|
{ |
|
"best_metric": 0.23117779195308685, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.7168458781362007, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0035842293906810036, |
|
"grad_norm": 8.764748573303223, |
|
"learning_rate": 1e-05, |
|
"loss": 3.1609, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0035842293906810036, |
|
"eval_loss": 0.9623667597770691, |
|
"eval_runtime": 49.6487, |
|
"eval_samples_per_second": 9.467, |
|
"eval_steps_per_second": 2.377, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.007168458781362007, |
|
"grad_norm": 9.91032600402832, |
|
"learning_rate": 2e-05, |
|
"loss": 3.2565, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.010752688172043012, |
|
"grad_norm": 10.527271270751953, |
|
"learning_rate": 3e-05, |
|
"loss": 3.3377, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.014336917562724014, |
|
"grad_norm": 13.104958534240723, |
|
"learning_rate": 4e-05, |
|
"loss": 3.7896, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.017921146953405017, |
|
"grad_norm": 11.248802185058594, |
|
"learning_rate": 5e-05, |
|
"loss": 3.1969, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.021505376344086023, |
|
"grad_norm": 11.298332214355469, |
|
"learning_rate": 6e-05, |
|
"loss": 3.5184, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.025089605734767026, |
|
"grad_norm": 11.769418716430664, |
|
"learning_rate": 7e-05, |
|
"loss": 2.882, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.02867383512544803, |
|
"grad_norm": 8.718571662902832, |
|
"learning_rate": 8e-05, |
|
"loss": 2.4813, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.03225806451612903, |
|
"grad_norm": 9.661246299743652, |
|
"learning_rate": 9e-05, |
|
"loss": 2.3433, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.035842293906810034, |
|
"grad_norm": 9.185708999633789, |
|
"learning_rate": 0.0001, |
|
"loss": 1.885, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.03942652329749104, |
|
"grad_norm": 10.113519668579102, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 2.0179, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.043010752688172046, |
|
"grad_norm": 13.618560791015625, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 2.741, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.04659498207885305, |
|
"grad_norm": 10.647560119628906, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 1.5398, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.05017921146953405, |
|
"grad_norm": 10.532403945922852, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 1.572, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.053763440860215055, |
|
"grad_norm": 11.85290813446045, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 1.3534, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.05734767025089606, |
|
"grad_norm": 9.96902084350586, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 1.2538, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.06093189964157706, |
|
"grad_norm": 15.623945236206055, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 1.6005, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.06451612903225806, |
|
"grad_norm": 22.614683151245117, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 1.3881, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.06810035842293907, |
|
"grad_norm": 10.335607528686523, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 0.9605, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.07168458781362007, |
|
"grad_norm": 14.423209190368652, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 1.2163, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.07526881720430108, |
|
"grad_norm": 14.910286903381348, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 1.7089, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.07885304659498207, |
|
"grad_norm": 12.511384963989258, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 1.7404, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.08243727598566308, |
|
"grad_norm": 7.866326332092285, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 1.3669, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.08602150537634409, |
|
"grad_norm": 7.211355686187744, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 1.1017, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.08960573476702509, |
|
"grad_norm": 9.416728973388672, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 1.4343, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.0931899641577061, |
|
"grad_norm": 10.1671781539917, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 1.4434, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.0967741935483871, |
|
"grad_norm": 7.222290992736816, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 1.3245, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.1003584229390681, |
|
"grad_norm": 6.214794158935547, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 1.0822, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.1039426523297491, |
|
"grad_norm": 6.57796049118042, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 1.0686, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.10752688172043011, |
|
"grad_norm": 5.079577922821045, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 0.8213, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1111111111111111, |
|
"grad_norm": 5.848414421081543, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 1.0584, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.11469534050179211, |
|
"grad_norm": 7.867873668670654, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 1.3737, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.11827956989247312, |
|
"grad_norm": 7.689173698425293, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 1.0373, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.12186379928315412, |
|
"grad_norm": 6.600484848022461, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 0.9088, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.12544802867383512, |
|
"grad_norm": 6.862224578857422, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 0.9346, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.12903225806451613, |
|
"grad_norm": 6.382403373718262, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 0.9656, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.13261648745519714, |
|
"grad_norm": 7.677615642547607, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 0.9647, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.13620071684587814, |
|
"grad_norm": 7.210923194885254, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 0.8935, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.13978494623655913, |
|
"grad_norm": 6.139534950256348, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 0.8763, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.14336917562724014, |
|
"grad_norm": 7.213724613189697, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 0.5432, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.14695340501792115, |
|
"grad_norm": 7.949337005615234, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 1.0147, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.15053763440860216, |
|
"grad_norm": 7.482588291168213, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 1.1104, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.15412186379928317, |
|
"grad_norm": 7.986391067504883, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 0.884, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.15770609318996415, |
|
"grad_norm": 7.138528347015381, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 0.6474, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.16129032258064516, |
|
"grad_norm": 6.815484523773193, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 0.8735, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.16487455197132617, |
|
"grad_norm": 6.020720481872559, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 0.8604, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.16845878136200718, |
|
"grad_norm": 5.6506500244140625, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 0.7477, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.17204301075268819, |
|
"grad_norm": 7.093197345733643, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 0.781, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.17562724014336917, |
|
"grad_norm": 7.793400764465332, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 0.9194, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.17921146953405018, |
|
"grad_norm": 5.8821282386779785, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 0.3512, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.17921146953405018, |
|
"eval_loss": 0.28004929423332214, |
|
"eval_runtime": 49.94, |
|
"eval_samples_per_second": 9.411, |
|
"eval_steps_per_second": 2.363, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.1827956989247312, |
|
"grad_norm": 9.191633224487305, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 1.8294, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.1863799283154122, |
|
"grad_norm": 7.976558208465576, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 2.0961, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.18996415770609318, |
|
"grad_norm": 6.3274946212768555, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 2.3197, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.1935483870967742, |
|
"grad_norm": 6.010959148406982, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 2.0092, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.1971326164874552, |
|
"grad_norm": 5.882427215576172, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 1.8697, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.2007168458781362, |
|
"grad_norm": 6.208774089813232, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 1.6801, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.20430107526881722, |
|
"grad_norm": 5.348367214202881, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 1.6368, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.2078853046594982, |
|
"grad_norm": 5.442905902862549, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 1.3868, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.2114695340501792, |
|
"grad_norm": 5.595448970794678, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 1.2856, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.21505376344086022, |
|
"grad_norm": 10.32099723815918, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 1.5686, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.21863799283154123, |
|
"grad_norm": 8.23583698272705, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 1.416, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.2222222222222222, |
|
"grad_norm": 12.7261962890625, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 1.6503, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.22580645161290322, |
|
"grad_norm": 5.927293300628662, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 0.8517, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.22939068100358423, |
|
"grad_norm": 6.635047435760498, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 0.9006, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.23297491039426524, |
|
"grad_norm": 8.029170989990234, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 0.799, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.23655913978494625, |
|
"grad_norm": 9.035547256469727, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 1.4406, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.24014336917562723, |
|
"grad_norm": 9.572111129760742, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 0.968, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.24372759856630824, |
|
"grad_norm": 7.323725700378418, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 0.7307, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.24731182795698925, |
|
"grad_norm": 8.532878875732422, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 0.921, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.25089605734767023, |
|
"grad_norm": 12.004900932312012, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 0.9137, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.25448028673835127, |
|
"grad_norm": 9.63646125793457, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 0.9467, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.25806451612903225, |
|
"grad_norm": 7.666362285614014, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 1.2921, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.2616487455197133, |
|
"grad_norm": 7.735444068908691, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 1.4973, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.26523297491039427, |
|
"grad_norm": 14.301405906677246, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 1.2532, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.26881720430107525, |
|
"grad_norm": 5.494683265686035, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 1.1198, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.2724014336917563, |
|
"grad_norm": 5.784592151641846, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 0.8771, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.27598566308243727, |
|
"grad_norm": 6.8970770835876465, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 1.2429, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.27956989247311825, |
|
"grad_norm": 5.237691402435303, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 0.9329, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.2831541218637993, |
|
"grad_norm": 8.241721153259277, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 1.3392, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.2867383512544803, |
|
"grad_norm": 5.223793029785156, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 0.9224, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.2903225806451613, |
|
"grad_norm": 7.292931079864502, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 1.2648, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.2939068100358423, |
|
"grad_norm": 6.255661964416504, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 1.1487, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.2974910394265233, |
|
"grad_norm": 5.561267375946045, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 1.1417, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.3010752688172043, |
|
"grad_norm": 5.368224620819092, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 0.8612, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.3046594982078853, |
|
"grad_norm": 6.687611103057861, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 1.0543, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.30824372759856633, |
|
"grad_norm": 6.311729431152344, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 1.0505, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.3118279569892473, |
|
"grad_norm": 5.3506598472595215, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 1.0001, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.3154121863799283, |
|
"grad_norm": 6.4501543045043945, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 1.1149, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.31899641577060933, |
|
"grad_norm": 7.826871395111084, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 1.0674, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.3225806451612903, |
|
"grad_norm": 6.225786209106445, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 0.8601, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.32616487455197135, |
|
"grad_norm": 4.410377025604248, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 0.6826, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.32974910394265233, |
|
"grad_norm": 5.833151817321777, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 0.6293, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.3333333333333333, |
|
"grad_norm": 6.048341751098633, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 0.775, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.33691756272401435, |
|
"grad_norm": 5.176039695739746, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 0.7012, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.34050179211469533, |
|
"grad_norm": 6.657656669616699, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 0.9241, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.34408602150537637, |
|
"grad_norm": 5.278158187866211, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 0.6386, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.34767025089605735, |
|
"grad_norm": 8.443243026733398, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 0.8443, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.35125448028673834, |
|
"grad_norm": 6.784634113311768, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 0.542, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.3548387096774194, |
|
"grad_norm": 7.89312744140625, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 1.0249, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.35842293906810035, |
|
"grad_norm": 9.279082298278809, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 0.953, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.35842293906810035, |
|
"eval_loss": 0.25865477323532104, |
|
"eval_runtime": 49.9482, |
|
"eval_samples_per_second": 9.41, |
|
"eval_steps_per_second": 2.362, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.36200716845878134, |
|
"grad_norm": 8.081390380859375, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 2.0773, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.3655913978494624, |
|
"grad_norm": 5.93018102645874, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 1.414, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.36917562724014336, |
|
"grad_norm": 6.167929649353027, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 1.9909, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.3727598566308244, |
|
"grad_norm": 5.644564151763916, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 1.9097, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.3763440860215054, |
|
"grad_norm": 5.8302693367004395, |
|
"learning_rate": 5e-05, |
|
"loss": 1.8426, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.37992831541218636, |
|
"grad_norm": 5.293248176574707, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 1.4679, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.3835125448028674, |
|
"grad_norm": 6.272945404052734, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 1.3422, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.3870967741935484, |
|
"grad_norm": 5.84699010848999, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 1.3006, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.3906810035842294, |
|
"grad_norm": 6.156829833984375, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 1.3395, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.3942652329749104, |
|
"grad_norm": 7.0067572593688965, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 1.3715, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.3978494623655914, |
|
"grad_norm": 6.35042142868042, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 1.2026, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.4014336917562724, |
|
"grad_norm": 6.0808587074279785, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 0.8454, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.4050179211469534, |
|
"grad_norm": 6.202322006225586, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 0.7719, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.40860215053763443, |
|
"grad_norm": 6.051723957061768, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 0.7517, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.4121863799283154, |
|
"grad_norm": 6.158517360687256, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 0.8183, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.4157706093189964, |
|
"grad_norm": 6.515998363494873, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 0.7646, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.41935483870967744, |
|
"grad_norm": 7.12328577041626, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 0.8475, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.4229390681003584, |
|
"grad_norm": 6.197124004364014, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 0.7205, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.4265232974910394, |
|
"grad_norm": 7.827014923095703, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 0.8189, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.43010752688172044, |
|
"grad_norm": 23.93094825744629, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 1.5903, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.4336917562724014, |
|
"grad_norm": 7.068710803985596, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 1.2603, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.43727598566308246, |
|
"grad_norm": 8.994744300842285, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 1.3092, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.44086021505376344, |
|
"grad_norm": 6.050983428955078, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 1.069, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.4444444444444444, |
|
"grad_norm": 5.181216716766357, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 0.8529, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.44802867383512546, |
|
"grad_norm": 6.688392162322998, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 1.5573, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.45161290322580644, |
|
"grad_norm": 5.401554584503174, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 1.0815, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.4551971326164875, |
|
"grad_norm": 5.66651725769043, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 0.9981, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.45878136200716846, |
|
"grad_norm": 4.585411548614502, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 0.8789, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.46236559139784944, |
|
"grad_norm": 5.330341339111328, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 1.0456, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.4659498207885305, |
|
"grad_norm": 5.617556095123291, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 1.2597, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.46953405017921146, |
|
"grad_norm": 4.747445106506348, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 0.9564, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.4731182795698925, |
|
"grad_norm": 5.659011363983154, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 1.0499, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.4767025089605735, |
|
"grad_norm": 7.251832962036133, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 1.0936, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.48028673835125446, |
|
"grad_norm": 6.353315830230713, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 1.0363, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.4838709677419355, |
|
"grad_norm": 5.097989082336426, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 0.8755, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.4874551971326165, |
|
"grad_norm": 4.705554008483887, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 0.8627, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.4910394265232975, |
|
"grad_norm": 4.368675231933594, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 0.7064, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.4946236559139785, |
|
"grad_norm": 5.788303375244141, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 1.0093, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.4982078853046595, |
|
"grad_norm": 6.125124454498291, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 0.9555, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.5017921146953405, |
|
"grad_norm": 4.498645305633545, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 0.8736, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.5053763440860215, |
|
"grad_norm": 5.132452964782715, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 0.8421, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.5089605734767025, |
|
"grad_norm": 5.978065490722656, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 0.8383, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.5125448028673835, |
|
"grad_norm": 4.4699811935424805, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 0.7035, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.5161290322580645, |
|
"grad_norm": 5.342233180999756, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 0.7621, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.5197132616487455, |
|
"grad_norm": 4.167507648468018, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 0.5562, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.5232974910394266, |
|
"grad_norm": 6.9307990074157715, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 1.0277, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.5268817204301075, |
|
"grad_norm": 5.642251491546631, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 0.822, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.5304659498207885, |
|
"grad_norm": 5.521220684051514, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 0.5681, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.5340501792114696, |
|
"grad_norm": 5.848841667175293, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 0.536, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.5376344086021505, |
|
"grad_norm": 6.174679756164551, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 0.5516, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.5376344086021505, |
|
"eval_loss": 0.2374449074268341, |
|
"eval_runtime": 50.1908, |
|
"eval_samples_per_second": 9.364, |
|
"eval_steps_per_second": 2.351, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.5412186379928315, |
|
"grad_norm": 4.161494255065918, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 1.6372, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.5448028673835126, |
|
"grad_norm": 4.966536045074463, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 1.7186, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.5483870967741935, |
|
"grad_norm": 5.3374505043029785, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 2.0429, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.5519713261648745, |
|
"grad_norm": 5.27268123626709, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 1.8136, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.5555555555555556, |
|
"grad_norm": 4.8616204261779785, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 1.1982, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.5591397849462365, |
|
"grad_norm": 5.509308338165283, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 1.8703, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.5627240143369175, |
|
"grad_norm": 4.640356540679932, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 1.1671, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.5663082437275986, |
|
"grad_norm": 6.552391529083252, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 1.459, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.5698924731182796, |
|
"grad_norm": 5.22314453125, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 1.1348, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.5734767025089605, |
|
"grad_norm": 6.343968868255615, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 1.3796, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.5770609318996416, |
|
"grad_norm": 6.153079509735107, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 1.1584, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.5806451612903226, |
|
"grad_norm": 5.950961112976074, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 0.8715, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.5842293906810035, |
|
"grad_norm": 5.237393856048584, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 0.8242, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.5878136200716846, |
|
"grad_norm": 5.861473083496094, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 1.1719, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.5913978494623656, |
|
"grad_norm": 6.166896343231201, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 0.9401, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.5949820788530465, |
|
"grad_norm": 5.746236324310303, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 0.9308, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.5985663082437276, |
|
"grad_norm": 5.8852386474609375, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 0.7628, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.6021505376344086, |
|
"grad_norm": 7.991213798522949, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 0.5941, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.6057347670250897, |
|
"grad_norm": 6.561001300811768, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 0.6071, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.6093189964157706, |
|
"grad_norm": 11.279608726501465, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 0.6636, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.6129032258064516, |
|
"grad_norm": 7.360454082489014, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 0.7445, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.6164874551971327, |
|
"grad_norm": 10.051535606384277, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 0.8789, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.6200716845878136, |
|
"grad_norm": 6.2288103103637695, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 1.3492, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.6236559139784946, |
|
"grad_norm": 6.141717433929443, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 1.1329, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.6272401433691757, |
|
"grad_norm": 4.577736854553223, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 0.8618, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.6308243727598566, |
|
"grad_norm": 6.113799095153809, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 0.994, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.6344086021505376, |
|
"grad_norm": 6.326624870300293, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 1.085, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.6379928315412187, |
|
"grad_norm": 5.638182640075684, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 0.7309, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.6415770609318996, |
|
"grad_norm": 6.623641490936279, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 1.3905, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.6451612903225806, |
|
"grad_norm": 5.113369941711426, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 0.9076, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.6487455197132617, |
|
"grad_norm": 5.99696683883667, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 1.1393, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.6523297491039427, |
|
"grad_norm": 6.0718994140625, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 1.1142, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.6559139784946236, |
|
"grad_norm": 5.084056854248047, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 1.0265, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.6594982078853047, |
|
"grad_norm": 4.511726379394531, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 0.8624, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.6630824372759857, |
|
"grad_norm": 5.503136157989502, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 1.2258, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.6666666666666666, |
|
"grad_norm": 3.832946300506592, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 0.4943, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.6702508960573477, |
|
"grad_norm": 5.375131607055664, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 1.2102, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.6738351254480287, |
|
"grad_norm": 5.761864185333252, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 0.9625, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.6774193548387096, |
|
"grad_norm": 5.710801601409912, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 0.8581, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.6810035842293907, |
|
"grad_norm": 5.041393756866455, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 0.8639, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.6845878136200717, |
|
"grad_norm": 3.7694079875946045, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 0.5182, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.6881720430107527, |
|
"grad_norm": 4.875755310058594, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 0.6256, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.6917562724014337, |
|
"grad_norm": 6.2275919914245605, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 1.0543, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.6953405017921147, |
|
"grad_norm": 4.438046455383301, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 0.7448, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.6989247311827957, |
|
"grad_norm": 4.657058238983154, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 0.7395, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.7025089605734767, |
|
"grad_norm": 4.163156509399414, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 0.6258, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.7060931899641577, |
|
"grad_norm": 5.096886157989502, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 0.6781, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.7096774193548387, |
|
"grad_norm": 5.7775774002075195, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 0.8462, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.7132616487455197, |
|
"grad_norm": 9.385880470275879, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 0.9194, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.7168458781362007, |
|
"grad_norm": 4.655171871185303, |
|
"learning_rate": 0.0, |
|
"loss": 0.3828, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.7168458781362007, |
|
"eval_loss": 0.23117779195308685, |
|
"eval_runtime": 49.945, |
|
"eval_samples_per_second": 9.41, |
|
"eval_steps_per_second": 2.363, |
|
"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": 4.094423145035858e+17, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|