{"event": "baseline", "step": 0, "val_loss": 8.86757787443655}
{"event": "train", "step": 1, "loss": 8.862750053405762, "lr": 5.999999999999999e-06, "seconds": 1.48, "steps_per_second": 0.674}
{"event": "train", "step": 25, "loss": 7.909865856170654, "lr": 0.00015, "seconds": 3.21, "steps_per_second": 7.777}
{"event": "train", "step": 50, "loss": 6.686218738555908, "lr": 0.0003, "seconds": 4.93, "steps_per_second": 10.151}
{"event": "train", "step": 75, "loss": 6.235850811004639, "lr": 0.00045, "seconds": 6.66, "steps_per_second": 11.263}
{"event": "train", "step": 100, "loss": 6.117806434631348, "lr": 0.0006, "seconds": 8.36, "steps_per_second": 11.962}
{"event": "train", "step": 125, "loss": 6.043346405029297, "lr": 0.0005999774067082243, "seconds": 10.08, "steps_per_second": 12.4}
{"event": "train", "step": 150, "loss": 6.012631416320801, "lr": 0.0005999096308364803, "seconds": 11.81, "steps_per_second": 12.702}
{"event": "train", "step": 175, "loss": 5.991415500640869, "lr": 0.0005997966843948076, "seconds": 13.64, "steps_per_second": 12.835}
{"event": "train", "step": 200, "loss": 5.933244705200195, "lr": 0.0005996385873975743, "seconds": 15.34, "steps_per_second": 13.034}
{"event": "train", "step": 225, "loss": 5.874759197235107, "lr": 0.0005994353678599297, "seconds": 17.23, "steps_per_second": 13.062}
{"event": "train", "step": 250, "loss": 5.748709201812744, "lr": 0.0005991870617928412, "seconds": 18.93, "steps_per_second": 13.204}
{"event": "train", "step": 275, "loss": 5.781266212463379, "lr": 0.0005988937131967122, "seconds": 20.72, "steps_per_second": 13.272}
{"event": "train", "step": 300, "loss": 5.764250755310059, "lr": 0.0005985553740535849, "seconds": 22.51, "steps_per_second": 13.325}
{"event": "validation", "step": 300, "val_loss": 5.7249987501408555, "best_val": 5.7249987501408555, "peak_process_rss_mib": 600.4, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 325, "loss": 5.69815731048584, "lr": 0.0005981721043179301, "seconds": 24.92, "steps_per_second": 13.044}
{"event": "train", "step": 350, "loss": 5.618410110473633, "lr": 0.0005977439719060218, "seconds": 26.74, "steps_per_second": 13.088}
{"event": "train", "step": 375, "loss": 5.561914920806885, "lr": 0.0005972710526839033, "seconds": 28.54, "steps_per_second": 13.138}
{"event": "train", "step": 400, "loss": 5.442761421203613, "lr": 0.0005967534304539428, "seconds": 30.28, "steps_per_second": 13.212}
{"event": "train", "step": 425, "loss": 5.444773197174072, "lr": 0.0005961911969399842, "seconds": 32.1, "steps_per_second": 13.241}
{"event": "train", "step": 450, "loss": 5.470274925231934, "lr": 0.0005955844517710928, "seconds": 33.82, "steps_per_second": 13.305}
{"event": "train", "step": 475, "loss": 5.434249401092529, "lr": 0.0005949333024639009, "seconds": 35.61, "steps_per_second": 13.34}
{"event": "train", "step": 500, "loss": 5.4980010986328125, "lr": 0.000594237864403556, "seconds": 37.36, "steps_per_second": 13.384}
{"event": "train", "step": 525, "loss": 5.477673530578613, "lr": 0.0005934982608232736, "seconds": 39.16, "steps_per_second": 13.407}
{"event": "train", "step": 550, "loss": 5.306995868682861, "lr": 0.000592714622782501, "seconds": 40.98, "steps_per_second": 13.42}
{"event": "train", "step": 575, "loss": 5.3248724937438965, "lr": 0.0005918870891436921, "seconds": 42.79, "steps_per_second": 13.438}
{"event": "train", "step": 600, "loss": 5.230597972869873, "lr": 0.0005910158065477015, "seconds": 44.49, "steps_per_second": 13.488}
{"event": "validation", "step": 600, "val_loss": 5.328562572774224, "best_val": 5.328562572774224, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 625, "loss": 5.38937520980835, "lr": 0.0005901009293877991, "seconds": 46.89, "steps_per_second": 13.328}
{"event": "train", "step": 650, "loss": 5.27610445022583, "lr": 0.0005891426197823113, "seconds": 48.7, "steps_per_second": 13.347}
{"event": "train", "step": 675, "loss": 5.286613941192627, "lr": 0.0005881410475458933, "seconds": 50.46, "steps_per_second": 13.377}
{"event": "train", "step": 700, "loss": 5.363029479980469, "lr": 0.0005870963901594372, "seconds": 52.26, "steps_per_second": 13.393}
{"event": "train", "step": 725, "loss": 5.274033069610596, "lr": 0.0005860088327386227, "seconds": 54.06, "steps_per_second": 13.41}
{"event": "train", "step": 750, "loss": 5.218993186950684, "lr": 0.0005848785680011135, "seconds": 55.78, "steps_per_second": 13.445}
{"event": "train", "step": 775, "loss": 5.185213565826416, "lr": 0.0005837057962324076, "seconds": 57.61, "steps_per_second": 13.453}
{"event": "train", "step": 800, "loss": 5.204285144805908, "lr": 0.0005824907252503463, "seconds": 59.35, "steps_per_second": 13.48}
{"event": "train", "step": 825, "loss": 5.17167854309082, "lr": 0.0005812335703682881, "seconds": 61.17, "steps_per_second": 13.486}
{"event": "train", "step": 850, "loss": 5.144318103790283, "lr": 0.0005799345543569552, "seconds": 62.91, "steps_per_second": 13.511}
{"event": "train", "step": 875, "loss": 5.14930534362793, "lr": 0.0005785939074049575, "seconds": 64.72, "steps_per_second": 13.52}
{"event": "train", "step": 900, "loss": 5.408031940460205, "lr": 0.0005772118670780028, "seconds": 66.58, "steps_per_second": 13.517}
{"event": "validation", "step": 900, "val_loss": 5.152780435136334, "best_val": 5.152780435136334, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 925, "loss": 5.2644524574279785, "lr": 0.0005757886782767998, "seconds": 69.01, "steps_per_second": 13.403}
{"event": "train", "step": 950, "loss": 5.243491172790527, "lr": 0.0005743245931936608, "seconds": 70.85, "steps_per_second": 13.409}
{"event": "train", "step": 975, "loss": 5.210367202758789, "lr": 0.0005728198712678132, "seconds": 72.69, "steps_per_second": 13.414}
{"event": "train", "step": 1000, "loss": 5.228847503662109, "lr": 0.0005712747791394257, "seconds": 74.49, "steps_per_second": 13.425}
{"event": "train", "step": 1025, "loss": 5.2135162353515625, "lr": 0.0005696895906023592, "seconds": 76.31, "steps_per_second": 13.431}
{"event": "train", "step": 1050, "loss": 5.1477227210998535, "lr": 0.00056806458655565, "seconds": 78.12, "steps_per_second": 13.441}
{"event": "train", "step": 1075, "loss": 5.0525689125061035, "lr": 0.0005664000549537343, "seconds": 79.88, "steps_per_second": 13.458}
{"event": "train", "step": 1100, "loss": 5.024498462677002, "lr": 0.0005646962907554204, "seconds": 81.72, "steps_per_second": 13.46}
{"event": "train", "step": 1125, "loss": 5.086690425872803, "lr": 0.000562953595871624, "seconds": 83.52, "steps_per_second": 13.471}
{"event": "train", "step": 1150, "loss": 5.081254959106445, "lr": 0.0005611722791118661, "seconds": 85.31, "steps_per_second": 13.481}
{"event": "train", "step": 1175, "loss": 5.041465759277344, "lr": 0.000559352656129553, "seconds": 87.12, "steps_per_second": 13.487}
{"event": "train", "step": 1200, "loss": 5.020495414733887, "lr": 0.0005574950493660413, "seconds": 88.91, "steps_per_second": 13.498}
{"event": "validation", "step": 1200, "val_loss": 5.0133851896263515, "best_val": 5.0133851896263515, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 1225, "loss": 4.8674116134643555, "lr": 0.0005555997879935005, "seconds": 91.43, "steps_per_second": 13.398}
{"event": "train", "step": 1250, "loss": 4.93159294128418, "lr": 0.0005536672078565831, "seconds": 93.25, "steps_per_second": 13.405}
{"event": "train", "step": 1275, "loss": 5.009015083312988, "lr": 0.0005516976514129118, "seconds": 95.05, "steps_per_second": 13.414}
{"event": "train", "step": 1300, "loss": 5.015984058380127, "lr": 0.0005496914676723956, "seconds": 96.88, "steps_per_second": 13.418}
{"event": "train", "step": 1325, "loss": 4.8693928718566895, "lr": 0.0005476490121353846, "seconds": 98.7, "steps_per_second": 13.424}
{"event": "train", "step": 1350, "loss": 5.036653518676758, "lr": 0.0005455706467296738, "seconds": 100.49, "steps_per_second": 13.435}
{"event": "train", "step": 1375, "loss": 5.005246162414551, "lr": 0.0005434567397463693, "seconds": 102.3, "steps_per_second": 13.44}
{"event": "train", "step": 1400, "loss": 4.975721836090088, "lr": 0.0005413076657746261, "seconds": 104.11, "steps_per_second": 13.448}
{"event": "train", "step": 1425, "loss": 4.826275825500488, "lr": 0.0005391238056352699, "seconds": 105.9, "steps_per_second": 13.456}
{"event": "train", "step": 1450, "loss": 5.026703834533691, "lr": 0.0005369055463133153, "seconds": 107.65, "steps_per_second": 13.469}
{"event": "train", "step": 1475, "loss": 4.9742960929870605, "lr": 0.0005346532808893903, "seconds": 109.36, "steps_per_second": 13.487}
{"event": "train", "step": 1500, "loss": 5.001156806945801, "lr": 0.000532367408470082, "seconds": 111.09, "steps_per_second": 13.502}
{"event": "validation", "step": 1500, "val_loss": 4.907590739432252, "best_val": 4.907590739432252, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 1525, "loss": 4.979414939880371, "lr": 0.0005300483341172146, "seconds": 113.52, "steps_per_second": 13.434}
{"event": "train", "step": 1550, "loss": 4.917886734008789, "lr": 0.0005276964687760703, "seconds": 115.27, "steps_per_second": 13.447}
{"event": "train", "step": 1575, "loss": 4.851538181304932, "lr": 0.0005253122292025696, "seconds": 117.1, "steps_per_second": 13.45}
{"event": "train", "step": 1600, "loss": 4.8470916748046875, "lr": 0.0005228960378894211, "seconds": 118.82, "steps_per_second": 13.465}
{"event": "train", "step": 1625, "loss": 4.938422203063965, "lr": 0.0005204483229912548, "seconds": 120.53, "steps_per_second": 13.482}
{"event": "train", "step": 1650, "loss": 4.8419389724731445, "lr": 0.000517969518248752, "seconds": 122.32, "steps_per_second": 13.489}
{"event": "train", "step": 1675, "loss": 4.926115036010742, "lr": 0.0005154600629117859, "seconds": 124.02, "steps_per_second": 13.506}
{"event": "train", "step": 1700, "loss": 4.82956075668335, "lr": 0.0005129204016615848, "seconds": 125.81, "steps_per_second": 13.513}
{"event": "train", "step": 1725, "loss": 4.911472320556641, "lr": 0.0005103509845319341, "seconds": 127.61, "steps_per_second": 13.518}
{"event": "train", "step": 1750, "loss": 4.7536211013793945, "lr": 0.0005077522668294291, "seconds": 129.43, "steps_per_second": 13.521}
{"event": "train", "step": 1775, "loss": 4.8346147537231445, "lr": 0.0005051247090527937, "seconds": 131.22, "steps_per_second": 13.527}
{"event": "train", "step": 1800, "loss": 4.825801372528076, "lr": 0.0005024687768112789, "seconds": 132.95, "steps_per_second": 13.539}
{"event": "validation", "step": 1800, "val_loss": 4.818951788611056, "best_val": 4.818951788611056, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 1825, "loss": 4.760134220123291, "lr": 0.0004997849407421559, "seconds": 135.25, "steps_per_second": 13.493}
{"event": "train", "step": 1850, "loss": 4.76740837097168, "lr": 0.0004970736764273178, "seconds": 137.08, "steps_per_second": 13.496}
{"event": "train", "step": 1875, "loss": 4.8198041915893555, "lr": 0.0004943354643090059, "seconds": 138.8, "steps_per_second": 13.509}
{"event": "train", "step": 1900, "loss": 4.680688858032227, "lr": 0.0004915707896046737, "seconds": 140.58, "steps_per_second": 13.516}
{"event": "train", "step": 1925, "loss": 4.790152549743652, "lr": 0.0004887801422210057, "seconds": 142.57, "steps_per_second": 13.502}
{"event": "train", "step": 1950, "loss": 4.766994953155518, "lr": 0.00048596401666710425, "seconds": 144.53, "steps_per_second": 13.492}
{"event": "train", "step": 1975, "loss": 4.795710563659668, "lr": 0.0004831229119668623, "seconds": 146.53, "steps_per_second": 13.478}
{"event": "train", "step": 2000, "loss": 4.698494911193848, "lr": 0.0004802573315705343, "seconds": 148.39, "steps_per_second": 13.478}
{"event": "train", "step": 2025, "loss": 4.633999824523926, "lr": 0.0004773677832655242, "seconds": 150.23, "steps_per_second": 13.48}
{"event": "train", "step": 2050, "loss": 4.817287921905518, "lr": 0.00047445477908640414, "seconds": 152.06, "steps_per_second": 13.482}
{"event": "train", "step": 2075, "loss": 4.6998515129089355, "lr": 0.00047151883522418103, "seconds": 153.9, "steps_per_second": 13.483}
{"event": "train", "step": 2100, "loss": 4.736083507537842, "lr": 0.00046856047193482577, "seconds": 155.87, "steps_per_second": 13.473}
{"event": "validation", "step": 2100, "val_loss": 4.743831286632426, "best_val": 4.743831286632426, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 2125, "loss": 4.749931335449219, "lr": 0.00046558021344708347, "seconds": 158.46, "steps_per_second": 13.41}
{"event": "train", "step": 2150, "loss": 4.645644187927246, "lr": 0.00046257858786957805, "seconds": 160.45, "steps_per_second": 13.4}
{"event": "train", "step": 2175, "loss": 4.8218770027160645, "lr": 0.0004595561270972311, "seconds": 162.45, "steps_per_second": 13.389}
{"event": "train", "step": 2200, "loss": 4.7491068840026855, "lr": 0.00045651336671700813, "seconds": 164.45, "steps_per_second": 13.378}
{"event": "train", "step": 2225, "loss": 4.56228494644165, "lr": 0.0004534508459130116, "seconds": 166.5, "steps_per_second": 13.363}
{"event": "train", "step": 2250, "loss": 4.753208160400391, "lr": 0.0004503691073709364, "seconds": 168.44, "steps_per_second": 13.358}
{"event": "train", "step": 2275, "loss": 4.745820999145508, "lr": 0.0004472686971819045, "seconds": 170.36, "steps_per_second": 13.354}
{"event": "train", "step": 2300, "loss": 4.77801513671875, "lr": 0.0004441501647456968, "seconds": 172.23, "steps_per_second": 13.355}
{"event": "train", "step": 2325, "loss": 4.556344985961914, "lr": 0.00044101406267339706, "seconds": 174.21, "steps_per_second": 13.346}
{"event": "train", "step": 2350, "loss": 4.711927890777588, "lr": 0.00043786094668946964, "seconds": 176.23, "steps_per_second": 13.335}
{"event": "train", "step": 2375, "loss": 4.769251346588135, "lr": 0.0004346913755332824, "seconds": 178.2, "steps_per_second": 13.328}
{"event": "train", "step": 2400, "loss": 4.575535297393799, "lr": 0.00043150591086009763, "seconds": 180.2, "steps_per_second": 13.319}
{"event": "validation", "step": 2400, "val_loss": 4.684960592360723, "best_val": 4.684960592360723, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 2425, "loss": 4.65748929977417, "lr": 0.00042830511714154524, "seconds": 182.73, "steps_per_second": 13.271}
{"event": "train", "step": 2450, "loss": 4.526055335998535, "lr": 0.00042508956156559694, "seconds": 184.65, "steps_per_second": 13.268}
{"event": "train", "step": 2475, "loss": 4.757646560668945, "lr": 0.0004218598139360593, "seconds": 186.71, "steps_per_second": 13.256}
{"event": "train", "step": 2500, "loss": 4.677148818969727, "lr": 0.0004186164465716033, "seconds": 188.87, "steps_per_second": 13.237}
{"event": "train", "step": 2525, "loss": 4.677950382232666, "lr": 0.0004153600342043483, "seconds": 191.11, "steps_per_second": 13.213}
{"event": "train", "step": 2550, "loss": 4.576961517333984, "lr": 0.00041209115387801787, "seconds": 193.25, "steps_per_second": 13.196}
{"event": "train", "step": 2575, "loss": 4.767956733703613, "lr": 0.0004088103848456864, "seconds": 195.24, "steps_per_second": 13.189}
{"event": "train", "step": 2600, "loss": 4.683537483215332, "lr": 0.00040551830846713427, "seconds": 197.29, "steps_per_second": 13.178}
{"event": "train", "step": 2625, "loss": 4.717271327972412, "lr": 0.0004022155081058295, "seconds": 199.28, "steps_per_second": 13.173}
{"event": "train", "step": 2650, "loss": 4.718204021453857, "lr": 0.00039890256902555413, "seconds": 201.41, "steps_per_second": 13.157}
{"event": "train", "step": 2675, "loss": 4.665269374847412, "lr": 0.00039558007828669503, "seconds": 203.59, "steps_per_second": 13.139}
{"event": "train", "step": 2700, "loss": 4.6568098068237305, "lr": 0.00039224862464221414, "seconds": 205.75, "steps_per_second": 13.122}
{"event": "validation", "step": 2700, "val_loss": 4.62970750057942, "best_val": 4.62970750057942, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 2725, "loss": 4.566028594970703, "lr": 0.00038890879843332123, "seconds": 208.45, "steps_per_second": 13.073}
{"event": "train", "step": 2750, "loss": 4.409126281738281, "lr": 0.0003855611914848633, "seconds": 210.59, "steps_per_second": 13.059}
{"event": "train", "step": 2775, "loss": 4.626798629760742, "lr": 0.0003822063970004527, "seconds": 212.84, "steps_per_second": 13.038}
{"event": "train", "step": 2800, "loss": 4.413450717926025, "lr": 0.00037884500945734907, "seconds": 215.09, "steps_per_second": 13.018}
{"event": "train", "step": 2825, "loss": 4.668604373931885, "lr": 0.00037547762450111747, "seconds": 217.21, "steps_per_second": 13.006}
{"event": "train", "step": 2850, "loss": 4.568082332611084, "lr": 0.00037210483884007813, "seconds": 219.07, "steps_per_second": 13.009}
{"event": "train", "step": 2875, "loss": 4.657015323638916, "lr": 0.0003687272501395686, "seconds": 220.9, "steps_per_second": 13.015}
{"event": "train", "step": 2900, "loss": 4.702998161315918, "lr": 0.00036534545691603614, "seconds": 222.84, "steps_per_second": 13.014}
{"event": "train", "step": 2925, "loss": 4.658095836639404, "lr": 0.00036196005843097884, "seconds": 224.85, "steps_per_second": 13.009}
{"event": "train", "step": 2950, "loss": 4.707334995269775, "lr": 0.0003585716545847555, "seconds": 226.95, "steps_per_second": 12.998}
{"event": "train", "step": 2975, "loss": 4.515985488891602, "lr": 0.00035518084581028197, "seconds": 228.99, "steps_per_second": 12.992}
{"event": "train", "step": 3000, "loss": 4.612889766693115, "lr": 0.00035178823296663254, "seconds": 230.93, "steps_per_second": 12.991}
{"event": "validation", "step": 3000, "val_loss": 4.572894648830669, "best_val": 4.572894648830669, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 3025, "loss": 4.54715633392334, "lr": 0.00034839441723256704, "seconds": 233.42, "steps_per_second": 12.959}
{"event": "train", "step": 3050, "loss": 4.739036560058594, "lr": 0.00034499999999999993, "seconds": 235.31, "steps_per_second": 12.962}
{"event": "train", "step": 3075, "loss": 4.4800519943237305, "lr": 0.000341605582767433, "seconds": 237.3, "steps_per_second": 12.958}
{"event": "train", "step": 3100, "loss": 4.635876655578613, "lr": 0.00033821176703336737, "seconds": 239.32, "steps_per_second": 12.953}
{"event": "train", "step": 3125, "loss": 4.641169548034668, "lr": 0.00033481915418971805, "seconds": 241.4, "steps_per_second": 12.945}
{"event": "train", "step": 3150, "loss": 4.516171455383301, "lr": 0.0003314283454152445, "seconds": 243.34, "steps_per_second": 12.945}
{"event": "train", "step": 3175, "loss": 4.49936056137085, "lr": 0.00032803994156902107, "seconds": 245.25, "steps_per_second": 12.946}
{"event": "train", "step": 3200, "loss": 4.4284772872924805, "lr": 0.00032465454308396377, "seconds": 247.18, "steps_per_second": 12.946}
{"event": "train", "step": 3225, "loss": 4.557837963104248, "lr": 0.00032127274986043123, "seconds": 249.09, "steps_per_second": 12.947}
{"event": "train", "step": 3250, "loss": 4.559176921844482, "lr": 0.00031789516115992184, "seconds": 251.11, "steps_per_second": 12.943}
{"event": "train", "step": 3275, "loss": 4.546382904052734, "lr": 0.0003145223754988825, "seconds": 253.15, "steps_per_second": 12.937}
{"event": "train", "step": 3300, "loss": 4.598040580749512, "lr": 0.0003111549905426509, "seconds": 255.05, "steps_per_second": 12.939}
{"event": "validation", "step": 3300, "val_loss": 4.524982397605259, "best_val": 4.524982397605259, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 3325, "loss": 4.469085216522217, "lr": 0.0003077936029995472, "seconds": 257.49, "steps_per_second": 12.913}
{"event": "train", "step": 3350, "loss": 4.477004528045654, "lr": 0.00030443880851513665, "seconds": 259.25, "steps_per_second": 12.922}
{"event": "train", "step": 3375, "loss": 4.515832424163818, "lr": 0.0003010912015666788, "seconds": 261.17, "steps_per_second": 12.923}
{"event": "train", "step": 3400, "loss": 4.373744487762451, "lr": 0.0002977513753577859, "seconds": 263.15, "steps_per_second": 12.92}
{"event": "train", "step": 3425, "loss": 4.551100254058838, "lr": 0.0002944199217133049, "seconds": 265.08, "steps_per_second": 12.921}
{"event": "train", "step": 3450, "loss": 4.725627899169922, "lr": 0.00029109743097444573, "seconds": 267.04, "steps_per_second": 12.92}
{"event": "train", "step": 3475, "loss": 4.365353584289551, "lr": 0.00028778449189417055, "seconds": 268.93, "steps_per_second": 12.922}
{"event": "train", "step": 3500, "loss": 4.372343063354492, "lr": 0.00028448169153286564, "seconds": 270.85, "steps_per_second": 12.922}
{"event": "train", "step": 3525, "loss": 4.453754425048828, "lr": 0.00028118961515431355, "seconds": 272.79, "steps_per_second": 12.922}
{"event": "train", "step": 3550, "loss": 4.4735822677612305, "lr": 0.0002779088461219821, "seconds": 274.87, "steps_per_second": 12.915}
{"event": "train", "step": 3575, "loss": 4.483319282531738, "lr": 0.0002746399657956516, "seconds": 277.25, "steps_per_second": 12.895}
{"event": "train", "step": 3600, "loss": 4.485356330871582, "lr": 0.0002713835534283967, "seconds": 280.14, "steps_per_second": 12.851}
{"event": "validation", "step": 3600, "val_loss": 4.479871991525419, "best_val": 4.479871991525419, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 3625, "loss": 4.415444850921631, "lr": 0.0002681401860639407, "seconds": 282.7, "steps_per_second": 12.823}
{"event": "train", "step": 3650, "loss": 4.42017126083374, "lr": 0.0002649104384344031, "seconds": 284.52, "steps_per_second": 12.829}
{"event": "train", "step": 3675, "loss": 4.519322395324707, "lr": 0.0002616948828584546, "seconds": 286.29, "steps_per_second": 12.837}
{"event": "train", "step": 3700, "loss": 4.469679832458496, "lr": 0.00025849408913990223, "seconds": 288.02, "steps_per_second": 12.846}
{"event": "train", "step": 3725, "loss": 4.3826093673706055, "lr": 0.0002553086244667175, "seconds": 289.9, "steps_per_second": 12.849}
{"event": "train", "step": 3750, "loss": 4.438538074493408, "lr": 0.0002521390533105303, "seconds": 291.77, "steps_per_second": 12.853}
{"event": "train", "step": 3775, "loss": 4.302765846252441, "lr": 0.0002489859373266029, "seconds": 293.62, "steps_per_second": 12.857}
{"event": "train", "step": 3800, "loss": 4.349156856536865, "lr": 0.0002458498352543032, "seconds": 295.42, "steps_per_second": 12.863}
{"event": "train", "step": 3825, "loss": 4.415363788604736, "lr": 0.00024273130281809535, "seconds": 297.24, "steps_per_second": 12.868}
{"event": "train", "step": 3850, "loss": 4.53059196472168, "lr": 0.0002396308926290636, "seconds": 299.01, "steps_per_second": 12.876}
{"event": "train", "step": 3875, "loss": 4.329322814941406, "lr": 0.0002365491540869884, "seconds": 300.82, "steps_per_second": 12.881}
{"event": "train", "step": 3900, "loss": 4.529847621917725, "lr": 0.00023348663328299186, "seconds": 302.62, "steps_per_second": 12.887}
{"event": "validation", "step": 3900, "val_loss": 4.446831559447443, "best_val": 4.446831559447443, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 3925, "loss": 4.45183801651001, "lr": 0.00023044387290276883, "seconds": 304.93, "steps_per_second": 12.872}
{"event": "train", "step": 3950, "loss": 4.3321452140808105, "lr": 0.00022742141213042183, "seconds": 306.66, "steps_per_second": 12.881}
{"event": "train", "step": 3975, "loss": 4.391538143157959, "lr": 0.0002244197865529166, "seconds": 308.47, "steps_per_second": 12.886}
{"event": "train", "step": 4000, "loss": 4.5124945640563965, "lr": 0.00022143952806517422, "seconds": 310.55, "steps_per_second": 12.88}
{"event": "train", "step": 4025, "loss": 4.436868667602539, "lr": 0.00021848116477581896, "seconds": 312.67, "steps_per_second": 12.873}
{"event": "train", "step": 4050, "loss": 4.393537998199463, "lr": 0.00021554522091359589, "seconds": 314.57, "steps_per_second": 12.875}
{"event": "train", "step": 4075, "loss": 4.501061916351318, "lr": 0.00021263221673447578, "seconds": 316.32, "steps_per_second": 12.883}
{"event": "train", "step": 4100, "loss": 4.398018836975098, "lr": 0.00020974266842946565, "seconds": 318.15, "steps_per_second": 12.887}
{"event": "train", "step": 4125, "loss": 4.383106708526611, "lr": 0.00020687708803313762, "seconds": 319.81, "steps_per_second": 12.898}
{"event": "train", "step": 4150, "loss": 4.456511974334717, "lr": 0.00020403598333289575, "seconds": 321.52, "steps_per_second": 12.908}
{"event": "train", "step": 4175, "loss": 4.337074279785156, "lr": 0.00020121985777899422, "seconds": 323.25, "steps_per_second": 12.916}
{"event": "train", "step": 4200, "loss": 4.38740348815918, "lr": 0.00019842921039532624, "seconds": 325.03, "steps_per_second": 12.922}
{"event": "validation", "step": 4200, "val_loss": 4.412905934076738, "best_val": 4.412905934076738, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 4225, "loss": 4.218251705169678, "lr": 0.000195664535690994, "seconds": 327.35, "steps_per_second": 12.907}
{"event": "train", "step": 4250, "loss": 4.347802639007568, "lr": 0.00019292632357268208, "seconds": 329.03, "steps_per_second": 12.917}
{"event": "train", "step": 4275, "loss": 4.304572582244873, "lr": 0.0001902150592578441, "seconds": 330.79, "steps_per_second": 12.924}
{"event": "train", "step": 4300, "loss": 4.413467884063721, "lr": 0.00018753122318872106, "seconds": 332.46, "steps_per_second": 12.934}
{"event": "train", "step": 4325, "loss": 4.425302505493164, "lr": 0.00018487529094720622, "seconds": 334.13, "steps_per_second": 12.944}
{"event": "train", "step": 4350, "loss": 4.357510089874268, "lr": 0.00018224773317057093, "seconds": 335.81, "steps_per_second": 12.954}
{"event": "train", "step": 4375, "loss": 4.316355228424072, "lr": 0.00017964901546806592, "seconds": 337.62, "steps_per_second": 12.958}
{"event": "train", "step": 4400, "loss": 4.380866050720215, "lr": 0.00017707959833841515, "seconds": 339.29, "steps_per_second": 12.968}
{"event": "train", "step": 4425, "loss": 4.449340343475342, "lr": 0.00017453993708821402, "seconds": 341.04, "steps_per_second": 12.975}
{"event": "train", "step": 4450, "loss": 4.304454803466797, "lr": 0.00017203048175124785, "seconds": 342.81, "steps_per_second": 12.981}
{"event": "train", "step": 4475, "loss": 4.362473011016846, "lr": 0.00016955167700874517, "seconds": 344.48, "steps_per_second": 12.991}
{"event": "train", "step": 4500, "loss": 4.363343715667725, "lr": 0.00016710396211057894, "seconds": 346.22, "steps_per_second": 12.998}
{"event": "validation", "step": 4500, "val_loss": 4.388296186897941, "best_val": 4.388296186897941, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 4525, "loss": 4.172045707702637, "lr": 0.00016468777079743036, "seconds": 348.46, "steps_per_second": 12.986}
{"event": "train", "step": 4550, "loss": 4.363791465759277, "lr": 0.00016230353122392963, "seconds": 350.19, "steps_per_second": 12.993}
{"event": "train", "step": 4575, "loss": 4.310970306396484, "lr": 0.00015995166588278535, "seconds": 351.98, "steps_per_second": 12.998}
{"event": "train", "step": 4600, "loss": 4.263629913330078, "lr": 0.00015763259152991787, "seconds": 353.72, "steps_per_second": 13.005}
{"event": "train", "step": 4625, "loss": 4.340151786804199, "lr": 0.0001553467191106097, "seconds": 355.5, "steps_per_second": 13.01}
{"event": "train", "step": 4650, "loss": 4.273427486419678, "lr": 0.0001530944536866847, "seconds": 357.18, "steps_per_second": 13.019}
{"event": "train", "step": 4675, "loss": 4.3984761238098145, "lr": 0.00015087619436472997, "seconds": 358.9, "steps_per_second": 13.026}
{"event": "train", "step": 4700, "loss": 4.2440900802612305, "lr": 0.00014869233422537385, "seconds": 360.68, "steps_per_second": 13.031}
{"event": "train", "step": 4725, "loss": 4.452341556549072, "lr": 0.0001465432602536307, "seconds": 362.47, "steps_per_second": 13.035}
{"event": "train", "step": 4750, "loss": 4.270356178283691, "lr": 0.00014442935327032614, "seconds": 364.12, "steps_per_second": 13.045}
{"event": "train", "step": 4775, "loss": 4.324588298797607, "lr": 0.0001423509878646153, "seconds": 365.88, "steps_per_second": 13.051}
{"event": "train", "step": 4800, "loss": 4.317485809326172, "lr": 0.00014030853232760432, "seconds": 367.62, "steps_per_second": 13.057}
{"event": "validation", "step": 4800, "val_loss": 4.366913439249461, "best_val": 4.366913439249461, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 4825, "loss": 4.349158763885498, "lr": 0.00013830234858708818, "seconds": 369.86, "steps_per_second": 13.045}
{"event": "train", "step": 4850, "loss": 4.359145164489746, "lr": 0.00013633279214341688, "seconds": 371.58, "steps_per_second": 13.052}
{"event": "train", "step": 4875, "loss": 4.171675682067871, "lr": 0.0001344002120064995, "seconds": 373.34, "steps_per_second": 13.058}
{"event": "train", "step": 4900, "loss": 4.356762886047363, "lr": 0.00013250495063395865, "seconds": 375.1, "steps_per_second": 13.063}
{"event": "train", "step": 4925, "loss": 4.219712257385254, "lr": 0.00013064734387044702, "seconds": 376.79, "steps_per_second": 13.071}
{"event": "train", "step": 4950, "loss": 4.373247146606445, "lr": 0.00012882772088813391, "seconds": 378.47, "steps_per_second": 13.079}
{"event": "train", "step": 4975, "loss": 4.288434028625488, "lr": 0.00012704640412837594, "seconds": 380.14, "steps_per_second": 13.087}
{"event": "train", "step": 5000, "loss": 4.282129764556885, "lr": 0.0001253037092445794, "seconds": 381.85, "steps_per_second": 13.094}
{"event": "train", "step": 5025, "loss": 4.361968517303467, "lr": 0.00012359994504626578, "seconds": 383.5, "steps_per_second": 13.103}
{"event": "train", "step": 5050, "loss": 4.353524684906006, "lr": 0.00012193541344434986, "seconds": 385.2, "steps_per_second": 13.11}
{"event": "train", "step": 5075, "loss": 4.32540225982666, "lr": 0.00012031040939764075, "seconds": 386.99, "steps_per_second": 13.114}
{"event": "train", "step": 5100, "loss": 4.232648849487305, "lr": 0.00011872522086057427, "seconds": 388.74, "steps_per_second": 13.119}
{"event": "validation", "step": 5100, "val_loss": 4.348691192033667, "best_val": 4.348691192033667, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 5125, "loss": 4.268082618713379, "lr": 0.00011718012873218671, "seconds": 391.12, "steps_per_second": 13.103}
{"event": "train", "step": 5150, "loss": 4.353888034820557, "lr": 0.00011567540680633911, "seconds": 392.81, "steps_per_second": 13.111}
{"event": "train", "step": 5175, "loss": 4.241316318511963, "lr": 0.00011421132172320017, "seconds": 394.53, "steps_per_second": 13.117}
{"event": "train", "step": 5200, "loss": 4.137839317321777, "lr": 0.00011278813292199712, "seconds": 396.31, "steps_per_second": 13.121}
{"event": "train", "step": 5225, "loss": 4.2639617919921875, "lr": 0.00011140609259504246, "seconds": 398.07, "steps_per_second": 13.126}
{"event": "train", "step": 5250, "loss": 4.321348190307617, "lr": 0.00011006544564304468, "seconds": 399.74, "steps_per_second": 13.133}
{"event": "train", "step": 5275, "loss": 4.269964218139648, "lr": 0.00010876642963171178, "seconds": 401.51, "steps_per_second": 13.138}
{"event": "train", "step": 5300, "loss": 4.319515228271484, "lr": 0.0001075092747496537, "seconds": 403.25, "steps_per_second": 13.143}
{"event": "train", "step": 5325, "loss": 4.371119022369385, "lr": 0.0001062942037675924, "seconds": 404.92, "steps_per_second": 13.151}
{"event": "train", "step": 5350, "loss": 4.348846912384033, "lr": 0.0001051214319988865, "seconds": 406.63, "steps_per_second": 13.157}
{"event": "train", "step": 5375, "loss": 4.344863414764404, "lr": 0.00010399116726137729, "seconds": 408.34, "steps_per_second": 13.163}
{"event": "train", "step": 5400, "loss": 4.360861778259277, "lr": 0.00010290360984056277, "seconds": 409.99, "steps_per_second": 13.171}
{"event": "validation", "step": 5400, "val_loss": 4.331280831844851, "best_val": 4.331280831844851, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 5425, "loss": 4.160545825958252, "lr": 0.00010185895245410669, "seconds": 412.31, "steps_per_second": 13.158}
{"event": "train", "step": 5450, "loss": 4.122735977172852, "lr": 0.00010085738021768859, "seconds": 413.97, "steps_per_second": 13.165}
{"event": "train", "step": 5475, "loss": 4.324886798858643, "lr": 9.989907061220079e-05, "seconds": 415.71, "steps_per_second": 13.17}
{"event": "train", "step": 5500, "loss": 4.266539096832275, "lr": 9.898419345229842e-05, "seconds": 417.47, "steps_per_second": 13.175}
{"event": "train", "step": 5525, "loss": 4.258175373077393, "lr": 9.811291085630787e-05, "seconds": 419.23, "steps_per_second": 13.179}
{"event": "train", "step": 5550, "loss": 4.411218643188477, "lr": 9.728537721749898e-05, "seconds": 420.99, "steps_per_second": 13.183}
{"event": "train", "step": 5575, "loss": 4.155972480773926, "lr": 9.65017391767263e-05, "seconds": 422.67, "steps_per_second": 13.19}
{"event": "train", "step": 5600, "loss": 4.225550651550293, "lr": 9.5762135596444e-05, "seconds": 424.32, "steps_per_second": 13.198}
{"event": "train", "step": 5625, "loss": 4.2327046394348145, "lr": 9.506669753609903e-05, "seconds": 426.03, "steps_per_second": 13.203}
{"event": "train", "step": 5650, "loss": 4.336045742034912, "lr": 9.441554822890715e-05, "seconds": 427.75, "steps_per_second": 13.209}
{"event": "train", "step": 5675, "loss": 4.3178019523620605, "lr": 9.380880306001571e-05, "seconds": 429.41, "steps_per_second": 13.216}
{"event": "train", "step": 5700, "loss": 4.257144451141357, "lr": 9.324656954605712e-05, "seconds": 431.18, "steps_per_second": 13.219}
{"event": "validation", "step": 5700, "val_loss": 4.3215583498793535, "best_val": 4.3215583498793535, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 5725, "loss": 4.319136619567871, "lr": 9.272894731609671e-05, "seconds": 433.45, "steps_per_second": 13.208}
{"event": "train", "step": 5750, "loss": 4.205748081207275, "lr": 9.225602809397817e-05, "seconds": 435.11, "steps_per_second": 13.215}
{"event": "train", "step": 5775, "loss": 4.267698287963867, "lr": 9.182789568206988e-05, "seconds": 436.86, "steps_per_second": 13.219}
{"event": "train", "step": 5800, "loss": 4.277442455291748, "lr": 9.144462594641502e-05, "seconds": 438.68, "steps_per_second": 13.221}
{"event": "train", "step": 5825, "loss": 4.251745700836182, "lr": 9.11062868032878e-05, "seconds": 440.39, "steps_per_second": 13.227}
{"event": "train", "step": 5850, "loss": 4.317497253417969, "lr": 9.08129382071587e-05, "seconds": 442.11, "steps_per_second": 13.232}
{"event": "train", "step": 5875, "loss": 4.1376190185546875, "lr": 9.056463214007024e-05, "seconds": 443.8, "steps_per_second": 13.238}
{"event": "train", "step": 5900, "loss": 4.302074432373047, "lr": 9.03614126024257e-05, "seconds": 445.51, "steps_per_second": 13.243}
{"event": "train", "step": 5925, "loss": 4.2693562507629395, "lr": 9.020331560519225e-05, "seconds": 447.3, "steps_per_second": 13.246}
{"event": "train", "step": 5950, "loss": 4.267000198364258, "lr": 9.009036916351956e-05, "seconds": 449.06, "steps_per_second": 13.25}
{"event": "train", "step": 5975, "loss": 4.202955722808838, "lr": 9.002259329177564e-05, "seconds": 450.77, "steps_per_second": 13.255}
{"event": "train", "step": 6000, "loss": 4.224427700042725, "lr": 8.999999999999999e-05, "seconds": 452.56, "steps_per_second": 13.258}
{"event": "validation", "step": 6000, "val_loss": 4.315424796221505, "best_val": 4.315424796221505, "peak_process_rss_mib": 600.8, "mps_driver_allocated_mib": 1217.5}
{"event": "train", "step": 6025, "loss": 4.4019694328308105, "lr": 0.00047411151650635894, "seconds": 455.29, "steps_per_second": 11.776}
{"event": "train", "step": 6050, "loss": 4.2288618087768555, "lr": 0.00047314542963911233, "seconds": 457.0, "steps_per_second": 13.027}
{"event": "train", "step": 6075, "loss": 4.218117713928223, "lr": 0.0004721768757322383, "seconds": 458.69, "steps_per_second": 13.571}
{"event": "train", "step": 6100, "loss": 4.450053691864014, "lr": 0.00047120587343221407, "seconds": 460.49, "steps_per_second": 13.661}
{"event": "train", "step": 6125, "loss": 4.379146099090576, "lr": 0.0004702324414326536, "seconds": 462.15, "steps_per_second": 13.92}
{"event": "train", "step": 6150, "loss": 4.29838228225708, "lr": 0.0004692565984739466, "seconds": 463.84, "steps_per_second": 14.056}
{"event": "train", "step": 6175, "loss": 4.348118782043457, "lr": 0.00046827836334289855, "seconds": 465.58, "steps_per_second": 14.103}
{"event": "train", "step": 6200, "loss": 4.348628997802734, "lr": 0.0004672977548723688, "seconds": 467.25, "steps_per_second": 14.2}
{"event": "train", "step": 6225, "loss": 4.339539051055908, "lr": 0.00046631479194090745, "seconds": 469.01, "steps_per_second": 14.199}
{"event": "train", "step": 6250, "loss": 4.217878818511963, "lr": 0.0004653294934723929, "seconds": 470.68, "steps_per_second": 14.272}
{"event": "train", "step": 6275, "loss": 4.415380954742432, "lr": 0.00046434187843566677, "seconds": 472.35, "steps_per_second": 14.333}
{"event": "train", "step": 6300, "loss": 4.291977405548096, "lr": 0.00046335196584416916, "seconds": 474.04, "steps_per_second": 14.372}
{"event": "validation", "step": 6300, "val_loss": 4.365927377711638, "best_val": 4.315424796221505, "peak_process_rss_mib": 650.8, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 6325, "loss": 4.300050258636475, "lr": 0.00046235977475557217, "seconds": 476.25, "steps_per_second": 14.079}
{"event": "train", "step": 6350, "loss": 4.268784046173096, "lr": 0.0004613653242714133, "seconds": 477.96, "steps_per_second": 14.117}
{"event": "train", "step": 6375, "loss": 4.4009108543396, "lr": 0.0004603686335367277, "seconds": 479.8, "steps_per_second": 14.082}
{"event": "train", "step": 6400, "loss": 4.253422260284424, "lr": 0.00045936972173967947, "seconds": 481.53, "steps_per_second": 14.105}
{"event": "train", "step": 6425, "loss": 4.320215702056885, "lr": 0.0004583686081111925, "seconds": 483.22, "steps_per_second": 14.144}
{"event": "train", "step": 6450, "loss": 4.316038131713867, "lr": 0.0004573653119245798, "seconds": 484.98, "steps_per_second": 14.146}
{"event": "train", "step": 6475, "loss": 4.236755847930908, "lr": 0.00045635985249517296, "seconds": 486.69, "steps_per_second": 14.171}
{"event": "train", "step": 6500, "loss": 4.272172451019287, "lr": 0.0004553522491799498, "seconds": 488.44, "steps_per_second": 14.175}
{"event": "train", "step": 6525, "loss": 4.255345821380615, "lr": 0.0004543425213771622, "seconds": 490.23, "steps_per_second": 14.164}
{"event": "train", "step": 6550, "loss": 4.4007158279418945, "lr": 0.0004533306885259623, "seconds": 491.9, "steps_per_second": 14.199}
{"event": "train", "step": 6575, "loss": 4.426392555236816, "lr": 0.0004523167701060283, "seconds": 493.59, "steps_per_second": 14.225}
{"event": "train", "step": 6600, "loss": 4.272359848022461, "lr": 0.00045130078563718955, "seconds": 495.35, "steps_per_second": 14.225}
{"event": "validation", "step": 6600, "val_loss": 4.354282400815338, "best_val": 4.315424796221505, "peak_process_rss_mib": 671.8, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 6625, "loss": 4.362105369567871, "lr": 0.0004502827546790508, "seconds": 497.6, "steps_per_second": 14.067}
{"event": "train", "step": 6650, "loss": 4.2544026374816895, "lr": 0.00044926269683061553, "seconds": 499.29, "steps_per_second": 14.093}
{"event": "train", "step": 6675, "loss": 4.409705638885498, "lr": 0.0004482406317299085, "seconds": 501.03, "steps_per_second": 14.102}
{"event": "train", "step": 6700, "loss": 4.216238975524902, "lr": 0.0004472165790535979, "seconds": 502.81, "steps_per_second": 14.1}
{"event": "train", "step": 6725, "loss": 4.287245750427246, "lr": 0.00044619055851661664, "seconds": 504.57, "steps_per_second": 14.104}
{"event": "train", "step": 6750, "loss": 4.390729904174805, "lr": 0.00044516258987178264, "seconds": 506.27, "steps_per_second": 14.124}
{"event": "train", "step": 6775, "loss": 4.411861419677734, "lr": 0.0004441326929094182, "seconds": 507.96, "steps_per_second": 14.143}
{"event": "train", "step": 6800, "loss": 4.202812194824219, "lr": 0.00044310088745696966, "seconds": 509.72, "steps_per_second": 14.145}
{"event": "train", "step": 6825, "loss": 4.223495006561279, "lr": 0.00044206719337862527, "seconds": 511.4, "steps_per_second": 14.167}
{"event": "train", "step": 6850, "loss": 4.17448091506958, "lr": 0.0004410316305749328, "seconds": 513.14, "steps_per_second": 14.173}
{"event": "train", "step": 6875, "loss": 4.302827835083008, "lr": 0.00043999421898241676, "seconds": 514.93, "steps_per_second": 14.168}
{"event": "train", "step": 6900, "loss": 4.2554931640625, "lr": 0.00043895497857319397, "seconds": 516.65, "steps_per_second": 14.178}
{"event": "validation", "step": 6900, "val_loss": 4.336633523305257, "best_val": 4.315424796221505, "peak_process_rss_mib": 684.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 6925, "loss": 4.42609167098999, "lr": 0.00043791392935458944, "seconds": 518.89, "steps_per_second": 14.075}
{"event": "train", "step": 6950, "loss": 4.401597499847412, "lr": 0.0004368710913687515, "seconds": 520.6, "steps_per_second": 14.088}
{"event": "train", "step": 6975, "loss": 4.346197605133057, "lr": 0.0004358264846922652, "seconds": 522.33, "steps_per_second": 14.098}
{"event": "train", "step": 7000, "loss": 4.152881622314453, "lr": 0.00043478012943576637, "seconds": 524.1, "steps_per_second": 14.097}
{"event": "train", "step": 7025, "loss": 4.193181037902832, "lr": 0.0004337320457435542, "seconds": 525.79, "steps_per_second": 14.114}
{"event": "train", "step": 7050, "loss": 4.299526214599609, "lr": 0.0004326822537932037, "seconds": 527.45, "steps_per_second": 14.134}
{"event": "train", "step": 7075, "loss": 4.336905479431152, "lr": 0.00043163077379517676, "seconds": 529.25, "steps_per_second": 14.13}
{"event": "train", "step": 7100, "loss": 4.338744163513184, "lr": 0.00043057762599243375, "seconds": 530.93, "steps_per_second": 14.146}
{"event": "train", "step": 7125, "loss": 4.4149580001831055, "lr": 0.00042952283066004286, "seconds": 532.62, "steps_per_second": 14.159}
{"event": "train", "step": 7150, "loss": 4.324179172515869, "lr": 0.0004284664081047908, "seconds": 534.38, "steps_per_second": 14.16}
{"event": "train", "step": 7175, "loss": 4.143990993499756, "lr": 0.00042740837866479114, "seconds": 536.08, "steps_per_second": 14.171}
{"event": "train", "step": 7200, "loss": 4.140163421630859, "lr": 0.0004263487627090931, "seconds": 537.76, "steps_per_second": 14.186}
{"event": "validation", "step": 7200, "val_loss": 4.308409187563542, "best_val": 4.308409187563542, "peak_process_rss_mib": 696.3, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 7225, "loss": 4.272471904754639, "lr": 0.0004252875806372891, "seconds": 540.09, "steps_per_second": 14.093}
{"event": "train", "step": 7250, "loss": 4.200552463531494, "lr": 0.0004242248528791225, "seconds": 541.76, "steps_per_second": 14.109}
{"event": "train", "step": 7275, "loss": 4.318378448486328, "lr": 0.00042316059989409377, "seconds": 543.44, "steps_per_second": 14.124}
{"event": "train", "step": 7300, "loss": 4.257970809936523, "lr": 0.000422094842171067, "seconds": 545.12, "steps_per_second": 14.138}
{"event": "train", "step": 7325, "loss": 4.230489253997803, "lr": 0.00042102760022787547, "seconds": 546.79, "steps_per_second": 14.153}
{"event": "train", "step": 7350, "loss": 4.392727375030518, "lr": 0.000419958894610926, "seconds": 548.53, "steps_per_second": 14.156}
{"event": "train", "step": 7375, "loss": 4.206121444702148, "lr": 0.0004188887458948044, "seconds": 550.33, "steps_per_second": 14.151}
{"event": "train", "step": 7400, "loss": 4.300406455993652, "lr": 0.0004178171746818783, "seconds": 551.97, "steps_per_second": 14.169}
{"event": "train", "step": 7425, "loss": 4.3299736976623535, "lr": 0.00041674420160190155, "seconds": 553.63, "steps_per_second": 14.184}
{"event": "train", "step": 7450, "loss": 4.424450874328613, "lr": 0.00041566984731161645, "seconds": 555.32, "steps_per_second": 14.195}
{"event": "train", "step": 7475, "loss": 4.280956745147705, "lr": 0.00041459413249435604, "seconds": 557.04, "steps_per_second": 14.2}
{"event": "train", "step": 7500, "loss": 4.1527862548828125, "lr": 0.00041351707785964605, "seconds": 558.81, "steps_per_second": 14.199}
{"event": "validation", "step": 7500, "val_loss": 4.298205887714085, "best_val": 4.298205887714085, "peak_process_rss_mib": 701.8, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 7525, "loss": 4.087921142578125, "lr": 0.00041243870414280634, "seconds": 561.16, "steps_per_second": 14.122}
{"event": "train", "step": 7550, "loss": 4.1327104568481445, "lr": 0.0004113590321045516, "seconds": 562.95, "steps_per_second": 14.119}
{"event": "train", "step": 7575, "loss": 4.171742916107178, "lr": 0.00041027808253059144, "seconds": 564.69, "steps_per_second": 14.123}
{"event": "train", "step": 7600, "loss": 4.2381439208984375, "lr": 0.0004091958762312306, "seconds": 566.37, "steps_per_second": 14.134}
{"event": "train", "step": 7625, "loss": 4.197778224945068, "lr": 0.00040811243404096796, "seconds": 568.06, "steps_per_second": 14.144}
{"event": "train", "step": 7650, "loss": 4.22402811050415, "lr": 0.00040702777681809587, "seconds": 569.87, "steps_per_second": 14.138}
{"event": "train", "step": 7675, "loss": 4.108954429626465, "lr": 0.0004059419254442983, "seconds": 571.66, "steps_per_second": 14.136}
{"event": "train", "step": 7700, "loss": 4.192224502563477, "lr": 0.0004048549008242485, "seconds": 573.41, "steps_per_second": 14.138}
{"event": "train", "step": 7725, "loss": 4.289620876312256, "lr": 0.00040376672388520727, "seconds": 575.16, "steps_per_second": 14.14}
{"event": "train", "step": 7750, "loss": 4.11630392074585, "lr": 0.00040267741557661966, "seconds": 576.85, "steps_per_second": 14.149}
{"event": "train", "step": 7775, "loss": 4.1434645652771, "lr": 0.00040158699686971167, "seconds": 578.68, "steps_per_second": 14.141}
{"event": "train", "step": 7800, "loss": 4.303722858428955, "lr": 0.0004004954887570865, "seconds": 580.44, "steps_per_second": 14.143}
{"event": "validation", "step": 7800, "val_loss": 4.284741671025987, "best_val": 4.284741671025987, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 7825, "loss": 4.266723155975342, "lr": 0.0003994029122523205, "seconds": 582.73, "steps_per_second": 14.086}
{"event": "train", "step": 7850, "loss": 4.234335422515869, "lr": 0.00039830928838955845, "seconds": 584.46, "steps_per_second": 14.091}
{"event": "train", "step": 7875, "loss": 4.303323268890381, "lr": 0.000397214638223109, "seconds": 586.27, "steps_per_second": 14.087}
{"event": "train", "step": 7900, "loss": 4.365428447723389, "lr": 0.0003961189828270388, "seconds": 587.99, "steps_per_second": 14.092}
{"event": "train", "step": 7925, "loss": 4.06534481048584, "lr": 0.0003950223432947672, "seconds": 589.81, "steps_per_second": 14.088}
{"event": "train", "step": 7950, "loss": 4.172029972076416, "lr": 0.00039392474073866, "seconds": 591.49, "steps_per_second": 14.098}
{"event": "train", "step": 7975, "loss": 4.154287815093994, "lr": 0.0003928261962896229, "seconds": 593.16, "steps_per_second": 14.108}
{"event": "train", "step": 8000, "loss": 4.080924987792969, "lr": 0.000391726731096695, "seconds": 594.88, "steps_per_second": 14.113}
{"event": "train", "step": 8025, "loss": 4.16148042678833, "lr": 0.0003906263663266413, "seconds": 596.59, "steps_per_second": 14.119}
{"event": "train", "step": 8050, "loss": 4.209174633026123, "lr": 0.0003895251231635453, "seconds": 598.29, "steps_per_second": 14.126}
{"event": "train", "step": 8075, "loss": 4.108459949493408, "lr": 0.0003884230228084014, "seconds": 599.98, "steps_per_second": 14.134}
{"event": "train", "step": 8100, "loss": 4.228477954864502, "lr": 0.0003873200864787065, "seconds": 601.7, "steps_per_second": 14.139}
{"event": "validation", "step": 8100, "val_loss": 4.277414137955678, "best_val": 4.277414137955678, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 8125, "loss": 4.202079772949219, "lr": 0.00038621633540805137, "seconds": 604.1, "steps_per_second": 14.079}
{"event": "train", "step": 8150, "loss": 4.1236653327941895, "lr": 0.00038511179084571237, "seconds": 605.9, "steps_per_second": 14.077}
{"event": "train", "step": 8175, "loss": 4.22900915145874, "lr": 0.0003840064740562418, "seconds": 607.58, "steps_per_second": 14.085}
{"event": "train", "step": 8200, "loss": 3.9981071949005127, "lr": 0.00038290040631905895, "seconds": 609.33, "steps_per_second": 14.088}
{"event": "train", "step": 8225, "loss": 4.229708194732666, "lr": 0.00038179360892804027, "seconds": 611.13, "steps_per_second": 14.085}
{"event": "train", "step": 8250, "loss": 4.3174333572387695, "lr": 0.00038068610319110923, "seconds": 612.84, "steps_per_second": 14.092}
{"event": "train", "step": 8275, "loss": 4.228780746459961, "lr": 0.0003795779104298265, "seconds": 614.56, "steps_per_second": 14.096}
{"event": "train", "step": 8300, "loss": 4.196824073791504, "lr": 0.00037846905197897916, "seconds": 616.25, "steps_per_second": 14.104}
{"event": "train", "step": 8325, "loss": 4.188594341278076, "lr": 0.00037735954918617, "seconds": 617.91, "steps_per_second": 14.113}
{"event": "train", "step": 8350, "loss": 4.074880123138428, "lr": 0.0003762494234114067, "seconds": 619.65, "steps_per_second": 14.115}
{"event": "train", "step": 8375, "loss": 4.160496234893799, "lr": 0.0003751386960266905, "seconds": 621.35, "steps_per_second": 14.122}
{"event": "train", "step": 8400, "loss": 4.217929363250732, "lr": 0.00037402738841560456, "seconds": 623.09, "steps_per_second": 14.124}
{"event": "validation", "step": 8400, "val_loss": 4.259638369890632, "best_val": 4.259638369890632, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 8425, "loss": 4.2234086990356445, "lr": 0.0003729155219729029, "seconds": 625.45, "steps_per_second": 14.076}
{"event": "train", "step": 8450, "loss": 4.174516677856445, "lr": 0.0003718031181040978, "seconds": 627.21, "steps_per_second": 14.077}
{"event": "train", "step": 8475, "loss": 4.211775302886963, "lr": 0.000370690198225048, "seconds": 628.91, "steps_per_second": 14.083}
{"event": "train", "step": 8500, "loss": 4.187228202819824, "lr": 0.0003695767837615465, "seconds": 630.65, "steps_per_second": 14.086}
{"event": "train", "step": 8525, "loss": 4.238003730773926, "lr": 0.0003684628961489078, "seconds": 632.32, "steps_per_second": 14.094}
{"event": "train", "step": 8550, "loss": 4.151711463928223, "lr": 0.0003673485568315557, "seconds": 633.99, "steps_per_second": 14.102}
{"event": "train", "step": 8575, "loss": 4.192493438720703, "lr": 0.00036623378726261, "seconds": 635.68, "steps_per_second": 14.108}
{"event": "train", "step": 8600, "loss": 4.00264310836792, "lr": 0.0003651186089034735, "seconds": 637.34, "steps_per_second": 14.117}
{"event": "train", "step": 8625, "loss": 4.044900894165039, "lr": 0.00036400304322341926, "seconds": 639.1, "steps_per_second": 14.118}
{"event": "train", "step": 8650, "loss": 4.189377307891846, "lr": 0.0003628871116991769, "seconds": 640.86, "steps_per_second": 14.118}
{"event": "train", "step": 8675, "loss": 4.139388561248779, "lr": 0.0003617708358145191, "seconds": 642.62, "steps_per_second": 14.12}
{"event": "train", "step": 8700, "loss": 4.11690092086792, "lr": 0.0003606542370598483, "seconds": 644.31, "steps_per_second": 14.125}
{"event": "validation", "step": 8700, "val_loss": 4.246952190832469, "best_val": 4.246952190832469, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 8725, "loss": 4.154616832733154, "lr": 0.00035953733693178255, "seconds": 646.58, "steps_per_second": 14.089}
{"event": "train", "step": 8750, "loss": 4.120707988739014, "lr": 0.0003584201569327421, "seconds": 648.35, "steps_per_second": 14.089}
{"event": "train", "step": 8775, "loss": 4.124118328094482, "lr": 0.0003573027185705354, "seconds": 650.15, "steps_per_second": 14.088}
{"event": "train", "step": 8800, "loss": 4.132100582122803, "lr": 0.00035618504335794453, "seconds": 651.82, "steps_per_second": 14.095}
{"event": "train", "step": 8825, "loss": 4.018430233001709, "lr": 0.00035506715281231134, "seconds": 653.5, "steps_per_second": 14.101}
{"event": "train", "step": 8850, "loss": 4.152836799621582, "lr": 0.0003539490684551238, "seconds": 655.23, "steps_per_second": 14.104}
{"event": "train", "step": 8875, "loss": 4.149251937866211, "lr": 0.00035283081181160047, "seconds": 657.0, "steps_per_second": 14.104}
{"event": "train", "step": 8900, "loss": 4.1866021156311035, "lr": 0.0003517124044102772, "seconds": 658.69, "steps_per_second": 14.11}
{"event": "train", "step": 8925, "loss": 4.1404829025268555, "lr": 0.00035059386778259194, "seconds": 660.37, "steps_per_second": 14.116}
{"event": "train", "step": 8950, "loss": 4.156425952911377, "lr": 0.00034947522346247056, "seconds": 662.05, "steps_per_second": 14.123}
{"event": "train", "step": 8975, "loss": 4.244732856750488, "lr": 0.00034835649298591226, "seconds": 663.82, "steps_per_second": 14.123}
{"event": "train", "step": 9000, "loss": 4.09048318862915, "lr": 0.000347237697890575, "seconds": 665.56, "steps_per_second": 14.125}
{"event": "validation", "step": 9000, "val_loss": 4.239511043028625, "best_val": 4.239511043028625, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 9025, "loss": 4.132788181304932, "lr": 0.0003461188597153604, "seconds": 667.89, "steps_per_second": 14.088}
{"event": "train", "step": 9050, "loss": 4.034650802612305, "lr": 0.00034499999999999993, "seconds": 669.67, "steps_per_second": 14.087}
{"event": "train", "step": 9075, "loss": 4.024207592010498, "lr": 0.0003438811402846395, "seconds": 671.44, "steps_per_second": 14.088}
{"event": "train", "step": 9100, "loss": 4.144346237182617, "lr": 0.00034276230210942497, "seconds": 673.1, "steps_per_second": 14.095}
{"event": "train", "step": 9125, "loss": 4.103811264038086, "lr": 0.00034164350701408765, "seconds": 674.87, "steps_per_second": 14.096}
{"event": "train", "step": 9150, "loss": 4.241456985473633, "lr": 0.00034052477653752936, "seconds": 676.8, "steps_per_second": 14.086}
{"event": "train", "step": 9175, "loss": 4.230361461639404, "lr": 0.0003394061322174081, "seconds": 678.51, "steps_per_second": 14.09}
{"event": "train", "step": 9200, "loss": 4.124135494232178, "lr": 0.0003382875955897228, "seconds": 680.2, "steps_per_second": 14.095}
{"event": "train", "step": 9225, "loss": 4.181117534637451, "lr": 0.00033716918818839944, "seconds": 681.96, "steps_per_second": 14.096}
{"event": "train", "step": 9250, "loss": 4.222672939300537, "lr": 0.0003360509315448762, "seconds": 683.61, "steps_per_second": 14.103}
{"event": "train", "step": 9275, "loss": 4.1080451011657715, "lr": 0.0003349328471876886, "seconds": 685.31, "steps_per_second": 14.108}
{"event": "train", "step": 9300, "loss": 4.057185649871826, "lr": 0.0003338149566420555, "seconds": 687.06, "steps_per_second": 14.109}
{"event": "validation", "step": 9300, "val_loss": 4.226806644999653, "best_val": 4.226806644999653, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 9325, "loss": 4.0982255935668945, "lr": 0.00033269728142946455, "seconds": 689.42, "steps_per_second": 14.074}
{"event": "train", "step": 9350, "loss": 4.180872917175293, "lr": 0.0003315798430672578, "seconds": 691.18, "steps_per_second": 14.075}
{"event": "train", "step": 9375, "loss": 4.213932991027832, "lr": 0.0003304626630682174, "seconds": 692.94, "steps_per_second": 14.076}
{"event": "train", "step": 9400, "loss": 4.106753826141357, "lr": 0.00032934576294015174, "seconds": 694.66, "steps_per_second": 14.079}
{"event": "train", "step": 9425, "loss": 4.172580242156982, "lr": 0.0003282291641854809, "seconds": 696.35, "steps_per_second": 14.084}
{"event": "train", "step": 9450, "loss": 4.214262962341309, "lr": 0.00032711288830082304, "seconds": 698.02, "steps_per_second": 14.09}
{"event": "train", "step": 9475, "loss": 4.128949165344238, "lr": 0.0003259969567765807, "seconds": 699.79, "steps_per_second": 14.09}
{"event": "train", "step": 9500, "loss": 4.30516242980957, "lr": 0.0003248813910965265, "seconds": 701.56, "steps_per_second": 14.09}
{"event": "train", "step": 9525, "loss": 4.117997646331787, "lr": 0.00032376621273738995, "seconds": 703.22, "steps_per_second": 14.097}
{"event": "train", "step": 9550, "loss": 4.1567463874816895, "lr": 0.00032265144316844424, "seconds": 704.94, "steps_per_second": 14.1}
{"event": "train", "step": 9575, "loss": 4.10636568069458, "lr": 0.0003215371038510921, "seconds": 706.64, "steps_per_second": 14.104}
{"event": "train", "step": 9600, "loss": 4.118015289306641, "lr": 0.00032042321623845343, "seconds": 708.32, "steps_per_second": 14.109}
{"event": "validation", "step": 9600, "val_loss": 4.219176273879238, "best_val": 4.219176273879238, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 9625, "loss": 4.056607723236084, "lr": 0.000319309801774952, "seconds": 710.62, "steps_per_second": 14.08}
{"event": "train", "step": 9650, "loss": 4.133216381072998, "lr": 0.0003181968818959021, "seconds": 712.33, "steps_per_second": 14.084}
{"event": "train", "step": 9675, "loss": 4.114788055419922, "lr": 0.00031708447802709705, "seconds": 714.03, "steps_per_second": 14.088}
{"event": "train", "step": 9700, "loss": 4.168825149536133, "lr": 0.0003159726115843954, "seconds": 715.72, "steps_per_second": 14.093}
{"event": "train", "step": 9725, "loss": 4.122509479522705, "lr": 0.0003148613039733095, "seconds": 717.43, "steps_per_second": 14.096}
{"event": "train", "step": 9750, "loss": 4.104125499725342, "lr": 0.0003137505765885933, "seconds": 719.09, "steps_per_second": 14.102}
{"event": "train", "step": 9775, "loss": 4.162664890289307, "lr": 0.00031264045081383, "seconds": 720.89, "steps_per_second": 14.1}
{"event": "train", "step": 9800, "loss": 4.0755085945129395, "lr": 0.0003115309480210208, "seconds": 722.56, "steps_per_second": 14.106}
{"event": "train", "step": 9825, "loss": 4.0489068031311035, "lr": 0.00031042208957017336, "seconds": 724.27, "steps_per_second": 14.109}
{"event": "train", "step": 9850, "loss": 4.1681952476501465, "lr": 0.00030931389680889074, "seconds": 726.05, "steps_per_second": 14.108}
{"event": "train", "step": 9875, "loss": 3.9784302711486816, "lr": 0.0003082063910719597, "seconds": 727.83, "steps_per_second": 14.108}
{"event": "train", "step": 9900, "loss": 4.191086292266846, "lr": 0.000307099593680941, "seconds": 729.6, "steps_per_second": 14.108}
{"event": "validation", "step": 9900, "val_loss": 4.207919547838017, "best_val": 4.207919547838017, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 9925, "loss": 4.319974899291992, "lr": 0.0003059935259437581, "seconds": 731.89, "steps_per_second": 14.082}
{"event": "train", "step": 9950, "loss": 4.144454479217529, "lr": 0.0003048882091542876, "seconds": 733.66, "steps_per_second": 14.082}
{"event": "train", "step": 9975, "loss": 4.020567893981934, "lr": 0.0003037836645919486, "seconds": 735.44, "steps_per_second": 14.082}
{"event": "train", "step": 10000, "loss": 3.973332643508911, "lr": 0.0003026799135212935, "seconds": 737.2, "steps_per_second": 14.083}
{"event": "train", "step": 10025, "loss": 4.074890613555908, "lr": 0.0003015769771915985, "seconds": 738.94, "steps_per_second": 14.085}
{"event": "train", "step": 10050, "loss": 4.0865325927734375, "lr": 0.00030047487683645464, "seconds": 740.77, "steps_per_second": 14.082}
{"event": "train", "step": 10075, "loss": 4.022426605224609, "lr": 0.0002993736336733587, "seconds": 742.44, "steps_per_second": 14.087}
{"event": "train", "step": 10100, "loss": 4.1984782218933105, "lr": 0.000298273268903305, "seconds": 744.14, "steps_per_second": 14.091}
{"event": "train", "step": 10125, "loss": 4.166333198547363, "lr": 0.00029717380371037714, "seconds": 745.86, "steps_per_second": 14.093}
{"event": "train", "step": 10150, "loss": 3.8590967655181885, "lr": 0.00029607525926133996, "seconds": 747.65, "steps_per_second": 14.092}
{"event": "train", "step": 10175, "loss": 4.118969440460205, "lr": 0.00029497765670523276, "seconds": 749.35, "steps_per_second": 14.096}
{"event": "train", "step": 10200, "loss": 4.1939921379089355, "lr": 0.0002938810171729612, "seconds": 751.07, "steps_per_second": 14.098}
{"event": "validation", "step": 10200, "val_loss": 4.19988202340429, "best_val": 4.19988202340429, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 10225, "loss": 4.2140984535217285, "lr": 0.00029278536177689094, "seconds": 753.33, "steps_per_second": 14.076}
{"event": "train", "step": 10250, "loss": 3.9841346740722656, "lr": 0.00029169071161044147, "seconds": 755.08, "steps_per_second": 14.077}
{"event": "train", "step": 10275, "loss": 3.9864840507507324, "lr": 0.0002905970877476794, "seconds": 756.74, "steps_per_second": 14.082}
{"event": "train", "step": 10300, "loss": 4.000270366668701, "lr": 0.0002895045112429135, "seconds": 758.52, "steps_per_second": 14.082}
{"event": "train", "step": 10325, "loss": 3.9792346954345703, "lr": 0.0002884130031302883, "seconds": 760.2, "steps_per_second": 14.087}
{"event": "train", "step": 10350, "loss": 4.0830769538879395, "lr": 0.00028732258442338025, "seconds": 761.87, "steps_per_second": 14.091}
{"event": "train", "step": 10375, "loss": 4.118250846862793, "lr": 0.0002862332761147927, "seconds": 763.66, "steps_per_second": 14.091}
{"event": "train", "step": 10400, "loss": 4.165230751037598, "lr": 0.00028514509917575154, "seconds": 765.41, "steps_per_second": 14.092}
{"event": "train", "step": 10425, "loss": 4.156717777252197, "lr": 0.0002840580745557017, "seconds": 767.15, "steps_per_second": 14.093}
{"event": "train", "step": 10450, "loss": 4.073525428771973, "lr": 0.00028297222318190404, "seconds": 768.81, "steps_per_second": 14.098}
{"event": "train", "step": 10475, "loss": 3.9492907524108887, "lr": 0.000281887565959032, "seconds": 770.43, "steps_per_second": 14.105}
{"event": "train", "step": 10500, "loss": 4.159026622772217, "lr": 0.0002808041237687693, "seconds": 771.89, "steps_per_second": 14.119}
{"event": "validation", "step": 10500, "val_loss": 4.1872429816469765, "best_val": 4.1872429816469765, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 10525, "loss": 4.125779628753662, "lr": 0.0002797219174694086, "seconds": 773.91, "steps_per_second": 14.108}
{"event": "train", "step": 10550, "loss": 4.03872013092041, "lr": 0.0002786409678954484, "seconds": 775.38, "steps_per_second": 14.121}
{"event": "train", "step": 10575, "loss": 4.038529396057129, "lr": 0.0002775612958571936, "seconds": 776.76, "steps_per_second": 14.138}
{"event": "train", "step": 10600, "loss": 4.120691299438477, "lr": 0.0002764829221403539, "seconds": 778.13, "steps_per_second": 14.155}
{"event": "train", "step": 10625, "loss": 4.155698299407959, "lr": 0.0002754058675056439, "seconds": 779.54, "steps_per_second": 14.171}
{"event": "train", "step": 10650, "loss": 4.1112847328186035, "lr": 0.00027433015268838346, "seconds": 780.92, "steps_per_second": 14.187}
{"event": "train", "step": 10675, "loss": 3.8501110076904297, "lr": 0.00027325579839809836, "seconds": 782.29, "steps_per_second": 14.204}
{"event": "train", "step": 10700, "loss": 4.025851726531982, "lr": 0.0002721828253181217, "seconds": 783.67, "steps_per_second": 14.221}
{"event": "train", "step": 10725, "loss": 3.903761625289917, "lr": 0.0002711112541051956, "seconds": 785.09, "steps_per_second": 14.235}
{"event": "train", "step": 10750, "loss": 4.006229400634766, "lr": 0.00027004110538907404, "seconds": 786.45, "steps_per_second": 14.252}
{"event": "train", "step": 10775, "loss": 4.188235282897949, "lr": 0.00026897239977212456, "seconds": 787.82, "steps_per_second": 14.268}
{"event": "train", "step": 10800, "loss": 3.9729199409484863, "lr": 0.0002679051578289329, "seconds": 789.22, "steps_per_second": 14.284}
{"event": "validation", "step": 10800, "val_loss": 4.176027377758968, "best_val": 4.176027377758968, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 10825, "loss": 3.9741594791412354, "lr": 0.00026683940010590625, "seconds": 791.1, "steps_per_second": 14.278}
{"event": "train", "step": 10850, "loss": 4.0365095138549805, "lr": 0.0002657751471208775, "seconds": 792.47, "steps_per_second": 14.294}
{"event": "train", "step": 10875, "loss": 4.124606132507324, "lr": 0.00026471241936271085, "seconds": 793.86, "steps_per_second": 14.309}
{"event": "train", "step": 10900, "loss": 4.184157848358154, "lr": 0.0002636512372909069, "seconds": 795.26, "steps_per_second": 14.324}
{"event": "train", "step": 10925, "loss": 4.02307653427124, "lr": 0.0002625916213352087, "seconds": 796.63, "steps_per_second": 14.339}
{"event": "train", "step": 10950, "loss": 3.9579830169677734, "lr": 0.0002615335918952091, "seconds": 798.0, "steps_per_second": 14.355}
{"event": "train", "step": 10975, "loss": 4.1235761642456055, "lr": 0.0002604771693399571, "seconds": 799.39, "steps_per_second": 14.369}
{"event": "train", "step": 11000, "loss": 4.044044017791748, "lr": 0.00025942237400756633, "seconds": 800.78, "steps_per_second": 14.384}
{"event": "train", "step": 11025, "loss": 4.020712852478027, "lr": 0.00025836922620482315, "seconds": 802.15, "steps_per_second": 14.399}
{"event": "train", "step": 11050, "loss": 3.9674909114837646, "lr": 0.00025731774620679624, "seconds": 803.52, "steps_per_second": 14.414}
{"event": "train", "step": 11075, "loss": 4.072983264923096, "lr": 0.0002562679542564458, "seconds": 804.98, "steps_per_second": 14.425}
{"event": "train", "step": 11100, "loss": 4.026339054107666, "lr": 0.0002552198705642336, "seconds": 806.36, "steps_per_second": 14.44}
{"event": "validation", "step": 11100, "val_loss": 4.174401251183372, "best_val": 4.174401251183372, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 11125, "loss": 3.9229583740234375, "lr": 0.00025417351530773475, "seconds": 808.24, "steps_per_second": 14.434}
{"event": "train", "step": 11150, "loss": 3.943262815475464, "lr": 0.00025312890863124845, "seconds": 809.64, "steps_per_second": 14.447}
{"event": "train", "step": 11175, "loss": 3.930509328842163, "lr": 0.0002520860706454104, "seconds": 811.03, "steps_per_second": 14.461}
{"event": "train", "step": 11200, "loss": 4.042850017547607, "lr": 0.00025104502142680606, "seconds": 812.4, "steps_per_second": 14.475}
{"event": "train", "step": 11225, "loss": 3.954432964324951, "lr": 0.00025000578101758326, "seconds": 813.79, "steps_per_second": 14.489}
{"event": "train", "step": 11250, "loss": 3.952489137649536, "lr": 0.0002489683694250671, "seconds": 815.19, "steps_per_second": 14.502}
{"event": "train", "step": 11275, "loss": 4.0661797523498535, "lr": 0.0002479328066213747, "seconds": 816.56, "steps_per_second": 14.516}
{"event": "train", "step": 11300, "loss": 4.070044994354248, "lr": 0.0002468991125430303, "seconds": 817.93, "steps_per_second": 14.53}
{"event": "train", "step": 11325, "loss": 4.07045316696167, "lr": 0.00024586730709058173, "seconds": 819.33, "steps_per_second": 14.543}
{"event": "train", "step": 11350, "loss": 3.9518063068389893, "lr": 0.0002448374101282173, "seconds": 820.72, "steps_per_second": 14.556}
{"event": "train", "step": 11375, "loss": 4.050046443939209, "lr": 0.0002438094414833832, "seconds": 822.1, "steps_per_second": 14.569}
{"event": "train", "step": 11400, "loss": 4.117152214050293, "lr": 0.0002427834209464021, "seconds": 823.47, "steps_per_second": 14.583}
{"event": "validation", "step": 11400, "val_loss": 4.161108590880561, "best_val": 4.161108590880561, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 11425, "loss": 4.039085388183594, "lr": 0.00024175936827009156, "seconds": 825.37, "steps_per_second": 14.575}
{"event": "train", "step": 11450, "loss": 3.9308724403381348, "lr": 0.00024073730316938454, "seconds": 826.74, "steps_per_second": 14.589}
{"event": "train", "step": 11475, "loss": 3.917080879211426, "lr": 0.00023971724532094915, "seconds": 828.1, "steps_per_second": 14.603}
{"event": "train", "step": 11500, "loss": 4.013641357421875, "lr": 0.00023869921436281037, "seconds": 829.5, "steps_per_second": 14.615}
{"event": "train", "step": 11525, "loss": 4.08280086517334, "lr": 0.00023768322989397165, "seconds": 830.89, "steps_per_second": 14.627}
{"event": "train", "step": 11550, "loss": 4.007204055786133, "lr": 0.00023666931147403768, "seconds": 832.26, "steps_per_second": 14.64}
{"event": "train", "step": 11575, "loss": 4.010478496551514, "lr": 0.00023565747862283774, "seconds": 833.64, "steps_per_second": 14.653}
{"event": "train", "step": 11600, "loss": 4.162961006164551, "lr": 0.00023464775082005014, "seconds": 835.05, "steps_per_second": 14.664}
{"event": "train", "step": 11625, "loss": 4.026665210723877, "lr": 0.00023364014750482706, "seconds": 836.42, "steps_per_second": 14.677}
{"event": "train", "step": 11650, "loss": 4.020020961761475, "lr": 0.00023263468807542014, "seconds": 837.79, "steps_per_second": 14.69}
{"event": "train", "step": 11675, "loss": 3.921314239501953, "lr": 0.00023163139188880745, "seconds": 839.19, "steps_per_second": 14.701}
{"event": "train", "step": 11700, "loss": 4.106010437011719, "lr": 0.00023063027826032047, "seconds": 840.58, "steps_per_second": 14.713}
{"event": "validation", "step": 11700, "val_loss": 4.155644437849079, "best_val": 4.155644437849079, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 11725, "loss": 4.135347843170166, "lr": 0.00022963136646327235, "seconds": 842.51, "steps_per_second": 14.704}
{"event": "train", "step": 11750, "loss": 4.075051307678223, "lr": 0.00022863467572858669, "seconds": 843.9, "steps_per_second": 14.716}
{"event": "train", "step": 11775, "loss": 4.006618022918701, "lr": 0.00022764022524442783, "seconds": 845.3, "steps_per_second": 14.727}
{"event": "train", "step": 11800, "loss": 4.00972843170166, "lr": 0.00022664803415583075, "seconds": 846.67, "steps_per_second": 14.739}
{"event": "train", "step": 11825, "loss": 4.052868366241455, "lr": 0.0002256581215643331, "seconds": 848.04, "steps_per_second": 14.752}
{"event": "train", "step": 11850, "loss": 3.9338817596435547, "lr": 0.00022467050652760713, "seconds": 849.44, "steps_per_second": 14.763}
{"event": "train", "step": 11875, "loss": 4.01691198348999, "lr": 0.00022368520805909258, "seconds": 850.82, "steps_per_second": 14.774}
{"event": "train", "step": 11900, "loss": 4.100257396697998, "lr": 0.0002227022451276313, "seconds": 852.19, "steps_per_second": 14.786}
{"event": "train", "step": 11925, "loss": 4.021249771118164, "lr": 0.00022172163665710133, "seconds": 853.57, "steps_per_second": 14.798}
{"event": "train", "step": 11950, "loss": 3.976917266845703, "lr": 0.00022074340152605332, "seconds": 854.97, "steps_per_second": 14.808}
{"event": "train", "step": 11975, "loss": 4.029576778411865, "lr": 0.00021976755856734635, "seconds": 856.35, "steps_per_second": 14.82}
{"event": "train", "step": 12000, "loss": 4.0975661277771, "lr": 0.00021879412656778584, "seconds": 857.74, "steps_per_second": 14.831}
{"event": "validation", "step": 12000, "val_loss": 4.1537969840463465, "best_val": 4.1537969840463465, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 12025, "loss": 4.037871360778809, "lr": 0.00021782312426776174, "seconds": 859.65, "steps_per_second": 14.822}
{"event": "train", "step": 12050, "loss": 4.014781475067139, "lr": 0.00021685457036088764, "seconds": 861.03, "steps_per_second": 14.833}
{"event": "train", "step": 12075, "loss": 4.154693603515625, "lr": 0.00021588848349364103, "seconds": 862.4, "steps_per_second": 14.845}
{"event": "train", "step": 12100, "loss": 3.9709935188293457, "lr": 0.00021492488226500434, "seconds": 863.78, "steps_per_second": 14.856}
{"event": "train", "step": 12125, "loss": 4.048872470855713, "lr": 0.0002139637852261067, "seconds": 865.24, "steps_per_second": 14.864}
{"event": "train", "step": 12150, "loss": 4.102145195007324, "lr": 0.00021300521087986684, "seconds": 866.61, "steps_per_second": 14.875}
{"event": "train", "step": 12175, "loss": 4.020147800445557, "lr": 0.00021204917768063715, "seconds": 867.98, "steps_per_second": 14.886}
{"event": "train", "step": 12200, "loss": 4.0142340660095215, "lr": 0.00021109570403384783, "seconds": 869.38, "steps_per_second": 14.896}
{"event": "train", "step": 12225, "loss": 4.002983570098877, "lr": 0.0002101448082956533, "seconds": 870.77, "steps_per_second": 14.907}
{"event": "train", "step": 12250, "loss": 4.097994327545166, "lr": 0.00020919650877257827, "seconds": 872.15, "steps_per_second": 14.917}
{"event": "train", "step": 12275, "loss": 4.068997859954834, "lr": 0.00020825082372116527, "seconds": 873.54, "steps_per_second": 14.927}
{"event": "train", "step": 12300, "loss": 4.141541957855225, "lr": 0.00020730777134762355, "seconds": 874.95, "steps_per_second": 14.937}
{"event": "validation", "step": 12300, "val_loss": 4.142380691709972, "best_val": 4.142380691709972, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 12325, "loss": 4.016695022583008, "lr": 0.00020636736980747818, "seconds": 876.82, "steps_per_second": 14.93}
{"event": "train", "step": 12350, "loss": 3.9711034297943115, "lr": 0.00020542963720522085, "seconds": 878.19, "steps_per_second": 14.94}
{"event": "train", "step": 12375, "loss": 3.948007345199585, "lr": 0.00020449459159396134, "seconds": 879.59, "steps_per_second": 14.95}
{"event": "train", "step": 12400, "loss": 3.942763090133667, "lr": 0.0002035622509750796, "seconds": 880.97, "steps_per_second": 14.96}
{"event": "train", "step": 12425, "loss": 3.9649791717529297, "lr": 0.00020263263329787948, "seconds": 882.34, "steps_per_second": 14.971}
{"event": "train", "step": 12450, "loss": 3.8987715244293213, "lr": 0.00020170575645924303, "seconds": 883.72, "steps_per_second": 14.981}
{"event": "train", "step": 12475, "loss": 3.988473415374756, "lr": 0.00020078163830328615, "seconds": 885.13, "steps_per_second": 14.99}
{"event": "train", "step": 12500, "loss": 4.081761837005615, "lr": 0.00019986029662101478, "seconds": 886.51, "steps_per_second": 15.0}
{"event": "train", "step": 12525, "loss": 4.056788921356201, "lr": 0.0001989417491499829, "seconds": 887.87, "steps_per_second": 15.01}
{"event": "train", "step": 12550, "loss": 4.020052433013916, "lr": 0.00019802601357395017, "seconds": 889.27, "steps_per_second": 15.019}
{"event": "train", "step": 12575, "loss": 3.9784090518951416, "lr": 0.0001971131075225422, "seconds": 890.67, "steps_per_second": 15.029}
{"event": "train", "step": 12600, "loss": 4.114835739135742, "lr": 0.0001962030485709112, "seconds": 892.03, "steps_per_second": 15.039}
{"event": "validation", "step": 12600, "val_loss": 4.139590696665646, "best_val": 4.139590696665646, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 12625, "loss": 3.925321102142334, "lr": 0.00019529585423939692, "seconds": 893.92, "steps_per_second": 15.031}
{"event": "train", "step": 12650, "loss": 3.9052674770355225, "lr": 0.0001943915419931903, "seconds": 895.37, "steps_per_second": 15.038}
{"event": "train", "step": 12675, "loss": 3.958151340484619, "lr": 0.00019349012924199644, "seconds": 896.75, "steps_per_second": 15.048}
{"event": "train", "step": 12700, "loss": 3.9729714393615723, "lr": 0.0001925916333396997, "seconds": 898.12, "steps_per_second": 15.058}
{"event": "train", "step": 12725, "loss": 3.8115499019622803, "lr": 0.00019169607158402984, "seconds": 899.52, "steps_per_second": 15.067}
{"event": "train", "step": 12750, "loss": 4.022032260894775, "lr": 0.00019080346121622853, "seconds": 900.9, "steps_per_second": 15.076}
{"event": "train", "step": 12775, "loss": 4.136622428894043, "lr": 0.000189913819420718, "seconds": 902.27, "steps_per_second": 15.086}
{"event": "train", "step": 12800, "loss": 3.981607437133789, "lr": 0.00018902716332476957, "seconds": 903.65, "steps_per_second": 15.095}
{"event": "train", "step": 12825, "loss": 4.0626912117004395, "lr": 0.00018814350999817457, "seconds": 905.05, "steps_per_second": 15.103}
{"event": "train", "step": 12850, "loss": 3.9644434452056885, "lr": 0.00018726287645291512, "seconds": 906.43, "steps_per_second": 15.113}
{"event": "train", "step": 12875, "loss": 4.045288562774658, "lr": 0.00018638527964283714, "seconds": 907.8, "steps_per_second": 15.122}
{"event": "train", "step": 12900, "loss": 3.9891390800476074, "lr": 0.00018551073646332346, "seconds": 909.19, "steps_per_second": 15.131}
{"event": "validation", "step": 12900, "val_loss": 4.137555099877273, "best_val": 4.137555099877273, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 12925, "loss": 4.049253940582275, "lr": 0.00018463926375096893, "seconds": 911.08, "steps_per_second": 15.123}
{"event": "train", "step": 12950, "loss": 3.9198155403137207, "lr": 0.00018377087828325625, "seconds": 912.45, "steps_per_second": 15.132}
{"event": "train", "step": 12975, "loss": 3.9329543113708496, "lr": 0.0001829055967782328, "seconds": 913.83, "steps_per_second": 15.141}
{"event": "train", "step": 13000, "loss": 3.8612966537475586, "lr": 0.00018204343589418875, "seconds": 915.23, "steps_per_second": 15.149}
{"event": "train", "step": 13025, "loss": 3.979663133621216, "lr": 0.0001811844122293365, "seconds": 916.6, "steps_per_second": 15.159}
{"event": "train", "step": 13050, "loss": 3.986945629119873, "lr": 0.00018032854232149115, "seconds": 917.97, "steps_per_second": 15.168}
{"event": "train", "step": 13075, "loss": 3.9816672801971436, "lr": 0.0001794758426477522, "seconds": 919.36, "steps_per_second": 15.176}
{"event": "train", "step": 13100, "loss": 3.9616899490356445, "lr": 0.00017862632962418608, "seconds": 920.75, "steps_per_second": 15.184}
{"event": "train", "step": 13125, "loss": 3.9084160327911377, "lr": 0.00017778001960551014, "seconds": 922.12, "steps_per_second": 15.194}
{"event": "train", "step": 13150, "loss": 3.8702752590179443, "lr": 0.00017693692888477803, "seconds": 923.49, "steps_per_second": 15.202}
{"event": "train", "step": 13175, "loss": 3.918766975402832, "lr": 0.00017609707369306581, "seconds": 924.9, "steps_per_second": 15.21}
{"event": "train", "step": 13200, "loss": 3.9189038276672363, "lr": 0.00017526047019915955, "seconds": 926.28, "steps_per_second": 15.218}
{"event": "validation", "step": 13200, "val_loss": 4.129040297278674, "best_val": 4.129040297278674, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 13225, "loss": 3.9694714546203613, "lr": 0.00017442713450924405, "seconds": 928.14, "steps_per_second": 15.211}
{"event": "train", "step": 13250, "loss": 3.8792941570281982, "lr": 0.0001735970826665928, "seconds": 929.55, "steps_per_second": 15.219}
{"event": "train", "step": 13275, "loss": 3.97754168510437, "lr": 0.00017277033065125905, "seconds": 930.93, "steps_per_second": 15.227}
{"event": "train", "step": 13300, "loss": 4.001214027404785, "lr": 0.00017194689437976819, "seconds": 932.31, "steps_per_second": 15.236}
{"event": "train", "step": 13325, "loss": 3.961507797241211, "lr": 0.00017112678970481116, "seconds": 933.69, "steps_per_second": 15.244}
{"event": "train", "step": 13350, "loss": 3.956632375717163, "lr": 0.0001703100324149399, "seconds": 935.09, "steps_per_second": 15.251}
{"event": "train", "step": 13375, "loss": 3.895660638809204, "lr": 0.0001694966382342623, "seconds": 936.47, "steps_per_second": 15.26}
{"event": "train", "step": 13400, "loss": 4.078010559082031, "lr": 0.00016868662282214063, "seconds": 937.83, "steps_per_second": 15.268}
{"event": "train", "step": 13425, "loss": 3.886416435241699, "lr": 0.00016788000177288932, "seconds": 939.23, "steps_per_second": 15.276}
{"event": "train", "step": 13450, "loss": 3.9593160152435303, "lr": 0.00016707679061547464, "seconds": 940.62, "steps_per_second": 15.283}
{"event": "train", "step": 13475, "loss": 3.88328218460083, "lr": 0.0001662770048132166, "seconds": 941.99, "steps_per_second": 15.292}
{"event": "train", "step": 13500, "loss": 3.9253427982330322, "lr": 0.0001654806597634904, "seconds": 943.37, "steps_per_second": 15.3}
{"event": "validation", "step": 13500, "val_loss": 4.129648849608961, "best_val": 4.129040297278674, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 13525, "loss": 3.752774715423584, "lr": 0.00016468777079743036, "seconds": 945.22, "steps_per_second": 15.293}
{"event": "train", "step": 13550, "loss": 3.981440305709839, "lr": 0.00016389835317963478, "seconds": 946.6, "steps_per_second": 15.301}
{"event": "train", "step": 13575, "loss": 3.855221748352051, "lr": 0.0001631124221078718, "seconds": 947.96, "steps_per_second": 15.309}
{"event": "train", "step": 13600, "loss": 3.874601125717163, "lr": 0.00016232999271278728, "seconds": 949.36, "steps_per_second": 15.317}
{"event": "train", "step": 13625, "loss": 3.9991536140441895, "lr": 0.000161551080057613, "seconds": 950.75, "steps_per_second": 15.324}
{"event": "train", "step": 13650, "loss": 3.8567564487457275, "lr": 0.0001607756991378773, "seconds": 952.12, "steps_per_second": 15.332}
{"event": "train", "step": 13675, "loss": 4.0319437980651855, "lr": 0.00016000386488111535, "seconds": 953.49, "steps_per_second": 15.34}
{"event": "train", "step": 13700, "loss": 4.0534563064575195, "lr": 0.00015923559214658304, "seconds": 954.9, "steps_per_second": 15.347}
{"event": "train", "step": 13725, "loss": 4.015703201293945, "lr": 0.00015847089572496994, "seconds": 956.28, "steps_per_second": 15.355}
{"event": "train", "step": 13750, "loss": 3.9133460521698, "lr": 0.00015770979033811485, "seconds": 957.65, "steps_per_second": 15.362}
{"event": "train", "step": 13775, "loss": 3.8987298011779785, "lr": 0.00015695229063872286, "seconds": 959.04, "steps_per_second": 15.369}
{"event": "train", "step": 13800, "loss": 4.040350914001465, "lr": 0.00015619841121008229, "seconds": 960.44, "steps_per_second": 15.376}
{"event": "validation", "step": 13800, "val_loss": 4.123351766012183, "best_val": 4.123351766012183, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 13825, "loss": 4.000515937805176, "lr": 0.00015544816656578474, "seconds": 962.3, "steps_per_second": 15.369}
{"event": "train", "step": 13850, "loss": 3.8551037311553955, "lr": 0.0001547015711494454, "seconds": 963.68, "steps_per_second": 15.377}
{"event": "train", "step": 13875, "loss": 4.011075019836426, "lr": 0.00015395863933442488, "seconds": 965.08, "steps_per_second": 15.383}
{"event": "train", "step": 13900, "loss": 3.9526467323303223, "lr": 0.00015321938542355268, "seconds": 966.46, "steps_per_second": 15.391}
{"event": "train", "step": 13925, "loss": 3.9622597694396973, "lr": 0.00015248382364885173, "seconds": 967.82, "steps_per_second": 15.399}
{"event": "train", "step": 13950, "loss": 3.976205825805664, "lr": 0.00015175196817126437, "seconds": 969.21, "steps_per_second": 15.406}
{"event": "train", "step": 13975, "loss": 3.9355404376983643, "lr": 0.0001510238330803799, "seconds": 970.6, "steps_per_second": 15.413}
{"event": "train", "step": 14000, "loss": 3.8906071186065674, "lr": 0.0001502994323941631, "seconds": 971.97, "steps_per_second": 15.42}
{"event": "train", "step": 14025, "loss": 3.8716602325439453, "lr": 0.00014957878005868438, "seconds": 973.34, "steps_per_second": 15.428}
{"event": "train", "step": 14050, "loss": 4.0244140625, "lr": 0.00014886188994785174, "seconds": 974.75, "steps_per_second": 15.434}
{"event": "train", "step": 14075, "loss": 4.016683578491211, "lr": 0.00014814877586314285, "seconds": 976.13, "steps_per_second": 15.441}
{"event": "train", "step": 14100, "loss": 4.051560401916504, "lr": 0.00014743945153334017, "seconds": 977.5, "steps_per_second": 15.448}
{"event": "validation", "step": 14100, "val_loss": 4.120772243014795, "best_val": 4.120772243014795, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 14125, "loss": 3.8667123317718506, "lr": 0.000146733930614266, "seconds": 979.4, "steps_per_second": 15.44}
{"event": "train", "step": 14150, "loss": 3.955408811569214, "lr": 0.00014603222668851988, "seconds": 980.79, "steps_per_second": 15.447}
{"event": "train", "step": 14175, "loss": 3.8536643981933594, "lr": 0.00014533435326521735, "seconds": 982.15, "steps_per_second": 15.454}
{"event": "train", "step": 14200, "loss": 3.9318997859954834, "lr": 0.00014464032377972934, "seconds": 983.54, "steps_per_second": 15.461}
{"event": "train", "step": 14225, "loss": 3.8986382484436035, "lr": 0.0001439501515934239, "seconds": 984.94, "steps_per_second": 15.467}
{"event": "train", "step": 14250, "loss": 3.9918274879455566, "lr": 0.00014326384999340872, "seconds": 986.32, "steps_per_second": 15.474}
{"event": "train", "step": 14275, "loss": 3.9722113609313965, "lr": 0.00014258143219227556, "seconds": 987.68, "steps_per_second": 15.481}
{"event": "train", "step": 14300, "loss": 3.937690496444702, "lr": 0.00014190291132784577, "seconds": 989.07, "steps_per_second": 15.488}
{"event": "train", "step": 14325, "loss": 3.869316816329956, "lr": 0.00014122830046291732, "seconds": 990.57, "steps_per_second": 15.491}
{"event": "train", "step": 14350, "loss": 4.0430827140808105, "lr": 0.00014055761258501334, "seconds": 992.0, "steps_per_second": 15.497}
{"event": "train", "step": 14375, "loss": 4.0295867919921875, "lr": 0.00013989086060613225, "seconds": 993.37, "steps_per_second": 15.503}
{"event": "train", "step": 14400, "loss": 3.9939982891082764, "lr": 0.00013922805736249887, "seconds": 994.77, "steps_per_second": 15.509}
{"event": "validation", "step": 14400, "val_loss": 4.1164134503869105, "best_val": 4.1164134503869105, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 14425, "loss": 3.9024441242218018, "lr": 0.00013856921561431759, "seconds": 996.64, "steps_per_second": 15.502}
{"event": "train", "step": 14450, "loss": 4.022858142852783, "lr": 0.00013791434804552637, "seconds": 998.01, "steps_per_second": 15.509}
{"event": "train", "step": 14475, "loss": 3.9903409481048584, "lr": 0.00013726346726355317, "seconds": 999.4, "steps_per_second": 15.515}
{"event": "train", "step": 14500, "loss": 3.8263838291168213, "lr": 0.00013661658579907252, "seconds": 1000.79, "steps_per_second": 15.522}
{"event": "train", "step": 14525, "loss": 3.9471678733825684, "lr": 0.00013597371610576462, "seconds": 1002.16, "steps_per_second": 15.528}
{"event": "train", "step": 14550, "loss": 4.033421516418457, "lr": 0.00013533487056007577, "seconds": 1003.54, "steps_per_second": 15.535}
{"event": "train", "step": 14575, "loss": 3.8231101036071777, "lr": 0.00013470006146097945, "seconds": 1004.95, "steps_per_second": 15.541}
{"event": "train", "step": 14600, "loss": 4.0644121170043945, "lr": 0.0001340693010297405, "seconds": 1006.32, "steps_per_second": 15.547}
{"event": "train", "step": 14625, "loss": 4.148835182189941, "lr": 0.00013344260140967903, "seconds": 1007.69, "steps_per_second": 15.554}
{"event": "train", "step": 14650, "loss": 3.893329620361328, "lr": 0.00013281997466593694, "seconds": 1009.08, "steps_per_second": 15.56}
{"event": "train", "step": 14675, "loss": 4.049973964691162, "lr": 0.00013220143278524574, "seconds": 1010.48, "steps_per_second": 15.566}
{"event": "train", "step": 14700, "loss": 3.8926432132720947, "lr": 0.0001315869876756955, "seconds": 1011.85, "steps_per_second": 15.572}
{"event": "validation", "step": 14700, "val_loss": 4.113153930729057, "best_val": 4.113153930729057, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 14725, "loss": 3.9321088790893555, "lr": 0.0001309766511665059, "seconds": 1013.73, "steps_per_second": 15.565}
{"event": "train", "step": 14750, "loss": 3.9131717681884766, "lr": 0.0001303704350077984, "seconds": 1015.13, "steps_per_second": 15.57}
{"event": "train", "step": 14775, "loss": 3.9610042572021484, "lr": 0.00012976835087036992, "seconds": 1016.5, "steps_per_second": 15.577}
{"event": "train", "step": 14800, "loss": 3.986732244491577, "lr": 0.00012917041034546815, "seconds": 1017.87, "steps_per_second": 15.583}
{"event": "train", "step": 14825, "loss": 3.91976261138916, "lr": 0.00012857662494456863, "seconds": 1019.26, "steps_per_second": 15.589}
{"event": "train", "step": 14850, "loss": 3.85955810546875, "lr": 0.00012798700609915285, "seconds": 1020.66, "steps_per_second": 15.595}
{"event": "train", "step": 14875, "loss": 3.9830238819122314, "lr": 0.00012740156516048834, "seconds": 1022.03, "steps_per_second": 15.601}
{"event": "train", "step": 14900, "loss": 3.9140448570251465, "lr": 0.0001268203133994102, "seconds": 1023.4, "steps_per_second": 15.608}
{"event": "train", "step": 14925, "loss": 3.86126446723938, "lr": 0.0001262432620061039, "seconds": 1024.8, "steps_per_second": 15.613}
{"event": "train", "step": 14950, "loss": 3.995065689086914, "lr": 0.00012567042208988998, "seconds": 1026.17, "steps_per_second": 15.619}
{"event": "train", "step": 14975, "loss": 3.8629207611083984, "lr": 0.00012510180467901016, "seconds": 1027.54, "steps_per_second": 15.626}
{"event": "train", "step": 15000, "loss": 4.002325057983398, "lr": 0.00012453742072041497, "seconds": 1028.93, "steps_per_second": 15.631}
{"event": "validation", "step": 15000, "val_loss": 4.104870636632944, "best_val": 4.104870636632944, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 15025, "loss": 3.872110366821289, "lr": 0.0001239772810795532, "seconds": 1030.82, "steps_per_second": 15.623}
{"event": "train", "step": 15050, "loss": 3.9170007705688477, "lr": 0.0001234213965401624, "seconds": 1032.19, "steps_per_second": 15.63}
{"event": "train", "step": 15075, "loss": 3.957706928253174, "lr": 0.0001228697778040616, "seconds": 1033.57, "steps_per_second": 15.636}
{"event": "train", "step": 15100, "loss": 3.773390293121338, "lr": 0.0001223224354909449, "seconds": 1034.97, "steps_per_second": 15.641}
{"event": "train", "step": 15125, "loss": 3.928734302520752, "lr": 0.00012177938013817751, "seconds": 1036.34, "steps_per_second": 15.647}
{"event": "train", "step": 15150, "loss": 4.0284423828125, "lr": 0.00012124062220059241, "seconds": 1037.72, "steps_per_second": 15.653}
{"event": "train", "step": 15175, "loss": 3.948892116546631, "lr": 0.00012070617205028947, "seconds": 1039.12, "steps_per_second": 15.658}
{"event": "train", "step": 15200, "loss": 3.9482157230377197, "lr": 0.00012017603997643543, "seconds": 1040.51, "steps_per_second": 15.664}
{"event": "train", "step": 15225, "loss": 3.8670356273651123, "lr": 0.00011965023618506607, "seconds": 1041.88, "steps_per_second": 15.67}
{"event": "train", "step": 15250, "loss": 3.88989520072937, "lr": 0.00011912877079888951, "seconds": 1043.26, "steps_per_second": 15.676}
{"event": "train", "step": 15275, "loss": 3.9598140716552734, "lr": 0.00011861165385709147, "seconds": 1044.71, "steps_per_second": 15.679}
{"event": "train", "step": 15300, "loss": 3.8782742023468018, "lr": 0.00011809889531514211, "seconds": 1046.09, "steps_per_second": 15.685}
{"event": "validation", "step": 15300, "val_loss": 4.1045674159048735, "best_val": 4.1045674159048735, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 15325, "loss": 3.932661533355713, "lr": 0.0001175905050446041, "seconds": 1047.95, "steps_per_second": 15.678}
{"event": "train", "step": 15350, "loss": 3.8676347732543945, "lr": 0.00011708649283294268, "seconds": 1049.34, "steps_per_second": 15.683}
{"event": "train", "step": 15375, "loss": 3.7389674186706543, "lr": 0.00011658686838333735, "seconds": 1050.73, "steps_per_second": 15.689}
{"event": "train", "step": 15400, "loss": 3.8691658973693848, "lr": 0.00011609164131449486, "seconds": 1052.1, "steps_per_second": 15.694}
{"event": "train", "step": 15425, "loss": 3.9369187355041504, "lr": 0.00011560082116046424, "seconds": 1053.48, "steps_per_second": 15.7}
{"event": "train", "step": 15450, "loss": 3.771369695663452, "lr": 0.00011511441737045304, "seconds": 1054.94, "steps_per_second": 15.704}
{"event": "train", "step": 15475, "loss": 3.82472825050354, "lr": 0.00011463243930864569, "seconds": 1056.31, "steps_per_second": 15.709}
{"event": "train", "step": 15500, "loss": 3.8801560401916504, "lr": 0.0001141548962540229, "seconds": 1057.68, "steps_per_second": 15.715}
{"event": "train", "step": 15525, "loss": 3.961092948913574, "lr": 0.00011368179740018325, "seconds": 1059.06, "steps_per_second": 15.72}
{"event": "train", "step": 15550, "loss": 3.9295175075531006, "lr": 0.00011321315185516618, "seconds": 1060.46, "steps_per_second": 15.725}
{"event": "train", "step": 15575, "loss": 3.860691547393799, "lr": 0.0001127489686412764, "seconds": 1061.83, "steps_per_second": 15.731}
{"event": "train", "step": 15600, "loss": 3.7876617908477783, "lr": 0.0001122892566949107, "seconds": 1063.2, "steps_per_second": 15.737}
{"event": "validation", "step": 15600, "val_loss": 4.100860886096746, "best_val": 4.100860886096746, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 15625, "loss": 3.9611332416534424, "lr": 0.00011183402486638529, "seconds": 1065.11, "steps_per_second": 15.729}
{"event": "train", "step": 15650, "loss": 3.9775142669677734, "lr": 0.00011138328191976593, "seconds": 1066.49, "steps_per_second": 15.734}
{"event": "train", "step": 15675, "loss": 3.841099977493286, "lr": 0.00011093703653269886, "seconds": 1067.86, "steps_per_second": 15.74}
{"event": "train", "step": 15700, "loss": 3.9532198905944824, "lr": 0.00011049529729624383, "seconds": 1069.26, "steps_per_second": 15.744}
{"event": "train", "step": 15725, "loss": 3.8634376525878906, "lr": 0.00011005807271470894, "seconds": 1070.64, "steps_per_second": 15.75}
{"event": "train", "step": 15750, "loss": 4.021152019500732, "lr": 0.00010962537120548662, "seconds": 1072.01, "steps_per_second": 15.755}
{"event": "train", "step": 15775, "loss": 3.9089467525482178, "lr": 0.00010919720109889161, "seconds": 1073.38, "steps_per_second": 15.761}
{"event": "train", "step": 15800, "loss": 3.8740389347076416, "lr": 0.00010877357063800076, "seconds": 1074.8, "steps_per_second": 15.765}
{"event": "train", "step": 15825, "loss": 3.948869228363037, "lr": 0.0001083544879784942, "seconds": 1076.17, "steps_per_second": 15.77}
{"event": "train", "step": 15850, "loss": 3.9106132984161377, "lr": 0.00010793996118849831, "seconds": 1077.54, "steps_per_second": 15.776}
{"event": "train", "step": 15875, "loss": 3.822164297103882, "lr": 0.00010752999824843065, "seconds": 1078.93, "steps_per_second": 15.781}
{"event": "train", "step": 15900, "loss": 3.9409430027008057, "lr": 0.00010712460705084586, "seconds": 1080.33, "steps_per_second": 15.785}
{"event": "validation", "step": 15900, "val_loss": 4.095645437494747, "best_val": 4.095645437494747, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 15925, "loss": 3.9004459381103516, "lr": 0.00010672379540028404, "seconds": 1082.19, "steps_per_second": 15.778}
{"event": "train", "step": 15950, "loss": 4.05908727645874, "lr": 0.00010632757101312051, "seconds": 1083.57, "steps_per_second": 15.784}
{"event": "train", "step": 15975, "loss": 3.8845467567443848, "lr": 0.00010593594151741711, "seconds": 1084.97, "steps_per_second": 15.788}
{"event": "train", "step": 16000, "loss": 3.836775302886963, "lr": 0.00010554891445277552, "seconds": 1086.34, "steps_per_second": 15.793}
{"event": "train", "step": 16025, "loss": 4.009103298187256, "lr": 0.00010516649727019182, "seconds": 1087.71, "steps_per_second": 15.799}
{"event": "train", "step": 16050, "loss": 3.9073476791381836, "lr": 0.00010478869733191334, "seconds": 1089.1, "steps_per_second": 15.803}
{"event": "train", "step": 16075, "loss": 3.8868367671966553, "lr": 0.0001044155219112968, "seconds": 1090.5, "steps_per_second": 15.808}
{"event": "train", "step": 16100, "loss": 3.9330832958221436, "lr": 0.00010404697819266822, "seconds": 1091.87, "steps_per_second": 15.813}
{"event": "train", "step": 16125, "loss": 3.9034957885742188, "lr": 0.00010368307327118474, "seconds": 1093.25, "steps_per_second": 15.818}
{"event": "train", "step": 16150, "loss": 3.915233850479126, "lr": 0.00010332381415269798, "seconds": 1094.65, "steps_per_second": 15.823}
{"event": "train", "step": 16175, "loss": 3.897836685180664, "lr": 0.00010296920775361909, "seconds": 1096.03, "steps_per_second": 15.828}
{"event": "train", "step": 16200, "loss": 3.8090710639953613, "lr": 0.0001026192609007857, "seconds": 1097.4, "steps_per_second": 15.833}
{"event": "validation", "step": 16200, "val_loss": 4.094592776242277, "best_val": 4.094592776242277, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 16225, "loss": 4.041939735412598, "lr": 0.0001022739803313304, "seconds": 1099.3, "steps_per_second": 15.825}
{"event": "train", "step": 16250, "loss": 3.812150239944458, "lr": 0.00010193337269255115, "seconds": 1100.7, "steps_per_second": 15.829}
{"event": "train", "step": 16275, "loss": 3.7652857303619385, "lr": 0.00010159744454178307, "seconds": 1102.07, "steps_per_second": 15.834}
{"event": "train", "step": 16300, "loss": 3.9070560932159424, "lr": 0.00010126620234627269, "seconds": 1103.45, "steps_per_second": 15.839}
{"event": "train", "step": 16325, "loss": 3.796112060546875, "lr": 0.0001009396524830528, "seconds": 1104.89, "steps_per_second": 15.843}
{"event": "train", "step": 16350, "loss": 3.8311285972595215, "lr": 0.00010061780123882015, "seconds": 1106.27, "steps_per_second": 15.847}
{"event": "train", "step": 16375, "loss": 3.875951051712036, "lr": 0.00010030065480981423, "seconds": 1107.64, "steps_per_second": 15.852}
{"event": "train", "step": 16400, "loss": 4.002318382263184, "lr": 9.998821930169802e-05, "seconds": 1109.03, "steps_per_second": 15.857}
{"event": "train", "step": 16425, "loss": 3.9195008277893066, "lr": 9.968050072944054e-05, "seconds": 1110.43, "steps_per_second": 15.861}
{"event": "train", "step": 16450, "loss": 4.050766944885254, "lr": 9.937750501720082e-05, "seconds": 1111.81, "steps_per_second": 15.866}
{"event": "train", "step": 16475, "loss": 3.849130630493164, "lr": 9.907923799821408e-05, "seconds": 1113.18, "steps_per_second": 15.871}
{"event": "train", "step": 16500, "loss": 3.7549922466278076, "lr": 9.878570541467928e-05, "seconds": 1114.58, "steps_per_second": 15.875}
{"event": "validation", "step": 16500, "val_loss": 4.08533077460806, "best_val": 4.08533077460806, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 16525, "loss": 3.847132682800293, "lr": 9.849691291764865e-05, "seconds": 1116.45, "steps_per_second": 15.868}
{"event": "train", "step": 16550, "loss": 3.99604868888855, "lr": 9.821286606691888e-05, "seconds": 1117.83, "steps_per_second": 15.873}
{"event": "train", "step": 16575, "loss": 3.926633834838867, "lr": 9.793357033092403e-05, "seconds": 1119.22, "steps_per_second": 15.877}
{"event": "train", "step": 16600, "loss": 3.8205406665802, "lr": 9.765903108663039e-05, "seconds": 1120.62, "steps_per_second": 15.881}
{"event": "train", "step": 16625, "loss": 3.840679168701172, "lr": 9.738925361943276e-05, "seconds": 1121.99, "steps_per_second": 15.886}
{"event": "train", "step": 16650, "loss": 3.8746302127838135, "lr": 9.712424312305294e-05, "seconds": 1123.36, "steps_per_second": 15.891}
{"event": "train", "step": 16675, "loss": 3.890223503112793, "lr": 9.686400469943943e-05, "seconds": 1124.76, "steps_per_second": 15.895}
{"event": "train", "step": 16700, "loss": 4.060847759246826, "lr": 9.660854335866962e-05, "seconds": 1126.14, "steps_per_second": 15.899}
{"event": "train", "step": 16725, "loss": 3.809231996536255, "lr": 9.635786401885288e-05, "seconds": 1127.51, "steps_per_second": 15.904}
{"event": "train", "step": 16750, "loss": 3.9807050228118896, "lr": 9.611197150603629e-05, "seconds": 1128.9, "steps_per_second": 15.909}
{"event": "train", "step": 16775, "loss": 3.9767873287200928, "lr": 9.587087055411143e-05, "seconds": 1130.3, "steps_per_second": 15.913}
{"event": "train", "step": 16800, "loss": 4.0203351974487305, "lr": 9.56345658047233e-05, "seconds": 1131.67, "steps_per_second": 15.917}
{"event": "validation", "step": 16800, "val_loss": 4.089837651525642, "best_val": 4.08533077460806, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 16825, "loss": 3.9471278190612793, "lr": 9.540306180718113e-05, "seconds": 1133.49, "steps_per_second": 15.912}
{"event": "train", "step": 16850, "loss": 3.771120309829712, "lr": 9.517636301837067e-05, "seconds": 1134.89, "steps_per_second": 15.915}
{"event": "train", "step": 16875, "loss": 3.8203306198120117, "lr": 9.495447380266839e-05, "seconds": 1136.27, "steps_per_second": 15.92}
{"event": "train", "step": 16900, "loss": 3.971783399581909, "lr": 9.473739843185737e-05, "seconds": 1137.64, "steps_per_second": 15.925}
{"event": "train", "step": 16925, "loss": 3.95509934425354, "lr": 9.452514108504529e-05, "seconds": 1139.03, "steps_per_second": 15.929}
{"event": "train", "step": 16950, "loss": 3.8940773010253906, "lr": 9.43177058485838e-05, "seconds": 1140.43, "steps_per_second": 15.933}
{"event": "train", "step": 16975, "loss": 3.878980875015259, "lr": 9.411509671598982e-05, "seconds": 1141.8, "steps_per_second": 15.937}
{"event": "train", "step": 17000, "loss": 3.766451120376587, "lr": 9.391731758786886e-05, "seconds": 1143.16, "steps_per_second": 15.942}
{"event": "train", "step": 17025, "loss": 3.8173441886901855, "lr": 9.372437227183962e-05, "seconds": 1144.56, "steps_per_second": 15.946}
{"event": "train", "step": 17050, "loss": 3.9150195121765137, "lr": 9.353626448246102e-05, "seconds": 1145.94, "steps_per_second": 15.95}
{"event": "train", "step": 17075, "loss": 3.8120532035827637, "lr": 9.335299784116043e-05, "seconds": 1147.31, "steps_per_second": 15.955}
{"event": "train", "step": 17100, "loss": 3.8543012142181396, "lr": 9.317457587616412e-05, "seconds": 1148.69, "steps_per_second": 15.959}
{"event": "validation", "step": 17100, "val_loss": 4.085307088362801, "best_val": 4.085307088362801, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 17125, "loss": 3.8818204402923584, "lr": 9.30010020224292e-05, "seconds": 1150.59, "steps_per_second": 15.952}
{"event": "train", "step": 17150, "loss": 3.932236909866333, "lr": 9.283227962157757e-05, "seconds": 1151.95, "steps_per_second": 15.956}
{"event": "train", "step": 17175, "loss": 3.876344680786133, "lr": 9.266841192183165e-05, "seconds": 1153.32, "steps_per_second": 15.961}
{"event": "train", "step": 17200, "loss": 3.757443904876709, "lr": 9.250940207795163e-05, "seconds": 1154.72, "steps_per_second": 15.965}
{"event": "train", "step": 17225, "loss": 4.004288196563721, "lr": 9.235525315117496e-05, "seconds": 1156.09, "steps_per_second": 15.969}
{"event": "train", "step": 17250, "loss": 3.9574475288391113, "lr": 9.220596810915732e-05, "seconds": 1157.47, "steps_per_second": 15.973}
{"event": "train", "step": 17275, "loss": 3.784423828125, "lr": 9.206154982591547e-05, "seconds": 1158.85, "steps_per_second": 15.977}
{"event": "train", "step": 17300, "loss": 3.658177614212036, "lr": 9.192200108177195e-05, "seconds": 1160.25, "steps_per_second": 15.981}
{"event": "train", "step": 17325, "loss": 3.9313998222351074, "lr": 9.178732456330164e-05, "seconds": 1161.62, "steps_per_second": 15.986}
{"event": "train", "step": 17350, "loss": 4.045326232910156, "lr": 9.165752286327975e-05, "seconds": 1162.98, "steps_per_second": 15.99}
{"event": "train", "step": 17375, "loss": 3.9883222579956055, "lr": 9.153259848063227e-05, "seconds": 1164.41, "steps_per_second": 15.993}
{"event": "train", "step": 17400, "loss": 3.891400098800659, "lr": 9.141255382038766e-05, "seconds": 1165.83, "steps_per_second": 15.996}
{"event": "validation", "step": 17400, "val_loss": 4.085205409765348, "best_val": 4.085205409765348, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 17425, "loss": 3.8507139682769775, "lr": 9.129739119363053e-05, "seconds": 1167.69, "steps_per_second": 15.99}
{"event": "train", "step": 17450, "loss": 3.8692872524261475, "lr": 9.118711281745723e-05, "seconds": 1169.08, "steps_per_second": 15.994}
{"event": "train", "step": 17475, "loss": 3.9868533611297607, "lr": 9.108172081493314e-05, "seconds": 1170.48, "steps_per_second": 15.997}
{"event": "train", "step": 17500, "loss": 3.78629994392395, "lr": 9.098121721505173e-05, "seconds": 1171.85, "steps_per_second": 16.002}
{"event": "train", "step": 17525, "loss": 3.7532293796539307, "lr": 9.088560395269568e-05, "seconds": 1173.22, "steps_per_second": 16.006}
{"event": "train", "step": 17550, "loss": 3.855069398880005, "lr": 9.079488286859942e-05, "seconds": 1174.62, "steps_per_second": 16.009}
{"event": "train", "step": 17575, "loss": 3.948608160018921, "lr": 9.070905570931382e-05, "seconds": 1176.01, "steps_per_second": 16.013}
{"event": "train", "step": 17600, "loss": 3.88506817817688, "lr": 9.062812412717246e-05, "seconds": 1177.38, "steps_per_second": 16.017}
{"event": "train", "step": 17625, "loss": 3.9566142559051514, "lr": 9.055208968026001e-05, "seconds": 1178.76, "steps_per_second": 16.021}
{"event": "train", "step": 17650, "loss": 3.9522716999053955, "lr": 9.048095383238198e-05, "seconds": 1180.16, "steps_per_second": 16.025}
{"event": "train", "step": 17675, "loss": 3.856187343597412, "lr": 9.041471795303674e-05, "seconds": 1181.53, "steps_per_second": 16.029}
{"event": "train", "step": 17700, "loss": 3.8537068367004395, "lr": 9.035338331738908e-05, "seconds": 1182.89, "steps_per_second": 16.033}
{"event": "validation", "step": 17700, "val_loss": 4.084866033573242, "best_val": 4.084866033573242, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
{"event": "train", "step": 17725, "loss": 3.902099370956421, "lr": 9.029695110624567e-05, "seconds": 1184.8, "steps_per_second": 16.026}
{"event": "train", "step": 17750, "loss": 3.8475663661956787, "lr": 9.024542240603231e-05, "seconds": 1186.18, "steps_per_second": 16.03}
{"event": "train", "step": 17775, "loss": 3.948460817337036, "lr": 9.019879820877301e-05, "seconds": 1187.55, "steps_per_second": 16.034}
{"event": "train", "step": 17800, "loss": 3.8508007526397705, "lr": 9.015707941207095e-05, "seconds": 1188.93, "steps_per_second": 16.038}
{"event": "train", "step": 17825, "loss": 3.8876309394836426, "lr": 9.012026681909112e-05, "seconds": 1190.33, "steps_per_second": 16.041}
{"event": "train", "step": 17850, "loss": 3.8818981647491455, "lr": 9.008836113854491e-05, "seconds": 1191.71, "steps_per_second": 16.045}
{"event": "train", "step": 17875, "loss": 3.8962366580963135, "lr": 9.006136298467647e-05, "seconds": 1193.07, "steps_per_second": 16.049}
{"event": "train", "step": 17900, "loss": 3.873304843902588, "lr": 9.003927287725083e-05, "seconds": 1194.47, "steps_per_second": 16.053}
{"event": "train", "step": 17925, "loss": 3.945727825164795, "lr": 9.002209124154397e-05, "seconds": 1195.86, "steps_per_second": 16.056}
{"event": "train", "step": 17950, "loss": 3.8438541889190674, "lr": 9.000981840833455e-05, "seconds": 1197.23, "steps_per_second": 16.06}
{"event": "train", "step": 17975, "loss": 3.830571174621582, "lr": 9.000245461389759e-05, "seconds": 1198.61, "steps_per_second": 16.064}
{"event": "train", "step": 18000, "loss": 3.8565893173217773, "lr": 8.999999999999999e-05, "seconds": 1200.01, "steps_per_second": 16.068}
{"event": "validation", "step": 18000, "val_loss": 4.084603522531118, "best_val": 4.084603522531118, "peak_process_rss_mib": 704.5, "mps_driver_allocated_mib": 1281.5}
