| { | |
| "best_metric": 0.7135888934135437, | |
| "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": 0.22390227019786835, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7891, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00019689399719426053, | |
| "grad_norm": 0.2734512686729431, | |
| "learning_rate": 2e-05, | |
| "loss": 0.9206, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0002953409957913908, | |
| "grad_norm": 0.269786536693573, | |
| "learning_rate": 3e-05, | |
| "loss": 0.9964, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00039378799438852106, | |
| "grad_norm": 0.2748248279094696, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9028, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0004922349929856513, | |
| "grad_norm": 0.2731742560863495, | |
| "learning_rate": 5e-05, | |
| "loss": 0.923, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005906819915827816, | |
| "grad_norm": 0.28288453817367554, | |
| "learning_rate": 6e-05, | |
| "loss": 1.017, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0006891289901799119, | |
| "grad_norm": 0.2824510931968689, | |
| "learning_rate": 7e-05, | |
| "loss": 1.02, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0007875759887770421, | |
| "grad_norm": 0.24208571016788483, | |
| "learning_rate": 8e-05, | |
| "loss": 0.8618, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0008860229873741724, | |
| "grad_norm": 0.21325424313545227, | |
| "learning_rate": 9e-05, | |
| "loss": 0.9321, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0009844699859713026, | |
| "grad_norm": 0.2290913462638855, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8679, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.001082916984568433, | |
| "grad_norm": 0.2823718786239624, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.9826, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0011813639831655632, | |
| "grad_norm": 0.3018488585948944, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.9849, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0012798109817626934, | |
| "grad_norm": 0.27093932032585144, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.907, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0013782579803598238, | |
| "grad_norm": 0.36700931191444397, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.771, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.001476704978956954, | |
| "grad_norm": 0.22497357428073883, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.8369, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0015751519775540842, | |
| "grad_norm": 0.2194261997938156, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.8784, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0016735989761512146, | |
| "grad_norm": 0.23490823805332184, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9204, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0017720459747483448, | |
| "grad_norm": 0.27081185579299927, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9428, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.001870492973345475, | |
| "grad_norm": 0.2338041216135025, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.8619, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0019689399719426052, | |
| "grad_norm": 0.3045388162136078, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.8159, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.002067386970539736, | |
| "grad_norm": 0.24665868282318115, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.8717, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.002165833969136866, | |
| "grad_norm": 0.22522355616092682, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.8371, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0022642809677339963, | |
| "grad_norm": 0.25770172476768494, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.854, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0023627279663311265, | |
| "grad_norm": 0.2478024661540985, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.8942, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0024611749649282567, | |
| "grad_norm": 0.2563309967517853, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.8957, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002559621963525387, | |
| "grad_norm": 0.24122270941734314, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.8243, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0026580689621225175, | |
| "grad_norm": 0.2688164710998535, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.7973, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0027565159607196477, | |
| "grad_norm": 0.271624892950058, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.9289, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.002854962959316778, | |
| "grad_norm": 0.2858806550502777, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.8637, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.002953409957913908, | |
| "grad_norm": 0.2943406403064728, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.8075, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0030518569565110383, | |
| "grad_norm": 0.30592080950737, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.9114, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0031503039551081685, | |
| "grad_norm": 0.3051643669605255, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.9143, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.003248750953705299, | |
| "grad_norm": 0.29056602716445923, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.8308, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0033471979523024293, | |
| "grad_norm": 0.3166995048522949, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.8669, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0034456449508995595, | |
| "grad_norm": 0.32888221740722656, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.8385, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0035440919494966897, | |
| "grad_norm": 0.31846779584884644, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.8716, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.00364253894809382, | |
| "grad_norm": 0.32777661085128784, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.9108, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.00374098594669095, | |
| "grad_norm": 0.32800793647766113, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.8412, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0038394329452880807, | |
| "grad_norm": 0.35635143518447876, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.7376, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0039378799438852105, | |
| "grad_norm": 0.37796255946159363, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.7852, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.004036326942482341, | |
| "grad_norm": 0.3594129979610443, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.8376, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004134773941079472, | |
| "grad_norm": 0.3548661768436432, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.8234, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0042332209396766015, | |
| "grad_norm": 0.4492071568965912, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.8187, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.004331667938273732, | |
| "grad_norm": 0.45113977789878845, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.92, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.004430114936870862, | |
| "grad_norm": 0.38140079379081726, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.7688, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0045285619354679925, | |
| "grad_norm": 0.3992875814437866, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.8098, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.004627008934065122, | |
| "grad_norm": 0.3943997919559479, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.7958, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004725455932662253, | |
| "grad_norm": 0.5221825838088989, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.827, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0048239029312593835, | |
| "grad_norm": 0.6273248195648193, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.7594, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.004922349929856513, | |
| "grad_norm": 0.5878886580467224, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.7383, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004922349929856513, | |
| "eval_loss": 0.8364587426185608, | |
| "eval_runtime": 909.6137, | |
| "eval_samples_per_second": 18.808, | |
| "eval_steps_per_second": 4.702, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005020796928453644, | |
| "grad_norm": 0.3476126194000244, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.7464, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.005119243927050774, | |
| "grad_norm": 0.33102530241012573, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.7377, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.005217690925647904, | |
| "grad_norm": 0.305660605430603, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8961, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.005316137924245035, | |
| "grad_norm": 0.2405320405960083, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.899, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.005414584922842165, | |
| "grad_norm": 0.19649557769298553, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.7635, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.005513031921439295, | |
| "grad_norm": 0.19191588461399078, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.7818, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.005611478920036425, | |
| "grad_norm": 0.18887370824813843, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.8376, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.005709925918633556, | |
| "grad_norm": 0.2009602189064026, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.8391, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0058083729172306855, | |
| "grad_norm": 0.1989857256412506, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.8894, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.005906819915827816, | |
| "grad_norm": 0.20614689588546753, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.7723, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.006005266914424947, | |
| "grad_norm": 0.24282915890216827, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.8136, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0061037139130220765, | |
| "grad_norm": 0.20389336347579956, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.91, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.006202160911619207, | |
| "grad_norm": 0.26750174164772034, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.8506, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.006300607910216337, | |
| "grad_norm": 0.19222524762153625, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.7848, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0063990549088134676, | |
| "grad_norm": 0.21804963052272797, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.741, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.006497501907410598, | |
| "grad_norm": 0.21584440767765045, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.847, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.006595948906007728, | |
| "grad_norm": 0.21278530359268188, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.7379, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.006694395904604859, | |
| "grad_norm": 0.20563411712646484, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.7461, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.006792842903201988, | |
| "grad_norm": 0.21897177398204803, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8705, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.006891289901799119, | |
| "grad_norm": 0.2351602166891098, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.8263, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.006989736900396249, | |
| "grad_norm": 0.2217237949371338, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.7559, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.007088183898993379, | |
| "grad_norm": 0.26899853348731995, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8191, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.00718663089759051, | |
| "grad_norm": 0.28284358978271484, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.8854, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.00728507789618764, | |
| "grad_norm": 0.2405179738998413, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8319, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.00738352489478477, | |
| "grad_norm": 0.2575754225254059, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.7436, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0074819718933819, | |
| "grad_norm": 0.2507788836956024, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.7965, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.007580418891979031, | |
| "grad_norm": 0.2707754969596863, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.8741, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.007678865890576161, | |
| "grad_norm": 0.24839383363723755, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.7669, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.007777312889173291, | |
| "grad_norm": 0.2674270570278168, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.6862, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.007875759887770421, | |
| "grad_norm": 0.2609316110610962, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.8209, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.007974206886367552, | |
| "grad_norm": 0.2987496554851532, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.8206, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.008072653884964682, | |
| "grad_norm": 0.2906162440776825, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.8512, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.008171100883561813, | |
| "grad_norm": 0.3020494282245636, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.8226, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.008269547882158943, | |
| "grad_norm": 0.29151833057403564, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.822, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.008367994880756072, | |
| "grad_norm": 0.3084469735622406, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.768, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.008466441879353203, | |
| "grad_norm": 0.3298437297344208, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.8502, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.008564888877950334, | |
| "grad_norm": 0.3054105341434479, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.7599, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.008663335876547464, | |
| "grad_norm": 0.3527895212173462, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.7729, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.008761782875144595, | |
| "grad_norm": 0.3258419930934906, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.6376, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.008860229873741724, | |
| "grad_norm": 0.3712911307811737, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.7394, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.008958676872338854, | |
| "grad_norm": 0.39494508504867554, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.7892, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.009057123870935985, | |
| "grad_norm": 0.3761928677558899, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.6972, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.009155570869533116, | |
| "grad_norm": 0.3725806474685669, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.6187, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.009254017868130245, | |
| "grad_norm": 0.38239786028862, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.6379, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.009352464866727375, | |
| "grad_norm": 0.45231762528419495, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.7822, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.009450911865324506, | |
| "grad_norm": 0.44761863350868225, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.6857, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.009549358863921636, | |
| "grad_norm": 0.47627896070480347, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.7845, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.009647805862518767, | |
| "grad_norm": 0.5246194005012512, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.6705, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.009746252861115896, | |
| "grad_norm": 0.5054622888565063, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.634, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.009844699859713027, | |
| "grad_norm": 0.6298301219940186, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.6048, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009844699859713027, | |
| "eval_loss": 0.7698514461517334, | |
| "eval_runtime": 909.2952, | |
| "eval_samples_per_second": 18.815, | |
| "eval_steps_per_second": 4.704, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.009943146858310157, | |
| "grad_norm": 0.20888717472553253, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.6472, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.010041593856907288, | |
| "grad_norm": 0.286599725484848, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.7566, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.010140040855504419, | |
| "grad_norm": 0.2587265372276306, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.6995, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.010238487854101547, | |
| "grad_norm": 0.2665272355079651, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.7562, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.010336934852698678, | |
| "grad_norm": 0.22268541157245636, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6623, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.010435381851295809, | |
| "grad_norm": 0.2853225767612457, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.7285, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.01053382884989294, | |
| "grad_norm": 0.21325519680976868, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.7669, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.01063227584849007, | |
| "grad_norm": 0.19689135253429413, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.732, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.010730722847087199, | |
| "grad_norm": 0.19993603229522705, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.7978, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.01082916984568433, | |
| "grad_norm": 0.21663668751716614, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.6984, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01092761684428146, | |
| "grad_norm": 0.21133282780647278, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.7156, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.01102606384287859, | |
| "grad_norm": 0.2072429060935974, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.8255, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.011124510841475721, | |
| "grad_norm": 0.22252115607261658, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.8123, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.01122295784007285, | |
| "grad_norm": 0.23027317225933075, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.8224, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.011321404838669981, | |
| "grad_norm": 0.2570335566997528, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.7697, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.011419851837267112, | |
| "grad_norm": 0.23723390698432922, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.7695, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.011518298835864242, | |
| "grad_norm": 0.23075798153877258, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.7022, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.011616745834461371, | |
| "grad_norm": 0.25582340359687805, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7882, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.011715192833058502, | |
| "grad_norm": 0.24433791637420654, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.7664, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.011813639831655632, | |
| "grad_norm": 0.22956714034080505, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.7661, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.011912086830252763, | |
| "grad_norm": 0.28372588753700256, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.7934, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.012010533828849894, | |
| "grad_norm": 0.27238729596138, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.8414, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.012108980827447022, | |
| "grad_norm": 0.24188391864299774, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.7355, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.012207427826044153, | |
| "grad_norm": 0.264715313911438, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.8409, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.012305874824641284, | |
| "grad_norm": 0.25678229331970215, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.7658, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.012404321823238414, | |
| "grad_norm": 0.273660272359848, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.7664, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.012502768821835545, | |
| "grad_norm": 0.2639061212539673, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.782, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.012601215820432674, | |
| "grad_norm": 0.2950485050678253, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.7399, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.012699662819029804, | |
| "grad_norm": 0.26575687527656555, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.7733, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.012798109817626935, | |
| "grad_norm": 0.3353694677352905, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.8266, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.012896556816224066, | |
| "grad_norm": 0.35672709345817566, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.8076, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.012995003814821196, | |
| "grad_norm": 0.2986244559288025, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.7258, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.013093450813418325, | |
| "grad_norm": 0.3568984866142273, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.7695, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.013191897812015456, | |
| "grad_norm": 0.30244019627571106, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.7806, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.013290344810612587, | |
| "grad_norm": 0.3304792642593384, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.7205, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.013388791809209717, | |
| "grad_norm": 0.3327755033969879, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.7679, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.013487238807806848, | |
| "grad_norm": 0.33431971073150635, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.7209, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.013585685806403977, | |
| "grad_norm": 0.34154167771339417, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.6884, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.013684132805001107, | |
| "grad_norm": 0.34425023198127747, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.6645, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.013782579803598238, | |
| "grad_norm": 0.3643842339515686, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.6696, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.013881026802195369, | |
| "grad_norm": 0.3612792193889618, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.6917, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.013979473800792497, | |
| "grad_norm": 0.35165005922317505, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.6392, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.014077920799389628, | |
| "grad_norm": 0.3895745277404785, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.6977, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.014176367797986759, | |
| "grad_norm": 0.4308588206768036, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.6688, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01427481479658389, | |
| "grad_norm": 0.3935404121875763, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.6176, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.01437326179518102, | |
| "grad_norm": 0.4410896599292755, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.5208, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.014471708793778149, | |
| "grad_norm": 0.4134191572666168, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.6132, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.01457015579237528, | |
| "grad_norm": 0.5402004718780518, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.5779, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01466860279097241, | |
| "grad_norm": 0.5370400547981262, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.5875, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01476704978956954, | |
| "grad_norm": 0.6376131176948547, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.5031, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01476704978956954, | |
| "eval_loss": 0.7231428027153015, | |
| "eval_runtime": 910.7367, | |
| "eval_samples_per_second": 18.785, | |
| "eval_steps_per_second": 4.696, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.014865496788166671, | |
| "grad_norm": 0.11930377781391144, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.5386, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0149639437867638, | |
| "grad_norm": 0.16247321665287018, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.6219, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.015062390785360931, | |
| "grad_norm": 0.20031054317951202, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.7861, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.015160837783958062, | |
| "grad_norm": 0.20611421763896942, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.7678, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.015259284782555192, | |
| "grad_norm": 0.22807583212852478, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.7018, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.015357731781152323, | |
| "grad_norm": 0.22471506893634796, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.6784, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.015456178779749452, | |
| "grad_norm": 0.26132553815841675, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.6765, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.015554625778346582, | |
| "grad_norm": 0.23332355916500092, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.7644, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.01565307277694371, | |
| "grad_norm": 0.24738949537277222, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.8344, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.015751519775540842, | |
| "grad_norm": 0.2664797306060791, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.7323, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.015849966774137973, | |
| "grad_norm": 0.27797332406044006, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.7011, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.015948413772735103, | |
| "grad_norm": 0.25038793683052063, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.7564, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.016046860771332234, | |
| "grad_norm": 0.24009868502616882, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.7167, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.016145307769929364, | |
| "grad_norm": 0.28158503770828247, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.7722, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.016243754768526495, | |
| "grad_norm": 0.2530911862850189, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.6774, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.016342201767123626, | |
| "grad_norm": 0.24448774755001068, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.7117, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.016440648765720756, | |
| "grad_norm": 0.2513751685619354, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.6804, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.016539095764317887, | |
| "grad_norm": 0.33193540573120117, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.7996, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.016637542762915014, | |
| "grad_norm": 0.29664281010627747, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.7504, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.016735989761512145, | |
| "grad_norm": 0.27495744824409485, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.7443, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.016834436760109275, | |
| "grad_norm": 0.2867414355278015, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.8144, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.016932883758706406, | |
| "grad_norm": 0.27185213565826416, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.7643, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.017031330757303537, | |
| "grad_norm": 0.3129732012748718, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.8846, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.017129777755900667, | |
| "grad_norm": 0.30969539284706116, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.761, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.017228224754497798, | |
| "grad_norm": 0.29353201389312744, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.814, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01732667175309493, | |
| "grad_norm": 0.2705335021018982, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.7505, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.01742511875169206, | |
| "grad_norm": 0.30902302265167236, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.7311, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.01752356575028919, | |
| "grad_norm": 0.30058932304382324, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.7389, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.017622012748886317, | |
| "grad_norm": 0.3121153712272644, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.7228, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.017720459747483448, | |
| "grad_norm": 0.3204442858695984, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.6967, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.017818906746080578, | |
| "grad_norm": 0.3386498987674713, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.7068, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.01791735374467771, | |
| "grad_norm": 0.31919077038764954, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.7103, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.01801580074327484, | |
| "grad_norm": 0.36073577404022217, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.7672, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01811424774187197, | |
| "grad_norm": 0.31926870346069336, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.6566, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0182126947404691, | |
| "grad_norm": 0.3300493359565735, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.7093, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01831114173906623, | |
| "grad_norm": 0.3521917462348938, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.7202, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.018409588737663362, | |
| "grad_norm": 0.4383225739002228, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.7847, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.01850803573626049, | |
| "grad_norm": 0.3749508559703827, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.6671, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01860648273485762, | |
| "grad_norm": 0.34106940031051636, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.7952, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.01870492973345475, | |
| "grad_norm": 0.38784539699554443, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.6736, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01880337673205188, | |
| "grad_norm": 0.37535935640335083, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.6749, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01890182373064901, | |
| "grad_norm": 0.37157732248306274, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.6325, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.019000270729246142, | |
| "grad_norm": 0.40247318148612976, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.6671, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.019098717727843273, | |
| "grad_norm": 0.40654775500297546, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.651, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.019197164726440404, | |
| "grad_norm": 0.3884858191013336, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.5634, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.019295611725037534, | |
| "grad_norm": 0.4609086215496063, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.6149, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.019394058723634665, | |
| "grad_norm": 0.549687385559082, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.6001, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.019492505722231792, | |
| "grad_norm": 0.5061210989952087, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.6994, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.019590952720828923, | |
| "grad_norm": 0.5341589450836182, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.6888, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.019689399719426053, | |
| "grad_norm": 0.6175804734230042, | |
| "learning_rate": 0.0, | |
| "loss": 0.4914, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.019689399719426053, | |
| "eval_loss": 0.7135888934135437, | |
| "eval_runtime": 910.6706, | |
| "eval_samples_per_second": 18.786, | |
| "eval_steps_per_second": 4.697, | |
| "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 | |
| } | |