| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.01588057805304113, | |
| "eval_steps": 50, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00010587052035360753, | |
| "grad_norm": 4.395105497678742e-05, | |
| "learning_rate": 1e-05, | |
| "loss": 46.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00010587052035360753, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 71.5228, | |
| "eval_samples_per_second": 222.419, | |
| "eval_steps_per_second": 55.605, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00021174104070721507, | |
| "grad_norm": 4.58413123851642e-05, | |
| "learning_rate": 2e-05, | |
| "loss": 46.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0003176115610608226, | |
| "grad_norm": 4.140105738770217e-05, | |
| "learning_rate": 3e-05, | |
| "loss": 46.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00042348208141443014, | |
| "grad_norm": 4.382082261145115e-05, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005293526017680377, | |
| "grad_norm": 4.889183765044436e-05, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006352231221216452, | |
| "grad_norm": 4.410705514601432e-05, | |
| "learning_rate": 6e-05, | |
| "loss": 46.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0007410936424752528, | |
| "grad_norm": 4.27704508183524e-05, | |
| "learning_rate": 7e-05, | |
| "loss": 46.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0008469641628288603, | |
| "grad_norm": 5.128671182319522e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 46.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0009528346831824678, | |
| "grad_norm": 4.963524406775832e-05, | |
| "learning_rate": 9e-05, | |
| "loss": 46.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0010587052035360754, | |
| "grad_norm": 5.1579288992797956e-05, | |
| "learning_rate": 0.0001, | |
| "loss": 46.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001164575723889683, | |
| "grad_norm": 5.476976730278693e-05, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 46.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0012704462442432905, | |
| "grad_norm": 6.490119267255068e-05, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 46.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.001376316764596898, | |
| "grad_norm": 6.41516744508408e-05, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 46.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0014821872849505055, | |
| "grad_norm": 6.963236228330061e-05, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 46.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.001588057805304113, | |
| "grad_norm": 7.260955317178741e-05, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 46.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0016939283256577206, | |
| "grad_norm": 8.003199764061719e-05, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 46.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.001799798846011328, | |
| "grad_norm": 6.157662573968992e-05, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 46.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0019056693663649356, | |
| "grad_norm": 9.91981869447045e-05, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 46.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0020115398867185433, | |
| "grad_norm": 8.243980119004846e-05, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 46.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.002117410407072151, | |
| "grad_norm": 9.023436723509803e-05, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 46.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0022232809274257584, | |
| "grad_norm": 0.000126901242765598, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 46.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002329151447779366, | |
| "grad_norm": 8.287958917208016e-05, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 46.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0024350219681329734, | |
| "grad_norm": 7.875176379457116e-05, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 46.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002540892488486581, | |
| "grad_norm": 8.90982846613042e-05, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 46.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0026467630088401885, | |
| "grad_norm": 9.042149758897722e-05, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 46.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002752633529193796, | |
| "grad_norm": 0.00010223303979728371, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 46.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0028585040495474035, | |
| "grad_norm": 0.0001294900430366397, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 46.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.002964374569901011, | |
| "grad_norm": 9.50166504480876e-05, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 46.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0030702450902546185, | |
| "grad_norm": 0.00010819704766618088, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 46.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.003176115610608226, | |
| "grad_norm": 7.831957918824628e-05, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 46.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0032819861309618336, | |
| "grad_norm": 0.0001063505478668958, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 46.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.003387856651315441, | |
| "grad_norm": 0.00011041116522392258, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 46.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0034937271716690486, | |
| "grad_norm": 0.0001492316514486447, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 46.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.003599597692022656, | |
| "grad_norm": 0.00012391034397296607, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 46.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0037054682123762637, | |
| "grad_norm": 0.00012100706953788176, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 46.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.003811338732729871, | |
| "grad_norm": 0.00023128939210437238, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 46.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.003917209253083479, | |
| "grad_norm": 0.00038155875517986715, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 46.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.004023079773437087, | |
| "grad_norm": 0.00035766095970757306, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 46.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.004128950293790694, | |
| "grad_norm": 0.0011300703044980764, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 46.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.004234820814144302, | |
| "grad_norm": 0.0015645480016246438, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 46.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004340691334497909, | |
| "grad_norm": 0.001370024518109858, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 46.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004446561854851517, | |
| "grad_norm": 0.0015453952364623547, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 46.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.004552432375205124, | |
| "grad_norm": 0.0013086320832371712, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 46.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004658302895558732, | |
| "grad_norm": 0.001663604867644608, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 46.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004764173415912339, | |
| "grad_norm": 0.0016766958869993687, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 46.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.004870043936265947, | |
| "grad_norm": 0.0018082212191075087, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 46.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004975914456619554, | |
| "grad_norm": 0.0017413385212421417, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 46.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.005081784976973162, | |
| "grad_norm": 0.0017923465929925442, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 46.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.005187655497326769, | |
| "grad_norm": 0.0023123230785131454, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 46.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005293526017680377, | |
| "grad_norm": 0.0017731377156451344, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 46.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005293526017680377, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 71.3773, | |
| "eval_samples_per_second": 222.872, | |
| "eval_steps_per_second": 55.718, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0053993965380339844, | |
| "grad_norm": 0.00021086225751787424, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 46.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.005505267058387592, | |
| "grad_norm": 0.0002711622219067067, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 46.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0056111375787411995, | |
| "grad_norm": 0.0002894172794185579, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 46.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005717008099094807, | |
| "grad_norm": 0.0002906731388065964, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 46.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0058228786194484145, | |
| "grad_norm": 0.00022666183940600604, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 46.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.005928749139802022, | |
| "grad_norm": 0.0002809033030644059, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 46.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0060346196601556296, | |
| "grad_norm": 0.0003183241351507604, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 46.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.006140490180509237, | |
| "grad_norm": 0.00024040187417995185, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 46.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.006246360700862845, | |
| "grad_norm": 0.0002741496136877686, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 46.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006352231221216452, | |
| "grad_norm": 0.0002534056839067489, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 46.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.00645810174157006, | |
| "grad_norm": 0.0002639411250129342, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 46.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.006563972261923667, | |
| "grad_norm": 0.0002514915249776095, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 46.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006669842782277275, | |
| "grad_norm": 0.00023417950433213264, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 46.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006775713302630882, | |
| "grad_norm": 0.0002676513104233891, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 46.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.00688158382298449, | |
| "grad_norm": 0.00027644241345115006, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 46.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006987454343338097, | |
| "grad_norm": 0.0003549778484739363, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 46.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.007093324863691705, | |
| "grad_norm": 0.0001558908261358738, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 46.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.007199195384045312, | |
| "grad_norm": 0.0002607716014608741, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 46.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.00730506590439892, | |
| "grad_norm": 0.0003103368217125535, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 46.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.007410936424752527, | |
| "grad_norm": 0.0002718765754252672, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 46.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.007516806945106135, | |
| "grad_norm": 0.00035626141470856965, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 46.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007622677465459742, | |
| "grad_norm": 0.0003415030369069427, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 46.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.00772854798581335, | |
| "grad_norm": 0.0002935390220955014, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 46.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.007834418506166958, | |
| "grad_norm": 0.0002605564659461379, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 46.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.007940289026520565, | |
| "grad_norm": 0.0002996485563926399, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 46.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.008046159546874173, | |
| "grad_norm": 0.00019948974659200758, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 46.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.00815203006722778, | |
| "grad_norm": 0.00022629639715887606, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 46.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.008257900587581388, | |
| "grad_norm": 0.0002324309607502073, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 46.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.008363771107934995, | |
| "grad_norm": 0.00020576587121468037, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 46.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.008469641628288603, | |
| "grad_norm": 0.00023339713516179472, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 46.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00857551214864221, | |
| "grad_norm": 0.00019102358783129603, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 46.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008681382668995818, | |
| "grad_norm": 0.00019402090401854366, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 46.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.008787253189349425, | |
| "grad_norm": 0.000268847361439839, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 46.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008893123709703033, | |
| "grad_norm": 0.0002833133621606976, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 46.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.00899899423005664, | |
| "grad_norm": 0.0004431276465766132, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 46.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.009104864750410249, | |
| "grad_norm": 0.0005799728678539395, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 46.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.009210735270763855, | |
| "grad_norm": 0.0009890341898426414, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 46.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.009316605791117464, | |
| "grad_norm": 0.0031588415149599314, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 46.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.00942247631147107, | |
| "grad_norm": 0.0032078688964247704, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 46.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.009528346831824679, | |
| "grad_norm": 0.003129588672891259, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 46.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.009634217352178285, | |
| "grad_norm": 0.0027315125335007906, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 46.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.009740087872531894, | |
| "grad_norm": 0.0024609374813735485, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 46.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0098459583928855, | |
| "grad_norm": 0.002924118423834443, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 46.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.009951828913239109, | |
| "grad_norm": 0.003406592644751072, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 46.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.010057699433592715, | |
| "grad_norm": 0.003292385721579194, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 46.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.010163569953946324, | |
| "grad_norm": 0.003402570728212595, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 46.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.01026944047429993, | |
| "grad_norm": 0.0033618933521211147, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 46.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.010375310994653539, | |
| "grad_norm": 0.003268142230808735, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 46.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.010481181515007145, | |
| "grad_norm": 0.003985446412116289, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 46.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.010587052035360754, | |
| "grad_norm": 0.003124967450276017, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 46.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.010587052035360754, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 71.4188, | |
| "eval_samples_per_second": 222.742, | |
| "eval_steps_per_second": 55.686, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01069292255571436, | |
| "grad_norm": 0.0004976472700946033, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 46.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010798793076067969, | |
| "grad_norm": 0.00048162229359149933, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 46.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.010904663596421577, | |
| "grad_norm": 0.00040496836300008, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 46.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.011010534116775184, | |
| "grad_norm": 0.000533498649019748, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 46.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.011116404637128792, | |
| "grad_norm": 0.0004485460522118956, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.011222275157482399, | |
| "grad_norm": 0.00041863738442771137, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 46.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.011328145677836007, | |
| "grad_norm": 0.0003908349317498505, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 46.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.011434016198189614, | |
| "grad_norm": 0.0004363873740658164, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 46.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.011539886718543222, | |
| "grad_norm": 0.0004144172416999936, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 46.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.011645757238896829, | |
| "grad_norm": 0.0004587766306940466, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 46.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.011751627759250437, | |
| "grad_norm": 0.0005415769992396235, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 46.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.011857498279604044, | |
| "grad_norm": 0.0005461342516355217, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 46.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.011963368799957652, | |
| "grad_norm": 0.0007212693453766406, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 46.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.012069239320311259, | |
| "grad_norm": 0.0005091916536912322, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 46.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.012175109840664868, | |
| "grad_norm": 0.0005328023689799011, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 46.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.012280980361018474, | |
| "grad_norm": 0.00036970811197534204, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 46.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.012386850881372083, | |
| "grad_norm": 0.00046359014231711626, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 46.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.01249272140172569, | |
| "grad_norm": 0.0004078349156770855, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 46.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.012598591922079298, | |
| "grad_norm": 0.0004568718431983143, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 46.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.012704462442432904, | |
| "grad_norm": 0.0004690240602940321, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 46.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.012810332962786513, | |
| "grad_norm": 0.0003936483699362725, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 46.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.01291620348314012, | |
| "grad_norm": 0.0004120618395972997, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 46.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.013022074003493728, | |
| "grad_norm": 0.00038696115370839834, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 46.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.013127944523847334, | |
| "grad_norm": 0.00036272944998927414, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 46.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.013233815044200943, | |
| "grad_norm": 0.0005612139357253909, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 46.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01333968556455455, | |
| "grad_norm": 0.0004239322151988745, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 46.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.013445556084908158, | |
| "grad_norm": 0.0002516970853321254, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 46.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.013551426605261764, | |
| "grad_norm": 0.0004707489279098809, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 46.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.013657297125615373, | |
| "grad_norm": 0.00036602691398002207, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 46.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.01376316764596898, | |
| "grad_norm": 0.00039477430982515216, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 46.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.013869038166322588, | |
| "grad_norm": 0.0003384229203220457, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 46.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.013974908686676194, | |
| "grad_norm": 0.00037501484621316195, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 46.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.014080779207029803, | |
| "grad_norm": 0.00032246869523078203, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 46.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.01418664972738341, | |
| "grad_norm": 0.00045177320134826005, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 46.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.014292520247737018, | |
| "grad_norm": 0.00041344534838572145, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 46.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.014398390768090625, | |
| "grad_norm": 0.0006918155122548342, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 46.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.014504261288444233, | |
| "grad_norm": 0.0009688546415418386, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 46.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.01461013180879784, | |
| "grad_norm": 0.0015239521162584424, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 46.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.014716002329151448, | |
| "grad_norm": 0.003637400222942233, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 46.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.014821872849505055, | |
| "grad_norm": 0.004702338948845863, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 46.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.014927743369858663, | |
| "grad_norm": 0.0037672354374080896, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 46.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01503361389021227, | |
| "grad_norm": 0.00495262211188674, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 46.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.015139484410565878, | |
| "grad_norm": 0.004686033818870783, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 46.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.015245354930919485, | |
| "grad_norm": 0.0046455031260848045, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 46.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.015351225451273093, | |
| "grad_norm": 0.004600158892571926, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 46.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.0154570959716267, | |
| "grad_norm": 0.004478240851312876, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 46.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.015562966491980308, | |
| "grad_norm": 0.00454744091257453, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 46.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.015668837012333917, | |
| "grad_norm": 0.004399455618113279, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 46.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01577470753268752, | |
| "grad_norm": 0.004890009760856628, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 46.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01588057805304113, | |
| "grad_norm": 0.004117410164326429, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 46.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01588057805304113, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 71.4043, | |
| "eval_samples_per_second": 222.788, | |
| "eval_steps_per_second": 55.697, | |
| "step": 150 | |
| } | |
| ], | |
| "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": 2 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 24239441313792.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |