| { | |
| "best_metric": 1.757194995880127, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.018896893823077833, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 9.448446911538915e-05, | |
| "grad_norm": 1.9333150386810303, | |
| "learning_rate": 1e-05, | |
| "loss": 2.9126, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 9.448446911538915e-05, | |
| "eval_loss": 3.2697126865386963, | |
| "eval_runtime": 579.8103, | |
| "eval_samples_per_second": 30.745, | |
| "eval_steps_per_second": 7.687, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0001889689382307783, | |
| "grad_norm": 2.8160860538482666, | |
| "learning_rate": 2e-05, | |
| "loss": 3.4601, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00028345340734616746, | |
| "grad_norm": 4.041690826416016, | |
| "learning_rate": 3e-05, | |
| "loss": 3.5655, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0003779378764615566, | |
| "grad_norm": 2.4956092834472656, | |
| "learning_rate": 4e-05, | |
| "loss": 3.3305, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0004724223455769458, | |
| "grad_norm": 2.446577548980713, | |
| "learning_rate": 5e-05, | |
| "loss": 2.8449, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005669068146923349, | |
| "grad_norm": 2.102759838104248, | |
| "learning_rate": 6e-05, | |
| "loss": 2.6481, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0006613912838077241, | |
| "grad_norm": 2.016897678375244, | |
| "learning_rate": 7e-05, | |
| "loss": 2.6101, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0007558757529231132, | |
| "grad_norm": 1.851047158241272, | |
| "learning_rate": 8e-05, | |
| "loss": 2.4416, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0008503602220385024, | |
| "grad_norm": 1.752279281616211, | |
| "learning_rate": 9e-05, | |
| "loss": 2.3294, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0009448446911538916, | |
| "grad_norm": 1.6278218030929565, | |
| "learning_rate": 0.0001, | |
| "loss": 2.1997, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0010393291602692808, | |
| "grad_norm": 1.5926096439361572, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.1491, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0011338136293846698, | |
| "grad_norm": 2.0885679721832275, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.1646, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.001228298098500059, | |
| "grad_norm": 2.2613275051116943, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.1292, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0013227825676154481, | |
| "grad_norm": 1.5518369674682617, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.9997, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0014172670367308374, | |
| "grad_norm": 1.6676948070526123, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.9554, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0015117515058462264, | |
| "grad_norm": 1.2931146621704102, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.0118, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0016062359749616157, | |
| "grad_norm": 1.3132572174072266, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.829, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0017007204440770047, | |
| "grad_norm": 1.330643892288208, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.9375, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001795204913192394, | |
| "grad_norm": 1.1122055053710938, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.8101, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0018896893823077833, | |
| "grad_norm": 1.5229848623275757, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.849, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0019841738514231725, | |
| "grad_norm": 1.305350422859192, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.868, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0020786583205385616, | |
| "grad_norm": 1.3570218086242676, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.9399, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0021731427896539506, | |
| "grad_norm": 1.251835823059082, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.9291, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0022676272587693396, | |
| "grad_norm": 1.2934904098510742, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.8466, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.002362111727884729, | |
| "grad_norm": 1.42888343334198, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.8205, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002456596197000118, | |
| "grad_norm": 1.5526320934295654, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.931, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002551080666115507, | |
| "grad_norm": 1.4184902906417847, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.9254, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0026455651352308963, | |
| "grad_norm": 1.3838050365447998, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.7265, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0027400496043462857, | |
| "grad_norm": 2.047424793243408, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.8256, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0028345340734616748, | |
| "grad_norm": 1.7719075679779053, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.8126, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002929018542577064, | |
| "grad_norm": 1.8356136083602905, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.8655, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003023503011692453, | |
| "grad_norm": 1.4337571859359741, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.7771, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0031179874808078423, | |
| "grad_norm": 1.4906586408615112, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.7153, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0032124719499232314, | |
| "grad_norm": 1.4137147665023804, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.6966, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0033069564190386204, | |
| "grad_norm": 1.519652247428894, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.9404, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0034014408881540095, | |
| "grad_norm": 1.4712717533111572, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.773, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.003495925357269399, | |
| "grad_norm": 1.910886287689209, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.8614, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.003590409826384788, | |
| "grad_norm": 1.5718668699264526, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.7963, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.003684894295500177, | |
| "grad_norm": 1.5304378271102905, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.8913, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0037793787646155665, | |
| "grad_norm": 3.7432029247283936, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.8994, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0038738632337309556, | |
| "grad_norm": 1.7016750574111938, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.8223, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.003968347702846345, | |
| "grad_norm": 1.7171598672866821, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.1064, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004062832171961734, | |
| "grad_norm": 1.5463639497756958, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.8284, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004157316641077123, | |
| "grad_norm": 4.164541721343994, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.9552, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004251801110192512, | |
| "grad_norm": 2.117732286453247, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.0919, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004346285579307901, | |
| "grad_norm": 2.1728408336639404, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.97, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004440770048423291, | |
| "grad_norm": 1.6911128759384155, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.8933, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004535254517538679, | |
| "grad_norm": 2.073023796081543, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.8497, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.004629738986654069, | |
| "grad_norm": 2.810434103012085, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.895, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.004724223455769458, | |
| "grad_norm": 2.8073461055755615, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.0466, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004724223455769458, | |
| "eval_loss": 1.932782769203186, | |
| "eval_runtime": 581.4246, | |
| "eval_samples_per_second": 30.659, | |
| "eval_steps_per_second": 7.666, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004818707924884847, | |
| "grad_norm": 3.2429471015930176, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.2514, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.004913192394000236, | |
| "grad_norm": 2.7793526649475098, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.4193, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005007676863115625, | |
| "grad_norm": 1.7214088439941406, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.2655, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005102161332231014, | |
| "grad_norm": 1.9832342863082886, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.3851, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005196645801346404, | |
| "grad_norm": 1.2338030338287354, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.9084, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0052911302704617925, | |
| "grad_norm": 1.087687611579895, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.7588, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005385614739577182, | |
| "grad_norm": 0.9978936910629272, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.7094, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0054800992086925715, | |
| "grad_norm": 1.075069546699524, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.7796, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.00557458367780796, | |
| "grad_norm": 0.9746108651161194, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.7147, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0056690681469233496, | |
| "grad_norm": 1.0000801086425781, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.8339, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.005763552616038739, | |
| "grad_norm": 0.9693276882171631, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.7672, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.005858037085154128, | |
| "grad_norm": 0.9413057565689087, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.6684, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.005952521554269517, | |
| "grad_norm": 1.0468066930770874, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.7496, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006047006023384906, | |
| "grad_norm": 0.9703423380851746, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.7785, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.006141490492500295, | |
| "grad_norm": 1.025797724723816, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.7526, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006235974961615685, | |
| "grad_norm": 1.0511350631713867, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.7325, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006330459430731073, | |
| "grad_norm": 1.152138590812683, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.7647, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006424943899846463, | |
| "grad_norm": 0.9773648977279663, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.6467, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006519428368961852, | |
| "grad_norm": 1.1374561786651611, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.8972, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.006613912838077241, | |
| "grad_norm": 1.1633188724517822, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.6887, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.00670839730719263, | |
| "grad_norm": 1.052025556564331, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.732, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.006802881776308019, | |
| "grad_norm": 1.2245405912399292, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.8181, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.006897366245423408, | |
| "grad_norm": 1.1203999519348145, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.7995, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.006991850714538798, | |
| "grad_norm": 1.1522430181503296, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.7269, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0070863351836541865, | |
| "grad_norm": 1.3500783443450928, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.8551, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.007180819652769576, | |
| "grad_norm": 1.1485049724578857, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.7065, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0072753041218849655, | |
| "grad_norm": 1.2016279697418213, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.7265, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007369788591000354, | |
| "grad_norm": 1.2644776105880737, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.7724, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0074642730601157435, | |
| "grad_norm": 1.2304818630218506, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.7601, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.007558757529231133, | |
| "grad_norm": 1.290157437324524, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.8474, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007653241998346522, | |
| "grad_norm": 1.248424768447876, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.819, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.007747726467461911, | |
| "grad_norm": 1.2796978950500488, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.7402, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0078422109365773, | |
| "grad_norm": 1.2680574655532837, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.6708, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.00793669540569269, | |
| "grad_norm": 1.5167262554168701, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.7557, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008031179874808078, | |
| "grad_norm": 1.4523342847824097, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.9022, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008125664343923467, | |
| "grad_norm": 1.3876347541809082, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.7835, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008220148813038857, | |
| "grad_norm": 1.47008216381073, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.776, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.008314633282154246, | |
| "grad_norm": 1.495134949684143, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.8661, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.008409117751269636, | |
| "grad_norm": 1.4588102102279663, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.7715, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.008503602220385023, | |
| "grad_norm": 1.5281662940979004, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.9156, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008598086689500413, | |
| "grad_norm": 1.4420088529586792, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.8325, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.008692571158615802, | |
| "grad_norm": 1.4622676372528076, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.7403, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.008787055627731192, | |
| "grad_norm": 1.4662892818450928, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.8083, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.008881540096846581, | |
| "grad_norm": 4.761938095092773, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.9105, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.00897602456596197, | |
| "grad_norm": 1.6844340562820435, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.727, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009070509035077359, | |
| "grad_norm": 1.597159743309021, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.847, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009164993504192748, | |
| "grad_norm": 1.6171674728393555, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.995, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009259477973308138, | |
| "grad_norm": 1.5847182273864746, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.8432, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.009353962442423527, | |
| "grad_norm": 1.671203851699829, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.8723, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.009448446911538916, | |
| "grad_norm": 1.6499465703964233, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.919, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009448446911538916, | |
| "eval_loss": 1.7929612398147583, | |
| "eval_runtime": 581.3831, | |
| "eval_samples_per_second": 30.661, | |
| "eval_steps_per_second": 7.666, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009542931380654304, | |
| "grad_norm": 1.1900925636291504, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.0002, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.009637415849769694, | |
| "grad_norm": 1.246945858001709, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.9059, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.009731900318885083, | |
| "grad_norm": 1.3612993955612183, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.0033, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.009826384788000473, | |
| "grad_norm": 1.561272144317627, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.0228, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.009920869257115862, | |
| "grad_norm": 0.9032238721847534, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6668, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.01001535372623125, | |
| "grad_norm": 1.0913615226745605, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.8146, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01010983819534664, | |
| "grad_norm": 0.936185359954834, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.8548, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.010204322664462029, | |
| "grad_norm": 0.9609870910644531, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.6464, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.010298807133577418, | |
| "grad_norm": 0.899878978729248, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.7359, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.010393291602692808, | |
| "grad_norm": 0.9809157848358154, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.6935, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.010487776071808197, | |
| "grad_norm": 0.9736418724060059, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.6766, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.010582260540923585, | |
| "grad_norm": 0.9102429747581482, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.704, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.010676745010038974, | |
| "grad_norm": 0.9363366365432739, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.6961, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.010771229479154364, | |
| "grad_norm": 0.9577345848083496, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.72, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.010865713948269753, | |
| "grad_norm": 1.033258318901062, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.6714, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.010960198417385143, | |
| "grad_norm": 0.9051771759986877, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.7117, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.01105468288650053, | |
| "grad_norm": 0.996087908744812, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.7179, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.01114916735561592, | |
| "grad_norm": 0.9880028367042542, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.6529, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.01124365182473131, | |
| "grad_norm": 1.0762004852294922, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.7738, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.011338136293846699, | |
| "grad_norm": 1.1333122253417969, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.6419, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.011432620762962089, | |
| "grad_norm": 0.9502783417701721, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.5949, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.011527105232077478, | |
| "grad_norm": 1.1124191284179688, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.8083, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.011621589701192866, | |
| "grad_norm": 1.213707447052002, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.7856, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.011716074170308255, | |
| "grad_norm": 1.2123509645462036, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.7704, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.011810558639423645, | |
| "grad_norm": 1.1718707084655762, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.6658, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.011905043108539034, | |
| "grad_norm": 1.242866039276123, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.6367, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.011999527577654424, | |
| "grad_norm": 1.2593830823898315, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.6954, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012094012046769811, | |
| "grad_norm": 1.1635302305221558, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.6607, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.012188496515885201, | |
| "grad_norm": 1.2189949750900269, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.7548, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.01228298098500059, | |
| "grad_norm": 1.4098221063613892, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.7796, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01237746545411598, | |
| "grad_norm": 1.4896501302719116, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.8862, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.01247194992323137, | |
| "grad_norm": 1.426547884941101, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.9333, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.012566434392346759, | |
| "grad_norm": 1.343456506729126, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.7487, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.012660918861462147, | |
| "grad_norm": 1.5635097026824951, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.8904, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.012755403330577536, | |
| "grad_norm": 1.3845025300979614, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.6632, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.012849887799692926, | |
| "grad_norm": 1.3566688299179077, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7332, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.012944372268808315, | |
| "grad_norm": 1.3113293647766113, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.7121, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.013038856737923704, | |
| "grad_norm": 1.3649717569351196, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.6727, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.013133341207039092, | |
| "grad_norm": 1.6140693426132202, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.7886, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.013227825676154482, | |
| "grad_norm": 1.4083361625671387, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.8824, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.013322310145269871, | |
| "grad_norm": 1.4232109785079956, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.8112, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01341679461438526, | |
| "grad_norm": 1.5610580444335938, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.686, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.01351127908350065, | |
| "grad_norm": 1.460500955581665, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.6369, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.013605763552616038, | |
| "grad_norm": 1.4880610704421997, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.8207, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.013700248021731427, | |
| "grad_norm": 1.4838835000991821, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.8259, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.013794732490846817, | |
| "grad_norm": 1.3945232629776, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.6692, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.013889216959962206, | |
| "grad_norm": 1.4665021896362305, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.8147, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.013983701429077596, | |
| "grad_norm": 1.4989385604858398, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.7704, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.014078185898192985, | |
| "grad_norm": 1.4875540733337402, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.6716, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.014172670367308373, | |
| "grad_norm": 1.9659397602081299, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.8831, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014172670367308373, | |
| "eval_loss": 1.7651623487472534, | |
| "eval_runtime": 581.394, | |
| "eval_samples_per_second": 30.661, | |
| "eval_steps_per_second": 7.666, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014267154836423762, | |
| "grad_norm": 1.266187071800232, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.9396, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.014361639305539152, | |
| "grad_norm": 1.3311487436294556, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.1726, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.014456123774654541, | |
| "grad_norm": 1.6654537916183472, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.2763, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.014550608243769931, | |
| "grad_norm": 1.3147859573364258, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.0177, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.014645092712885319, | |
| "grad_norm": 0.9603314399719238, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.7243, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.014739577182000708, | |
| "grad_norm": 0.8720261454582214, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.5657, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.014834061651116098, | |
| "grad_norm": 0.912075400352478, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.7066, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.014928546120231487, | |
| "grad_norm": 1.0191247463226318, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.7639, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.015023030589346877, | |
| "grad_norm": 0.973504900932312, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.6683, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.015117515058462266, | |
| "grad_norm": 1.2498688697814941, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.6786, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.015211999527577654, | |
| "grad_norm": 1.0481356382369995, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.773, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.015306483996693043, | |
| "grad_norm": 0.9499745965003967, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.6872, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.015400968465808433, | |
| "grad_norm": 0.9340441823005676, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.6789, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.015495452934923822, | |
| "grad_norm": 0.9798192381858826, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.7058, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.015589937404039212, | |
| "grad_norm": 0.9891881942749023, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.765, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0156844218731546, | |
| "grad_norm": 1.0343868732452393, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.7166, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.01577890634226999, | |
| "grad_norm": 1.042275071144104, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.6897, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.01587339081138538, | |
| "grad_norm": 1.0385280847549438, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.6589, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.015967875280500768, | |
| "grad_norm": 1.2660365104675293, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.6698, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.016062359749616156, | |
| "grad_norm": 1.1566091775894165, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.676, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.016156844218731547, | |
| "grad_norm": 1.0901968479156494, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.7482, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.016251328687846935, | |
| "grad_norm": 1.038339376449585, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.6376, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.016345813156962326, | |
| "grad_norm": 1.1729512214660645, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.6658, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.016440297626077714, | |
| "grad_norm": 1.1359233856201172, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.7034, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0165347820951931, | |
| "grad_norm": 1.184664249420166, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.626, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.016629266564308492, | |
| "grad_norm": 1.0900330543518066, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.6683, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01672375103342388, | |
| "grad_norm": 1.121760606765747, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.7167, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01681823550253927, | |
| "grad_norm": 1.1315759420394897, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.7935, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01691271997165466, | |
| "grad_norm": 1.5151325464248657, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.6812, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.017007204440770047, | |
| "grad_norm": 1.2991371154785156, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.5557, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.017101688909885438, | |
| "grad_norm": 1.275704026222229, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.7451, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.017196173379000826, | |
| "grad_norm": 1.4051334857940674, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.7479, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.017290657848116217, | |
| "grad_norm": 1.2633906602859497, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.7556, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.017385142317231605, | |
| "grad_norm": 1.363315224647522, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.8108, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.017479626786346993, | |
| "grad_norm": 1.347550868988037, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.747, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.017574111255462384, | |
| "grad_norm": 1.4271029233932495, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.8299, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01766859572457777, | |
| "grad_norm": 1.3581104278564453, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.6718, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.017763080193693163, | |
| "grad_norm": 1.3167551755905151, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.6193, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01785756466280855, | |
| "grad_norm": 1.3643068075180054, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.6294, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01795204913192394, | |
| "grad_norm": 1.6632636785507202, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.7939, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01804653360103933, | |
| "grad_norm": 1.9095407724380493, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.9309, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.018141018070154717, | |
| "grad_norm": 1.4025647640228271, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.6995, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01823550253927011, | |
| "grad_norm": 1.3318939208984375, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.6632, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.018329987008385496, | |
| "grad_norm": 1.7360457181930542, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.7647, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.018424471477500887, | |
| "grad_norm": 1.5479822158813477, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.7412, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.018518955946616275, | |
| "grad_norm": 1.7617862224578857, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.7251, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.018613440415731663, | |
| "grad_norm": 1.6031512022018433, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.7793, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.018707924884847054, | |
| "grad_norm": 1.5139026641845703, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.7204, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.018802409353962442, | |
| "grad_norm": 1.5834425687789917, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.7269, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.018896893823077833, | |
| "grad_norm": 1.8426460027694702, | |
| "learning_rate": 0.0, | |
| "loss": 2.0622, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.018896893823077833, | |
| "eval_loss": 1.757194995880127, | |
| "eval_runtime": 581.5346, | |
| "eval_samples_per_second": 30.653, | |
| "eval_steps_per_second": 7.664, | |
| "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": 5.966773220278272e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |