| { | |
| "best_metric": 1.6819664239883423, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.0176772140710624, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 8.838607035531201e-05, | |
| "grad_norm": 6.53464412689209, | |
| "learning_rate": 1e-05, | |
| "loss": 3.8502, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 8.838607035531201e-05, | |
| "eval_loss": 5.7815704345703125, | |
| "eval_runtime": 2020.2323, | |
| "eval_samples_per_second": 9.432, | |
| "eval_steps_per_second": 2.358, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00017677214071062401, | |
| "grad_norm": 6.685802936553955, | |
| "learning_rate": 2e-05, | |
| "loss": 4.1071, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00026515821106593603, | |
| "grad_norm": 7.507639408111572, | |
| "learning_rate": 3e-05, | |
| "loss": 4.2408, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00035354428142124803, | |
| "grad_norm": 6.150238990783691, | |
| "learning_rate": 4e-05, | |
| "loss": 3.8954, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00044193035177656, | |
| "grad_norm": 5.556862831115723, | |
| "learning_rate": 5e-05, | |
| "loss": 3.3462, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005303164221318721, | |
| "grad_norm": 5.092200756072998, | |
| "learning_rate": 6e-05, | |
| "loss": 3.074, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.000618702492487184, | |
| "grad_norm": 5.379222869873047, | |
| "learning_rate": 7e-05, | |
| "loss": 2.7401, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0007070885628424961, | |
| "grad_norm": 4.70679235458374, | |
| "learning_rate": 8e-05, | |
| "loss": 2.5116, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.000795474633197808, | |
| "grad_norm": 5.143443584442139, | |
| "learning_rate": 9e-05, | |
| "loss": 2.3826, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00088386070355312, | |
| "grad_norm": 3.7451655864715576, | |
| "learning_rate": 0.0001, | |
| "loss": 2.3914, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.000972246773908432, | |
| "grad_norm": 3.7099339962005615, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 2.2408, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0010606328442637441, | |
| "grad_norm": 3.3339648246765137, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.2038, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.001149018914619056, | |
| "grad_norm": 3.031165599822998, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 2.0562, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.001237404984974368, | |
| "grad_norm": 3.627713918685913, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 2.3146, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00132579105532968, | |
| "grad_norm": 3.2613565921783447, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 2.2534, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0014141771256849921, | |
| "grad_norm": 3.0103759765625, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 2.0051, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.001502563196040304, | |
| "grad_norm": 3.3258683681488037, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.2906, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.001590949266395616, | |
| "grad_norm": 2.946044445037842, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 2.2303, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001679335336750928, | |
| "grad_norm": 2.8645474910736084, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.0739, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00176772140710624, | |
| "grad_norm": 3.159550905227661, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 2.1103, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0018561074774615521, | |
| "grad_norm": 2.9757511615753174, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.0949, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.001944493547816864, | |
| "grad_norm": 3.2422940731048584, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.1918, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0020328796181721762, | |
| "grad_norm": 3.1693954467773438, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.1189, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0021212656885274883, | |
| "grad_norm": 3.409475088119507, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.151, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0022096517588828, | |
| "grad_norm": 3.3714005947113037, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.1188, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002298037829238112, | |
| "grad_norm": 3.51196026802063, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.4361, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.002386423899593424, | |
| "grad_norm": 3.1889004707336426, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.1662, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.002474809969948736, | |
| "grad_norm": 2.8497188091278076, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.0601, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.002563196040304048, | |
| "grad_norm": 2.6012513637542725, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.9917, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.00265158211065936, | |
| "grad_norm": 2.9800682067871094, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.3159, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002739968181014672, | |
| "grad_norm": 2.791994333267212, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.0652, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0028283542513699842, | |
| "grad_norm": 3.644646406173706, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.3555, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0029167403217252963, | |
| "grad_norm": 3.3598499298095703, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.2242, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.003005126392080608, | |
| "grad_norm": 3.3691976070404053, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.0999, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.00309351246243592, | |
| "grad_norm": 3.119875192642212, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.1714, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.003181898532791232, | |
| "grad_norm": 2.598971366882324, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.0018, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.003270284603146544, | |
| "grad_norm": 2.5810036659240723, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.1097, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.003358670673501856, | |
| "grad_norm": 3.0927035808563232, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.1079, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.003447056743857168, | |
| "grad_norm": 2.4966320991516113, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.0592, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00353544281421248, | |
| "grad_norm": 3.064293384552002, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.3416, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0036238288845677922, | |
| "grad_norm": 4.061313629150391, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.2301, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0037122149549231043, | |
| "grad_norm": 3.5145466327667236, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.1789, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0038006010252784163, | |
| "grad_norm": 2.5114963054656982, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.9405, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.003888987095633728, | |
| "grad_norm": 2.589874744415283, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.1264, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.00397737316598904, | |
| "grad_norm": 2.691175937652588, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.1812, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0040657592363443525, | |
| "grad_norm": 2.5702614784240723, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.2359, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0041541453066996645, | |
| "grad_norm": 2.8497369289398193, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.3361, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0042425313770549766, | |
| "grad_norm": 3.4419796466827393, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.4921, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.004330917447410288, | |
| "grad_norm": 3.733684539794922, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.4058, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0044193035177656, | |
| "grad_norm": 5.64238166809082, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.791, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0044193035177656, | |
| "eval_loss": 2.769718647003174, | |
| "eval_runtime": 2027.785, | |
| "eval_samples_per_second": 9.397, | |
| "eval_steps_per_second": 2.349, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004507689588120912, | |
| "grad_norm": 6.559182167053223, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.9342, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.004596075658476224, | |
| "grad_norm": 6.170121669769287, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.7304, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.004684461728831536, | |
| "grad_norm": 4.403921127319336, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.3755, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.004772847799186848, | |
| "grad_norm": 3.2437002658843994, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.9101, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.00486123386954216, | |
| "grad_norm": 3.1315319538116455, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.776, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.004949619939897472, | |
| "grad_norm": 2.976041555404663, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.807, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005038006010252784, | |
| "grad_norm": 3.4600260257720947, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.1235, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005126392080608096, | |
| "grad_norm": 2.90805721282959, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.9054, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.005214778150963408, | |
| "grad_norm": 2.7653403282165527, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.7711, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.00530316422131872, | |
| "grad_norm": 2.4785687923431396, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.8015, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.005391550291674032, | |
| "grad_norm": 2.7270736694335938, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.8218, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.005479936362029344, | |
| "grad_norm": 2.521132469177246, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.8076, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.005568322432384656, | |
| "grad_norm": 2.979375123977661, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.6767, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0056567085027399685, | |
| "grad_norm": 2.416459560394287, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.7175, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0057450945730952805, | |
| "grad_norm": 2.827824115753174, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.7098, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0058334806434505925, | |
| "grad_norm": 2.4488983154296875, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.5795, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.005921866713805905, | |
| "grad_norm": 2.954071283340454, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.7408, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006010252784161216, | |
| "grad_norm": 3.441884756088257, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.8142, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006098638854516528, | |
| "grad_norm": 3.259413719177246, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.932, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.00618702492487184, | |
| "grad_norm": 2.7689199447631836, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.792, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.006275410995227152, | |
| "grad_norm": 2.511742353439331, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.8017, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.006363797065582464, | |
| "grad_norm": 2.755967617034912, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.8016, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.006452183135937776, | |
| "grad_norm": 3.482419490814209, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 2.1019, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.006540569206293088, | |
| "grad_norm": 2.745054006576538, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 2.0913, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0066289552766484, | |
| "grad_norm": 2.840120553970337, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.9103, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.006717341347003712, | |
| "grad_norm": 2.36970591545105, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.7377, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.006805727417359024, | |
| "grad_norm": 3.196659564971924, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.1502, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.006894113487714336, | |
| "grad_norm": 3.0459177494049072, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.9075, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.006982499558069648, | |
| "grad_norm": 2.935102939605713, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.9224, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.00707088562842496, | |
| "grad_norm": 3.206444501876831, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.0231, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007159271698780272, | |
| "grad_norm": 2.17788028717041, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.7113, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0072476577691355844, | |
| "grad_norm": 2.160576343536377, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.8688, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0073360438394908965, | |
| "grad_norm": 2.738328456878662, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.0045, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0074244299098462085, | |
| "grad_norm": 2.4200971126556396, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.8772, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.007512815980201521, | |
| "grad_norm": 2.4582793712615967, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.8056, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.007601202050556833, | |
| "grad_norm": 2.3232736587524414, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.7085, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.007689588120912144, | |
| "grad_norm": 2.6430585384368896, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.8968, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.007777974191267456, | |
| "grad_norm": 2.9717512130737305, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.9066, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.007866360261622768, | |
| "grad_norm": 3.4684290885925293, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.0091, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.00795474633197808, | |
| "grad_norm": 2.8795735836029053, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.955, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008043132402333392, | |
| "grad_norm": 3.475484609603882, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.1407, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.008131518472688705, | |
| "grad_norm": 3.168266773223877, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.1106, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.008219904543044016, | |
| "grad_norm": 2.6973724365234375, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.9513, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.008308290613399329, | |
| "grad_norm": 2.9799537658691406, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.0633, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.00839667668375464, | |
| "grad_norm": 2.6777544021606445, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.1921, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.008485062754109953, | |
| "grad_norm": 2.5322084426879883, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.1012, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.008573448824465264, | |
| "grad_norm": 2.460355043411255, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.0223, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.008661834894820575, | |
| "grad_norm": 2.5451252460479736, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.1772, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.008750220965175888, | |
| "grad_norm": 2.993802070617676, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.3068, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0088386070355312, | |
| "grad_norm": 3.484532117843628, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.3767, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0088386070355312, | |
| "eval_loss": 2.2520580291748047, | |
| "eval_runtime": 2027.0061, | |
| "eval_samples_per_second": 9.401, | |
| "eval_steps_per_second": 2.35, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.008926993105886512, | |
| "grad_norm": 4.646420001983643, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.5802, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.009015379176241824, | |
| "grad_norm": 4.246099948883057, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.5119, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.009103765246597137, | |
| "grad_norm": 3.8484346866607666, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.1894, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.009192151316952448, | |
| "grad_norm": 3.377668857574463, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.0767, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.00928053738730776, | |
| "grad_norm": 3.1087446212768555, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9457, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.009368923457663072, | |
| "grad_norm": 2.5138378143310547, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.7825, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.009457309528018385, | |
| "grad_norm": 2.6978957653045654, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.7397, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.009545695598373696, | |
| "grad_norm": 2.6779136657714844, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.7387, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.009634081668729009, | |
| "grad_norm": 2.667104959487915, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.7842, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.00972246773908432, | |
| "grad_norm": 2.772145986557007, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.6166, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.009810853809439633, | |
| "grad_norm": 3.478252410888672, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.7501, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.009899239879794944, | |
| "grad_norm": 3.0266244411468506, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.6469, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.009987625950150257, | |
| "grad_norm": 2.9716689586639404, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.6877, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.010076012020505568, | |
| "grad_norm": 2.7877931594848633, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.8618, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.010164398090860881, | |
| "grad_norm": 2.4523091316223145, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.6637, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.010252784161216192, | |
| "grad_norm": 2.798417091369629, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.6339, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.010341170231571504, | |
| "grad_norm": 2.793260335922241, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.798, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.010429556301926816, | |
| "grad_norm": 2.5718994140625, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.6182, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.010517942372282128, | |
| "grad_norm": 2.7119593620300293, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.7379, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.01060632844263744, | |
| "grad_norm": 2.3828659057617188, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.7351, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.010694714512992752, | |
| "grad_norm": 3.208634376525879, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.7868, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.010783100583348065, | |
| "grad_norm": 2.6685726642608643, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.68, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.010871486653703376, | |
| "grad_norm": 2.2974953651428223, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.642, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.010959872724058689, | |
| "grad_norm": 2.9607999324798584, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.7523, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.011048258794414, | |
| "grad_norm": 2.7029364109039307, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.7139, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.011136644864769313, | |
| "grad_norm": 3.0020194053649902, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.8548, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.011225030935124624, | |
| "grad_norm": 3.1021790504455566, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.9282, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.011313417005479937, | |
| "grad_norm": 3.1041276454925537, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.9804, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.011401803075835248, | |
| "grad_norm": 2.8402466773986816, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.8393, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.011490189146190561, | |
| "grad_norm": 2.5224215984344482, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.6515, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.011578575216545872, | |
| "grad_norm": 2.701436996459961, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.7917, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.011666961286901185, | |
| "grad_norm": 2.6748011112213135, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.7698, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.011755347357256496, | |
| "grad_norm": 2.7161076068878174, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.7945, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.01184373342761181, | |
| "grad_norm": 3.0575504302978516, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.8757, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.01193211949796712, | |
| "grad_norm": 2.325249433517456, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.7374, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.012020505568322432, | |
| "grad_norm": 2.388791799545288, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.7831, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.012108891638677744, | |
| "grad_norm": 2.4928314685821533, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.7944, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.012197277709033056, | |
| "grad_norm": 2.78300404548645, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.9688, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.012285663779388369, | |
| "grad_norm": 2.606663227081299, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.8228, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.01237404984974368, | |
| "grad_norm": 2.757668972015381, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.9387, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.012462435920098993, | |
| "grad_norm": 2.9312398433685303, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.8566, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.012550821990454304, | |
| "grad_norm": 2.509982109069824, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.7954, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.012639208060809617, | |
| "grad_norm": 2.791313886642456, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.9832, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.012727594131164928, | |
| "grad_norm": 2.56941556930542, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.9498, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01281598020152024, | |
| "grad_norm": 2.8514652252197266, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.0436, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.012904366271875552, | |
| "grad_norm": 2.640472888946533, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.9536, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.012992752342230865, | |
| "grad_norm": 2.3976221084594727, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.9227, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.013081138412586176, | |
| "grad_norm": 3.0007519721984863, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.2496, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.013169524482941489, | |
| "grad_norm": 3.3027806282043457, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.3187, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0132579105532968, | |
| "grad_norm": 3.5042450428009033, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.374, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0132579105532968, | |
| "eval_loss": 1.7811901569366455, | |
| "eval_runtime": 2026.2724, | |
| "eval_samples_per_second": 9.404, | |
| "eval_steps_per_second": 2.351, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.013346296623652113, | |
| "grad_norm": 2.9841861724853516, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.709, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.013434682694007424, | |
| "grad_norm": 2.9830799102783203, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.7547, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.013523068764362737, | |
| "grad_norm": 3.237985372543335, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.8634, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.013611454834718048, | |
| "grad_norm": 3.117237091064453, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.9148, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.01369984090507336, | |
| "grad_norm": 3.335922956466675, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.9442, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.013788226975428673, | |
| "grad_norm": 2.8202741146087646, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.6752, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.013876613045783984, | |
| "grad_norm": 2.995234727859497, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.6851, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.013964999116139297, | |
| "grad_norm": 3.0984115600585938, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.8314, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.014053385186494608, | |
| "grad_norm": 3.166579246520996, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.6618, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01414177125684992, | |
| "grad_norm": 2.3838489055633545, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.5229, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.014230157327205232, | |
| "grad_norm": 2.6802868843078613, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.5825, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.014318543397560545, | |
| "grad_norm": 2.5786678791046143, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.66, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.014406929467915856, | |
| "grad_norm": 2.610809087753296, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.6396, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.014495315538271169, | |
| "grad_norm": 2.23087739944458, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.5383, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.01458370160862648, | |
| "grad_norm": 2.3542158603668213, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.6023, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.014672087678981793, | |
| "grad_norm": 2.2035248279571533, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.6049, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.014760473749337104, | |
| "grad_norm": 1.968978762626648, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.419, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.014848859819692417, | |
| "grad_norm": 2.527904748916626, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.7361, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.014937245890047728, | |
| "grad_norm": 2.420118808746338, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.4985, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.015025631960403041, | |
| "grad_norm": 2.4440720081329346, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.6054, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.015114018030758352, | |
| "grad_norm": 2.413752555847168, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.7365, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.015202404101113665, | |
| "grad_norm": 2.3553225994110107, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.4568, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.015290790171468976, | |
| "grad_norm": 2.3717799186706543, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.5006, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.015379176241824288, | |
| "grad_norm": 2.442169189453125, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.6709, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0154675623121796, | |
| "grad_norm": 2.752803325653076, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.6393, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.015555948382534912, | |
| "grad_norm": 2.4288506507873535, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.6867, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.015644334452890225, | |
| "grad_norm": 2.4188172817230225, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.6484, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.015732720523245536, | |
| "grad_norm": 2.445518970489502, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.6903, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.015821106593600847, | |
| "grad_norm": 2.8650357723236084, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.7948, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.01590949266395616, | |
| "grad_norm": 2.5260698795318604, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.7497, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.015997878734311473, | |
| "grad_norm": 2.724684000015259, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.7562, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.016086264804666784, | |
| "grad_norm": 2.5071051120758057, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.7037, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.016174650875022095, | |
| "grad_norm": 2.478604793548584, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.6496, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01626303694537741, | |
| "grad_norm": 2.50051212310791, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.7892, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.01635142301573272, | |
| "grad_norm": 2.701234817504883, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.8549, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.016439809086088032, | |
| "grad_norm": 2.644554615020752, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.7135, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.016528195156443343, | |
| "grad_norm": 2.41847562789917, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.7421, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.016616581226798658, | |
| "grad_norm": 2.704580783843994, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.8237, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01670496729715397, | |
| "grad_norm": 2.8257317543029785, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.9417, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01679335336750928, | |
| "grad_norm": 2.8316500186920166, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.8592, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01688173943786459, | |
| "grad_norm": 2.6244895458221436, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.9357, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.016970125508219906, | |
| "grad_norm": 2.379777431488037, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.7802, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.017058511578575217, | |
| "grad_norm": 3.0008177757263184, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.0144, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.01714689764893053, | |
| "grad_norm": 2.5710954666137695, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.9029, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.01723528371928584, | |
| "grad_norm": 2.841198444366455, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.1394, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.01732366978964115, | |
| "grad_norm": 2.8154985904693604, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.853, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.017412055859996466, | |
| "grad_norm": 2.8547253608703613, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.0402, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.017500441930351777, | |
| "grad_norm": 2.837372303009033, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.1644, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.017588828000707088, | |
| "grad_norm": 2.9967141151428223, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.9786, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.0176772140710624, | |
| "grad_norm": 3.9809532165527344, | |
| "learning_rate": 0.0, | |
| "loss": 2.4716, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0176772140710624, | |
| "eval_loss": 1.6819664239883423, | |
| "eval_runtime": 2026.9883, | |
| "eval_samples_per_second": 9.401, | |
| "eval_steps_per_second": 2.35, | |
| "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": 3.358172450390016e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |