| { | |
| "best_metric": 0.008051020093262196, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.9002375296912115, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009501187648456057, | |
| "grad_norm": 10.634076118469238, | |
| "learning_rate": 1e-05, | |
| "loss": 1.634, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009501187648456057, | |
| "eval_loss": 2.2654600143432617, | |
| "eval_runtime": 3.1156, | |
| "eval_samples_per_second": 57.132, | |
| "eval_steps_per_second": 14.443, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.019002375296912115, | |
| "grad_norm": 10.325587272644043, | |
| "learning_rate": 2e-05, | |
| "loss": 1.6844, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.028503562945368172, | |
| "grad_norm": 13.5761079788208, | |
| "learning_rate": 3e-05, | |
| "loss": 1.9244, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03800475059382423, | |
| "grad_norm": 11.778326034545898, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2022, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.047505938242280284, | |
| "grad_norm": 7.98240852355957, | |
| "learning_rate": 5e-05, | |
| "loss": 1.3515, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.057007125890736345, | |
| "grad_norm": 9.219783782958984, | |
| "learning_rate": 6e-05, | |
| "loss": 1.0971, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0665083135391924, | |
| "grad_norm": 7.73058557510376, | |
| "learning_rate": 7e-05, | |
| "loss": 0.8157, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07600950118764846, | |
| "grad_norm": 7.745980739593506, | |
| "learning_rate": 8e-05, | |
| "loss": 0.982, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0855106888361045, | |
| "grad_norm": 9.983631134033203, | |
| "learning_rate": 9e-05, | |
| "loss": 0.7368, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09501187648456057, | |
| "grad_norm": 6.394253730773926, | |
| "learning_rate": 0.0001, | |
| "loss": 0.3611, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10451306413301663, | |
| "grad_norm": 4.376123905181885, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.2768, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11401425178147269, | |
| "grad_norm": 4.922821521759033, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.3543, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12351543942992874, | |
| "grad_norm": 4.584092617034912, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.634, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1330166270783848, | |
| "grad_norm": 3.142916202545166, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.2842, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14251781472684086, | |
| "grad_norm": 2.926008939743042, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.1415, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15201900237529692, | |
| "grad_norm": 3.285600185394287, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.26, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.16152019002375298, | |
| "grad_norm": 3.1018033027648926, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.3467, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.171021377672209, | |
| "grad_norm": 1.7889859676361084, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.1505, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18052256532066507, | |
| "grad_norm": 2.104616641998291, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.2665, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19002375296912113, | |
| "grad_norm": 1.4343472719192505, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.1549, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1995249406175772, | |
| "grad_norm": 1.4845852851867676, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.2139, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.20902612826603326, | |
| "grad_norm": 1.5346105098724365, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.1356, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.21852731591448932, | |
| "grad_norm": 2.5972952842712402, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.2574, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.22802850356294538, | |
| "grad_norm": 3.826779365539551, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.2551, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.2375296912114014, | |
| "grad_norm": 5.629703521728516, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.1369, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24703087885985747, | |
| "grad_norm": 5.624516487121582, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3514, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.25653206650831356, | |
| "grad_norm": 2.812879800796509, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0959, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.2660332541567696, | |
| "grad_norm": 2.1208438873291016, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.2576, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.2755344418052256, | |
| "grad_norm": 1.8859907388687134, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.1634, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.2850356294536817, | |
| "grad_norm": 1.177573561668396, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0828, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.29453681710213775, | |
| "grad_norm": 1.028439998626709, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0252, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.30403800475059384, | |
| "grad_norm": 1.0798872709274292, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.0812, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.31353919239904987, | |
| "grad_norm": 1.484721064567566, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.1003, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.32304038004750596, | |
| "grad_norm": 0.5597230792045593, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0156, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.332541567695962, | |
| "grad_norm": 1.41647469997406, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0863, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.342042755344418, | |
| "grad_norm": 0.5221539735794067, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0242, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.3515439429928741, | |
| "grad_norm": 0.9052144289016724, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0488, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.36104513064133015, | |
| "grad_norm": 1.344610571861267, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.095, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.37054631828978624, | |
| "grad_norm": 1.3425592184066772, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0503, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.38004750593824227, | |
| "grad_norm": 1.131867527961731, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0687, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.38954869358669836, | |
| "grad_norm": 0.635826826095581, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.0196, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.3990498812351544, | |
| "grad_norm": 2.282115936279297, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.1152, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.4085510688836104, | |
| "grad_norm": 0.7146430611610413, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.0091, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.4180522565320665, | |
| "grad_norm": 0.8013799786567688, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.0185, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.42755344418052255, | |
| "grad_norm": 2.0490562915802, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.2301, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.43705463182897863, | |
| "grad_norm": 1.2392498254776, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.0626, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.44655581947743467, | |
| "grad_norm": 1.7980742454528809, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.1476, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.45605700712589076, | |
| "grad_norm": 2.345305919647217, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.1909, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.4655581947743468, | |
| "grad_norm": 0.6705112457275391, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0184, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.4750593824228028, | |
| "grad_norm": 0.32240453362464905, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0072, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4750593824228028, | |
| "eval_loss": 0.0333050861954689, | |
| "eval_runtime": 3.1094, | |
| "eval_samples_per_second": 57.245, | |
| "eval_steps_per_second": 14.472, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4845605700712589, | |
| "grad_norm": 0.4481276869773865, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0094, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.49406175771971494, | |
| "grad_norm": 0.5273180603981018, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0134, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.503562945368171, | |
| "grad_norm": 8.373512268066406, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0505, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5130641330166271, | |
| "grad_norm": 0.4383963644504547, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.0095, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5225653206650831, | |
| "grad_norm": 0.6486546993255615, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0127, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5320665083135392, | |
| "grad_norm": 0.6159929037094116, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0092, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5415676959619953, | |
| "grad_norm": 0.8543300032615662, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0464, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5510688836104513, | |
| "grad_norm": 0.15231961011886597, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0026, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5605700712589073, | |
| "grad_norm": 0.7067123651504517, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.0339, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5700712589073634, | |
| "grad_norm": 0.7852799892425537, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.0336, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5795724465558195, | |
| "grad_norm": 0.11485043168067932, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0018, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5890736342042755, | |
| "grad_norm": 0.25540459156036377, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0073, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.5985748218527316, | |
| "grad_norm": 0.6394416093826294, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0085, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6080760095011877, | |
| "grad_norm": 0.8363385796546936, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0277, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6175771971496437, | |
| "grad_norm": 0.532666027545929, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0085, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6270783847980997, | |
| "grad_norm": 0.5964115262031555, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0309, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6365795724465558, | |
| "grad_norm": 0.7682636380195618, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0082, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6460807600950119, | |
| "grad_norm": 1.2844709157943726, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0432, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6555819477434679, | |
| "grad_norm": 0.40052053332328796, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0088, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.665083135391924, | |
| "grad_norm": 1.1109915971755981, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0781, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.6745843230403801, | |
| "grad_norm": 0.3353760540485382, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.0065, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.684085510688836, | |
| "grad_norm": 2.3546252250671387, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0632, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.6935866983372921, | |
| "grad_norm": 0.8902141451835632, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0154, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.7030878859857482, | |
| "grad_norm": 1.0232930183410645, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.058, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.7125890736342043, | |
| "grad_norm": 2.897697687149048, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0739, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7220902612826603, | |
| "grad_norm": 0.5444605350494385, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0126, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7315914489311164, | |
| "grad_norm": 0.8834287524223328, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0278, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7410926365795725, | |
| "grad_norm": 2.945617198944092, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0763, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7505938242280285, | |
| "grad_norm": 0.29715582728385925, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0067, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7600950118764845, | |
| "grad_norm": 0.44521045684814453, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.036, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7695961995249406, | |
| "grad_norm": 0.15839532017707825, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.003, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.7790973871733967, | |
| "grad_norm": 1.0403664112091064, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0218, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7885985748218527, | |
| "grad_norm": 0.8506789207458496, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0363, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.7980997624703088, | |
| "grad_norm": 1.2762515544891357, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0492, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.8076009501187649, | |
| "grad_norm": 0.3407810926437378, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0142, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8171021377672208, | |
| "grad_norm": 0.1169484555721283, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.002, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8266033254156769, | |
| "grad_norm": 0.08654441684484482, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0021, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.836104513064133, | |
| "grad_norm": 0.6437569260597229, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0138, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8456057007125891, | |
| "grad_norm": 1.3247535228729248, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0196, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8551068883610451, | |
| "grad_norm": 0.5922229290008545, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.015, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8646080760095012, | |
| "grad_norm": 0.1911570280790329, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0084, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.8741092636579573, | |
| "grad_norm": 0.16789893805980682, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0033, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.8836104513064132, | |
| "grad_norm": 0.08045918494462967, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0024, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.8931116389548693, | |
| "grad_norm": 0.05782223492860794, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0011, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.9026128266033254, | |
| "grad_norm": 0.44856613874435425, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0069, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9121140142517815, | |
| "grad_norm": 0.08039326220750809, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0015, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9216152019002375, | |
| "grad_norm": 0.222639799118042, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0029, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.9311163895486936, | |
| "grad_norm": 0.3431375324726105, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0132, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9406175771971497, | |
| "grad_norm": 0.6533656716346741, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0093, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9501187648456056, | |
| "grad_norm": 0.5601425170898438, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0209, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9501187648456056, | |
| "eval_loss": 0.013422280550003052, | |
| "eval_runtime": 3.1093, | |
| "eval_samples_per_second": 57.248, | |
| "eval_steps_per_second": 14.473, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9596199524940617, | |
| "grad_norm": 0.5394858717918396, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.018, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.9691211401425178, | |
| "grad_norm": 0.19572888314723969, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0024, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.9786223277909739, | |
| "grad_norm": 0.03570723533630371, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0007, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9881235154394299, | |
| "grad_norm": 1.0345633029937744, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.026, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.997624703087886, | |
| "grad_norm": 0.20308668911457062, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0056, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.007125890736342, | |
| "grad_norm": 0.07612226903438568, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.001, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.0166270783847982, | |
| "grad_norm": 0.8962851166725159, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0155, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.0261282660332542, | |
| "grad_norm": 0.02778765559196472, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0006, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.03562945368171, | |
| "grad_norm": 0.21089020371437073, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.006, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.0451306413301662, | |
| "grad_norm": 0.011732185259461403, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0004, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0546318289786223, | |
| "grad_norm": 0.09432829171419144, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0043, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.0641330166270784, | |
| "grad_norm": 0.6465476155281067, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0192, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.0736342042755345, | |
| "grad_norm": 0.28978919982910156, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0099, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.0831353919239906, | |
| "grad_norm": 0.087046317756176, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0014, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.0926365795724466, | |
| "grad_norm": 0.2096734642982483, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.006, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1021377672209025, | |
| "grad_norm": 0.07756292819976807, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.001, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.1116389548693586, | |
| "grad_norm": 0.5461010336875916, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.014, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.1211401425178147, | |
| "grad_norm": 0.06456967443227768, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0013, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.1306413301662708, | |
| "grad_norm": 0.3732730448246002, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0111, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.1401425178147269, | |
| "grad_norm": 0.13659219443798065, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0016, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.149643705463183, | |
| "grad_norm": 0.4977100193500519, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0078, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.159144893111639, | |
| "grad_norm": 0.045318689197301865, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0014, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.168646080760095, | |
| "grad_norm": 0.2404695749282837, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0061, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.178147268408551, | |
| "grad_norm": 0.5813296437263489, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0236, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.187648456057007, | |
| "grad_norm": 0.6855394244194031, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0263, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.1971496437054632, | |
| "grad_norm": 0.4205869138240814, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0186, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.2066508313539193, | |
| "grad_norm": 0.028451422229409218, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0009, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.2161520190023754, | |
| "grad_norm": 0.31507155299186707, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0039, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.2256532066508314, | |
| "grad_norm": 0.011005699634552002, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0004, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.2351543942992875, | |
| "grad_norm": 0.010146229527890682, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0003, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.2446555819477434, | |
| "grad_norm": 0.9834302663803101, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0066, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.2541567695961995, | |
| "grad_norm": 0.12252087891101837, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0019, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.2636579572446556, | |
| "grad_norm": 0.17279238998889923, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0101, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.2731591448931117, | |
| "grad_norm": 0.15653680264949799, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0037, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.2826603325415677, | |
| "grad_norm": 0.3399457633495331, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0109, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.2921615201900236, | |
| "grad_norm": 0.015262948349118233, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0004, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.3016627078384797, | |
| "grad_norm": 0.3130092918872833, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0085, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.3111638954869358, | |
| "grad_norm": 0.12765063345432281, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.004, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.3206650831353919, | |
| "grad_norm": 0.08206624537706375, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0051, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.330166270783848, | |
| "grad_norm": 0.03269046172499657, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0006, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.339667458432304, | |
| "grad_norm": 0.17274565994739532, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0037, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.3491686460807601, | |
| "grad_norm": 0.0898587703704834, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0019, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.3586698337292162, | |
| "grad_norm": 0.16785341501235962, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0063, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.3681710213776723, | |
| "grad_norm": 0.18016614019870758, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0025, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.3776722090261282, | |
| "grad_norm": 0.331484854221344, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.005, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.3871733966745843, | |
| "grad_norm": 0.05079348012804985, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.001, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.3966745843230404, | |
| "grad_norm": 0.34630659222602844, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0039, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.4061757719714965, | |
| "grad_norm": 0.03176404908299446, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0009, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.4156769596199525, | |
| "grad_norm": 0.3133653998374939, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0111, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.4251781472684084, | |
| "grad_norm": 0.04033322259783745, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.001, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4251781472684084, | |
| "eval_loss": 0.008251837454736233, | |
| "eval_runtime": 3.1092, | |
| "eval_samples_per_second": 57.249, | |
| "eval_steps_per_second": 14.473, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4346793349168645, | |
| "grad_norm": 0.25339964032173157, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0036, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.4441805225653206, | |
| "grad_norm": 0.3924987018108368, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0104, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.4536817102137767, | |
| "grad_norm": 0.04515230283141136, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0011, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.4631828978622328, | |
| "grad_norm": 0.014813421294093132, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0004, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.4726840855106889, | |
| "grad_norm": 0.027775825932621956, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0005, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.482185273159145, | |
| "grad_norm": 0.029516329988837242, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0005, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.491686460807601, | |
| "grad_norm": 0.023214198648929596, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0005, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.5011876484560571, | |
| "grad_norm": 0.04656863212585449, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0006, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.5106888361045132, | |
| "grad_norm": 0.19086000323295593, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0047, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.520190023752969, | |
| "grad_norm": 0.21550394594669342, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0109, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5296912114014252, | |
| "grad_norm": 0.17475444078445435, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0052, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.5391923990498813, | |
| "grad_norm": 0.015265436843037605, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0004, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.5486935866983373, | |
| "grad_norm": 0.15023384988307953, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0038, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.5581947743467932, | |
| "grad_norm": 0.059008125215768814, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0012, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.5676959619952493, | |
| "grad_norm": 0.07769186049699783, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0072, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.5771971496437054, | |
| "grad_norm": 0.10617467761039734, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0023, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.5866983372921615, | |
| "grad_norm": 0.010569179430603981, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0003, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.5961995249406176, | |
| "grad_norm": 0.0083869444206357, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0003, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.6057007125890737, | |
| "grad_norm": 0.08926163613796234, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0023, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.6152019002375297, | |
| "grad_norm": 0.04555676132440567, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0009, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6247030878859858, | |
| "grad_norm": 0.016096578910946846, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0005, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.634204275534442, | |
| "grad_norm": 0.010908099822700024, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0004, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.643705463182898, | |
| "grad_norm": 0.03802594169974327, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0008, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.6532066508313539, | |
| "grad_norm": 0.013735340908169746, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0005, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.66270783847981, | |
| "grad_norm": 0.5358371734619141, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0063, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.672209026128266, | |
| "grad_norm": 0.1102970689535141, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.003, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.6817102137767221, | |
| "grad_norm": 0.023134298622608185, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0008, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.691211401425178, | |
| "grad_norm": 0.03622838109731674, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0013, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.700712589073634, | |
| "grad_norm": 0.23029854893684387, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0073, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.7102137767220902, | |
| "grad_norm": 0.5697638392448425, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0092, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.7197149643705463, | |
| "grad_norm": 0.010024786926805973, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0003, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.7292161520190024, | |
| "grad_norm": 0.03487780690193176, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0006, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.7387173396674585, | |
| "grad_norm": 0.012627474963665009, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0004, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.7482185273159145, | |
| "grad_norm": 0.6416562795639038, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0052, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.7577197149643706, | |
| "grad_norm": 0.08133221417665482, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0019, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7672209026128267, | |
| "grad_norm": 0.0081032058224082, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0003, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.7767220902612828, | |
| "grad_norm": 0.14601969718933105, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0036, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.7862232779097387, | |
| "grad_norm": 0.06469998508691788, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.001, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.7957244655581948, | |
| "grad_norm": 0.016130423173308372, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0005, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.8052256532066508, | |
| "grad_norm": 0.2341870367527008, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.016, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.814726840855107, | |
| "grad_norm": 0.08566369861364365, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0041, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.8242280285035628, | |
| "grad_norm": 0.015213421545922756, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0004, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.833729216152019, | |
| "grad_norm": 0.049922917038202286, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0007, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.843230403800475, | |
| "grad_norm": 0.0111550223082304, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0004, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.852731591448931, | |
| "grad_norm": 0.04146502539515495, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0049, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8622327790973872, | |
| "grad_norm": 0.1204209253191948, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.003, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.8717339667458432, | |
| "grad_norm": 0.01997855119407177, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0005, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.8812351543942993, | |
| "grad_norm": 0.12081720679998398, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0024, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.8907363420427554, | |
| "grad_norm": 0.021318506449460983, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0006, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.9002375296912115, | |
| "grad_norm": 0.015093390829861164, | |
| "learning_rate": 0.0, | |
| "loss": 0.0005, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9002375296912115, | |
| "eval_loss": 0.008051020093262196, | |
| "eval_runtime": 3.1089, | |
| "eval_samples_per_second": 57.255, | |
| "eval_steps_per_second": 14.475, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 1.5447304260550656e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |