| { | |
| "best_metric": 2.020968167926185e-06, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 2.588996763754045, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.012944983818770227, | |
| "grad_norm": 1.0203962326049805, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1373, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.012944983818770227, | |
| "eval_loss": 0.2981894910335541, | |
| "eval_runtime": 9.5643, | |
| "eval_samples_per_second": 13.592, | |
| "eval_steps_per_second": 3.45, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.025889967637540454, | |
| "grad_norm": 1.2321498394012451, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1639, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.038834951456310676, | |
| "grad_norm": 1.2425283193588257, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1687, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05177993527508091, | |
| "grad_norm": 1.0462325811386108, | |
| "learning_rate": 4e-05, | |
| "loss": 0.1525, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06472491909385113, | |
| "grad_norm": 0.6241796016693115, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0971, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07766990291262135, | |
| "grad_norm": 0.5437662601470947, | |
| "learning_rate": 6e-05, | |
| "loss": 0.0619, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.09061488673139159, | |
| "grad_norm": 0.6059942245483398, | |
| "learning_rate": 7e-05, | |
| "loss": 0.0309, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.10355987055016182, | |
| "grad_norm": 0.2637348473072052, | |
| "learning_rate": 8e-05, | |
| "loss": 0.0107, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.11650485436893204, | |
| "grad_norm": 0.12676547467708588, | |
| "learning_rate": 9e-05, | |
| "loss": 0.0039, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.12944983818770225, | |
| "grad_norm": 0.07366133481264114, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0013, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1423948220064725, | |
| "grad_norm": 0.06091802567243576, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.0008, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1553398058252427, | |
| "grad_norm": 0.09185962378978729, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.0005, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.16828478964401294, | |
| "grad_norm": 0.007255407050251961, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.0001, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.18122977346278318, | |
| "grad_norm": 0.19597330689430237, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.0009, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1941747572815534, | |
| "grad_norm": 0.002528818789869547, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.20711974110032363, | |
| "grad_norm": 0.0019764695316553116, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.22006472491909385, | |
| "grad_norm": 0.004141903482377529, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.0001, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.23300970873786409, | |
| "grad_norm": 0.24057364463806152, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.0001, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2459546925566343, | |
| "grad_norm": 0.007576645351946354, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.0001, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2588996763754045, | |
| "grad_norm": 0.11651884764432907, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.0008, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.27184466019417475, | |
| "grad_norm": 0.05754302814602852, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.0003, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.284789644012945, | |
| "grad_norm": 0.05541009455919266, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.0002, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.2977346278317152, | |
| "grad_norm": 0.002488422906026244, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3106796116504854, | |
| "grad_norm": 0.0021644814405590296, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.32362459546925565, | |
| "grad_norm": 0.003139674663543701, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3365695792880259, | |
| "grad_norm": 0.0062321946024894714, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.34951456310679613, | |
| "grad_norm": 0.0015530330128967762, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.36245954692556637, | |
| "grad_norm": 0.0009751790203154087, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.37540453074433655, | |
| "grad_norm": 0.0011889493325725198, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3883495145631068, | |
| "grad_norm": 0.0020521723199635744, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.40129449838187703, | |
| "grad_norm": 0.001421238761395216, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.41423948220064727, | |
| "grad_norm": 0.0026180348359048367, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.42718446601941745, | |
| "grad_norm": 0.001268848660402, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4401294498381877, | |
| "grad_norm": 0.0014048975426703691, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.45307443365695793, | |
| "grad_norm": 0.0029678246937692165, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.46601941747572817, | |
| "grad_norm": 0.0025685736909508705, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.47896440129449835, | |
| "grad_norm": 0.003125803079456091, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.4919093851132686, | |
| "grad_norm": 0.007677475456148386, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5048543689320388, | |
| "grad_norm": 0.0005735827726311982, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.517799352750809, | |
| "grad_norm": 0.01382033433765173, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5307443365695793, | |
| "grad_norm": 0.00015104783233255148, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.5436893203883495, | |
| "grad_norm": 0.0015415921807289124, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.5566343042071198, | |
| "grad_norm": 0.0002161674783565104, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.56957928802589, | |
| "grad_norm": 0.00024602553457953036, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.5825242718446602, | |
| "grad_norm": 0.0005134981474839151, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.5954692556634305, | |
| "grad_norm": 0.00012119917664676905, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6084142394822006, | |
| "grad_norm": 0.00011360419739503413, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.6213592233009708, | |
| "grad_norm": 0.0001089339202735573, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.6343042071197411, | |
| "grad_norm": 0.00019188143778592348, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.6472491909385113, | |
| "grad_norm": 0.0002130446519004181, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6472491909385113, | |
| "eval_loss": 5.114246505399933e-06, | |
| "eval_runtime": 9.8014, | |
| "eval_samples_per_second": 13.263, | |
| "eval_steps_per_second": 3.367, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6601941747572816, | |
| "grad_norm": 0.0008072551572695374, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.6731391585760518, | |
| "grad_norm": 0.0003115938452538103, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.686084142394822, | |
| "grad_norm": 0.0003803007130045444, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.6990291262135923, | |
| "grad_norm": 0.001679311622865498, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.7119741100323624, | |
| "grad_norm": 0.0014463587431237102, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.7249190938511327, | |
| "grad_norm": 0.00020687367941718549, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.7378640776699029, | |
| "grad_norm": 0.00012489533401094377, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.7508090614886731, | |
| "grad_norm": 0.0011946961749345064, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.7637540453074434, | |
| "grad_norm": 0.0034886086359620094, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.7766990291262136, | |
| "grad_norm": 0.019774286076426506, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.7896440129449838, | |
| "grad_norm": 0.00012018957932014018, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.8025889967637541, | |
| "grad_norm": 0.0008413722389377654, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.8155339805825242, | |
| "grad_norm": 7.062353688525036e-05, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.8284789644012945, | |
| "grad_norm": 0.0010793240508064628, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.8414239482200647, | |
| "grad_norm": 0.00023476831847801805, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.8543689320388349, | |
| "grad_norm": 0.00031458126613870263, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.8673139158576052, | |
| "grad_norm": 0.0001667094766162336, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.8802588996763754, | |
| "grad_norm": 0.00017667474457994103, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.8932038834951457, | |
| "grad_norm": 0.0005229483940638602, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.9061488673139159, | |
| "grad_norm": 0.00027375674108043313, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.919093851132686, | |
| "grad_norm": 0.00028505577938631177, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.9320388349514563, | |
| "grad_norm": 0.002035663463175297, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.9449838187702265, | |
| "grad_norm": 0.0007079575443640351, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.9579288025889967, | |
| "grad_norm": 0.0003011709777638316, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.970873786407767, | |
| "grad_norm": 0.0004272864025551826, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.9838187702265372, | |
| "grad_norm": 0.0001912742154672742, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.9967637540453075, | |
| "grad_norm": 0.0001855505834100768, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.0097087378640777, | |
| "grad_norm": 0.000368554494343698, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.022653721682848, | |
| "grad_norm": 0.0005397357745096087, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.035598705501618, | |
| "grad_norm": 9.197810140904039e-05, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0485436893203883, | |
| "grad_norm": 0.0047591556794941425, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.0614886731391586, | |
| "grad_norm": 9.622135985409841e-05, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.074433656957929, | |
| "grad_norm": 0.0002235884458059445, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.087378640776699, | |
| "grad_norm": 6.833249790361151e-05, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.1003236245954693, | |
| "grad_norm": 0.0005551163922064006, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.1132686084142396, | |
| "grad_norm": 0.00010846730583580211, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.1262135922330097, | |
| "grad_norm": 5.773642988060601e-05, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.13915857605178, | |
| "grad_norm": 0.00016505061648786068, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.1521035598705502, | |
| "grad_norm": 0.008876933716237545, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.1650485436893203, | |
| "grad_norm": 0.16270585358142853, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0007, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.1779935275080906, | |
| "grad_norm": 0.00022468912356998771, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.190938511326861, | |
| "grad_norm": 9.756765211932361e-05, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.203883495145631, | |
| "grad_norm": 0.00018813037604559213, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.2168284789644013, | |
| "grad_norm": 0.0003693510952871293, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.2297734627831716, | |
| "grad_norm": 0.0021898068953305483, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.2427184466019416, | |
| "grad_norm": 0.37865975499153137, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0006, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.255663430420712, | |
| "grad_norm": 0.0014068527380004525, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.2686084142394822, | |
| "grad_norm": 0.0009542718180455267, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.2815533980582523, | |
| "grad_norm": 0.09191913157701492, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0001, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.2944983818770226, | |
| "grad_norm": 0.004437869414687157, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.2944983818770226, | |
| "eval_loss": 7.178713531175163e-06, | |
| "eval_runtime": 9.7892, | |
| "eval_samples_per_second": 13.28, | |
| "eval_steps_per_second": 3.371, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.307443365695793, | |
| "grad_norm": 0.0008209234802052379, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.3203883495145632, | |
| "grad_norm": 0.054027754813432693, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0001, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.0007307673222385347, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.3462783171521036, | |
| "grad_norm": 0.005266569554805756, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.3592233009708738, | |
| "grad_norm": 0.0006765456637367606, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.3721682847896441, | |
| "grad_norm": 0.0358976311981678, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0001, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.3851132686084142, | |
| "grad_norm": 0.0003064029151573777, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.3980582524271845, | |
| "grad_norm": 0.00010614845086820424, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.4110032362459548, | |
| "grad_norm": 0.00012399001570884138, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.4239482200647249, | |
| "grad_norm": 0.003827552078291774, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.4368932038834952, | |
| "grad_norm": 0.0005444152629934251, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.4498381877022655, | |
| "grad_norm": 0.00020939875685144216, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.4627831715210355, | |
| "grad_norm": 0.00023036557831801474, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.4757281553398058, | |
| "grad_norm": 0.0022763346787542105, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.4886731391585761, | |
| "grad_norm": 0.00015546870417892933, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.5016181229773462, | |
| "grad_norm": 0.006108403671532869, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.5145631067961165, | |
| "grad_norm": 0.0018249741988256574, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.5275080906148868, | |
| "grad_norm": 9.492759272689e-05, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.5404530744336569, | |
| "grad_norm": 0.00016195476928260177, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.5533980582524272, | |
| "grad_norm": 0.0001234033115906641, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.5663430420711975, | |
| "grad_norm": 9.540676546748728e-05, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.5792880258899675, | |
| "grad_norm": 0.005599276628345251, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.5922330097087378, | |
| "grad_norm": 0.00025100537459366024, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.6051779935275081, | |
| "grad_norm": 0.0002124908205587417, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.6181229773462782, | |
| "grad_norm": 0.00010561349336057901, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.6310679611650487, | |
| "grad_norm": 0.00021081748127471656, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.6440129449838188, | |
| "grad_norm": 0.00012433409574441612, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.6569579288025889, | |
| "grad_norm": 9.320802928414196e-05, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.6699029126213594, | |
| "grad_norm": 9.787802991922945e-05, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.6828478964401294, | |
| "grad_norm": 0.0007250069174915552, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.6957928802588995, | |
| "grad_norm": 0.0006313190679065883, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.70873786407767, | |
| "grad_norm": 0.0002635714190546423, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.72168284789644, | |
| "grad_norm": 0.00019160538795404136, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.7346278317152104, | |
| "grad_norm": 0.0006658636266365647, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.7475728155339807, | |
| "grad_norm": 0.0035931889433413744, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.7605177993527508, | |
| "grad_norm": 0.00011197294952580705, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.773462783171521, | |
| "grad_norm": 0.0027357996441423893, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.7864077669902914, | |
| "grad_norm": 0.0010884864022955298, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.7993527508090614, | |
| "grad_norm": 0.00019315312965773046, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.8122977346278317, | |
| "grad_norm": 8.6668660514988e-05, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.825242718446602, | |
| "grad_norm": 5.9428366512292996e-05, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.838187702265372, | |
| "grad_norm": 0.10113563388586044, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0005, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.8511326860841424, | |
| "grad_norm": 0.002838822780176997, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.8640776699029127, | |
| "grad_norm": 0.0007735037361271679, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.8770226537216828, | |
| "grad_norm": 9.024533937918022e-05, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.889967637540453, | |
| "grad_norm": 0.0010357049759477377, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.9029126213592233, | |
| "grad_norm": 0.00022478362370748073, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.9158576051779934, | |
| "grad_norm": 5.8589426771504804e-05, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.9288025889967637, | |
| "grad_norm": 0.0005812260787934065, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.941747572815534, | |
| "grad_norm": 0.00010281232971465215, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.941747572815534, | |
| "eval_loss": 2.265722969241324e-06, | |
| "eval_runtime": 9.7942, | |
| "eval_samples_per_second": 13.273, | |
| "eval_steps_per_second": 3.369, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.954692556634304, | |
| "grad_norm": 0.0001606423466000706, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.9676375404530746, | |
| "grad_norm": 6.61443846183829e-05, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.9805825242718447, | |
| "grad_norm": 0.00035633280640468, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.9935275080906147, | |
| "grad_norm": 0.00015127830556593835, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.0064724919093853, | |
| "grad_norm": 0.018420366570353508, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.0194174757281553, | |
| "grad_norm": 8.060120308073238e-05, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.0323624595469254, | |
| "grad_norm": 4.847139643970877e-05, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.045307443365696, | |
| "grad_norm": 5.316246461006813e-05, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.058252427184466, | |
| "grad_norm": 0.01738496869802475, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.071197411003236, | |
| "grad_norm": 0.00018610119877848774, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.0841423948220066, | |
| "grad_norm": 4.633475327864289e-05, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.0970873786407767, | |
| "grad_norm": 5.792090450995602e-05, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.1100323624595467, | |
| "grad_norm": 3.176022437401116e-05, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.1229773462783172, | |
| "grad_norm": 6.537281296914443e-05, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.1359223300970873, | |
| "grad_norm": 7.813788397470489e-05, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.148867313915858, | |
| "grad_norm": 6.678163481410593e-05, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.161812297734628, | |
| "grad_norm": 6.245343683985993e-05, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.174757281553398, | |
| "grad_norm": 0.00010019366891356185, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.1877022653721685, | |
| "grad_norm": 8.844850526656955e-05, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.2006472491909386, | |
| "grad_norm": 9.018035780172795e-05, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.2135922330097086, | |
| "grad_norm": 5.3382853366201743e-05, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.226537216828479, | |
| "grad_norm": 5.0268845370737836e-05, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.2394822006472492, | |
| "grad_norm": 0.00014311954146251082, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.2524271844660193, | |
| "grad_norm": 6.574299914063886e-05, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.26537216828479, | |
| "grad_norm": 3.37300771207083e-05, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.27831715210356, | |
| "grad_norm": 0.00013506595860235393, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.29126213592233, | |
| "grad_norm": 0.00039620394818484783, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.3042071197411005, | |
| "grad_norm": 4.100550358998589e-05, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.3171521035598706, | |
| "grad_norm": 4.1235663957195356e-05, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.3300970873786406, | |
| "grad_norm": 0.0008670018287375569, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.343042071197411, | |
| "grad_norm": 5.1306054956512526e-05, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.355987055016181, | |
| "grad_norm": 0.00012276761117391288, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.3689320388349513, | |
| "grad_norm": 3.127683885395527e-05, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.381877022653722, | |
| "grad_norm": 0.00036061613354831934, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.394822006472492, | |
| "grad_norm": 5.2442424930632114e-05, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.407766990291262, | |
| "grad_norm": 3.742905391845852e-05, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.4207119741100325, | |
| "grad_norm": 0.0001615755318198353, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.4336569579288025, | |
| "grad_norm": 0.00017121432756539434, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.4466019417475726, | |
| "grad_norm": 7.700425339862704e-05, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.459546925566343, | |
| "grad_norm": 0.00012283331307116896, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.472491909385113, | |
| "grad_norm": 4.537297718343325e-05, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.4854368932038833, | |
| "grad_norm": 0.00021832490165252239, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.498381877022654, | |
| "grad_norm": 0.0002951624628622085, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.511326860841424, | |
| "grad_norm": 0.00016684699221514165, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.524271844660194, | |
| "grad_norm": 6.800913979532197e-05, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.5372168284789645, | |
| "grad_norm": 3.736185317393392e-05, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.5501618122977345, | |
| "grad_norm": 7.34529021428898e-05, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.5631067961165046, | |
| "grad_norm": 0.00027745647821575403, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.576051779935275, | |
| "grad_norm": 0.00021835808001924306, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.588996763754045, | |
| "grad_norm": 0.00012005464668618515, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.588996763754045, | |
| "eval_loss": 2.020968167926185e-06, | |
| "eval_runtime": 9.8069, | |
| "eval_samples_per_second": 13.256, | |
| "eval_steps_per_second": 3.365, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 3.01514151551828e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |