{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.074398830875515, "global_step": 23000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 9.982285208148806e-06, "loss": 1.2518, "step": 100 }, { "epoch": 0.04, "learning_rate": 9.964570416297609e-06, "loss": 1.2505, "step": 200 }, { "epoch": 0.05, "learning_rate": 9.946855624446414e-06, "loss": 1.2781, "step": 300 }, { "epoch": 0.07, "learning_rate": 9.929140832595217e-06, "loss": 1.2509, "step": 400 }, { "epoch": 0.09, "learning_rate": 9.911426040744022e-06, "loss": 1.2275, "step": 500 }, { "epoch": 0.11, "learning_rate": 9.893711248892827e-06, "loss": 1.2583, "step": 600 }, { "epoch": 0.12, "learning_rate": 9.87599645704163e-06, "loss": 1.2391, "step": 700 }, { "epoch": 0.14, "learning_rate": 9.858281665190435e-06, "loss": 1.2406, "step": 800 }, { "epoch": 0.16, "learning_rate": 9.840566873339238e-06, "loss": 1.2467, "step": 900 }, { "epoch": 0.18, "learning_rate": 9.822852081488043e-06, "loss": 1.251, "step": 1000 }, { "epoch": 0.19, "learning_rate": 9.805137289636848e-06, "loss": 1.2033, "step": 1100 }, { "epoch": 0.21, "learning_rate": 9.787422497785651e-06, "loss": 1.2345, "step": 1200 }, { "epoch": 0.23, "learning_rate": 9.769707705934456e-06, "loss": 1.2463, "step": 1300 }, { "epoch": 0.25, "learning_rate": 9.751992914083261e-06, "loss": 1.237, "step": 1400 }, { "epoch": 0.27, "learning_rate": 9.734278122232064e-06, "loss": 1.2276, "step": 1500 }, { "epoch": 0.28, "learning_rate": 9.71656333038087e-06, "loss": 1.2274, "step": 1600 }, { "epoch": 0.3, "learning_rate": 9.698848538529672e-06, "loss": 1.2487, "step": 1700 }, { "epoch": 0.32, "learning_rate": 9.681133746678477e-06, "loss": 1.2168, "step": 1800 }, { "epoch": 0.34, "learning_rate": 9.663418954827282e-06, "loss": 1.2465, "step": 1900 }, { "epoch": 0.35, "learning_rate": 9.645704162976086e-06, "loss": 1.2375, "step": 2000 }, { "epoch": 0.37, "learning_rate": 9.627989371124889e-06, "loss": 1.2178, "step": 2100 }, { "epoch": 0.39, "learning_rate": 9.610274579273694e-06, "loss": 1.2438, "step": 2200 }, { "epoch": 0.41, "learning_rate": 9.592559787422499e-06, "loss": 1.2365, "step": 2300 }, { "epoch": 0.43, "learning_rate": 9.574844995571303e-06, "loss": 1.2508, "step": 2400 }, { "epoch": 0.44, "learning_rate": 9.557130203720107e-06, "loss": 1.2316, "step": 2500 }, { "epoch": 0.46, "learning_rate": 9.539415411868912e-06, "loss": 1.2448, "step": 2600 }, { "epoch": 0.48, "learning_rate": 9.521700620017715e-06, "loss": 1.2259, "step": 2700 }, { "epoch": 0.5, "learning_rate": 9.50398582816652e-06, "loss": 1.2091, "step": 2800 }, { "epoch": 0.51, "learning_rate": 9.486271036315325e-06, "loss": 1.2292, "step": 2900 }, { "epoch": 0.53, "learning_rate": 9.46855624446413e-06, "loss": 1.2013, "step": 3000 }, { "epoch": 0.55, "learning_rate": 9.450841452612933e-06, "loss": 1.2216, "step": 3100 }, { "epoch": 0.57, "learning_rate": 9.433126660761736e-06, "loss": 1.2353, "step": 3200 }, { "epoch": 0.58, "learning_rate": 9.415411868910541e-06, "loss": 1.2189, "step": 3300 }, { "epoch": 0.6, "learning_rate": 9.397697077059346e-06, "loss": 1.2103, "step": 3400 }, { "epoch": 0.62, "learning_rate": 9.379982285208149e-06, "loss": 1.2514, "step": 3500 }, { "epoch": 0.64, "learning_rate": 9.362267493356954e-06, "loss": 1.2227, "step": 3600 }, { "epoch": 0.66, "learning_rate": 9.344552701505759e-06, "loss": 1.2287, "step": 3700 }, { "epoch": 0.67, "learning_rate": 9.326837909654562e-06, "loss": 1.2374, "step": 3800 }, { "epoch": 0.69, "learning_rate": 9.309123117803367e-06, "loss": 1.2078, "step": 3900 }, { "epoch": 0.71, "learning_rate": 9.29140832595217e-06, "loss": 1.2001, "step": 4000 }, { "epoch": 0.73, "learning_rate": 9.273693534100975e-06, "loss": 1.2325, "step": 4100 }, { "epoch": 0.74, "learning_rate": 9.25597874224978e-06, "loss": 1.2311, "step": 4200 }, { "epoch": 0.76, "learning_rate": 9.238263950398583e-06, "loss": 1.2056, "step": 4300 }, { "epoch": 0.78, "learning_rate": 9.220549158547388e-06, "loss": 1.1852, "step": 4400 }, { "epoch": 0.8, "learning_rate": 9.202834366696191e-06, "loss": 1.2417, "step": 4500 }, { "epoch": 0.81, "learning_rate": 9.185119574844996e-06, "loss": 1.2329, "step": 4600 }, { "epoch": 0.83, "learning_rate": 9.167404782993801e-06, "loss": 1.2252, "step": 4700 }, { "epoch": 0.85, "learning_rate": 9.149689991142604e-06, "loss": 1.2052, "step": 4800 }, { "epoch": 0.87, "learning_rate": 9.13197519929141e-06, "loss": 1.2238, "step": 4900 }, { "epoch": 0.89, "learning_rate": 9.114260407440213e-06, "loss": 1.2166, "step": 5000 }, { "epoch": 0.9, "learning_rate": 9.096545615589017e-06, "loss": 1.2269, "step": 5100 }, { "epoch": 0.92, "learning_rate": 9.078830823737822e-06, "loss": 1.232, "step": 5200 }, { "epoch": 0.94, "learning_rate": 9.061116031886627e-06, "loss": 1.2372, "step": 5300 }, { "epoch": 0.96, "learning_rate": 9.04340124003543e-06, "loss": 1.2281, "step": 5400 }, { "epoch": 0.97, "learning_rate": 9.025686448184234e-06, "loss": 1.2031, "step": 5500 }, { "epoch": 0.99, "learning_rate": 9.007971656333039e-06, "loss": 1.1884, "step": 5600 }, { "epoch": 1.01, "learning_rate": 8.990256864481844e-06, "loss": 1.1951, "step": 5700 }, { "epoch": 1.03, "learning_rate": 8.972542072630648e-06, "loss": 1.2428, "step": 5800 }, { "epoch": 1.05, "learning_rate": 8.954827280779452e-06, "loss": 1.2197, "step": 5900 }, { "epoch": 1.06, "learning_rate": 8.937112488928255e-06, "loss": 1.1972, "step": 6000 }, { "epoch": 1.08, "learning_rate": 8.91939769707706e-06, "loss": 1.1989, "step": 6100 }, { "epoch": 1.1, "learning_rate": 8.901682905225865e-06, "loss": 1.1922, "step": 6200 }, { "epoch": 1.12, "learning_rate": 8.883968113374668e-06, "loss": 1.2149, "step": 6300 }, { "epoch": 1.13, "learning_rate": 8.866253321523473e-06, "loss": 1.1956, "step": 6400 }, { "epoch": 1.15, "learning_rate": 8.848538529672278e-06, "loss": 1.2049, "step": 6500 }, { "epoch": 1.17, "learning_rate": 8.830823737821081e-06, "loss": 1.2201, "step": 6600 }, { "epoch": 1.19, "learning_rate": 8.813108945969886e-06, "loss": 1.206, "step": 6700 }, { "epoch": 1.2, "learning_rate": 8.795394154118689e-06, "loss": 1.2079, "step": 6800 }, { "epoch": 1.22, "learning_rate": 8.777679362267494e-06, "loss": 1.2033, "step": 6900 }, { "epoch": 1.24, "learning_rate": 8.759964570416299e-06, "loss": 1.1853, "step": 7000 }, { "epoch": 1.26, "learning_rate": 8.742249778565102e-06, "loss": 1.1822, "step": 7100 }, { "epoch": 1.28, "learning_rate": 8.724534986713907e-06, "loss": 1.1834, "step": 7200 }, { "epoch": 1.29, "learning_rate": 8.70682019486271e-06, "loss": 1.2146, "step": 7300 }, { "epoch": 1.31, "learning_rate": 8.689105403011515e-06, "loss": 1.1983, "step": 7400 }, { "epoch": 1.33, "learning_rate": 8.67139061116032e-06, "loss": 1.2084, "step": 7500 }, { "epoch": 1.35, "learning_rate": 8.653675819309125e-06, "loss": 1.2055, "step": 7600 }, { "epoch": 1.36, "learning_rate": 8.635961027457928e-06, "loss": 1.2199, "step": 7700 }, { "epoch": 1.38, "learning_rate": 8.618246235606731e-06, "loss": 1.1745, "step": 7800 }, { "epoch": 1.4, "learning_rate": 8.600531443755536e-06, "loss": 1.195, "step": 7900 }, { "epoch": 1.42, "learning_rate": 8.582816651904341e-06, "loss": 1.2223, "step": 8000 }, { "epoch": 1.43, "learning_rate": 8.565101860053146e-06, "loss": 1.1919, "step": 8100 }, { "epoch": 1.45, "learning_rate": 8.54738706820195e-06, "loss": 1.2043, "step": 8200 }, { "epoch": 1.47, "learning_rate": 8.529672276350753e-06, "loss": 1.2102, "step": 8300 }, { "epoch": 1.49, "learning_rate": 8.511957484499558e-06, "loss": 1.1839, "step": 8400 }, { "epoch": 1.51, "learning_rate": 8.494242692648362e-06, "loss": 1.1889, "step": 8500 }, { "epoch": 1.52, "learning_rate": 8.476527900797167e-06, "loss": 1.1858, "step": 8600 }, { "epoch": 1.54, "learning_rate": 8.45881310894597e-06, "loss": 1.1706, "step": 8700 }, { "epoch": 1.56, "learning_rate": 8.441098317094775e-06, "loss": 1.2071, "step": 8800 }, { "epoch": 1.58, "learning_rate": 8.423383525243579e-06, "loss": 1.1671, "step": 8900 }, { "epoch": 1.59, "learning_rate": 8.405668733392384e-06, "loss": 1.1837, "step": 9000 }, { "epoch": 1.61, "learning_rate": 8.387953941541187e-06, "loss": 1.1627, "step": 9100 }, { "epoch": 1.63, "learning_rate": 8.370239149689992e-06, "loss": 1.2027, "step": 9200 }, { "epoch": 1.65, "learning_rate": 8.352524357838797e-06, "loss": 1.1831, "step": 9300 }, { "epoch": 1.67, "learning_rate": 8.3348095659876e-06, "loss": 1.1941, "step": 9400 }, { "epoch": 1.68, "learning_rate": 8.317094774136405e-06, "loss": 1.2047, "step": 9500 }, { "epoch": 1.7, "learning_rate": 8.299379982285208e-06, "loss": 1.2045, "step": 9600 }, { "epoch": 1.72, "learning_rate": 8.281665190434013e-06, "loss": 1.2022, "step": 9700 }, { "epoch": 1.74, "learning_rate": 8.263950398582818e-06, "loss": 1.1934, "step": 9800 }, { "epoch": 1.75, "learning_rate": 8.246235606731621e-06, "loss": 1.1672, "step": 9900 }, { "epoch": 1.77, "learning_rate": 8.228520814880426e-06, "loss": 1.1802, "step": 10000 }, { "epoch": 1.79, "learning_rate": 8.210806023029229e-06, "loss": 1.1933, "step": 10100 }, { "epoch": 1.81, "learning_rate": 8.193091231178034e-06, "loss": 1.205, "step": 10200 }, { "epoch": 1.82, "learning_rate": 8.175376439326839e-06, "loss": 1.1844, "step": 10300 }, { "epoch": 1.84, "learning_rate": 8.157661647475644e-06, "loss": 1.197, "step": 10400 }, { "epoch": 1.86, "learning_rate": 8.139946855624447e-06, "loss": 1.1978, "step": 10500 }, { "epoch": 1.88, "learning_rate": 8.12223206377325e-06, "loss": 1.2017, "step": 10600 }, { "epoch": 1.9, "learning_rate": 8.104517271922055e-06, "loss": 1.2067, "step": 10700 }, { "epoch": 1.91, "learning_rate": 8.08680248007086e-06, "loss": 1.2042, "step": 10800 }, { "epoch": 1.93, "learning_rate": 8.069087688219665e-06, "loss": 1.206, "step": 10900 }, { "epoch": 1.95, "learning_rate": 8.051372896368468e-06, "loss": 1.1987, "step": 11000 }, { "epoch": 1.97, "learning_rate": 8.033658104517273e-06, "loss": 1.1688, "step": 11100 }, { "epoch": 1.98, "learning_rate": 8.015943312666076e-06, "loss": 1.1723, "step": 11200 }, { "epoch": 2.0, "learning_rate": 7.998228520814881e-06, "loss": 1.1676, "step": 11300 }, { "epoch": 2.02, "learning_rate": 7.980513728963686e-06, "loss": 1.1638, "step": 11400 }, { "epoch": 2.04, "learning_rate": 7.96279893711249e-06, "loss": 1.2048, "step": 11500 }, { "epoch": 2.05, "learning_rate": 7.945084145261294e-06, "loss": 1.1705, "step": 11600 }, { "epoch": 2.07, "learning_rate": 7.927369353410098e-06, "loss": 1.1734, "step": 11700 }, { "epoch": 2.09, "learning_rate": 7.909654561558902e-06, "loss": 1.1781, "step": 11800 }, { "epoch": 2.11, "learning_rate": 7.891939769707706e-06, "loss": 1.1313, "step": 11900 }, { "epoch": 2.13, "learning_rate": 7.87422497785651e-06, "loss": 1.1673, "step": 12000 }, { "epoch": 2.14, "learning_rate": 7.856510186005316e-06, "loss": 1.1926, "step": 12100 }, { "epoch": 2.16, "learning_rate": 7.838795394154119e-06, "loss": 1.1832, "step": 12200 }, { "epoch": 2.18, "learning_rate": 7.821080602302924e-06, "loss": 1.1921, "step": 12300 }, { "epoch": 2.2, "learning_rate": 7.803365810451727e-06, "loss": 1.1869, "step": 12400 }, { "epoch": 2.21, "learning_rate": 7.785651018600532e-06, "loss": 1.1816, "step": 12500 }, { "epoch": 2.23, "learning_rate": 7.767936226749337e-06, "loss": 1.1784, "step": 12600 }, { "epoch": 2.25, "learning_rate": 7.750221434898142e-06, "loss": 1.1809, "step": 12700 }, { "epoch": 2.27, "learning_rate": 7.732506643046945e-06, "loss": 1.1713, "step": 12800 }, { "epoch": 2.29, "learning_rate": 7.714791851195748e-06, "loss": 1.178, "step": 12900 }, { "epoch": 2.3, "learning_rate": 7.697077059344553e-06, "loss": 1.1915, "step": 13000 }, { "epoch": 2.32, "learning_rate": 7.679362267493358e-06, "loss": 1.1814, "step": 13100 }, { "epoch": 2.34, "learning_rate": 7.661647475642163e-06, "loss": 1.1598, "step": 13200 }, { "epoch": 2.36, "learning_rate": 7.643932683790966e-06, "loss": 1.192, "step": 13300 }, { "epoch": 2.37, "learning_rate": 7.62621789193977e-06, "loss": 1.1932, "step": 13400 }, { "epoch": 2.39, "learning_rate": 7.608503100088574e-06, "loss": 1.1586, "step": 13500 }, { "epoch": 2.41, "learning_rate": 7.590788308237379e-06, "loss": 1.1614, "step": 13600 }, { "epoch": 2.43, "learning_rate": 7.573073516386183e-06, "loss": 1.1527, "step": 13700 }, { "epoch": 2.44, "learning_rate": 7.555358724534987e-06, "loss": 1.181, "step": 13800 }, { "epoch": 2.46, "learning_rate": 7.537643932683791e-06, "loss": 1.1306, "step": 13900 }, { "epoch": 2.48, "learning_rate": 7.519929140832596e-06, "loss": 1.1611, "step": 14000 }, { "epoch": 2.5, "learning_rate": 7.5022143489814e-06, "loss": 1.1703, "step": 14100 }, { "epoch": 2.52, "learning_rate": 7.484499557130205e-06, "loss": 1.1642, "step": 14200 }, { "epoch": 2.53, "learning_rate": 7.466784765279008e-06, "loss": 1.1285, "step": 14300 }, { "epoch": 2.55, "learning_rate": 7.449069973427812e-06, "loss": 1.1627, "step": 14400 }, { "epoch": 2.57, "learning_rate": 7.431355181576617e-06, "loss": 1.1667, "step": 14500 }, { "epoch": 2.59, "learning_rate": 7.413640389725421e-06, "loss": 1.162, "step": 14600 }, { "epoch": 2.6, "learning_rate": 7.3959255978742254e-06, "loss": 1.1835, "step": 14700 }, { "epoch": 2.62, "learning_rate": 7.3782108060230295e-06, "loss": 1.1697, "step": 14800 }, { "epoch": 2.64, "learning_rate": 7.3604960141718344e-06, "loss": 1.18, "step": 14900 }, { "epoch": 2.66, "learning_rate": 7.3427812223206385e-06, "loss": 1.1594, "step": 15000 }, { "epoch": 2.67, "learning_rate": 7.325066430469443e-06, "loss": 1.165, "step": 15100 }, { "epoch": 2.69, "learning_rate": 7.307351638618247e-06, "loss": 1.1598, "step": 15200 }, { "epoch": 2.71, "learning_rate": 7.289636846767051e-06, "loss": 1.1412, "step": 15300 }, { "epoch": 2.73, "learning_rate": 7.271922054915856e-06, "loss": 1.1745, "step": 15400 }, { "epoch": 2.75, "learning_rate": 7.25420726306466e-06, "loss": 1.1961, "step": 15500 }, { "epoch": 2.76, "learning_rate": 7.2364924712134646e-06, "loss": 1.1557, "step": 15600 }, { "epoch": 2.78, "learning_rate": 7.218777679362268e-06, "loss": 1.171, "step": 15700 }, { "epoch": 2.8, "learning_rate": 7.201062887511072e-06, "loss": 1.146, "step": 15800 }, { "epoch": 2.82, "learning_rate": 7.183348095659877e-06, "loss": 1.157, "step": 15900 }, { "epoch": 2.83, "learning_rate": 7.165633303808681e-06, "loss": 1.1622, "step": 16000 }, { "epoch": 2.85, "learning_rate": 7.147918511957485e-06, "loss": 1.1661, "step": 16100 }, { "epoch": 2.87, "learning_rate": 7.130203720106289e-06, "loss": 1.1694, "step": 16200 }, { "epoch": 2.89, "learning_rate": 7.112488928255094e-06, "loss": 1.1689, "step": 16300 }, { "epoch": 2.91, "learning_rate": 7.094774136403898e-06, "loss": 1.1394, "step": 16400 }, { "epoch": 2.92, "learning_rate": 7.077059344552703e-06, "loss": 1.137, "step": 16500 }, { "epoch": 2.94, "learning_rate": 7.059344552701506e-06, "loss": 1.1639, "step": 16600 }, { "epoch": 2.96, "learning_rate": 7.04162976085031e-06, "loss": 1.1545, "step": 16700 }, { "epoch": 2.98, "learning_rate": 7.023914968999115e-06, "loss": 1.1742, "step": 16800 }, { "epoch": 2.99, "learning_rate": 7.006200177147919e-06, "loss": 1.161, "step": 16900 }, { "epoch": 3.01, "learning_rate": 6.988485385296724e-06, "loss": 1.1506, "step": 17000 }, { "epoch": 3.03, "learning_rate": 6.970770593445527e-06, "loss": 1.154, "step": 17100 }, { "epoch": 3.05, "learning_rate": 6.953055801594331e-06, "loss": 1.1414, "step": 17200 }, { "epoch": 3.06, "learning_rate": 6.935341009743136e-06, "loss": 1.1289, "step": 17300 }, { "epoch": 3.08, "learning_rate": 6.91762621789194e-06, "loss": 1.1664, "step": 17400 }, { "epoch": 3.1, "learning_rate": 6.899911426040744e-06, "loss": 1.1751, "step": 17500 }, { "epoch": 3.12, "learning_rate": 6.882196634189548e-06, "loss": 1.1514, "step": 17600 }, { "epoch": 3.14, "learning_rate": 6.864481842338353e-06, "loss": 1.1219, "step": 17700 }, { "epoch": 3.15, "learning_rate": 6.846767050487157e-06, "loss": 1.1343, "step": 17800 }, { "epoch": 3.17, "learning_rate": 6.829052258635962e-06, "loss": 1.1209, "step": 17900 }, { "epoch": 3.19, "learning_rate": 6.8113374667847655e-06, "loss": 1.1612, "step": 18000 }, { "epoch": 3.21, "learning_rate": 6.79362267493357e-06, "loss": 1.1427, "step": 18100 }, { "epoch": 3.22, "learning_rate": 6.7759078830823745e-06, "loss": 1.1621, "step": 18200 }, { "epoch": 3.24, "learning_rate": 6.7581930912311786e-06, "loss": 1.1434, "step": 18300 }, { "epoch": 3.26, "learning_rate": 6.7404782993799835e-06, "loss": 1.1324, "step": 18400 }, { "epoch": 3.28, "learning_rate": 6.722763507528787e-06, "loss": 1.1566, "step": 18500 }, { "epoch": 3.29, "learning_rate": 6.705048715677592e-06, "loss": 1.1393, "step": 18600 }, { "epoch": 3.31, "learning_rate": 6.687333923826396e-06, "loss": 1.1642, "step": 18700 }, { "epoch": 3.33, "learning_rate": 6.6696191319752006e-06, "loss": 1.1419, "step": 18800 }, { "epoch": 3.35, "learning_rate": 6.651904340124004e-06, "loss": 1.1293, "step": 18900 }, { "epoch": 3.37, "learning_rate": 6.634189548272808e-06, "loss": 1.1467, "step": 19000 }, { "epoch": 3.38, "learning_rate": 6.616474756421613e-06, "loss": 1.1619, "step": 19100 }, { "epoch": 3.4, "learning_rate": 6.598759964570417e-06, "loss": 1.1566, "step": 19200 }, { "epoch": 3.42, "learning_rate": 6.581045172719222e-06, "loss": 1.1741, "step": 19300 }, { "epoch": 3.44, "learning_rate": 6.563330380868025e-06, "loss": 1.1763, "step": 19400 }, { "epoch": 3.45, "learning_rate": 6.545615589016829e-06, "loss": 1.1628, "step": 19500 }, { "epoch": 3.47, "learning_rate": 6.527900797165634e-06, "loss": 1.1585, "step": 19600 }, { "epoch": 3.49, "learning_rate": 6.510186005314438e-06, "loss": 1.1546, "step": 19700 }, { "epoch": 3.51, "learning_rate": 6.492471213463243e-06, "loss": 1.1684, "step": 19800 }, { "epoch": 3.53, "learning_rate": 6.474756421612046e-06, "loss": 1.1274, "step": 19900 }, { "epoch": 3.54, "learning_rate": 6.457041629760851e-06, "loss": 1.1432, "step": 20000 }, { "epoch": 3.56, "learning_rate": 6.439326837909655e-06, "loss": 1.1391, "step": 20100 }, { "epoch": 3.58, "learning_rate": 6.42161204605846e-06, "loss": 1.154, "step": 20200 }, { "epoch": 3.6, "learning_rate": 6.403897254207263e-06, "loss": 1.128, "step": 20300 }, { "epoch": 3.61, "learning_rate": 6.386182462356067e-06, "loss": 1.1623, "step": 20400 }, { "epoch": 3.63, "learning_rate": 6.368467670504872e-06, "loss": 1.1661, "step": 20500 }, { "epoch": 3.65, "learning_rate": 6.350752878653676e-06, "loss": 1.1328, "step": 20600 }, { "epoch": 3.67, "learning_rate": 6.333038086802481e-06, "loss": 1.1457, "step": 20700 }, { "epoch": 3.68, "learning_rate": 6.315323294951284e-06, "loss": 1.1347, "step": 20800 }, { "epoch": 3.7, "learning_rate": 6.2976085031000885e-06, "loss": 1.1434, "step": 20900 }, { "epoch": 3.72, "learning_rate": 6.279893711248893e-06, "loss": 1.1342, "step": 21000 }, { "epoch": 3.74, "learning_rate": 6.2621789193976975e-06, "loss": 1.1377, "step": 21100 }, { "epoch": 3.76, "learning_rate": 6.244464127546502e-06, "loss": 1.1321, "step": 21200 }, { "epoch": 3.77, "learning_rate": 6.226749335695306e-06, "loss": 1.138, "step": 21300 }, { "epoch": 3.79, "learning_rate": 6.2090345438441105e-06, "loss": 1.1458, "step": 21400 }, { "epoch": 3.81, "learning_rate": 6.1913197519929146e-06, "loss": 1.1134, "step": 21500 }, { "epoch": 3.83, "learning_rate": 6.1736049601417195e-06, "loss": 1.1179, "step": 21600 }, { "epoch": 3.84, "learning_rate": 6.155890168290523e-06, "loss": 1.1396, "step": 21700 }, { "epoch": 3.86, "learning_rate": 6.138175376439327e-06, "loss": 1.122, "step": 21800 }, { "epoch": 3.88, "learning_rate": 6.120460584588132e-06, "loss": 1.1576, "step": 21900 }, { "epoch": 3.9, "learning_rate": 6.102745792736936e-06, "loss": 1.1513, "step": 22000 }, { "epoch": 3.91, "learning_rate": 6.085031000885741e-06, "loss": 1.14, "step": 22100 }, { "epoch": 3.93, "learning_rate": 6.067316209034544e-06, "loss": 1.1657, "step": 22200 }, { "epoch": 3.95, "learning_rate": 6.049601417183349e-06, "loss": 1.1452, "step": 22300 }, { "epoch": 3.97, "learning_rate": 6.031886625332153e-06, "loss": 1.1379, "step": 22400 }, { "epoch": 3.99, "learning_rate": 6.014171833480958e-06, "loss": 1.1167, "step": 22500 }, { "epoch": 4.0, "learning_rate": 5.996457041629762e-06, "loss": 1.1221, "step": 22600 }, { "epoch": 4.02, "learning_rate": 5.978742249778565e-06, "loss": 1.1274, "step": 22700 }, { "epoch": 4.04, "learning_rate": 5.96102745792737e-06, "loss": 1.1267, "step": 22800 }, { "epoch": 4.06, "learning_rate": 5.943312666076174e-06, "loss": 1.1295, "step": 22900 }, { "epoch": 4.07, "learning_rate": 5.925597874224979e-06, "loss": 1.1073, "step": 23000 } ], "max_steps": 56450, "num_train_epochs": 10, "total_flos": 4.338858046644173e+17, "trial_name": null, "trial_params": null }