05, | |
| "loss": 23.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.010631117111036112, | |
| "grad_norm": 6.574280996574089e-05, | |
| "learning_rate": 7.261e-05, | |
| "loss": 23.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.010799865001687479, | |
| "grad_norm": 7.882023783167824e-05, | |
| "learning_rate": 7.208e-05, | |
| "loss": 23.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.010968612892338846, | |
| "grad_norm": 7.584910781588405e-05, | |
| "learning_rate": 7.154999999999999e-05, | |
| "loss": 23.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.011137360782990213, | |
| "grad_norm": 8.913804776966572e-05, | |
| "learning_rate": 7.102e-05, | |
| "loss": 23.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.011306108673641579, | |
| "grad_norm": 6.0102716815890744e-05, | |
| "learning_rate": 7.049e-05, | |
| "loss": 23.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.011474856564292946, | |
| "grad_norm": 6.130318797659129e-05, | |
| "learning_rate": 6.996e-05, | |
| "loss": 23.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.011643604454944313, | |
| "grad_norm": 3.667854252853431e-05, | |
| "learning_rate": 6.943e-05, | |
| "loss": 23.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.01181235234559568, | |
| "grad_norm": 7.211296178866178e-05, | |
| "learning_rate": 6.89e-05, | |
| "loss": 23.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.011981100236247048, | |
| "grad_norm": 0.0001030169878504239, | |
| "learning_rate": 6.837e-05, | |
| "loss": 23.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.012149848126898413, | |
| "grad_norm": 0.0001360457536065951, | |
| "learning_rate": 6.784e-05, | |
| "loss": 23.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.01231859601754978, | |
| "grad_norm": 6.413022492779419e-05, | |
| "learning_rate": 6.730999999999999e-05, | |
| "loss": 23.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.012487343908201148, | |
| "grad_norm": 0.00011897562217200175, | |
| "learning_rate": 6.678e-05, | |
| "loss": 23.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.012656091798852515, | |
| "grad_norm": 6.470419612014666e-05, | |
| "learning_rate": 6.625e-05, | |
| "loss": 23.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01282483968950388, | |
| "grad_norm": 7.391779217869043e-05, | |
| "learning_rate": 6.572e-05, | |
| "loss": 23.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.012993587580155248, | |
| "grad_norm": 0.00013688707258552313, | |
| "learning_rate": 6.519e-05, | |
| "loss": 23.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.013162335470806615, | |
| "grad_norm": 0.00010867404489545152, | |
| "learning_rate": 6.466e-05, | |
| "loss": 23.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.013331083361457982, | |
| "grad_norm": 7.905341772129759e-05, | |
| "learning_rate": 6.413e-05, | |
| "loss": 23.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.01349983125210935, | |
| "grad_norm": 7.693654333706945e-05, | |
| "learning_rate": 6.359999999999999e-05, | |
| "loss": 23.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.013668579142760715, | |
| "grad_norm": 0.00014570896746590734, | |
| "learning_rate": 6.306999999999999e-05, | |
| "loss": 23.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.013837327033412082, | |
| "grad_norm": 9.294889605371282e-05, | |
| "learning_rate": 6.254000000000001e-05, | |
| "loss": 23.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.01400607492406345, | |
| "grad_norm": 6.171363202156499e-05, | |
| "learning_rate": 6.201e-05, | |
| "loss": 23.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.014174822814714817, | |
| "grad_norm": 0.00016772393428254873, | |
| "learning_rate": 6.148e-05, | |
| "loss": 23.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.014343570705366182, | |
| "grad_norm": 9.405903983861208e-05, | |
| "learning_rate": 6.095e-05, | |
| "loss": 23.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.01451231859601755, | |
| "grad_norm": 0.00011767326213885099, | |
| "learning_rate": 6.0419999999999994e-05, | |
| "loss": 23.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.014681066486668917, | |
| "grad_norm": 6.386899622157216e-05, | |
| "learning_rate": 5.988999999999999e-05, | |
| "loss": 23.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.014849814377320284, | |
| "grad_norm": 0.00018532607646193355, | |
| "learning_rate": 5.9359999999999994e-05, | |
| "loss": 23.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.015018562267971651, | |
| "grad_norm": 7.472938159480691e-05, | |
| "learning_rate": 5.8830000000000004e-05, | |
| "loss": 23.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.015187310158623017, | |
| "grad_norm": 0.00012555929424706846, | |
| "learning_rate": 5.83e-05, | |
| "loss": 23.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.015356058049274384, | |
| "grad_norm": 9.398732800036669e-05, | |
| "learning_rate": 5.777e-05, | |
| "loss": 23.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.015524805939925751, | |
| "grad_norm": 0.00015068292850628495, | |
| "learning_rate": 5.7239999999999994e-05, | |
| "loss": 23.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.015693553830577116, | |
| "grad_norm": 0.0002185801713494584, | |
| "learning_rate": 5.671e-05, | |
| "loss": 23.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.015862301721228485, | |
| "grad_norm": 7.917396578704938e-05, | |
| "learning_rate": 5.6179999999999994e-05, | |
| "loss": 23.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.01603104961187985, | |
| "grad_norm": 0.00011329845438012853, | |
| "learning_rate": 5.5650000000000004e-05, | |
| "loss": 23.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.01619979750253122, | |
| "grad_norm": 0.00015697274648118764, | |
| "learning_rate": 5.512e-05, | |
| "loss": 23.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.016368545393182585, | |
| "grad_norm": 0.00018705315596889704, | |
| "learning_rate": 5.459e-05, | |
| "loss": 23.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.01653729328383395, | |
| "grad_norm": 0.00033546588383615017, | |
| "learning_rate": 5.406e-05, | |
| "loss": 23.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01670604117448532, | |
| "grad_norm": 0.00029555062064900994, | |
| "learning_rate": 5.353e-05, | |
| "loss": 23.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.016874789065136685, | |
| "grad_norm": 0.00036910088965669274, | |
| "learning_rate": 5.2999999999999994e-05, | |
| "loss": 23.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.016874789065136685, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 31.1248, | |
| "eval_samples_per_second": 80.161, | |
| "eval_steps_per_second": 20.048, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.017043536955788054, | |
| "grad_norm": 7.682123396079987e-05, | |
| "learning_rate": 5.246999999999999e-05, | |
| "loss": 23.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.01721228484643942, | |
| "grad_norm": 7.189896132331342e-05, | |
| "learning_rate": 5.194e-05, | |
| "loss": 23.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.017381032737090785, | |
| "grad_norm": 6.011120058246888e-05, | |
| "learning_rate": 5.141e-05, | |
| "loss": 23.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.017549780627742154, | |
| "grad_norm": 6.253538595046848e-05, | |
| "learning_rate": 5.088e-05, | |
| "loss": 23.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.01771852851839352, | |
| "grad_norm": 5.117239925311878e-05, | |
| "learning_rate": 5.035e-05, | |
| "loss": 23.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.017887276409044885, | |
| "grad_norm": 8.745577360969037e-05, | |
| "learning_rate": 4.9819999999999994e-05, | |
| "loss": 23.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.018056024299696254, | |
| "grad_norm": 0.00013368716463446617, | |
| "learning_rate": 4.929e-05, | |
| "loss": 23.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01822477219034762, | |
| "grad_norm": 9.314952330896631e-05, | |
| "learning_rate": 4.876e-05, | |
| "loss": 23.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.01839352008099899, | |
| "grad_norm": 0.0001433063152944669, | |
| "learning_rate": 4.823e-05, | |
| "loss": 23.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.018562267971650354, | |
| "grad_norm": 5.847116335644387e-05, | |
| "learning_rate": 4.7699999999999994e-05, | |
| "loss": 23.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01873101586230172, | |
| "grad_norm": 8.442316902801394e-05, | |
| "learning_rate": 4.717e-05, | |
| "loss": 23.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.01889976375295309, | |
| "grad_norm": 0.00013739695714320987, | |
| "learning_rate": 4.6639999999999994e-05, | |
| "loss": 23.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.019068511643604454, | |
| "grad_norm": 0.00012455254909582436, | |
| "learning_rate": 4.611e-05, | |
| "loss": 23.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.019237259534255823, | |
| "grad_norm": 5.286196028464474e-05, | |
| "learning_rate": 4.558e-05, | |
| "loss": 23.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.01940600742490719, | |
| "grad_norm": 0.00013484164082910866, | |
| "learning_rate": 4.505e-05, | |
| "loss": 23.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.019574755315558554, | |
| "grad_norm": 0.00012526288628578186, | |
| "learning_rate": 4.4519999999999994e-05, | |
| "loss": 23.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.019743503206209923, | |
| "grad_norm": 0.00020170622156001627, | |
| "learning_rate": 4.399e-05, | |
| "loss": 23.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.01991225109686129, | |
| "grad_norm": 8.426374552072957e-05, | |
| "learning_rate": 4.346e-05, | |
| "loss": 23.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.020080998987512658, | |
| "grad_norm": 0.00027226656675338745, | |
| "learning_rate": 4.293e-05, | |
| "loss": 23.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.020249746878164023, | |
| "grad_norm": 0.00015929453365970403, | |
| "learning_rate": 4.2399999999999994e-05, | |
| "loss": 23.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.02041849476881539, | |
| "grad_norm": 0.00010509557614568621, | |
| "learning_rate": 4.187e-05, | |
| "loss": 23.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.020587242659466758, | |
| "grad_norm": 0.0001663622388150543, | |
| "learning_rate": 4.134e-05, | |
| "loss": 23.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.020755990550118123, | |
| "grad_norm": 0.0001297971175517887, | |
| "learning_rate": 4.081e-05, | |
| "loss": 23.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.020924738440769492, | |
| "grad_norm": 0.0001566116261528805, | |
| "learning_rate": 4.028e-05, | |
| "loss": 23.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.021093486331420858, | |
| "grad_norm": 0.00023184812744148076, | |
| "learning_rate": 3.975e-05, | |
| "loss": 23.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.021262234222072223, | |
| "grad_norm": 0.0001358414883725345, | |
| "learning_rate": 3.9219999999999994e-05, | |
| "loss": 23.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.021430982112723592, | |
| "grad_norm": 0.00010918582847807556, | |
| "learning_rate": 3.869e-05, | |
| "loss": 23.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.021599730003374958, | |
| "grad_norm": 0.0001907457917695865, | |
| "learning_rate": 3.816e-05, | |
| "loss": 23.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.021768477894026323, | |
| "grad_norm": 0.00012014003732474521, | |
| "learning_rate": 3.763e-05, | |
| "loss": 23.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.021937225784677692, | |
| "grad_norm": 0.00033094800892286, | |
| "learning_rate": 3.7099999999999994e-05, | |
| "loss": 23.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.022105973675329058, | |
| "grad_norm": 0.0001388384698657319, | |
| "learning_rate": 3.657e-05, | |
| "loss": 23.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.022274721565980427, | |
| "grad_norm": 0.00015554000856354833, | |
| "learning_rate": 3.604e-05, | |
| "loss": 23.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.022443469456631792, | |
| "grad_norm": 0.0001033740772982128, | |
| "learning_rate": 3.551e-05, | |
| "loss": 23.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.022612217347283158, | |
| "grad_norm": 0.00017761386698111892, | |
| "learning_rate": 3.498e-05, | |
| "loss": 23.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.022780965237934526, | |
| "grad_norm": 0.0001914931199280545, | |
| "learning_rate": 3.445e-05, | |
| "loss": 23.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.022949713128585892, | |
| "grad_norm": 0.0001077301349141635, | |
| "learning_rate": 3.392e-05, | |
| "loss": 23.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.02311846101923726, | |
| "grad_norm": 0.00014837745402473956, | |
| "learning_rate": 3.339e-05, | |
| "loss": 23.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.023287208909888626, | |
| "grad_norm": 0.00013769303041044623, | |
| "learning_rate": 3.286e-05, | |
| "loss": 23.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.023455956800539992, | |
| "grad_norm": 0.00011395317415008321, | |
| "learning_rate": 3.233e-05, | |
| "loss": 23.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.02362470469119136, | |
| "grad_norm": 0.0002430876047583297, | |
| "learning_rate": 3.1799999999999994e-05, | |
| "loss": 23.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.023793452581842726, | |
| "grad_norm": 0.00016095052706077695, | |
| "learning_rate": 3.1270000000000004e-05, | |
| "loss": 23.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.023962200472494095, | |
| "grad_norm": 0.00023896507627796382, | |
| "learning_rate": 3.074e-05, | |
| "loss": 23.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.02413094836314546, | |
| "grad_norm": 0.0002305787493241951, | |
| "learning_rate": 3.0209999999999997e-05, | |
| "loss": 23.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.024299696253796826, | |
| "grad_norm": 0.00044004127266816795, | |
| "learning_rate": 2.9679999999999997e-05, | |
| "loss": 23.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.024468444144448195, | |
| "grad_norm": 0.0004754870606120676, | |
| "learning_rate": 2.915e-05, | |
| "loss": 23.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.02463719203509956, | |
| "grad_norm": 0.00039452221244573593, | |
| "learning_rate": 2.8619999999999997e-05, | |
| "loss": 23.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.02480593992575093, | |
| "grad_norm": 0.0002141989243682474, | |
| "learning_rate": 2.8089999999999997e-05, | |
| "loss": 23.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.024974687816402295, | |
| "grad_norm": 0.00038773109554313123, | |
| "learning_rate": 2.756e-05, | |
| "loss": 23.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.02514343570705366, | |
| "grad_norm": 0.00020906349527649581, | |
| "learning_rate": 2.703e-05, | |
| "loss": 23.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.02531218359770503, | |
| "grad_norm": 0.00029203391750343144, | |
| "learning_rate": 2.6499999999999997e-05, | |
| "loss": 23.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02531218359770503, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 31.0725, | |
| "eval_samples_per_second": 80.296, | |
| "eval_steps_per_second": 20.082, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.025480931488356395, | |
| "grad_norm": 8.79631916177459e-05, | |
| "learning_rate": 2.597e-05, | |
| "loss": 23.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.02564967937900776, | |
| "grad_norm": 0.00010261518764309585, | |
| "learning_rate": 2.544e-05, | |
| "loss": 23.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.02581842726965913, | |
| "grad_norm": 0.00018040614668279886, | |
| "learning_rate": 2.4909999999999997e-05, | |
| "loss": 23.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.025987175160310495, | |
| "grad_norm": 0.00017360891797579825, | |
| "learning_rate": 2.438e-05, | |
| "loss": 23.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.026155923050961864, | |
| "grad_norm": 7.864111830713227e-05, | |
| "learning_rate": 2.3849999999999997e-05, | |
| "loss": 23.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.02632467094161323, | |
| "grad_norm": 0.0001662558934185654, | |
| "learning_rate": 2.3319999999999997e-05, | |
| "loss": 23.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.026493418832264595, | |
| "grad_norm": 0.00011379940406186506, | |
| "learning_rate": 2.279e-05, | |
| "loss": 23.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.026662166722915964, | |
| "grad_norm": 0.00011387839913368225, | |
| "learning_rate": 2.2259999999999997e-05, | |
| "loss": 23.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.02683091461356733, | |
| "grad_norm": 0.0001579077506903559, | |
| "learning_rate": 2.173e-05, | |
| "loss": 23.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0269996625042187, | |
| "grad_norm": 0.00015800283290445805, | |
| "learning_rate": 2.1199999999999997e-05, | |
| "loss": 23.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.027168410394870064, | |
| "grad_norm": 7.721775182290003e-05, | |
| "learning_rate": 2.067e-05, | |
| "loss": 23.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.02733715828552143, | |
| "grad_norm": 6.047181159374304e-05, | |
| "learning_rate": 2.014e-05, | |
| "loss": 23.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.0275059061761728, | |
| "grad_norm": 7.38459566491656e-05, | |
| "learning_rate": 1.9609999999999997e-05, | |
| "loss": 23.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.027674654066824164, | |
| "grad_norm": 0.00011628502397798002, | |
| "learning_rate": 1.908e-05, | |
| "loss": 23.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.027843401957475533, | |
| "grad_norm": 0.00011288729729130864, | |
| "learning_rate": 1.8549999999999997e-05, | |
| "loss": 23.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0280121498481269, | |
| "grad_norm": 9.810877236304805e-05, | |
| "learning_rate": 1.802e-05, | |
| "loss": 23.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.028180897738778264, | |
| "grad_norm": 0.00023592002980876714, | |
| "learning_rate": 1.749e-05, | |
| "loss": 23.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.028349645629429633, | |
| "grad_norm": 0.00020759565813932568, | |
| "learning_rate": 1.696e-05, | |
| "loss": 23.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.028518393520081, | |
| "grad_norm": 0.0002557075640652329, | |
| "learning_rate": 1.643e-05, | |
| "loss": 23.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.028687141410732364, | |
| "grad_norm": 0.0001290784712182358, | |
| "learning_rate": 1.5899999999999997e-05, | |
| "loss": 23.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.028855889301383733, | |
| "grad_norm": 8.08468321338296e-05, | |
| "learning_rate": 1.537e-05, | |
| "loss": 23.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0290246371920351, | |
| "grad_norm": 0.00019896485900972039, | |
| "learning_rate": 1.4839999999999999e-05, | |
| "loss": 23.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.029193385082686468, | |
| "grad_norm": 0.0002528635086491704, | |
| "learning_rate": 1.4309999999999999e-05, | |
| "loss": 23.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.029362132973337833, | |
| "grad_norm": 0.00024162382760550827, | |
| "learning_rate": 1.378e-05, | |
| "loss": 23.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0295308808639892, | |
| "grad_norm": 0.00018234840536024421, | |
| "learning_rate": 1.3249999999999999e-05, | |
| "loss": 23.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.029699628754640568, | |
| "grad_norm": 0.00025969985290430486, | |
| "learning_rate": 1.272e-05, | |
| "loss": 23.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.029868376645291933, | |
| "grad_norm": 0.0002434734342386946, | |
| "learning_rate": 1.219e-05, | |
| "loss": 23.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.030037124535943302, | |
| "grad_norm": 0.00023295178834814578, | |
| "learning_rate": 1.1659999999999998e-05, | |
| "loss": 23.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.030205872426594668, | |
| "grad_norm": 9.934210538631305e-05, | |
| "learning_rate": 1.1129999999999998e-05, | |
| "loss": 23.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.030374620317246033, | |
| "grad_norm": 0.0003217124321963638, | |
| "learning_rate": 1.0599999999999998e-05, | |
| "loss": 23.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.030543368207897402, | |
| "grad_norm": 0.00017936142103280872, | |
| "learning_rate": 1.007e-05, | |
| "loss": 23.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.030712116098548767, | |
| "grad_norm": 0.00024236104218289256, | |
| "learning_rate": 9.54e-06, | |
| "loss": 23.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.030880863989200136, | |
| "grad_norm": 0.00013197006774134934, | |
| "learning_rate": 9.01e-06, | |
| "loss": 23.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.031049611879851502, | |
| "grad_norm": 0.00025421561440452933, | |
| "learning_rate": 8.48e-06, | |
| "loss": 23.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.031218359770502867, | |
| "grad_norm": 0.00021938957797829062, | |
| "learning_rate": 7.949999999999998e-06, | |
| "loss": 23.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.03138710766115423, | |
| "grad_norm": 0.00020342513744253665, | |
| "learning_rate": 7.419999999999999e-06, | |
| "loss": 23.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.0315558555518056, | |
| "grad_norm": 0.00019843995687551796, | |
| "learning_rate": 6.89e-06, | |
| "loss": 23.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.03172460344245697, | |
| "grad_norm": 0.0004309473733883351, | |
| "learning_rate": 6.36e-06, | |
| "loss": 23.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.03189335133310833, | |
| "grad_norm": 0.00014857637870591134, | |
| "learning_rate": 5.829999999999999e-06, | |
| "loss": 23.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0320620992237597, | |
| "grad_norm": 0.00024400238180533051, | |
| "learning_rate": 5.299999999999999e-06, | |
| "loss": 23.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.03223084711441107, | |
| "grad_norm": 0.0002994404057972133, | |
| "learning_rate": 4.77e-06, | |
| "loss": 23.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.03239959500506244, | |
| "grad_norm": 0.00024177887826226652, | |
| "learning_rate": 4.24e-06, | |
| "loss": 23.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0325683428957138, | |
| "grad_norm": 0.00022951627033762634, | |
| "learning_rate": 3.7099999999999996e-06, | |
| "loss": 23.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.03273709078636517, | |
| "grad_norm": 0.00020361649512778968, | |
| "learning_rate": 3.18e-06, | |
| "loss": 23.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.03290583867701654, | |
| "grad_norm": 0.00028880671015940607, | |
| "learning_rate": 2.6499999999999996e-06, | |
| "loss": 23.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0330745865676679, | |
| "grad_norm": 0.0002678855962585658, | |
| "learning_rate": 2.12e-06, | |
| "loss": 23.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.03324333445831927, | |
| "grad_norm": 0.00022941327188163996, | |
| "learning_rate": 1.59e-06, | |
| "loss": 23.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.03341208234897064, | |
| "grad_norm": 0.00020409403077792376, | |
| "learning_rate": 1.06e-06, | |
| "loss": 23.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.033580830239622, | |
| "grad_norm": 0.0004483718948904425, | |
| "learning_rate": 5.3e-07, | |
| "loss": 23.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.03374957813027337, | |
| "grad_norm": 0.00058153917780146, | |
| "learning_rate": 0.0, | |
| "loss": 23.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03374957813027337, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 31.0274, | |
| "eval_samples_per_second": 80.413, | |
| "eval_steps_per_second": 20.111, | |
| "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": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8039615692800.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |