| { | |
| "best_metric": 3.0394115447998047, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.333889816360601, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00333889816360601, | |
| "grad_norm": 37.723533630371094, | |
| "learning_rate": 1e-05, | |
| "loss": 8.6363, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00333889816360601, | |
| "eval_loss": 3.710127830505371, | |
| "eval_runtime": 35.9461, | |
| "eval_samples_per_second": 14.049, | |
| "eval_steps_per_second": 3.533, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00667779632721202, | |
| "grad_norm": 44.79377365112305, | |
| "learning_rate": 2e-05, | |
| "loss": 10.4558, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01001669449081803, | |
| "grad_norm": 35.595638275146484, | |
| "learning_rate": 3e-05, | |
| "loss": 10.9784, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01335559265442404, | |
| "grad_norm": 23.04273796081543, | |
| "learning_rate": 4e-05, | |
| "loss": 10.3057, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01669449081803005, | |
| "grad_norm": 19.645402908325195, | |
| "learning_rate": 5e-05, | |
| "loss": 9.6469, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02003338898163606, | |
| "grad_norm": 14.549028396606445, | |
| "learning_rate": 6e-05, | |
| "loss": 9.3464, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02337228714524207, | |
| "grad_norm": 13.930132865905762, | |
| "learning_rate": 7e-05, | |
| "loss": 9.6147, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02671118530884808, | |
| "grad_norm": 13.57673168182373, | |
| "learning_rate": 8e-05, | |
| "loss": 9.0649, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03005008347245409, | |
| "grad_norm": 15.489851951599121, | |
| "learning_rate": 9e-05, | |
| "loss": 8.9347, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0333889816360601, | |
| "grad_norm": 15.896312713623047, | |
| "learning_rate": 0.0001, | |
| "loss": 9.8896, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03672787979966611, | |
| "grad_norm": 15.528593063354492, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 9.1073, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04006677796327212, | |
| "grad_norm": 15.032848358154297, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 9.2407, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04340567612687813, | |
| "grad_norm": 14.364212036132812, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.1868, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04674457429048414, | |
| "grad_norm": 14.188446998596191, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 8.5398, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05008347245409015, | |
| "grad_norm": 13.074345588684082, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 9.3816, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05342237061769616, | |
| "grad_norm": 14.517337799072266, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 9.052, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.05676126878130217, | |
| "grad_norm": 15.009509086608887, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.1783, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06010016694490818, | |
| "grad_norm": 13.750170707702637, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 9.5757, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06343906510851419, | |
| "grad_norm": 14.414755821228027, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 9.667, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0667779632721202, | |
| "grad_norm": 14.613687515258789, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 9.9493, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07011686143572621, | |
| "grad_norm": 13.50129222869873, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 9.3863, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07345575959933222, | |
| "grad_norm": 13.425811767578125, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 10.0341, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07679465776293823, | |
| "grad_norm": 14.33968448638916, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 9.4556, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08013355592654424, | |
| "grad_norm": 14.002581596374512, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.3062, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.08347245409015025, | |
| "grad_norm": 14.369874000549316, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 9.2341, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08681135225375626, | |
| "grad_norm": 12.856378555297852, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.157, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09015025041736227, | |
| "grad_norm": 14.03198528289795, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 9.5912, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.09348914858096828, | |
| "grad_norm": 12.59056568145752, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 9.499, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.09682804674457429, | |
| "grad_norm": 13.37918758392334, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 9.5738, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1001669449081803, | |
| "grad_norm": 15.977365493774414, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.7106, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.10350584307178631, | |
| "grad_norm": 16.689184188842773, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.2359, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.10684474123539232, | |
| "grad_norm": 13.517721176147461, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 9.8871, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11018363939899833, | |
| "grad_norm": 17.465274810791016, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 9.877, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.11352253756260434, | |
| "grad_norm": 13.706908226013184, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 9.8173, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.11686143572621036, | |
| "grad_norm": 14.503616333007812, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.431, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12020033388981637, | |
| "grad_norm": 14.698240280151367, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 9.8721, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.12353923205342238, | |
| "grad_norm": 15.244645118713379, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 9.9001, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.12687813021702837, | |
| "grad_norm": 18.27094841003418, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.8394, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1302170283806344, | |
| "grad_norm": 15.800652503967285, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.0216, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1335559265442404, | |
| "grad_norm": 14.698711395263672, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.146, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13689482470784642, | |
| "grad_norm": 16.719526290893555, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.8504, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14023372287145242, | |
| "grad_norm": 14.47635555267334, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.4568, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.14357262103505844, | |
| "grad_norm": 18.760009765625, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.0805, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.14691151919866444, | |
| "grad_norm": 17.59091567993164, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.6397, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15025041736227046, | |
| "grad_norm": 18.24905014038086, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 10.7017, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.15358931552587646, | |
| "grad_norm": 20.4737491607666, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 9.1001, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.15692821368948248, | |
| "grad_norm": 26.5585880279541, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 10.7494, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.16026711185308848, | |
| "grad_norm": 170.9046630859375, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 8.1164, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1636060100166945, | |
| "grad_norm": 42.19401931762695, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 8.1099, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1669449081803005, | |
| "grad_norm": 59.42169189453125, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.5789, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1669449081803005, | |
| "eval_loss": 3.170802116394043, | |
| "eval_runtime": 36.7398, | |
| "eval_samples_per_second": 13.745, | |
| "eval_steps_per_second": 3.457, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17028380634390652, | |
| "grad_norm": 57.85205841064453, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.0838, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.17362270450751252, | |
| "grad_norm": 24.928632736206055, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.0305, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.17696160267111852, | |
| "grad_norm": 12.831768035888672, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 8.4044, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.18030050083472454, | |
| "grad_norm": 11.783881187438965, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 8.8682, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.18363939899833054, | |
| "grad_norm": 10.361265182495117, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 8.097, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.18697829716193656, | |
| "grad_norm": 9.282995223999023, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 7.9845, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.19031719532554256, | |
| "grad_norm": 9.993803977966309, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 8.0418, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.19365609348914858, | |
| "grad_norm": 9.868743896484375, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 8.4737, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.19699499165275458, | |
| "grad_norm": 9.891820907592773, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 8.5594, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2003338898163606, | |
| "grad_norm": 9.493250846862793, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 8.2122, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2036727879799666, | |
| "grad_norm": 10.277020454406738, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 8.6064, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.20701168614357263, | |
| "grad_norm": 12.072765350341797, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 8.8124, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.21035058430717862, | |
| "grad_norm": 11.113184928894043, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 8.8311, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.21368948247078465, | |
| "grad_norm": 10.37546157836914, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 8.9539, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.21702838063439064, | |
| "grad_norm": 11.003464698791504, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 8.4274, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.22036727879799667, | |
| "grad_norm": 11.640214920043945, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 8.732, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.22370617696160267, | |
| "grad_norm": 10.23813533782959, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 8.4723, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2270450751252087, | |
| "grad_norm": 10.505694389343262, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 8.5907, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2303839732888147, | |
| "grad_norm": 11.966251373291016, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 9.4347, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2337228714524207, | |
| "grad_norm": 11.130023002624512, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 9.242, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2370617696160267, | |
| "grad_norm": 11.233088493347168, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 9.2261, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.24040066777963273, | |
| "grad_norm": 10.693445205688477, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 8.998, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.24373956594323873, | |
| "grad_norm": 10.547587394714355, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 9.3157, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.24707846410684475, | |
| "grad_norm": 10.857179641723633, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 9.4256, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.25041736227045075, | |
| "grad_norm": 11.358246803283691, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 9.5806, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.25375626043405675, | |
| "grad_norm": 11.498934745788574, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 9.7731, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2570951585976628, | |
| "grad_norm": 11.578938484191895, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 9.9566, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2604340567612688, | |
| "grad_norm": 11.857230186462402, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 8.9907, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2637729549248748, | |
| "grad_norm": 11.511113166809082, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 8.8168, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2671118530884808, | |
| "grad_norm": 12.616565704345703, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 9.119, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2704507512520868, | |
| "grad_norm": 11.596989631652832, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 8.8922, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.27378964941569284, | |
| "grad_norm": 12.636907577514648, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 9.1289, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.27712854757929883, | |
| "grad_norm": 12.50519847869873, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 8.7335, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.28046744574290483, | |
| "grad_norm": 12.061328887939453, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 9.1102, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2838063439065108, | |
| "grad_norm": 17.33370018005371, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 9.3422, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2871452420701169, | |
| "grad_norm": 13.716808319091797, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 9.0279, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2904841402337229, | |
| "grad_norm": 15.019532203674316, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 9.8689, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2938230383973289, | |
| "grad_norm": 14.468006134033203, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 10.0441, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.29716193656093487, | |
| "grad_norm": 13.162923812866211, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 9.4476, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3005008347245409, | |
| "grad_norm": 14.644915580749512, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 9.3327, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3038397328881469, | |
| "grad_norm": 13.767836570739746, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 9.4936, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.3071786310517529, | |
| "grad_norm": 14.879925727844238, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 9.5901, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3105175292153589, | |
| "grad_norm": 16.64482307434082, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.1544, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.31385642737896496, | |
| "grad_norm": 15.542655944824219, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 9.7929, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.31719532554257096, | |
| "grad_norm": 17.215084075927734, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 10.8473, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.32053422370617696, | |
| "grad_norm": 16.660884857177734, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 10.3807, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.32387312186978295, | |
| "grad_norm": 20.6027774810791, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 10.0095, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.327212020033389, | |
| "grad_norm": 26.255468368530273, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 9.1675, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.330550918196995, | |
| "grad_norm": 26.772062301635742, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 7.9304, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.333889816360601, | |
| "grad_norm": 45.09166717529297, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 10.911, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.333889816360601, | |
| "eval_loss": 3.0394115447998047, | |
| "eval_runtime": 36.7677, | |
| "eval_samples_per_second": 13.735, | |
| "eval_steps_per_second": 3.454, | |
| "step": 100 | |
| } | |
| ], | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.431616484278272e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |