| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.6562887670575053, |
| "global_step": 4100, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.01, |
| "learning_rate": 5.1063829787234044e-05, |
| "loss": 2.2862, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00010425531914893618, |
| "loss": 1.2492, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00015744680851063832, |
| "loss": 0.9072, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019999865579077434, |
| "loss": 0.8917, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001999516122622535, |
| "loss": 0.8346, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019983739440605075, |
| "loss": 0.8663, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019965607898438662, |
| "loss": 0.8291, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019940778785364013, |
| "loss": 0.8034, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001990996433127183, |
| "loss": 0.7637, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001987192758461899, |
| "loss": 0.7684, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001982725622664361, |
| "loss": 0.7658, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019775980279557144, |
| "loss": 0.7453, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019718134204307187, |
| "loss": 0.7404, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019653756877417345, |
| "loss": 0.7236, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001958289156485956, |
| "loss": 0.7398, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001950558589297642, |
| "loss": 0.7089, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019421891816472967, |
| "loss": 0.7109, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001933186558349962, |
| "loss": 0.7114, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019235567697849525, |
| "loss": 0.7024, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00019133062878295894, |
| "loss": 0.6949, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001902442001509654, |
| "loss": 0.6939, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018909712123694916, |
| "loss": 0.7093, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018789016295648724, |
| "loss": 0.6677, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018662413646819127, |
| "loss": 0.642, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018529989262855328, |
| "loss": 0.6663, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018391832142011175, |
| "loss": 0.6644, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018248035135332269, |
| "loss": 0.6884, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018098694884253672, |
| "loss": 0.6672, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.00017943911755650266, |
| "loss": 0.7046, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00017783789774383332, |
| "loss": 0.6661, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.00017618436553388725, |
| "loss": 0.6367, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00017447963221353603, |
| "loss": 0.6542, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.0001727248434803035, |
| "loss": 0.6529, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.0001709211786723785, |
| "loss": 0.6667, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 0.00016906984997601892, |
| "loss": 0.6473, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 0.00016717210161087945, |
| "loss": 0.6355, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 0.00016522920899381086, |
| "loss": 0.6507, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 0.00016324247788169244, |
| "loss": 0.6422, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 0.000161213243493874, |
| "loss": 0.6394, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 0.00015914286961481717, |
| "loss": 0.6475, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 0.00015703274767753857, |
| "loss": 0.6392, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 0.00015488429582847192, |
| "loss": 0.6315, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 0.0001526989579743762, |
| "loss": 0.6151, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 0.0001504782028119315, |
| "loss": 0.6389, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 0.00014822352284067444, |
| "loss": 0.6428, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 0.0001459364333599362, |
| "loss": 0.6495, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 0.0001436184714504575, |
| "loss": 0.6367, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 0.00014127119494136527, |
| "loss": 0.6311, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 0.00013889618136320465, |
| "loss": 0.6084, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 0.00013649502688773053, |
| "loss": 0.6056, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 0.0001340693452551708, |
| "loss": 0.6308, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 0.0001316207666896824, |
| "loss": 0.6185, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 0.00012915093680372915, |
| "loss": 0.6564, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 0.00012666151549211733, |
| "loss": 0.6295, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 0.00012415417581643288, |
| "loss": 0.6144, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 0.0001216306028806293, |
| "loss": 0.6057, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 0.0001190924926985225, |
| "loss": 0.6307, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 0.00011654155105395334, |
| "loss": 0.6166, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 0.0001139794923543841, |
| "loss": 0.5842, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 0.0001114080384786993, |
| "loss": 0.6122, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 0.00010882891761998507, |
| "loss": 0.6494, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 0.00010624386312406523, |
| "loss": 0.6065, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 0.00010365461232457406, |
| "loss": 0.6116, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 0.00010106290537534917, |
| "loss": 0.6213, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.847048408092904e-05, |
| "loss": 0.609, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.587909072594096e-05, |
| "loss": 0.5936, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.329046690416654e-05, |
| "loss": 0.6047, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.070635234807153e-05, |
| "loss": 0.5944, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.812848375958627e-05, |
| "loss": 0.5857, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.555859364292315e-05, |
| "loss": 0.5901, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.299840914021524e-05, |
| "loss": 0.6086, |
| "step": 3550 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.044965087075841e-05, |
| "loss": 0.5972, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.791403177463736e-05, |
| "loss": 0.6021, |
| "step": 3650 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.539325596151254e-05, |
| "loss": 0.5976, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.288901756534196e-05, |
| "loss": 0.5941, |
| "step": 3750 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.040299960580694e-05, |
| "loss": 0.5974, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.793687285720765e-05, |
| "loss": 0.5985, |
| "step": 3850 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.549229472558864e-05, |
| "loss": 0.606, |
| "step": 3900 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.307090813484814e-05, |
| "loss": 0.6082, |
| "step": 3950 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.067434042258092e-05, |
| "loss": 0.5809, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.83042022463959e-05, |
| "loss": 0.6045, |
| "step": 4050 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.596208650144402e-05, |
| "loss": 0.58, |
| "step": 4100 |
| } |
| ], |
| "max_steps": 6247, |
| "num_train_epochs": 1, |
| "total_flos": 1883958567985152.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
