| { | |
| "best_metric": 0.7140485644340515, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.019689399719426053, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 9.844699859713026e-05, | |
| "grad_norm": 0.21939219534397125, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7888, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 9.844699859713026e-05, | |
| "eval_loss": 1.0887796878814697, | |
| "eval_runtime": 1231.2887, | |
| "eval_samples_per_second": 13.894, | |
| "eval_steps_per_second": 3.474, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00019689399719426053, | |
| "grad_norm": 0.2647113502025604, | |
| "learning_rate": 2e-05, | |
| "loss": 0.9205, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0002953409957913908, | |
| "grad_norm": 0.26799139380455017, | |
| "learning_rate": 3e-05, | |
| "loss": 0.9965, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00039378799438852106, | |
| "grad_norm": 0.24789780378341675, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9025, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0004922349929856513, | |
| "grad_norm": 0.2603679895401001, | |
| "learning_rate": 5e-05, | |
| "loss": 0.9232, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005906819915827816, | |
| "grad_norm": 0.2867107093334198, | |
| "learning_rate": 6e-05, | |
| "loss": 1.0167, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0006891289901799119, | |
| "grad_norm": 0.2654423713684082, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0207, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0007875759887770421, | |
| "grad_norm": 0.24241261184215546, | |
| "learning_rate": 8e-05, | |
| "loss": 0.8621, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0008860229873741724, | |
| "grad_norm": 0.2112324982881546, | |
| "learning_rate": 9e-05, | |
| "loss": 0.932, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0009844699859713026, | |
| "grad_norm": 0.22786235809326172, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8677, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001082916984568433, | |
| "grad_norm": 0.2848811447620392, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.9823, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0011813639831655632, | |
| "grad_norm": 0.304223895072937, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.9843, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0012798109817626934, | |
| "grad_norm": 0.26909175515174866, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.9055, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0013782579803598238, | |
| "grad_norm": 0.218538299202919, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.7707, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.001476704978956954, | |
| "grad_norm": 0.21921083331108093, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.8375, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0015751519775540842, | |
| "grad_norm": 0.2284502238035202, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.8785, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0016735989761512146, | |
| "grad_norm": 0.2384493201971054, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9195, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0017720459747483448, | |
| "grad_norm": 0.5345220565795898, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9418, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001870492973345475, | |
| "grad_norm": 0.22504675388336182, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.8606, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0019689399719426052, | |
| "grad_norm": 0.2344212681055069, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.8136, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002067386970539736, | |
| "grad_norm": 0.22572244703769684, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.8713, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002165833969136866, | |
| "grad_norm": 0.3545560836791992, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.8386, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0022642809677339963, | |
| "grad_norm": 0.24938194453716278, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.8536, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0023627279663311265, | |
| "grad_norm": 0.2670334577560425, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.8942, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0024611749649282567, | |
| "grad_norm": 0.24915124475955963, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.894, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002559621963525387, | |
| "grad_norm": 0.2405252903699875, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.8241, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0026580689621225175, | |
| "grad_norm": 0.27342650294303894, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7975, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0027565159607196477, | |
| "grad_norm": 0.27550604939460754, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.9299, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.002854962959316778, | |
| "grad_norm": 0.27711233496665955, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.8643, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.002953409957913908, | |
| "grad_norm": 0.29124537110328674, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.8075, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0030518569565110383, | |
| "grad_norm": 0.3193858563899994, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.9129, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0031503039551081685, | |
| "grad_norm": 0.30408740043640137, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9135, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.003248750953705299, | |
| "grad_norm": 0.292966365814209, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.8313, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0033471979523024293, | |
| "grad_norm": 0.31631526350975037, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.8659, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0034456449508995595, | |
| "grad_norm": 0.3310135304927826, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8383, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0035440919494966897, | |
| "grad_norm": 0.3195180296897888, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.8724, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.00364253894809382, | |
| "grad_norm": 0.3270772695541382, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.9098, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.00374098594669095, | |
| "grad_norm": 0.3305830657482147, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.8415, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0038394329452880807, | |
| "grad_norm": 0.35488954186439514, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.7382, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0039378799438852105, | |
| "grad_norm": 0.3825576603412628, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.7865, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004036326942482341, | |
| "grad_norm": 0.3629797697067261, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8379, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004134773941079472, | |
| "grad_norm": 0.3552936613559723, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.8216, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0042332209396766015, | |
| "grad_norm": 0.46250760555267334, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.8175, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004331667938273732, | |
| "grad_norm": 0.4543311297893524, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.919, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004430114936870862, | |
| "grad_norm": 0.37943360209465027, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.7684, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0045285619354679925, | |
| "grad_norm": 0.4000648558139801, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8123, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004627008934065122, | |
| "grad_norm": 0.39708080887794495, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.7945, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004725455932662253, | |
| "grad_norm": 0.5106943845748901, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.8255, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0048239029312593835, | |
| "grad_norm": 0.6115984320640564, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.758, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.004922349929856513, | |
| "grad_norm": 0.588426947593689, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.7376, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004922349929856513, | |
| "eval_loss": 0.8385166525840759, | |
| "eval_runtime": 1238.7288, | |
| "eval_samples_per_second": 13.811, | |
| "eval_steps_per_second": 3.453, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005020796928453644, | |
| "grad_norm": 0.36069297790527344, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.7487, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.005119243927050774, | |
| "grad_norm": 0.33008241653442383, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.7388, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005217690925647904, | |
| "grad_norm": 0.31474047899246216, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8969, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005316137924245035, | |
| "grad_norm": 0.24151387810707092, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.9006, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005414584922842165, | |
| "grad_norm": 0.19937486946582794, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.7642, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.005513031921439295, | |
| "grad_norm": 0.18645475804805756, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.7808, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005611478920036425, | |
| "grad_norm": 0.19090257585048676, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.8375, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005709925918633556, | |
| "grad_norm": 0.2419562190771103, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.8403, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0058083729172306855, | |
| "grad_norm": 0.20582005381584167, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.891, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.005906819915827816, | |
| "grad_norm": 0.2004568725824356, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.7726, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.006005266914424947, | |
| "grad_norm": 0.2447865605354309, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.8143, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0061037139130220765, | |
| "grad_norm": 0.20331642031669617, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.9104, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006202160911619207, | |
| "grad_norm": 0.246404230594635, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.8502, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006300607910216337, | |
| "grad_norm": 0.18997882306575775, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.7866, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0063990549088134676, | |
| "grad_norm": 0.21694961190223694, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.741, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006497501907410598, | |
| "grad_norm": 0.21334491670131683, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.8463, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006595948906007728, | |
| "grad_norm": 0.21204103529453278, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.7384, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006694395904604859, | |
| "grad_norm": 0.19957895576953888, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.7448, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006792842903201988, | |
| "grad_norm": 0.21239468455314636, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8697, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.006891289901799119, | |
| "grad_norm": 0.23870718479156494, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.8272, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.006989736900396249, | |
| "grad_norm": 0.2248721718788147, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.7556, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007088183898993379, | |
| "grad_norm": 0.24748794734477997, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8199, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.00718663089759051, | |
| "grad_norm": 0.28307509422302246, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.8861, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.00728507789618764, | |
| "grad_norm": 0.2441713958978653, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8327, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.00738352489478477, | |
| "grad_norm": 0.25680777430534363, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.7428, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0074819718933819, | |
| "grad_norm": 0.24944563210010529, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.7965, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.007580418891979031, | |
| "grad_norm": 0.28170955181121826, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.8745, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007678865890576161, | |
| "grad_norm": 0.25036072731018066, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.7678, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.007777312889173291, | |
| "grad_norm": 0.26784461736679077, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.6854, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.007875759887770421, | |
| "grad_norm": 0.2638746500015259, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.82, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007974206886367552, | |
| "grad_norm": 0.29421475529670715, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.821, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008072653884964682, | |
| "grad_norm": 0.28754812479019165, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.8482, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.008171100883561813, | |
| "grad_norm": 0.2944120168685913, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.8223, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008269547882158943, | |
| "grad_norm": 0.2951282560825348, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.8239, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008367994880756072, | |
| "grad_norm": 0.3089338541030884, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.7714, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008466441879353203, | |
| "grad_norm": 0.3272501230239868, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.8504, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008564888877950334, | |
| "grad_norm": 0.30962470173835754, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.7594, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.008663335876547464, | |
| "grad_norm": 0.35390883684158325, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.773, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.008761782875144595, | |
| "grad_norm": 0.3277943432331085, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6385, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.008860229873741724, | |
| "grad_norm": 0.34669122099876404, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7411, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008958676872338854, | |
| "grad_norm": 0.3893776535987854, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.7897, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.009057123870935985, | |
| "grad_norm": 0.37446585297584534, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.698, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.009155570869533116, | |
| "grad_norm": 0.3651737868785858, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.6184, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.009254017868130245, | |
| "grad_norm": 0.3889082670211792, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.6392, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.009352464866727375, | |
| "grad_norm": 0.4377343952655792, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.7809, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009450911865324506, | |
| "grad_norm": 0.4487859308719635, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.6837, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009549358863921636, | |
| "grad_norm": 0.47383832931518555, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7838, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009647805862518767, | |
| "grad_norm": 0.5235933661460876, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6708, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.009746252861115896, | |
| "grad_norm": 0.5028972029685974, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.636, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.009844699859713027, | |
| "grad_norm": 0.6277095675468445, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.6084, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009844699859713027, | |
| "eval_loss": 0.7725606560707092, | |
| "eval_runtime": 1239.4187, | |
| "eval_samples_per_second": 13.803, | |
| "eval_steps_per_second": 3.451, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009943146858310157, | |
| "grad_norm": 0.2167699933052063, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.6497, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010041593856907288, | |
| "grad_norm": 0.29755645990371704, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.7594, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.010140040855504419, | |
| "grad_norm": 0.269955039024353, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.7016, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.010238487854101547, | |
| "grad_norm": 0.27121371030807495, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.7579, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.010336934852698678, | |
| "grad_norm": 0.22451242804527283, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6634, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.010435381851295809, | |
| "grad_norm": 0.28514641523361206, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.7262, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01053382884989294, | |
| "grad_norm": 0.21499718725681305, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.7663, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01063227584849007, | |
| "grad_norm": 0.1963709592819214, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.7326, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.010730722847087199, | |
| "grad_norm": 0.19709207117557526, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.7984, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.01082916984568433, | |
| "grad_norm": 0.21196742355823517, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.6991, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01092761684428146, | |
| "grad_norm": 0.20844809710979462, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.717, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.01102606384287859, | |
| "grad_norm": 0.20329934358596802, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.8268, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.011124510841475721, | |
| "grad_norm": 0.2340536117553711, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.8134, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.01122295784007285, | |
| "grad_norm": 0.22988249361515045, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.8233, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.011321404838669981, | |
| "grad_norm": 0.26659059524536133, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7696, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.011419851837267112, | |
| "grad_norm": 0.22979311645030975, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7695, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.011518298835864242, | |
| "grad_norm": 0.2350446581840515, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.7028, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.011616745834461371, | |
| "grad_norm": 0.24908263981342316, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7874, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.011715192833058502, | |
| "grad_norm": 0.24260659515857697, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.7671, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.011813639831655632, | |
| "grad_norm": 0.22742068767547607, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.7657, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.011912086830252763, | |
| "grad_norm": 0.2886553406715393, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.796, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.012010533828849894, | |
| "grad_norm": 0.2719157934188843, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.8424, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.012108980827447022, | |
| "grad_norm": 0.23944056034088135, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.7366, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.012207427826044153, | |
| "grad_norm": 0.26301103830337524, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.842, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.012305874824641284, | |
| "grad_norm": 0.2571735680103302, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.7656, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.012404321823238414, | |
| "grad_norm": 0.2682114541530609, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.7663, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.012502768821835545, | |
| "grad_norm": 0.26435235142707825, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.7823, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012601215820432674, | |
| "grad_norm": 0.2912524342536926, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.7405, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.012699662819029804, | |
| "grad_norm": 0.2640072703361511, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.772, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.012798109817626935, | |
| "grad_norm": 0.33146077394485474, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.8261, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.012896556816224066, | |
| "grad_norm": 0.35161370038986206, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.8091, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.012995003814821196, | |
| "grad_norm": 0.2969111502170563, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.7278, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.013093450813418325, | |
| "grad_norm": 0.3931649923324585, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.7695, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.013191897812015456, | |
| "grad_norm": 0.3043035864830017, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.7816, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.013290344810612587, | |
| "grad_norm": 0.3296326696872711, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.7207, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.013388791809209717, | |
| "grad_norm": 0.3272622227668762, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.7689, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.013487238807806848, | |
| "grad_norm": 0.3301616311073303, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.7207, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.013585685806403977, | |
| "grad_norm": 0.3362463712692261, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.6888, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.013684132805001107, | |
| "grad_norm": 0.34245991706848145, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.6665, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.013782579803598238, | |
| "grad_norm": 0.3581242561340332, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.671, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.013881026802195369, | |
| "grad_norm": 0.3574250042438507, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.6918, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.013979473800792497, | |
| "grad_norm": 0.3510986566543579, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.6382, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.014077920799389628, | |
| "grad_norm": 0.39022693037986755, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.6997, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.014176367797986759, | |
| "grad_norm": 0.43161872029304504, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.6712, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01427481479658389, | |
| "grad_norm": 0.3904305398464203, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6176, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.01437326179518102, | |
| "grad_norm": 0.4287797808647156, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.5234, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.014471708793778149, | |
| "grad_norm": 0.4110037386417389, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.6162, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.01457015579237528, | |
| "grad_norm": 0.5409560203552246, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.5807, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01466860279097241, | |
| "grad_norm": 0.5107161998748779, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.5888, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01476704978956954, | |
| "grad_norm": 0.6170666217803955, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.5027, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01476704978956954, | |
| "eval_loss": 0.7233030200004578, | |
| "eval_runtime": 1238.9408, | |
| "eval_samples_per_second": 13.809, | |
| "eval_steps_per_second": 3.452, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014865496788166671, | |
| "grad_norm": 0.12099899351596832, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.5387, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0149639437867638, | |
| "grad_norm": 0.16086311638355255, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.6224, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.015062390785360931, | |
| "grad_norm": 0.19216428697109222, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.786, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.015160837783958062, | |
| "grad_norm": 0.20429813861846924, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.7682, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.015259284782555192, | |
| "grad_norm": 0.2234884351491928, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.7007, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.015357731781152323, | |
| "grad_norm": 0.22140982747077942, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.6793, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.015456178779749452, | |
| "grad_norm": 0.22914864122867584, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.6767, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.015554625778346582, | |
| "grad_norm": 0.23383435606956482, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.7644, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.01565307277694371, | |
| "grad_norm": 0.25456830859184265, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.8353, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.015751519775540842, | |
| "grad_norm": 0.2665367126464844, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.7327, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.015849966774137973, | |
| "grad_norm": 0.2768714129924774, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.7025, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.015948413772735103, | |
| "grad_norm": 0.24915184080600739, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.7571, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.016046860771332234, | |
| "grad_norm": 0.24189279973506927, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.7179, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.016145307769929364, | |
| "grad_norm": 0.2797166705131531, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.7716, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.016243754768526495, | |
| "grad_norm": 0.2578907608985901, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.6788, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.016342201767123626, | |
| "grad_norm": 0.25776296854019165, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.7127, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.016440648765720756, | |
| "grad_norm": 0.25119781494140625, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.6797, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.016539095764317887, | |
| "grad_norm": 0.3322225511074066, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.8004, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.016637542762915014, | |
| "grad_norm": 0.2811940908432007, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.7511, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.016735989761512145, | |
| "grad_norm": 0.2745896577835083, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.7449, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.016834436760109275, | |
| "grad_norm": 0.2845320403575897, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.8137, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.016932883758706406, | |
| "grad_norm": 0.27154210209846497, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.7647, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.017031330757303537, | |
| "grad_norm": 0.31270915269851685, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.8859, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.017129777755900667, | |
| "grad_norm": 0.2980620265007019, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.7621, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.017228224754497798, | |
| "grad_norm": 0.29272592067718506, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.8133, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01732667175309493, | |
| "grad_norm": 0.2685643136501312, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.749, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01742511875169206, | |
| "grad_norm": 0.3375403881072998, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.7332, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01752356575028919, | |
| "grad_norm": 0.2986336648464203, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.7408, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.017622012748886317, | |
| "grad_norm": 0.31312334537506104, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.7226, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.017720459747483448, | |
| "grad_norm": 0.3165228068828583, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.6979, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.017818906746080578, | |
| "grad_norm": 0.32625827193260193, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.7074, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.01791735374467771, | |
| "grad_norm": 0.3192249834537506, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.7116, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.01801580074327484, | |
| "grad_norm": 0.3574119210243225, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.7676, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01811424774187197, | |
| "grad_norm": 0.32246261835098267, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.6575, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0182126947404691, | |
| "grad_norm": 0.32882922887802124, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.7102, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01831114173906623, | |
| "grad_norm": 0.3541630804538727, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.7203, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.018409588737663362, | |
| "grad_norm": 0.435232549905777, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.784, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01850803573626049, | |
| "grad_norm": 0.36940282583236694, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.6672, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01860648273485762, | |
| "grad_norm": 0.34149521589279175, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.7983, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01870492973345475, | |
| "grad_norm": 0.38780590891838074, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.673, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01880337673205188, | |
| "grad_norm": 0.37067875266075134, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.6755, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01890182373064901, | |
| "grad_norm": 0.3726218342781067, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.6333, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.019000270729246142, | |
| "grad_norm": 0.40441393852233887, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.6701, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.019098717727843273, | |
| "grad_norm": 0.41199642419815063, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.6553, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.019197164726440404, | |
| "grad_norm": 0.38783925771713257, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.564, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.019295611725037534, | |
| "grad_norm": 0.45699724555015564, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.6171, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.019394058723634665, | |
| "grad_norm": 0.5459165573120117, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.5975, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.019492505722231792, | |
| "grad_norm": 0.5128494501113892, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.7016, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.019590952720828923, | |
| "grad_norm": 0.5270230174064636, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.6909, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.019689399719426053, | |
| "grad_norm": 0.6088939309120178, | |
| "learning_rate": 0.0, | |
| "loss": 0.4912, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.019689399719426053, | |
| "eval_loss": 0.7140485644340515, | |
| "eval_runtime": 1238.4309, | |
| "eval_samples_per_second": 13.814, | |
| "eval_steps_per_second": 3.454, | |
| "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": 2.6909537631574426e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |