| { | |
| "best_metric": 1.3342723846435547, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.07501875468867217, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00037509377344336085, | |
| "grad_norm": 4.712777614593506, | |
| "learning_rate": 1e-05, | |
| "loss": 5.4457, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00037509377344336085, | |
| "eval_loss": 1.9310895204544067, | |
| "eval_runtime": 92.1059, | |
| "eval_samples_per_second": 48.748, | |
| "eval_steps_per_second": 12.192, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0007501875468867217, | |
| "grad_norm": 5.964175224304199, | |
| "learning_rate": 2e-05, | |
| "loss": 6.2217, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0011252813203300824, | |
| "grad_norm": 5.687808513641357, | |
| "learning_rate": 3e-05, | |
| "loss": 6.5327, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0015003750937734434, | |
| "grad_norm": 5.759525299072266, | |
| "learning_rate": 4e-05, | |
| "loss": 7.1272, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0018754688672168042, | |
| "grad_norm": 5.854959011077881, | |
| "learning_rate": 5e-05, | |
| "loss": 7.1806, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002250562640660165, | |
| "grad_norm": 4.275196075439453, | |
| "learning_rate": 6e-05, | |
| "loss": 6.5565, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.002625656414103526, | |
| "grad_norm": 3.3260657787323, | |
| "learning_rate": 7e-05, | |
| "loss": 6.5913, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003000750187546887, | |
| "grad_norm": 3.2868363857269287, | |
| "learning_rate": 8e-05, | |
| "loss": 6.8081, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0033758439609902473, | |
| "grad_norm": 4.345282554626465, | |
| "learning_rate": 9e-05, | |
| "loss": 6.6061, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0037509377344336083, | |
| "grad_norm": 4.528377056121826, | |
| "learning_rate": 0.0001, | |
| "loss": 6.61, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004126031507876969, | |
| "grad_norm": 3.3214199542999268, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 6.6105, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00450112528132033, | |
| "grad_norm": 2.410160541534424, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 7.0012, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.004876219054763691, | |
| "grad_norm": 2.01269793510437, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 6.6236, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.005251312828207052, | |
| "grad_norm": 2.203320026397705, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 6.7032, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.005626406601650412, | |
| "grad_norm": 2.148984909057617, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 6.2351, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.006001500375093774, | |
| "grad_norm": 1.9245797395706177, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 6.6563, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.006376594148537134, | |
| "grad_norm": 1.9248274564743042, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 6.3262, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.006751687921980495, | |
| "grad_norm": 1.7623064517974854, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 6.128, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.007126781695423856, | |
| "grad_norm": 1.695133924484253, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 6.3588, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.007501875468867217, | |
| "grad_norm": 1.6393914222717285, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 5.7356, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.007876969242310577, | |
| "grad_norm": 1.8381825685501099, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 6.052, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.008252063015753939, | |
| "grad_norm": 1.9777308702468872, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 6.1458, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0086271567891973, | |
| "grad_norm": 1.7540639638900757, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 5.9839, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.00900225056264066, | |
| "grad_norm": 1.6651537418365479, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 5.6999, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.009377344336084021, | |
| "grad_norm": 1.6417546272277832, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 5.7478, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.009752438109527382, | |
| "grad_norm": 1.770182490348816, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 6.0332, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.010127531882970742, | |
| "grad_norm": 1.7568144798278809, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 5.8598, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.010502625656414103, | |
| "grad_norm": 1.814144492149353, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 5.832, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.010877719429857465, | |
| "grad_norm": 1.8401905298233032, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 5.9567, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.011252813203300824, | |
| "grad_norm": 1.815159797668457, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 5.6437, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.011627906976744186, | |
| "grad_norm": 1.9872208833694458, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 5.7721, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.012003000750187547, | |
| "grad_norm": 1.8732664585113525, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 5.672, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.012378094523630907, | |
| "grad_norm": 1.9048558473587036, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 5.4007, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.012753188297074268, | |
| "grad_norm": 1.9422022104263306, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 5.8894, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01312828207051763, | |
| "grad_norm": 2.1154680252075195, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 5.5981, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01350337584396099, | |
| "grad_norm": 2.073786735534668, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 5.8068, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01387846961740435, | |
| "grad_norm": 1.9941192865371704, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 5.7105, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.014253563390847712, | |
| "grad_norm": 2.099759101867676, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 5.8676, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.014628657164291074, | |
| "grad_norm": 2.225414514541626, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 6.0542, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.015003750937734433, | |
| "grad_norm": 2.408869504928589, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 5.9688, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.015378844711177795, | |
| "grad_norm": 2.297445774078369, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 5.9047, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.015753938484621154, | |
| "grad_norm": 2.5326590538024902, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 5.5829, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.016129032258064516, | |
| "grad_norm": 2.5996813774108887, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 5.5409, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.016504126031507877, | |
| "grad_norm": 2.4606714248657227, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 5.5461, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.01687921980495124, | |
| "grad_norm": 2.56042742729187, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 5.7991, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0172543135783946, | |
| "grad_norm": 2.932943105697632, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 6.0878, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.017629407351837958, | |
| "grad_norm": 2.945050001144409, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 5.5706, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.01800450112528132, | |
| "grad_norm": 2.9749271869659424, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 6.104, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.01837959489872468, | |
| "grad_norm": 3.3921656608581543, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 5.8879, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.018754688672168042, | |
| "grad_norm": 4.113914489746094, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 5.7999, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.018754688672168042, | |
| "eval_loss": 1.614846110343933, | |
| "eval_runtime": 93.1712, | |
| "eval_samples_per_second": 48.191, | |
| "eval_steps_per_second": 12.053, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.019129782445611403, | |
| "grad_norm": 25.759489059448242, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 5.9653, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.019504876219054765, | |
| "grad_norm": 22.479291915893555, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 5.763, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.019879969992498126, | |
| "grad_norm": 13.063644409179688, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 5.9359, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.020255063765941484, | |
| "grad_norm": 5.125077247619629, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 5.5952, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.020630157539384845, | |
| "grad_norm": 3.35514760017395, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 5.4908, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.021005251312828207, | |
| "grad_norm": 4.583368301391602, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 5.906, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02138034508627157, | |
| "grad_norm": 4.549632549285889, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 5.7888, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02175543885971493, | |
| "grad_norm": 2.504077911376953, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 5.9362, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.02213053263315829, | |
| "grad_norm": 2.554222822189331, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 5.617, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02250562640660165, | |
| "grad_norm": 3.2068235874176025, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 5.9524, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02288072018004501, | |
| "grad_norm": 2.7508137226104736, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 6.0756, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.023255813953488372, | |
| "grad_norm": 2.0300989151000977, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 6.1019, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.023630907726931733, | |
| "grad_norm": 2.272578239440918, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 5.7189, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.024006001500375095, | |
| "grad_norm": 2.3081860542297363, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 5.8933, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.024381095273818456, | |
| "grad_norm": 1.9610217809677124, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 5.8692, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.024756189047261814, | |
| "grad_norm": 1.7300370931625366, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 5.7148, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.025131282820705175, | |
| "grad_norm": 1.828964352607727, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 5.7911, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.025506376594148537, | |
| "grad_norm": 2.072789192199707, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 6.2862, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.025881470367591898, | |
| "grad_norm": 1.5874710083007812, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 5.4174, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.02625656414103526, | |
| "grad_norm": 1.8556567430496216, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 5.6568, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02663165791447862, | |
| "grad_norm": 1.784419298171997, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 5.6411, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.02700675168792198, | |
| "grad_norm": 1.863321304321289, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 5.8267, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.02738184546136534, | |
| "grad_norm": 1.756763219833374, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 5.5442, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0277569392348087, | |
| "grad_norm": 1.7847658395767212, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 5.3388, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.028132033008252063, | |
| "grad_norm": 1.8583643436431885, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 5.3687, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.028507126781695424, | |
| "grad_norm": 1.6691844463348389, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 5.2, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.028882220555138786, | |
| "grad_norm": 2.1105973720550537, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 5.6585, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.029257314328582147, | |
| "grad_norm": 1.9019383192062378, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 5.5041, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.029632408102025505, | |
| "grad_norm": 1.8200429677963257, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 5.4595, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.030007501875468866, | |
| "grad_norm": 1.7709747552871704, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 5.5544, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.030382595648912228, | |
| "grad_norm": 1.9791682958602905, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 5.6781, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.03075768942235559, | |
| "grad_norm": 2.0219204425811768, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 5.5342, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03113278319579895, | |
| "grad_norm": 2.041910171508789, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 5.2591, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03150787696924231, | |
| "grad_norm": 1.8891141414642334, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 5.5522, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.03188297074268567, | |
| "grad_norm": 1.939340353012085, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 5.4912, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03225806451612903, | |
| "grad_norm": 1.9492323398590088, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 5.297, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.032633158289572396, | |
| "grad_norm": 1.82587468624115, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 5.1813, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.033008252063015754, | |
| "grad_norm": 2.324018716812134, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 5.8357, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03338334583645911, | |
| "grad_norm": 2.013951539993286, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 5.1113, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.03375843960990248, | |
| "grad_norm": 2.1993765830993652, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 5.6682, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.034133533383345835, | |
| "grad_norm": 2.3582077026367188, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 5.6474, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0345086271567892, | |
| "grad_norm": 2.091481924057007, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 5.1883, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.03488372093023256, | |
| "grad_norm": 2.2112295627593994, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 5.4115, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.035258814703675916, | |
| "grad_norm": 2.2919301986694336, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 5.3728, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.03563390847711928, | |
| "grad_norm": 2.649906635284424, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 5.3693, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03600900225056264, | |
| "grad_norm": 2.68513560295105, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 5.5947, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.036384096024006, | |
| "grad_norm": 2.5721240043640137, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 5.329, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.03675918979744936, | |
| "grad_norm": 2.8568177223205566, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 5.3509, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.037134283570892726, | |
| "grad_norm": 3.0468573570251465, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 5.4196, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.037509377344336084, | |
| "grad_norm": 3.358666181564331, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 5.391, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.037509377344336084, | |
| "eval_loss": 1.4431664943695068, | |
| "eval_runtime": 93.0017, | |
| "eval_samples_per_second": 48.279, | |
| "eval_steps_per_second": 12.075, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03788447111777944, | |
| "grad_norm": 8.383264541625977, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 4.8479, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.03825956489122281, | |
| "grad_norm": 7.975872993469238, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 5.1303, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.038634658664666165, | |
| "grad_norm": 6.235596656799316, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 5.2497, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.03900975243810953, | |
| "grad_norm": 3.17594051361084, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 5.1602, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.03938484621155289, | |
| "grad_norm": 2.3974931240081787, | |
| "learning_rate": 5e-05, | |
| "loss": 5.4011, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.03975993998499625, | |
| "grad_norm": 2.0740904808044434, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 5.3506, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04013503375843961, | |
| "grad_norm": 2.0758721828460693, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 4.9752, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04051012753188297, | |
| "grad_norm": 2.624284267425537, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 5.6791, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04088522130532633, | |
| "grad_norm": 2.619621753692627, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 5.4424, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.04126031507876969, | |
| "grad_norm": 1.9772664308547974, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 5.4186, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.041635408852213056, | |
| "grad_norm": 1.9781255722045898, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 5.5893, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.042010502625656414, | |
| "grad_norm": 1.7533586025238037, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 5.7089, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.04238559639909977, | |
| "grad_norm": 1.676369547843933, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 5.6857, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.04276069017254314, | |
| "grad_norm": 2.161539077758789, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 5.8798, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.043135783945986494, | |
| "grad_norm": 1.9153467416763306, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 5.7885, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04351087771942986, | |
| "grad_norm": 1.8520301580429077, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 5.9956, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.04388597149287322, | |
| "grad_norm": 1.8187308311462402, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 5.7652, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.04426106526631658, | |
| "grad_norm": 1.7822813987731934, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 5.8474, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.04463615903975994, | |
| "grad_norm": 1.916996717453003, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 6.1343, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0450112528132033, | |
| "grad_norm": 1.9927321672439575, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 5.6692, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04538634658664666, | |
| "grad_norm": 1.8190302848815918, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 5.4014, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.04576144036009002, | |
| "grad_norm": 1.671014428138733, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 5.3868, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.046136534133533386, | |
| "grad_norm": 1.9057344198226929, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 5.6802, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.046511627906976744, | |
| "grad_norm": 1.850015640258789, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 5.4339, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.04688672168042011, | |
| "grad_norm": 1.7557549476623535, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 5.415, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.047261815453863466, | |
| "grad_norm": 1.6906707286834717, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 5.1949, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.047636909227306824, | |
| "grad_norm": 1.7078936100006104, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 5.1429, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.04801200300075019, | |
| "grad_norm": 1.7312103509902954, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 5.3206, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.04838709677419355, | |
| "grad_norm": 2.0543112754821777, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 5.598, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.04876219054763691, | |
| "grad_norm": 1.8180476427078247, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 5.4977, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04913728432108027, | |
| "grad_norm": 1.8911992311477661, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 5.4082, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.04951237809452363, | |
| "grad_norm": 1.8683141469955444, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 5.5103, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.04988747186796699, | |
| "grad_norm": 1.8732874393463135, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 5.1073, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.05026256564141035, | |
| "grad_norm": 1.9124647378921509, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 5.4139, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.050637659414853715, | |
| "grad_norm": 1.9962282180786133, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 5.1778, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.05101275318829707, | |
| "grad_norm": 1.8331456184387207, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 5.2381, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.05138784696174044, | |
| "grad_norm": 1.9578392505645752, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 5.2916, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.051762940735183796, | |
| "grad_norm": 1.9451302289962769, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 5.1925, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.052138034508627154, | |
| "grad_norm": 2.0928683280944824, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 5.4758, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.05251312828207052, | |
| "grad_norm": 2.143028974533081, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 5.3859, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05288822205551388, | |
| "grad_norm": 2.1343834400177, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 5.6273, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.05326331582895724, | |
| "grad_norm": 2.1455912590026855, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 5.4586, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.0536384096024006, | |
| "grad_norm": 2.219521999359131, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 5.1488, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.05401350337584396, | |
| "grad_norm": 2.245352268218994, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 5.0639, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.05438859714928732, | |
| "grad_norm": 2.540377140045166, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 5.3002, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.05476369092273068, | |
| "grad_norm": 2.4462618827819824, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 5.1012, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.055138784696174045, | |
| "grad_norm": 2.7136683464050293, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 5.6671, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.0555138784696174, | |
| "grad_norm": 2.9355623722076416, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 5.4329, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.05588897224306077, | |
| "grad_norm": 3.0026261806488037, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 5.463, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.056264066016504126, | |
| "grad_norm": 3.6827046871185303, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 5.4591, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.056264066016504126, | |
| "eval_loss": 1.3617476224899292, | |
| "eval_runtime": 93.1539, | |
| "eval_samples_per_second": 48.2, | |
| "eval_steps_per_second": 12.055, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.056639159789947484, | |
| "grad_norm": 4.144399642944336, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 4.3184, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.05701425356339085, | |
| "grad_norm": 4.340470314025879, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 4.9074, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.05738934733683421, | |
| "grad_norm": 4.358315467834473, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 5.2288, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.05776444111027757, | |
| "grad_norm": 3.5348572731018066, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 4.8488, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.05813953488372093, | |
| "grad_norm": 3.7469818592071533, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 5.348, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.058514628657164294, | |
| "grad_norm": 3.0768086910247803, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 5.4008, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.05888972243060765, | |
| "grad_norm": 2.3540875911712646, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 5.1607, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.05926481620405101, | |
| "grad_norm": 2.4944212436676025, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 5.4342, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.059639909977494375, | |
| "grad_norm": 1.9738702774047852, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 5.4822, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.06001500375093773, | |
| "grad_norm": 1.9951307773590088, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 5.6218, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0603900975243811, | |
| "grad_norm": 1.6997627019882202, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 5.2412, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.060765191297824456, | |
| "grad_norm": 1.8695182800292969, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 5.4832, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.061140285071267814, | |
| "grad_norm": 2.1702582836151123, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 5.6178, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.06151537884471118, | |
| "grad_norm": 2.4172706604003906, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 5.6853, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.061890472618154536, | |
| "grad_norm": 2.01557993888855, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 5.7229, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.0622655663915979, | |
| "grad_norm": 2.1539080142974854, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 5.5319, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.06264066016504126, | |
| "grad_norm": 2.3977644443511963, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 5.656, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.06301575393848462, | |
| "grad_norm": 2.3694355487823486, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 5.7526, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.06339084771192798, | |
| "grad_norm": 2.55816912651062, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 5.5902, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.06376594148537135, | |
| "grad_norm": 1.6467595100402832, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 5.3585, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0641410352588147, | |
| "grad_norm": 2.0331952571868896, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 5.4384, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.06451612903225806, | |
| "grad_norm": 2.1236963272094727, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 5.5012, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.06489122280570142, | |
| "grad_norm": 1.7954583168029785, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 5.3234, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.06526631657914479, | |
| "grad_norm": 1.792651891708374, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 5.3117, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.06564141035258815, | |
| "grad_norm": 1.9872102737426758, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 5.4393, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06601650412603151, | |
| "grad_norm": 1.7396267652511597, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 5.477, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.06639159789947487, | |
| "grad_norm": 1.856001853942871, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 5.3395, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.06676669167291822, | |
| "grad_norm": 1.8975470066070557, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 5.3487, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0671417854463616, | |
| "grad_norm": 1.7905710935592651, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 5.2846, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.06751687921980495, | |
| "grad_norm": 1.8551831245422363, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 5.2001, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06789197299324831, | |
| "grad_norm": 1.9693666696548462, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 5.5733, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.06826706676669167, | |
| "grad_norm": 1.843341588973999, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 5.2628, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.06864216054013503, | |
| "grad_norm": 1.8210816383361816, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 5.2898, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0690172543135784, | |
| "grad_norm": 1.9059969186782837, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 5.207, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.06939234808702176, | |
| "grad_norm": 2.046881675720215, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 5.2952, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06976744186046512, | |
| "grad_norm": 1.8661057949066162, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 5.2521, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.07014253563390847, | |
| "grad_norm": 2.149541139602661, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 5.4584, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.07051762940735183, | |
| "grad_norm": 2.0000863075256348, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 5.0171, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0708927231807952, | |
| "grad_norm": 1.9972003698349, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 5.435, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.07126781695423856, | |
| "grad_norm": 2.0368270874023438, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 5.1753, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07164291072768192, | |
| "grad_norm": 2.2272322177886963, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 5.1746, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.07201800450112528, | |
| "grad_norm": 2.2874584197998047, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 5.0281, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.07239309827456865, | |
| "grad_norm": 2.416949987411499, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 5.5412, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.072768192048012, | |
| "grad_norm": 2.594515800476074, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 5.2231, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.07314328582145536, | |
| "grad_norm": 2.465099573135376, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 5.4567, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.07351837959489872, | |
| "grad_norm": 2.674084424972534, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 5.7588, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.07389347336834208, | |
| "grad_norm": 2.663978099822998, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 5.0821, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.07426856714178545, | |
| "grad_norm": 3.477551221847534, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 5.5366, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.07464366091522881, | |
| "grad_norm": 3.856997013092041, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 5.7195, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.07501875468867217, | |
| "grad_norm": 4.119273662567139, | |
| "learning_rate": 0.0, | |
| "loss": 5.4221, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07501875468867217, | |
| "eval_loss": 1.3342723846435547, | |
| "eval_runtime": 93.1818, | |
| "eval_samples_per_second": 48.185, | |
| "eval_steps_per_second": 12.052, | |
| "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": 5.488516125425664e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |