| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.319096786053494, | |
| "global_step": 500000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.999999999999999e-06, | |
| "loss": 0.9345, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.1999999999999999e-05, | |
| "loss": 0.7514, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 0.9052248597145081, | |
| "eval_runtime": 0.5073, | |
| "eval_samples_per_second": 1971.284, | |
| "eval_steps_per_second": 31.541, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.7999999999999997e-05, | |
| "loss": 0.7408, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.3999999999999997e-05, | |
| "loss": 0.74, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 0.9058064818382263, | |
| "eval_runtime": 0.5123, | |
| "eval_samples_per_second": 1952.096, | |
| "eval_steps_per_second": 31.234, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.9999999999999997e-05, | |
| "loss": 0.7398, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.5999999999999994e-05, | |
| "loss": 0.7395, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 0.9068936705589294, | |
| "eval_runtime": 0.499, | |
| "eval_samples_per_second": 2003.855, | |
| "eval_steps_per_second": 32.062, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.7394, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7999999999999994e-05, | |
| "loss": 0.7392, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_loss": 0.9032273888587952, | |
| "eval_runtime": 0.5086, | |
| "eval_samples_per_second": 1966.372, | |
| "eval_steps_per_second": 31.462, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5.399999999999999e-05, | |
| "loss": 0.7389, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.7386, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_loss": 0.9034351706504822, | |
| "eval_runtime": 0.5101, | |
| "eval_samples_per_second": 1960.24, | |
| "eval_steps_per_second": 31.364, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.599999999999999e-05, | |
| "loss": 0.7382, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.199999999999999e-05, | |
| "loss": 0.7377, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_loss": 0.8661078810691833, | |
| "eval_runtime": 0.5119, | |
| "eval_samples_per_second": 1953.649, | |
| "eval_steps_per_second": 31.258, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.8e-05, | |
| "loss": 0.7375, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.7373, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_loss": 0.8658460974693298, | |
| "eval_runtime": 0.5198, | |
| "eval_samples_per_second": 1923.698, | |
| "eval_steps_per_second": 30.779, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.999999999999999e-05, | |
| "loss": 0.7367, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.599999999999999e-05, | |
| "loss": 0.7207, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_loss": 0.8675529956817627, | |
| "eval_runtime": 0.5261, | |
| "eval_samples_per_second": 1900.917, | |
| "eval_steps_per_second": 30.415, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000102, | |
| "loss": 0.6905, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00010799999999999998, | |
| "loss": 0.6746, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "eval_loss": 0.8773286938667297, | |
| "eval_runtime": 0.5159, | |
| "eval_samples_per_second": 1938.548, | |
| "eval_steps_per_second": 31.017, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00011399999999999999, | |
| "loss": 0.6617, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.6406, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "eval_loss": 0.8803548812866211, | |
| "eval_runtime": 0.5084, | |
| "eval_samples_per_second": 1966.828, | |
| "eval_steps_per_second": 31.469, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00012599999999999997, | |
| "loss": 0.6275, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00013199999999999998, | |
| "loss": 0.6079, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 0.8707549571990967, | |
| "eval_runtime": 0.5111, | |
| "eval_samples_per_second": 1956.647, | |
| "eval_steps_per_second": 31.306, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000138, | |
| "loss": 0.5977, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00014399999999999998, | |
| "loss": 0.5907, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "eval_loss": 0.8676372766494751, | |
| "eval_runtime": 0.5179, | |
| "eval_samples_per_second": 1930.992, | |
| "eval_steps_per_second": 30.896, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00015, | |
| "loss": 0.5834, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000156, | |
| "loss": 0.5764, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.8676320910453796, | |
| "eval_runtime": 0.5238, | |
| "eval_samples_per_second": 1909.011, | |
| "eval_steps_per_second": 30.544, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000162, | |
| "loss": 0.5696, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000168, | |
| "loss": 0.5648, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 0.860458254814148, | |
| "eval_runtime": 0.5396, | |
| "eval_samples_per_second": 1853.225, | |
| "eval_steps_per_second": 29.652, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00017399999999999997, | |
| "loss": 0.5602, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.556, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_loss": 0.8584501147270203, | |
| "eval_runtime": 0.5323, | |
| "eval_samples_per_second": 1878.593, | |
| "eval_steps_per_second": 30.057, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.000186, | |
| "loss": 0.5519, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00019199999999999998, | |
| "loss": 0.5459, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_loss": 0.8554501533508301, | |
| "eval_runtime": 0.5107, | |
| "eval_samples_per_second": 1957.988, | |
| "eval_steps_per_second": 31.328, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000198, | |
| "loss": 0.5412, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000204, | |
| "loss": 0.5374, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 0.8554509282112122, | |
| "eval_runtime": 0.5145, | |
| "eval_samples_per_second": 1943.819, | |
| "eval_steps_per_second": 31.101, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00020999999999999998, | |
| "loss": 0.5338, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00021599999999999996, | |
| "loss": 0.5305, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_loss": 0.8557196259498596, | |
| "eval_runtime": 0.5163, | |
| "eval_samples_per_second": 1936.818, | |
| "eval_steps_per_second": 30.989, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00022199999999999998, | |
| "loss": 0.5273, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00022799999999999999, | |
| "loss": 0.5243, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_loss": 0.8606428503990173, | |
| "eval_runtime": 0.5087, | |
| "eval_samples_per_second": 1965.911, | |
| "eval_steps_per_second": 31.455, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.000234, | |
| "loss": 0.5215, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.5188, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "eval_loss": 0.8626545071601868, | |
| "eval_runtime": 0.4996, | |
| "eval_samples_per_second": 2001.549, | |
| "eval_steps_per_second": 32.025, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00024599999999999996, | |
| "loss": 0.516, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00025199999999999995, | |
| "loss": 0.5136, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_loss": 0.8466567993164062, | |
| "eval_runtime": 0.521, | |
| "eval_samples_per_second": 1919.499, | |
| "eval_steps_per_second": 30.712, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000258, | |
| "loss": 0.5112, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00026399999999999997, | |
| "loss": 0.5089, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "eval_loss": 0.8465701937675476, | |
| "eval_runtime": 0.5015, | |
| "eval_samples_per_second": 1994.167, | |
| "eval_steps_per_second": 31.907, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00027, | |
| "loss": 0.5067, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000276, | |
| "loss": 0.5048, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.8367106914520264, | |
| "eval_runtime": 0.5157, | |
| "eval_samples_per_second": 1939.06, | |
| "eval_steps_per_second": 31.025, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00028199999999999997, | |
| "loss": 0.5029, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00028799999999999995, | |
| "loss": 0.5009, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "eval_loss": 0.8282895088195801, | |
| "eval_runtime": 0.5191, | |
| "eval_samples_per_second": 1926.314, | |
| "eval_steps_per_second": 30.821, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000294, | |
| "loss": 0.4986, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0003, | |
| "loss": 0.4968, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_loss": 0.8393005132675171, | |
| "eval_runtime": 0.5106, | |
| "eval_samples_per_second": 1958.371, | |
| "eval_steps_per_second": 31.334, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00029999920715161553, | |
| "loss": 0.495, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002999968286151326, | |
| "loss": 0.4932, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.8276752829551697, | |
| "eval_runtime": 0.5157, | |
| "eval_samples_per_second": 1939.179, | |
| "eval_steps_per_second": 31.027, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002999928644165624, | |
| "loss": 0.4916, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002999873145992569, | |
| "loss": 0.4899, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "eval_loss": 0.8151644468307495, | |
| "eval_runtime": 0.5093, | |
| "eval_samples_per_second": 1963.529, | |
| "eval_steps_per_second": 31.416, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.000299980179223908, | |
| "loss": 0.4883, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002999714583685469, | |
| "loss": 0.4868, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_loss": 0.8103233575820923, | |
| "eval_runtime": 0.5126, | |
| "eval_samples_per_second": 1950.786, | |
| "eval_steps_per_second": 31.213, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00029996115212854366, | |
| "loss": 0.4855, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00029994926061660554, | |
| "loss": 0.4841, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "eval_loss": 0.8084909319877625, | |
| "eval_runtime": 0.5298, | |
| "eval_samples_per_second": 1887.571, | |
| "eval_steps_per_second": 30.201, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002999357839627762, | |
| "loss": 0.4823, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00029992072231443425, | |
| "loss": 0.4806, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_loss": 0.8131279945373535, | |
| "eval_runtime": 0.5066, | |
| "eval_samples_per_second": 1974.102, | |
| "eval_steps_per_second": 31.586, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002999040758362914, | |
| "loss": 0.4791, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00029988584471039094, | |
| "loss": 0.4776, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_loss": 0.8034030795097351, | |
| "eval_runtime": 0.4996, | |
| "eval_samples_per_second": 2001.487, | |
| "eval_steps_per_second": 32.024, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002998660291361054, | |
| "loss": 0.476, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002998446293301349, | |
| "loss": 0.4741, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_loss": 0.8003847599029541, | |
| "eval_runtime": 0.5033, | |
| "eval_samples_per_second": 1986.864, | |
| "eval_steps_per_second": 31.79, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002998216455265042, | |
| "loss": 0.4724, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00029979707797656046, | |
| "loss": 0.4709, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "eval_loss": 0.7997156977653503, | |
| "eval_runtime": 0.5174, | |
| "eval_samples_per_second": 1932.776, | |
| "eval_steps_per_second": 30.924, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00029977092694897053, | |
| "loss": 0.4691, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0002997431927297178, | |
| "loss": 0.467, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_loss": 0.803820013999939, | |
| "eval_runtime": 0.5367, | |
| "eval_samples_per_second": 1863.102, | |
| "eval_steps_per_second": 29.81, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00029971387562209936, | |
| "loss": 0.4653, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00029968297594672226, | |
| "loss": 0.4638, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_loss": 0.8011023998260498, | |
| "eval_runtime": 0.5429, | |
| "eval_samples_per_second": 1842.037, | |
| "eval_steps_per_second": 29.473, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002996504940415005, | |
| "loss": 0.462, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00029961643026165096, | |
| "loss": 0.4601, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_loss": 0.8029471635818481, | |
| "eval_runtime": 0.5067, | |
| "eval_samples_per_second": 1973.42, | |
| "eval_steps_per_second": 31.575, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00029958078497968973, | |
| "loss": 0.4587, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002995435585854278, | |
| "loss": 0.457, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "eval_loss": 0.7997317910194397, | |
| "eval_runtime": 0.5216, | |
| "eval_samples_per_second": 1917.135, | |
| "eval_steps_per_second": 30.674, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0002995047514859671, | |
| "loss": 0.4556, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002994643641056959, | |
| "loss": 0.4541, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_loss": 0.8004159927368164, | |
| "eval_runtime": 0.5088, | |
| "eval_samples_per_second": 1965.38, | |
| "eval_steps_per_second": 31.446, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.000299422396886284, | |
| "loss": 0.4524, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0002993788502866783, | |
| "loss": 0.4514, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "eval_loss": 0.797309160232544, | |
| "eval_runtime": 0.5211, | |
| "eval_samples_per_second": 1918.834, | |
| "eval_steps_per_second": 30.701, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00029933372478309746, | |
| "loss": 0.45, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00029928702086902664, | |
| "loss": 0.4485, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "eval_loss": 0.7876560688018799, | |
| "eval_runtime": 0.5381, | |
| "eval_samples_per_second": 1858.547, | |
| "eval_steps_per_second": 29.737, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00029923873905521244, | |
| "loss": 0.4471, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.000299188879869657, | |
| "loss": 0.446, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_loss": 0.7949715852737427, | |
| "eval_runtime": 0.5157, | |
| "eval_samples_per_second": 1939.022, | |
| "eval_steps_per_second": 31.024, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00029913744385761244, | |
| "loss": 0.4446, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00029908443158157465, | |
| "loss": 0.4437, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_loss": 0.7941656112670898, | |
| "eval_runtime": 0.5261, | |
| "eval_samples_per_second": 1900.807, | |
| "eval_steps_per_second": 30.413, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0002990298436212775, | |
| "loss": 0.4422, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0002989736805736861, | |
| "loss": 0.4413, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_loss": 0.7861095666885376, | |
| "eval_runtime": 0.5221, | |
| "eval_samples_per_second": 1915.408, | |
| "eval_steps_per_second": 30.647, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00029891594305299065, | |
| "loss": 0.44, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00029885663169059926, | |
| "loss": 0.4391, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_loss": 0.7935870289802551, | |
| "eval_runtime": 0.5329, | |
| "eval_samples_per_second": 1876.586, | |
| "eval_steps_per_second": 30.025, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0002987957471351316, | |
| "loss": 0.4378, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00029873329005241137, | |
| "loss": 0.4366, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "eval_loss": 0.7878534197807312, | |
| "eval_runtime": 0.5149, | |
| "eval_samples_per_second": 1941.994, | |
| "eval_steps_per_second": 31.072, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00029866926112545925, | |
| "loss": 0.4355, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00029860366105448534, | |
| "loss": 0.4342, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "eval_loss": 0.7861126661300659, | |
| "eval_runtime": 0.5143, | |
| "eval_samples_per_second": 1944.264, | |
| "eval_steps_per_second": 31.108, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00029853649055688143, | |
| "loss": 0.4333, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00029846775036721337, | |
| "loss": 0.432, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "eval_loss": 0.7865832448005676, | |
| "eval_runtime": 0.5183, | |
| "eval_samples_per_second": 1929.412, | |
| "eval_steps_per_second": 30.871, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.0002983974412372129, | |
| "loss": 0.4308, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00029832556393576934, | |
| "loss": 0.4297, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "eval_loss": 0.777747392654419, | |
| "eval_runtime": 0.527, | |
| "eval_samples_per_second": 1897.653, | |
| "eval_steps_per_second": 30.362, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0002982521192489214, | |
| "loss": 0.4288, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0002981771079798483, | |
| "loss": 0.4278, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "eval_loss": 0.7763716578483582, | |
| "eval_runtime": 0.5211, | |
| "eval_samples_per_second": 1919.193, | |
| "eval_steps_per_second": 30.707, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00029810053094886136, | |
| "loss": 0.4269, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00029802238899339473, | |
| "loss": 0.4258, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_loss": 0.7902368307113647, | |
| "eval_runtime": 0.5206, | |
| "eval_samples_per_second": 1920.725, | |
| "eval_steps_per_second": 30.732, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0002979426829679962, | |
| "loss": 0.4247, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0002978614137443183, | |
| "loss": 0.424, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "eval_loss": 0.7871042490005493, | |
| "eval_runtime": 0.5261, | |
| "eval_samples_per_second": 1900.754, | |
| "eval_steps_per_second": 30.412, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.000297778582211108, | |
| "loss": 0.4232, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00029769418927419786, | |
| "loss": 0.422, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_loss": 0.7916954159736633, | |
| "eval_runtime": 0.5311, | |
| "eval_samples_per_second": 1882.884, | |
| "eval_steps_per_second": 30.126, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0002976082358564954, | |
| "loss": 0.4211, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00029752072289797353, | |
| "loss": 0.4202, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_loss": 0.7859019637107849, | |
| "eval_runtime": 0.5358, | |
| "eval_samples_per_second": 1866.426, | |
| "eval_steps_per_second": 29.863, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00029743165135565986, | |
| "loss": 0.4194, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00029734102220362654, | |
| "loss": 0.4184, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "eval_loss": 0.7869405150413513, | |
| "eval_runtime": 0.5363, | |
| "eval_samples_per_second": 1864.609, | |
| "eval_steps_per_second": 29.834, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00029724883643297937, | |
| "loss": 0.4177, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0002971550950518473, | |
| "loss": 0.4165, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "eval_loss": 0.7873055338859558, | |
| "eval_runtime": 0.5249, | |
| "eval_samples_per_second": 1905.112, | |
| "eval_steps_per_second": 30.482, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.000297059799085371, | |
| "loss": 0.4156, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00029696294957569196, | |
| "loss": 0.4149, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_loss": 0.7851352095603943, | |
| "eval_runtime": 0.5685, | |
| "eval_samples_per_second": 1758.943, | |
| "eval_steps_per_second": 28.143, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00029686454758194076, | |
| "loss": 0.4141, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00029676459418022594, | |
| "loss": 0.4136, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_loss": 0.7847647666931152, | |
| "eval_runtime": 0.5348, | |
| "eval_samples_per_second": 1869.958, | |
| "eval_steps_per_second": 29.919, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0002966630904636219, | |
| "loss": 0.4126, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0002965600375421569, | |
| "loss": 0.4119, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "eval_loss": 0.7846025228500366, | |
| "eval_runtime": 0.5441, | |
| "eval_samples_per_second": 1837.795, | |
| "eval_steps_per_second": 29.405, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0002964554365428013, | |
| "loss": 0.4107, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00029634928860945486, | |
| "loss": 0.4103, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_loss": 0.7957924604415894, | |
| "eval_runtime": 0.5171, | |
| "eval_samples_per_second": 1934.049, | |
| "eval_steps_per_second": 30.945, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0002962415949029343, | |
| "loss": 0.4099, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00029613235660096084, | |
| "loss": 0.409, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_loss": 0.7877106070518494, | |
| "eval_runtime": 0.5384, | |
| "eval_samples_per_second": 1857.26, | |
| "eval_steps_per_second": 29.716, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00029602157489814693, | |
| "loss": 0.4084, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00029590925100598365, | |
| "loss": 0.4077, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "eval_loss": 0.7871591448783875, | |
| "eval_runtime": 0.5252, | |
| "eval_samples_per_second": 1904.035, | |
| "eval_steps_per_second": 30.465, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.000295795386152827, | |
| "loss": 0.4068, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0002956799815838848, | |
| "loss": 0.4061, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "eval_loss": 0.7920863032341003, | |
| "eval_runtime": 0.5271, | |
| "eval_samples_per_second": 1897.169, | |
| "eval_steps_per_second": 30.355, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0002955630385612029, | |
| "loss": 0.4056, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0002954445583636515, | |
| "loss": 0.405, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "eval_loss": 0.7836641073226929, | |
| "eval_runtime": 0.5362, | |
| "eval_samples_per_second": 1864.939, | |
| "eval_steps_per_second": 29.839, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00029532454228691103, | |
| "loss": 0.4043, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0002952029916434581, | |
| "loss": 0.4036, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "eval_loss": 0.7819440960884094, | |
| "eval_runtime": 0.5223, | |
| "eval_samples_per_second": 1914.709, | |
| "eval_steps_per_second": 30.635, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00029507990776255107, | |
| "loss": 0.403, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00029495529199021555, | |
| "loss": 0.4023, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_loss": 0.7944305539131165, | |
| "eval_runtime": 0.5274, | |
| "eval_samples_per_second": 1896.133, | |
| "eval_steps_per_second": 30.338, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0002948291456892296, | |
| "loss": 0.4018, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.00029470147023910907, | |
| "loss": 0.4009, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_loss": 0.7896050214767456, | |
| "eval_runtime": 0.514, | |
| "eval_samples_per_second": 1945.63, | |
| "eval_steps_per_second": 31.13, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0002945722670360921, | |
| "loss": 0.4006, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.0002944415374931243, | |
| "loss": 0.4003, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "eval_loss": 0.7898439168930054, | |
| "eval_runtime": 0.5343, | |
| "eval_samples_per_second": 1871.754, | |
| "eval_steps_per_second": 29.948, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.00029430928303984295, | |
| "loss": 0.3993, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0002941755051225616, | |
| "loss": 0.3988, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_loss": 0.7866422533988953, | |
| "eval_runtime": 0.5273, | |
| "eval_samples_per_second": 1896.599, | |
| "eval_steps_per_second": 30.346, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00029404020520425417, | |
| "loss": 0.3984, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.0002939033847645388, | |
| "loss": 0.3979, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_loss": 0.7866395711898804, | |
| "eval_runtime": 0.5223, | |
| "eval_samples_per_second": 1914.777, | |
| "eval_steps_per_second": 30.636, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.00029376504529966195, | |
| "loss": 0.3974, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.00029362518832248184, | |
| "loss": 0.3968, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "eval_loss": 0.7869418263435364, | |
| "eval_runtime": 0.52, | |
| "eval_samples_per_second": 1922.99, | |
| "eval_steps_per_second": 30.768, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0002934838153624519, | |
| "loss": 0.3962, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00029334092796560427, | |
| "loss": 0.3958, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "eval_loss": 0.7898754477500916, | |
| "eval_runtime": 0.5162, | |
| "eval_samples_per_second": 1937.227, | |
| "eval_steps_per_second": 30.996, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0002931965276945326, | |
| "loss": 0.3951, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.0002930506161283751, | |
| "loss": 0.3947, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "eval_loss": 0.7828860878944397, | |
| "eval_runtime": 0.5502, | |
| "eval_samples_per_second": 1817.401, | |
| "eval_steps_per_second": 29.078, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.00029290319486279724, | |
| "loss": 0.3944, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0002927542655099744, | |
| "loss": 0.3935, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "eval_loss": 0.7836081385612488, | |
| "eval_runtime": 0.518, | |
| "eval_samples_per_second": 1930.433, | |
| "eval_steps_per_second": 30.887, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.00029260382969857417, | |
| "loss": 0.3936, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.00029245188907373845, | |
| "loss": 0.393, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "eval_loss": 0.7833809852600098, | |
| "eval_runtime": 0.5265, | |
| "eval_samples_per_second": 1899.502, | |
| "eval_steps_per_second": 30.392, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.0002922984452970655, | |
| "loss": 0.3923, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.000292143500046592, | |
| "loss": 0.392, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_loss": 0.7891106009483337, | |
| "eval_runtime": 0.549, | |
| "eval_samples_per_second": 1821.565, | |
| "eval_steps_per_second": 29.145, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.0002919870550167743, | |
| "loss": 0.3917, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.0002918291119184702, | |
| "loss": 0.3913, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "eval_loss": 0.7858054041862488, | |
| "eval_runtime": 0.5293, | |
| "eval_samples_per_second": 1889.27, | |
| "eval_steps_per_second": 30.228, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.0002916696724789201, | |
| "loss": 0.3906, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.00029150873844172823, | |
| "loss": 0.3904, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "eval_loss": 0.7869090437889099, | |
| "eval_runtime": 0.5095, | |
| "eval_samples_per_second": 1962.553, | |
| "eval_steps_per_second": 31.401, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00029134631156684334, | |
| "loss": 0.3898, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 0.0002911823936305398, | |
| "loss": 0.3893, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "eval_loss": 0.7837140560150146, | |
| "eval_runtime": 0.5128, | |
| "eval_samples_per_second": 1950.24, | |
| "eval_steps_per_second": 31.204, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.0002910169864253979, | |
| "loss": 0.3892, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.0002908500917602842, | |
| "loss": 0.3886, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.787075936794281, | |
| "eval_runtime": 0.5289, | |
| "eval_samples_per_second": 1890.813, | |
| "eval_steps_per_second": 30.253, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00029068171146033226, | |
| "loss": 0.3883, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 0.0002905118473669218, | |
| "loss": 0.3877, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "eval_loss": 0.7844729423522949, | |
| "eval_runtime": 0.512, | |
| "eval_samples_per_second": 1953.312, | |
| "eval_steps_per_second": 31.253, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.00029034050133765947, | |
| "loss": 0.3873, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.00029016767524635804, | |
| "loss": 0.3872, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "eval_loss": 0.7831702828407288, | |
| "eval_runtime": 0.5564, | |
| "eval_samples_per_second": 1797.202, | |
| "eval_steps_per_second": 28.755, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.00028999337098301585, | |
| "loss": 0.3865, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0002898175904537964, | |
| "loss": 0.3858, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "eval_loss": 0.7806258797645569, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.174, | |
| "eval_steps_per_second": 30.851, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.0002896403355810075, | |
| "loss": 0.3859, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.00028946160830307997, | |
| "loss": 0.3854, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "eval_loss": 0.7836428880691528, | |
| "eval_runtime": 0.545, | |
| "eval_samples_per_second": 1834.936, | |
| "eval_steps_per_second": 29.359, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.00028928141057454665, | |
| "loss": 0.3849, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0002890997443660211, | |
| "loss": 0.3848, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_loss": 0.7852834463119507, | |
| "eval_runtime": 0.5158, | |
| "eval_samples_per_second": 1938.61, | |
| "eval_steps_per_second": 31.018, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00028891661166417586, | |
| "loss": 0.3844, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.00028873201447172074, | |
| "loss": 0.3838, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "eval_loss": 0.7852866053581238, | |
| "eval_runtime": 0.5265, | |
| "eval_samples_per_second": 1899.356, | |
| "eval_steps_per_second": 30.39, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.0002885459548073812, | |
| "loss": 0.3837, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.0002883584347058758, | |
| "loss": 0.383, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_loss": 0.7794146537780762, | |
| "eval_runtime": 0.5219, | |
| "eval_samples_per_second": 1916.094, | |
| "eval_steps_per_second": 30.657, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.00028816945621789437, | |
| "loss": 0.3831, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0002879790214100753, | |
| "loss": 0.3823, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "eval_loss": 0.7845750451087952, | |
| "eval_runtime": 0.5135, | |
| "eval_samples_per_second": 1947.28, | |
| "eval_steps_per_second": 31.156, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.0002877871323649833, | |
| "loss": 0.382, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.0002875937911810861, | |
| "loss": 0.3818, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "eval_loss": 0.7793565988540649, | |
| "eval_runtime": 0.5472, | |
| "eval_samples_per_second": 1827.624, | |
| "eval_steps_per_second": 29.242, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.000287398999972732, | |
| "loss": 0.3817, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.00028720276087012636, | |
| "loss": 0.3809, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "eval_loss": 0.786903440952301, | |
| "eval_runtime": 0.522, | |
| "eval_samples_per_second": 1915.886, | |
| "eval_steps_per_second": 30.654, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0002870050760193086, | |
| "loss": 0.3806, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.00028680594758212854, | |
| "loss": 0.3805, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "eval_loss": 0.7848221659660339, | |
| "eval_runtime": 0.5114, | |
| "eval_samples_per_second": 1955.578, | |
| "eval_steps_per_second": 31.289, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.00028660537773622294, | |
| "loss": 0.3803, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.00028640336867499143, | |
| "loss": 0.3796, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_loss": 0.7904797196388245, | |
| "eval_runtime": 0.5244, | |
| "eval_samples_per_second": 1906.796, | |
| "eval_steps_per_second": 30.509, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0002861999226075728, | |
| "loss": 0.3798, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0002859950417588206, | |
| "loss": 0.3792, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "eval_loss": 0.7905736565589905, | |
| "eval_runtime": 0.5306, | |
| "eval_samples_per_second": 1884.737, | |
| "eval_steps_per_second": 30.156, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.00028578872836927904, | |
| "loss": 0.3788, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0002855809846951582, | |
| "loss": 0.3789, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "eval_loss": 0.7809098958969116, | |
| "eval_runtime": 0.5414, | |
| "eval_samples_per_second": 1847.16, | |
| "eval_steps_per_second": 29.555, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.00028537181300830963, | |
| "loss": 0.3782, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.0002851612155962014, | |
| "loss": 0.3784, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "eval_loss": 0.7892218232154846, | |
| "eval_runtime": 0.516, | |
| "eval_samples_per_second": 1938.108, | |
| "eval_steps_per_second": 31.01, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0002849491947618932, | |
| "loss": 0.3778, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.0002847357528240107, | |
| "loss": 0.3775, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "eval_loss": 0.7847021818161011, | |
| "eval_runtime": 0.5181, | |
| "eval_samples_per_second": 1930.051, | |
| "eval_steps_per_second": 30.881, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0002845208921167208, | |
| "loss": 0.3773, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.00028430461498970584, | |
| "loss": 0.3768, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "eval_loss": 0.78525710105896, | |
| "eval_runtime": 0.5373, | |
| "eval_samples_per_second": 1861.031, | |
| "eval_steps_per_second": 29.776, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.00028408692380813775, | |
| "loss": 0.3768, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.00028386782095265247, | |
| "loss": 0.3769, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "eval_loss": 0.7837897539138794, | |
| "eval_runtime": 0.5288, | |
| "eval_samples_per_second": 1891.025, | |
| "eval_steps_per_second": 30.256, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0002836473088193237, | |
| "loss": 0.3761, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.00028342538981963677, | |
| "loss": 0.3797, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.7850324511528015, | |
| "eval_runtime": 0.5376, | |
| "eval_samples_per_second": 1860.245, | |
| "eval_steps_per_second": 29.764, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 0.0002832020663804624, | |
| "loss": 0.3752, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 0.00028297734094402986, | |
| "loss": 0.3747, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "eval_loss": 0.780020534992218, | |
| "eval_runtime": 0.5281, | |
| "eval_samples_per_second": 1893.464, | |
| "eval_steps_per_second": 30.295, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.0002827512159679005, | |
| "loss": 0.375, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 0.00028252369392494086, | |
| "loss": 0.3746, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "eval_loss": 0.7855916619300842, | |
| "eval_runtime": 0.5271, | |
| "eval_samples_per_second": 1897.09, | |
| "eval_steps_per_second": 30.353, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 0.0002822947773032956, | |
| "loss": 0.3743, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.0002820644686063602, | |
| "loss": 0.3743, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "eval_loss": 0.7854102253913879, | |
| "eval_runtime": 0.541, | |
| "eval_samples_per_second": 1848.349, | |
| "eval_steps_per_second": 29.574, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 0.00028183277035275363, | |
| "loss": 0.374, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.0002815996850762909, | |
| "loss": 0.374, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "eval_loss": 0.7880498170852661, | |
| "eval_runtime": 0.5283, | |
| "eval_samples_per_second": 1892.948, | |
| "eval_steps_per_second": 30.287, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.00028136521532595515, | |
| "loss": 0.3737, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 0.00028112936366587023, | |
| "loss": 0.3734, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "eval_loss": 0.7864383459091187, | |
| "eval_runtime": 0.5228, | |
| "eval_samples_per_second": 1912.628, | |
| "eval_steps_per_second": 30.602, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 0.00028089213267527184, | |
| "loss": 0.3733, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.0002806535249484803, | |
| "loss": 0.3731, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "eval_loss": 0.783866822719574, | |
| "eval_runtime": 0.5321, | |
| "eval_samples_per_second": 1879.482, | |
| "eval_steps_per_second": 30.072, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 0.00028041354309487135, | |
| "loss": 0.3728, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0002801721897388482, | |
| "loss": 0.3723, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "eval_loss": 0.784348726272583, | |
| "eval_runtime": 0.5158, | |
| "eval_samples_per_second": 1938.703, | |
| "eval_steps_per_second": 31.019, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 0.0002799294675198124, | |
| "loss": 0.3723, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 0.00027968537909213524, | |
| "loss": 0.372, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_loss": 0.7889582514762878, | |
| "eval_runtime": 0.5199, | |
| "eval_samples_per_second": 1923.296, | |
| "eval_steps_per_second": 30.773, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 0.0002794399271251287, | |
| "loss": 0.3721, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 0.0002791931143030162, | |
| "loss": 0.3714, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_loss": 0.7859266400337219, | |
| "eval_runtime": 0.5189, | |
| "eval_samples_per_second": 1926.987, | |
| "eval_steps_per_second": 30.832, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 0.00027894494332490315, | |
| "loss": 0.3715, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 0.0002786954169047476, | |
| "loss": 0.3711, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "eval_loss": 0.7798612117767334, | |
| "eval_runtime": 0.5242, | |
| "eval_samples_per_second": 1907.581, | |
| "eval_steps_per_second": 30.521, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 0.0002784445377713306, | |
| "loss": 0.3711, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 0.0002781923086682261, | |
| "loss": 0.371, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "eval_loss": 0.7840728759765625, | |
| "eval_runtime": 0.5137, | |
| "eval_samples_per_second": 1946.758, | |
| "eval_steps_per_second": 31.148, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 0.0002779387323537711, | |
| "loss": 0.3704, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.0002776838116010356, | |
| "loss": 0.3705, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "eval_loss": 0.7811622023582458, | |
| "eval_runtime": 0.5403, | |
| "eval_samples_per_second": 1850.736, | |
| "eval_steps_per_second": 29.612, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 0.0002774275491977922, | |
| "loss": 0.3703, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 0.0002771699479464853, | |
| "loss": 0.3701, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "eval_loss": 0.7903389930725098, | |
| "eval_runtime": 0.5256, | |
| "eval_samples_per_second": 1902.556, | |
| "eval_steps_per_second": 30.441, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 0.00027691101066420104, | |
| "loss": 0.3694, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 0.0002766507401826361, | |
| "loss": 0.3692, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "eval_loss": 0.7830519080162048, | |
| "eval_runtime": 0.5162, | |
| "eval_samples_per_second": 1937.313, | |
| "eval_steps_per_second": 30.997, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 0.0002763891393480666, | |
| "loss": 0.3694, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 0.0002761262110213175, | |
| "loss": 0.3686, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "eval_loss": 0.7800700664520264, | |
| "eval_runtime": 0.5106, | |
| "eval_samples_per_second": 1958.342, | |
| "eval_steps_per_second": 31.333, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.00027586195807773083, | |
| "loss": 0.3689, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 0.00027559638340713435, | |
| "loss": 0.3685, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "eval_loss": 0.7914212942123413, | |
| "eval_runtime": 0.5261, | |
| "eval_samples_per_second": 1900.85, | |
| "eval_steps_per_second": 30.414, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 0.00027532948991381025, | |
| "loss": 0.3686, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 0.00027506128051646287, | |
| "loss": 0.3681, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "eval_loss": 0.7931650280952454, | |
| "eval_runtime": 0.5617, | |
| "eval_samples_per_second": 1780.334, | |
| "eval_steps_per_second": 28.485, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 0.00027479175814818733, | |
| "loss": 0.368, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 0.000274520925756437, | |
| "loss": 0.3679, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "eval_loss": 0.7921015620231628, | |
| "eval_runtime": 0.5338, | |
| "eval_samples_per_second": 1873.268, | |
| "eval_steps_per_second": 29.972, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.00027424878630299157, | |
| "loss": 0.3673, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 0.0002739753427639244, | |
| "loss": 0.3673, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "eval_loss": 0.7863946557044983, | |
| "eval_runtime": 0.528, | |
| "eval_samples_per_second": 1893.847, | |
| "eval_steps_per_second": 30.302, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 0.0002737005981295704, | |
| "loss": 0.3673, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 0.0002734245554044927, | |
| "loss": 0.3668, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "eval_loss": 0.7858835458755493, | |
| "eval_runtime": 0.5221, | |
| "eval_samples_per_second": 1915.181, | |
| "eval_steps_per_second": 30.643, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 0.0002731472176074504, | |
| "loss": 0.3667, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 0.0002728685877713653, | |
| "loss": 0.3669, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "eval_loss": 0.786138653755188, | |
| "eval_runtime": 0.5283, | |
| "eval_samples_per_second": 1893.025, | |
| "eval_steps_per_second": 30.288, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 0.0002725886689432884, | |
| "loss": 0.3664, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 0.0002723074641843674, | |
| "loss": 0.3664, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "eval_loss": 0.7753082513809204, | |
| "eval_runtime": 0.539, | |
| "eval_samples_per_second": 1855.209, | |
| "eval_steps_per_second": 29.683, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 0.0002720249765698123, | |
| "loss": 0.3661, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 0.0002717412091888626, | |
| "loss": 0.3663, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "eval_loss": 0.7859072089195251, | |
| "eval_runtime": 0.5189, | |
| "eval_samples_per_second": 1927.188, | |
| "eval_steps_per_second": 30.835, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00027145616514475274, | |
| "loss": 0.3656, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.0002711698475546788, | |
| "loss": 0.3656, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "eval_loss": 0.7800880074501038, | |
| "eval_runtime": 0.5426, | |
| "eval_samples_per_second": 1842.936, | |
| "eval_steps_per_second": 29.487, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.00027088225954976407, | |
| "loss": 0.3653, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.0002705934042750249, | |
| "loss": 0.3653, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "eval_loss": 0.7793842554092407, | |
| "eval_runtime": 0.5318, | |
| "eval_samples_per_second": 1880.356, | |
| "eval_steps_per_second": 30.086, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 0.00027030328488933625, | |
| "loss": 0.3649, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 0.00027001190456539726, | |
| "loss": 0.3648, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "eval_loss": 0.7881133556365967, | |
| "eval_runtime": 0.507, | |
| "eval_samples_per_second": 1972.529, | |
| "eval_steps_per_second": 31.56, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.0002697192664896965, | |
| "loss": 0.3648, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00026942537386247706, | |
| "loss": 0.3644, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "eval_loss": 0.7862609624862671, | |
| "eval_runtime": 0.5236, | |
| "eval_samples_per_second": 1909.846, | |
| "eval_steps_per_second": 30.558, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 0.0002691302298977016, | |
| "loss": 0.3643, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.0002688338378230173, | |
| "loss": 0.3642, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "eval_loss": 0.7819104194641113, | |
| "eval_runtime": 0.5098, | |
| "eval_samples_per_second": 1961.492, | |
| "eval_steps_per_second": 31.384, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 0.00026853620087972035, | |
| "loss": 0.3636, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 0.00026823732232272065, | |
| "loss": 0.3638, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "eval_loss": 0.7886289358139038, | |
| "eval_runtime": 0.5488, | |
| "eval_samples_per_second": 1822.211, | |
| "eval_steps_per_second": 29.155, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.0002679372054205063, | |
| "loss": 0.3637, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 0.0002676358534551076, | |
| "loss": 0.3637, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "eval_loss": 0.7856259942054749, | |
| "eval_runtime": 0.5448, | |
| "eval_samples_per_second": 1835.504, | |
| "eval_steps_per_second": 29.368, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 0.00026733326972206133, | |
| "loss": 0.3631, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.0002670294575303748, | |
| "loss": 0.3631, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "eval_loss": 0.7774640917778015, | |
| "eval_runtime": 0.5451, | |
| "eval_samples_per_second": 1834.446, | |
| "eval_steps_per_second": 29.351, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.0002667244202024894, | |
| "loss": 0.3629, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 0.00026641816107424453, | |
| "loss": 0.363, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "eval_loss": 0.7824729084968567, | |
| "eval_runtime": 0.5158, | |
| "eval_samples_per_second": 1938.903, | |
| "eval_steps_per_second": 31.022, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.0002661106834948409, | |
| "loss": 0.3626, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.0002658019908268041, | |
| "loss": 0.3625, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "eval_loss": 0.7918245792388916, | |
| "eval_runtime": 0.521, | |
| "eval_samples_per_second": 1919.333, | |
| "eval_steps_per_second": 30.709, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 0.00026549208644594766, | |
| "loss": 0.3622, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 0.00026518097374133627, | |
| "loss": 0.3622, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "eval_loss": 0.7886719107627869, | |
| "eval_runtime": 0.5166, | |
| "eval_samples_per_second": 1935.893, | |
| "eval_steps_per_second": 30.974, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.00026486865611524853, | |
| "loss": 0.3622, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.00026455513698314003, | |
| "loss": 0.3616, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "eval_loss": 0.7906731963157654, | |
| "eval_runtime": 0.5202, | |
| "eval_samples_per_second": 1922.502, | |
| "eval_steps_per_second": 30.76, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.0002642404197736058, | |
| "loss": 0.3618, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 0.0002639245079283428, | |
| "loss": 0.3646, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "eval_loss": 0.7901037931442261, | |
| "eval_runtime": 0.5304, | |
| "eval_samples_per_second": 1885.342, | |
| "eval_steps_per_second": 30.165, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 0.00026360740490211234, | |
| "loss": 0.3612, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 0.0002632891141627023, | |
| "loss": 0.3604, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "eval_loss": 0.7916610836982727, | |
| "eval_runtime": 0.5131, | |
| "eval_samples_per_second": 1948.858, | |
| "eval_steps_per_second": 31.182, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00026296963919088923, | |
| "loss": 0.3603, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 0.00026264898348040024, | |
| "loss": 0.3604, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "eval_loss": 0.7712200284004211, | |
| "eval_runtime": 0.519, | |
| "eval_samples_per_second": 1926.697, | |
| "eval_steps_per_second": 30.827, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 0.0002623271505378748, | |
| "loss": 0.3608, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.00026200414388282637, | |
| "loss": 0.361, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "eval_loss": 0.7850333452224731, | |
| "eval_runtime": 0.5415, | |
| "eval_samples_per_second": 1846.815, | |
| "eval_steps_per_second": 29.549, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 0.00026167996704760406, | |
| "loss": 0.3603, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 0.00026135462357735375, | |
| "loss": 0.3604, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "eval_loss": 0.7824327349662781, | |
| "eval_runtime": 0.5307, | |
| "eval_samples_per_second": 1884.218, | |
| "eval_steps_per_second": 30.147, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 0.0002610281170299795, | |
| "loss": 0.3602, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 0.00026070045097610465, | |
| "loss": 0.36, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "eval_loss": 0.787578821182251, | |
| "eval_runtime": 0.5306, | |
| "eval_samples_per_second": 1884.712, | |
| "eval_steps_per_second": 30.155, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.0002603716289990326, | |
| "loss": 0.3601, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 0.00026004165469470787, | |
| "loss": 0.3597, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "eval_loss": 0.7803857326507568, | |
| "eval_runtime": 0.5242, | |
| "eval_samples_per_second": 1907.849, | |
| "eval_steps_per_second": 30.526, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.0002597105316716766, | |
| "loss": 0.3598, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 0.000259378263551047, | |
| "loss": 0.3598, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "eval_loss": 0.7810735106468201, | |
| "eval_runtime": 0.5319, | |
| "eval_samples_per_second": 1879.976, | |
| "eval_steps_per_second": 30.08, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 0.0002590448539664501, | |
| "loss": 0.3595, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.00025871030656399966, | |
| "loss": 0.3585, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "eval_loss": 0.7808945775032043, | |
| "eval_runtime": 0.5345, | |
| "eval_samples_per_second": 1870.803, | |
| "eval_steps_per_second": 29.933, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 0.00025837462500225255, | |
| "loss": 0.3588, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 0.0002580378129521685, | |
| "loss": 0.3589, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "eval_loss": 0.7889230251312256, | |
| "eval_runtime": 0.5368, | |
| "eval_samples_per_second": 1862.837, | |
| "eval_steps_per_second": 29.805, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 0.0002576998740970701, | |
| "loss": 0.3587, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 0.00025736081213260253, | |
| "loss": 0.3588, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "eval_loss": 0.7862944602966309, | |
| "eval_runtime": 0.5148, | |
| "eval_samples_per_second": 1942.516, | |
| "eval_steps_per_second": 31.08, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 0.0002570206307666931, | |
| "loss": 0.3588, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.0002566793337195108, | |
| "loss": 0.3582, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "eval_loss": 0.7859060168266296, | |
| "eval_runtime": 0.5347, | |
| "eval_samples_per_second": 1870.153, | |
| "eval_steps_per_second": 29.922, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 0.0002563369247234254, | |
| "loss": 0.358, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 0.0002559934075229669, | |
| "loss": 0.3577, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "eval_loss": 0.7823663353919983, | |
| "eval_runtime": 0.5372, | |
| "eval_samples_per_second": 1861.564, | |
| "eval_steps_per_second": 29.785, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.0002556487858747843, | |
| "loss": 0.3578, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 0.00025530306354760464, | |
| "loss": 0.3577, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "eval_loss": 0.7786086201667786, | |
| "eval_runtime": 0.5231, | |
| "eval_samples_per_second": 1911.791, | |
| "eval_steps_per_second": 30.589, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.000254956244322192, | |
| "loss": 0.3578, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 0.00025460833199130595, | |
| "loss": 0.3575, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "eval_loss": 0.7865322232246399, | |
| "eval_runtime": 0.5303, | |
| "eval_samples_per_second": 1885.606, | |
| "eval_steps_per_second": 30.17, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 0.00025425933035965983, | |
| "loss": 0.3575, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 0.00025390924324387965, | |
| "loss": 0.3573, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "eval_loss": 0.787778913974762, | |
| "eval_runtime": 0.5202, | |
| "eval_samples_per_second": 1922.286, | |
| "eval_steps_per_second": 30.757, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 0.0002535580744724621, | |
| "loss": 0.3569, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 0.00025320582788573246, | |
| "loss": 0.357, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "eval_loss": 0.7871734499931335, | |
| "eval_runtime": 0.5167, | |
| "eval_samples_per_second": 1935.344, | |
| "eval_steps_per_second": 30.965, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 0.000252852507335803, | |
| "loss": 0.357, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 0.0002524981166865307, | |
| "loss": 0.3569, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "eval_loss": 0.785658597946167, | |
| "eval_runtime": 0.5353, | |
| "eval_samples_per_second": 1868.045, | |
| "eval_steps_per_second": 29.889, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.00025214265981347487, | |
| "loss": 0.3565, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 0.000251786140603855, | |
| "loss": 0.3562, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "eval_loss": 0.7895671129226685, | |
| "eval_runtime": 0.5212, | |
| "eval_samples_per_second": 1918.71, | |
| "eval_steps_per_second": 30.699, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 0.00025142856295650795, | |
| "loss": 0.3564, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 0.0002510699307818457, | |
| "loss": 0.3561, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "eval_loss": 0.7849823832511902, | |
| "eval_runtime": 0.5118, | |
| "eval_samples_per_second": 1954.042, | |
| "eval_steps_per_second": 31.265, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 0.00025071024800181214, | |
| "loss": 0.356, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.0002503495185498405, | |
| "loss": 0.3559, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "eval_loss": 0.779691755771637, | |
| "eval_runtime": 0.5352, | |
| "eval_samples_per_second": 1868.349, | |
| "eval_steps_per_second": 29.894, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 0.00024998774637081044, | |
| "loss": 0.3561, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 0.00024962493542100443, | |
| "loss": 0.3557, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 0.7861126661300659, | |
| "eval_runtime": 0.519, | |
| "eval_samples_per_second": 1926.643, | |
| "eval_steps_per_second": 30.826, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 0.0002492610896680649, | |
| "loss": 0.3554, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00024889621309095067, | |
| "loss": 0.3554, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "eval_loss": 0.7840675115585327, | |
| "eval_runtime": 0.525, | |
| "eval_samples_per_second": 1904.845, | |
| "eval_steps_per_second": 30.478, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 0.00024853030967989366, | |
| "loss": 0.3549, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 0.00024816338343635485, | |
| "loss": 0.3552, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "eval_loss": 0.776941180229187, | |
| "eval_runtime": 0.5336, | |
| "eval_samples_per_second": 1874.176, | |
| "eval_steps_per_second": 29.987, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 0.0002477954383729809, | |
| "loss": 0.3548, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 0.00024742647851355997, | |
| "loss": 0.3548, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "eval_loss": 0.7874701023101807, | |
| "eval_runtime": 0.5123, | |
| "eval_samples_per_second": 1952.132, | |
| "eval_steps_per_second": 31.234, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 0.0002470565078929781, | |
| "loss": 0.3548, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 0.00024668553055717465, | |
| "loss": 0.3548, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "eval_loss": 0.7873478531837463, | |
| "eval_runtime": 0.5309, | |
| "eval_samples_per_second": 1883.499, | |
| "eval_steps_per_second": 30.136, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 0.0002463135505630984, | |
| "loss": 0.3548, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 0.00024594057197866283, | |
| "loss": 0.356, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "eval_loss": 0.7846257090568542, | |
| "eval_runtime": 0.5192, | |
| "eval_samples_per_second": 1926.096, | |
| "eval_steps_per_second": 30.818, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 0.0002455665988827021, | |
| "loss": 0.3542, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.0002451916353649261, | |
| "loss": 0.3538, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "eval_loss": 0.7811290621757507, | |
| "eval_runtime": 0.5184, | |
| "eval_samples_per_second": 1929.028, | |
| "eval_steps_per_second": 30.864, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.00024481568552587566, | |
| "loss": 0.3539, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.0002444387534768781, | |
| "loss": 0.3544, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "eval_loss": 0.7849481105804443, | |
| "eval_runtime": 0.5349, | |
| "eval_samples_per_second": 1869.494, | |
| "eval_steps_per_second": 29.912, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 0.0002440608433400018, | |
| "loss": 0.354, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 0.00024368195924801158, | |
| "loss": 0.3537, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "eval_loss": 0.7821019887924194, | |
| "eval_runtime": 0.5319, | |
| "eval_samples_per_second": 1880.027, | |
| "eval_steps_per_second": 30.08, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 0.00024330210534432314, | |
| "loss": 0.3544, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 0.0002429212857829579, | |
| "loss": 0.3535, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "eval_loss": 0.7799093127250671, | |
| "eval_runtime": 0.5175, | |
| "eval_samples_per_second": 1932.464, | |
| "eval_steps_per_second": 30.919, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 0.00024253950472849758, | |
| "loss": 0.3529, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 0.0002421567663560386, | |
| "loss": 0.353, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "eval_loss": 0.787077784538269, | |
| "eval_runtime": 0.5165, | |
| "eval_samples_per_second": 1936.126, | |
| "eval_steps_per_second": 30.978, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.00024177307485114653, | |
| "loss": 0.3535, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 0.0002413884344098101, | |
| "loss": 0.3528, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "eval_loss": 0.7880011200904846, | |
| "eval_runtime": 0.5056, | |
| "eval_samples_per_second": 1977.97, | |
| "eval_steps_per_second": 31.648, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 0.00024100284923839568, | |
| "loss": 0.3529, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 0.0002406163235536008, | |
| "loss": 0.3528, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "eval_loss": 0.7795886397361755, | |
| "eval_runtime": 0.5086, | |
| "eval_samples_per_second": 1966.255, | |
| "eval_steps_per_second": 31.46, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 0.00024022886158240857, | |
| "loss": 0.3526, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 0.0002398404675620409, | |
| "loss": 0.353, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "eval_loss": 0.7831447124481201, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.417, | |
| "eval_steps_per_second": 30.855, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 0.0002394511457399126, | |
| "loss": 0.3524, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 0.00023906090037358478, | |
| "loss": 0.3519, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "eval_loss": 0.7807596325874329, | |
| "eval_runtime": 0.5293, | |
| "eval_samples_per_second": 1889.214, | |
| "eval_steps_per_second": 30.227, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 0.0002386697357307182, | |
| "loss": 0.3519, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 0.00023827765608902676, | |
| "loss": 0.3517, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "eval_loss": 0.7891790866851807, | |
| "eval_runtime": 0.5285, | |
| "eval_samples_per_second": 1892.002, | |
| "eval_steps_per_second": 30.272, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.0002378846657362306, | |
| "loss": 0.3518, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 0.00023749076897000928, | |
| "loss": 0.3518, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "eval_loss": 0.7876036167144775, | |
| "eval_runtime": 0.5137, | |
| "eval_samples_per_second": 1946.703, | |
| "eval_steps_per_second": 31.147, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 0.00023709597009795465, | |
| "loss": 0.3517, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 0.000236700273437524, | |
| "loss": 0.3515, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "eval_loss": 0.780876100063324, | |
| "eval_runtime": 0.5145, | |
| "eval_samples_per_second": 1943.544, | |
| "eval_steps_per_second": 31.097, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 0.0002363036833159925, | |
| "loss": 0.3509, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 0.00023590620407040633, | |
| "loss": 0.351, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "eval_loss": 0.7936240434646606, | |
| "eval_runtime": 0.5234, | |
| "eval_samples_per_second": 1910.641, | |
| "eval_steps_per_second": 30.57, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 0.00023550784004753471, | |
| "loss": 0.3512, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 0.0002351085956038229, | |
| "loss": 0.3508, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "eval_loss": 0.7791932225227356, | |
| "eval_runtime": 0.5202, | |
| "eval_samples_per_second": 1922.513, | |
| "eval_steps_per_second": 30.76, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 0.00023470847510534407, | |
| "loss": 0.3507, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 0.00023430748292775188, | |
| "loss": 0.3508, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "eval_loss": 0.783431887626648, | |
| "eval_runtime": 0.5231, | |
| "eval_samples_per_second": 1911.723, | |
| "eval_steps_per_second": 30.588, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 0.0002339056234562326, | |
| "loss": 0.3505, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 0.00023350290108545694, | |
| "loss": 0.3505, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "eval_loss": 0.787056028842926, | |
| "eval_runtime": 0.5298, | |
| "eval_samples_per_second": 1887.522, | |
| "eval_steps_per_second": 30.2, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 0.00023309932021953238, | |
| "loss": 0.3502, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 0.00023269488527195446, | |
| "loss": 0.3504, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "eval_loss": 0.7937799692153931, | |
| "eval_runtime": 0.5254, | |
| "eval_samples_per_second": 1903.32, | |
| "eval_steps_per_second": 30.453, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 0.00023228960066555907, | |
| "loss": 0.3515, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 0.00023188347083247365, | |
| "loss": 0.3499, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "eval_loss": 0.7750864028930664, | |
| "eval_runtime": 0.5058, | |
| "eval_samples_per_second": 1977.117, | |
| "eval_steps_per_second": 31.634, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 0.00023147650021406905, | |
| "loss": 0.3502, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 0.00023106869326091075, | |
| "loss": 0.3493, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "eval_loss": 0.7836278080940247, | |
| "eval_runtime": 0.5246, | |
| "eval_samples_per_second": 1906.272, | |
| "eval_steps_per_second": 30.5, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 0.00023066005443271017, | |
| "loss": 0.3496, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 0.00023025058819827618, | |
| "loss": 0.349, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "eval_loss": 0.7845665216445923, | |
| "eval_runtime": 0.5357, | |
| "eval_samples_per_second": 1866.879, | |
| "eval_steps_per_second": 29.87, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 0.00022984029903546578, | |
| "loss": 0.3492, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 0.00022942919143113572, | |
| "loss": 0.3493, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "eval_loss": 0.7854596376419067, | |
| "eval_runtime": 0.5507, | |
| "eval_samples_per_second": 1815.925, | |
| "eval_steps_per_second": 29.055, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 0.0002290172698810927, | |
| "loss": 0.3508, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 0.00022860453889004493, | |
| "loss": 0.3489, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "eval_loss": 0.7816638350486755, | |
| "eval_runtime": 0.5378, | |
| "eval_samples_per_second": 1859.411, | |
| "eval_steps_per_second": 29.751, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 0.00022819100297155235, | |
| "loss": 0.349, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 0.0002277766666479774, | |
| "loss": 0.3492, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "eval_loss": 0.7852403521537781, | |
| "eval_runtime": 0.5329, | |
| "eval_samples_per_second": 1876.357, | |
| "eval_steps_per_second": 30.022, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 0.00022736153445043595, | |
| "loss": 0.3487, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 0.00022694561091874706, | |
| "loss": 0.349, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "eval_loss": 0.7810923457145691, | |
| "eval_runtime": 0.5168, | |
| "eval_samples_per_second": 1934.966, | |
| "eval_steps_per_second": 30.959, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 0.00022652890060138387, | |
| "loss": 0.3486, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 0.00022611140805542366, | |
| "loss": 0.3483, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "eval_loss": 0.78509521484375, | |
| "eval_runtime": 0.541, | |
| "eval_samples_per_second": 1848.302, | |
| "eval_steps_per_second": 29.573, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 0.00022569313784649798, | |
| "loss": 0.3483, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.0002252740945487429, | |
| "loss": 0.3482, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "eval_loss": 0.7885275483131409, | |
| "eval_runtime": 0.5166, | |
| "eval_samples_per_second": 1935.833, | |
| "eval_steps_per_second": 30.973, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 0.00022485428274474867, | |
| "loss": 0.348, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 0.00022443370702551, | |
| "loss": 0.3479, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "eval_loss": 0.7895678877830505, | |
| "eval_runtime": 0.5259, | |
| "eval_samples_per_second": 1901.494, | |
| "eval_steps_per_second": 30.424, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 0.00022401237199037565, | |
| "loss": 0.3476, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 0.0002235902822469979, | |
| "loss": 0.3477, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "eval_loss": 0.7867687940597534, | |
| "eval_runtime": 0.5223, | |
| "eval_samples_per_second": 1914.598, | |
| "eval_steps_per_second": 30.634, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 0.00022316744241128268, | |
| "loss": 0.3476, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 0.00022274385710733855, | |
| "loss": 0.3489, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "eval_loss": 0.7812724709510803, | |
| "eval_runtime": 0.5275, | |
| "eval_samples_per_second": 1895.908, | |
| "eval_steps_per_second": 30.335, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 0.00022231953096742672, | |
| "loss": 0.3471, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 0.00022189446863190974, | |
| "loss": 0.3473, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "eval_loss": 0.7800177931785583, | |
| "eval_runtime": 0.5216, | |
| "eval_samples_per_second": 1917.115, | |
| "eval_steps_per_second": 30.674, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 0.00022146867474920118, | |
| "loss": 0.347, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 0.00022104215397571484, | |
| "loss": 0.3473, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "eval_loss": 0.7774147391319275, | |
| "eval_runtime": 0.5178, | |
| "eval_samples_per_second": 1931.312, | |
| "eval_steps_per_second": 30.901, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 0.0002206149109758135, | |
| "loss": 0.3472, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 0.00022018695042175818, | |
| "loss": 0.3482, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "eval_loss": 0.7872973680496216, | |
| "eval_runtime": 0.5262, | |
| "eval_samples_per_second": 1900.242, | |
| "eval_steps_per_second": 30.404, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 0.00021975827699365693, | |
| "loss": 0.3467, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 0.00021932889537941365, | |
| "loss": 0.3482, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "eval_loss": 0.7847963571548462, | |
| "eval_runtime": 0.5508, | |
| "eval_samples_per_second": 1815.574, | |
| "eval_steps_per_second": 29.049, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 0.0002188988102746769, | |
| "loss": 0.3467, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 0.0002184680263827885, | |
| "loss": 0.3484, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "eval_loss": 0.7891322374343872, | |
| "eval_runtime": 0.5219, | |
| "eval_samples_per_second": 1915.989, | |
| "eval_steps_per_second": 30.656, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 0.00021803654841473204, | |
| "loss": 0.346, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 0.00021760438108908142, | |
| "loss": 0.3457, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "eval_loss": 0.784561276435852, | |
| "eval_runtime": 0.5317, | |
| "eval_samples_per_second": 1880.904, | |
| "eval_steps_per_second": 30.094, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 0.0002171715291319494, | |
| "loss": 0.3479, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 0.0002167379972769355, | |
| "loss": 0.3465, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_loss": 0.7826551795005798, | |
| "eval_runtime": 0.519, | |
| "eval_samples_per_second": 1926.794, | |
| "eval_steps_per_second": 30.829, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 0.0002163037902650747, | |
| "loss": 0.3459, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.0002158689128447853, | |
| "loss": 0.3456, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "eval_loss": 0.7869973182678223, | |
| "eval_runtime": 0.5199, | |
| "eval_samples_per_second": 1923.629, | |
| "eval_steps_per_second": 30.778, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 0.00021543336977181704, | |
| "loss": 0.3456, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 0.00021499716580919933, | |
| "loss": 0.3458, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "eval_loss": 0.7797828912734985, | |
| "eval_runtime": 0.5178, | |
| "eval_samples_per_second": 1931.291, | |
| "eval_steps_per_second": 30.901, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 0.00021456030572718866, | |
| "loss": 0.3456, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 0.000214122794303217, | |
| "loss": 0.3454, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "eval_loss": 0.7822918891906738, | |
| "eval_runtime": 0.5124, | |
| "eval_samples_per_second": 1951.652, | |
| "eval_steps_per_second": 31.226, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 0.00021368463632183912, | |
| "loss": 0.3454, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00021324583657468055, | |
| "loss": 0.3454, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "eval_loss": 0.779866635799408, | |
| "eval_runtime": 0.5273, | |
| "eval_samples_per_second": 1896.43, | |
| "eval_steps_per_second": 30.343, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 0.000212806399860385, | |
| "loss": 0.3451, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 0.00021236633098456196, | |
| "loss": 0.3455, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "eval_loss": 0.783266544342041, | |
| "eval_runtime": 0.5227, | |
| "eval_samples_per_second": 1913.03, | |
| "eval_steps_per_second": 30.608, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 0.0002119256347597342, | |
| "loss": 0.3448, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 0.000211484316005285, | |
| "loss": 0.3454, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "eval_loss": 0.7877166867256165, | |
| "eval_runtime": 0.5202, | |
| "eval_samples_per_second": 1922.214, | |
| "eval_steps_per_second": 30.755, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 0.00021104237954740554, | |
| "loss": 0.3447, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 0.00021059983021904215, | |
| "loss": 0.3444, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "eval_loss": 0.7820906043052673, | |
| "eval_runtime": 0.5318, | |
| "eval_samples_per_second": 1880.364, | |
| "eval_steps_per_second": 30.086, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 0.00021015667285984336, | |
| "loss": 0.3444, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 0.00020971291231610707, | |
| "loss": 0.3445, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "eval_loss": 0.7963393330574036, | |
| "eval_runtime": 0.5097, | |
| "eval_samples_per_second": 1961.765, | |
| "eval_steps_per_second": 31.388, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 0.0002092685534407274, | |
| "loss": 0.3447, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 0.00020882360109314197, | |
| "loss": 0.3442, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "eval_loss": 0.7794355154037476, | |
| "eval_runtime": 0.539, | |
| "eval_samples_per_second": 1855.442, | |
| "eval_steps_per_second": 29.687, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 0.0002083780601392783, | |
| "loss": 0.3442, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 0.0002079319354515008, | |
| "loss": 0.3442, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "eval_loss": 0.7892481088638306, | |
| "eval_runtime": 0.5219, | |
| "eval_samples_per_second": 1916.151, | |
| "eval_steps_per_second": 30.658, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 0.00020748523190855772, | |
| "loss": 0.3439, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 0.0002070379543955273, | |
| "loss": 0.3437, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "eval_loss": 0.7817493081092834, | |
| "eval_runtime": 0.5139, | |
| "eval_samples_per_second": 1945.926, | |
| "eval_steps_per_second": 31.135, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 0.00020659010780376487, | |
| "loss": 0.3436, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.00020614169703084896, | |
| "loss": 0.3436, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "eval_loss": 0.7851976156234741, | |
| "eval_runtime": 0.5495, | |
| "eval_samples_per_second": 1819.926, | |
| "eval_steps_per_second": 29.119, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 0.0002056927269805279, | |
| "loss": 0.3439, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 0.00020524320256266635, | |
| "loss": 0.3435, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "eval_loss": 0.7840863466262817, | |
| "eval_runtime": 0.5295, | |
| "eval_samples_per_second": 1888.573, | |
| "eval_steps_per_second": 30.217, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 0.0002047931286931912, | |
| "loss": 0.3436, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 0.00020434251029403824, | |
| "loss": 0.3435, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "eval_loss": 0.7881864309310913, | |
| "eval_runtime": 0.5367, | |
| "eval_samples_per_second": 1863.216, | |
| "eval_steps_per_second": 29.811, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 0.00020389135229309803, | |
| "loss": 0.3433, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 0.00020343965962416229, | |
| "loss": 0.3438, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "eval_loss": 0.7825648784637451, | |
| "eval_runtime": 0.5312, | |
| "eval_samples_per_second": 1882.622, | |
| "eval_steps_per_second": 30.122, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 0.00020298743722686958, | |
| "loss": 0.3442, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 0.0002025346900466516, | |
| "loss": 0.3447, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "eval_loss": 0.7843804955482483, | |
| "eval_runtime": 0.5233, | |
| "eval_samples_per_second": 1910.972, | |
| "eval_steps_per_second": 30.576, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 0.0002020814230346791, | |
| "loss": 0.3428, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 0.00020162764114780733, | |
| "loss": 0.3427, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "eval_loss": 0.7835332155227661, | |
| "eval_runtime": 0.5169, | |
| "eval_samples_per_second": 1934.661, | |
| "eval_steps_per_second": 30.955, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 0.0002011733493485224, | |
| "loss": 0.3424, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 0.00020071855260488664, | |
| "loss": 0.3424, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "eval_loss": 0.7858285903930664, | |
| "eval_runtime": 0.5163, | |
| "eval_samples_per_second": 1936.764, | |
| "eval_steps_per_second": 30.988, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 0.0002002632558904843, | |
| "loss": 0.3424, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 0.00019980746418436736, | |
| "loss": 0.3427, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "eval_loss": 0.7807540893554688, | |
| "eval_runtime": 0.5276, | |
| "eval_samples_per_second": 1895.318, | |
| "eval_steps_per_second": 30.325, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.00019935118247100088, | |
| "loss": 0.3422, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 0.00019889441574020864, | |
| "loss": 0.3692, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "eval_loss": 0.9048200845718384, | |
| "eval_runtime": 0.5244, | |
| "eval_samples_per_second": 1906.815, | |
| "eval_steps_per_second": 30.509, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 0.0001984371689871183, | |
| "loss": 0.7372, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 0.00019797944721210725, | |
| "loss": 0.7368, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "eval_loss": 0.8997135162353516, | |
| "eval_runtime": 0.5259, | |
| "eval_samples_per_second": 1901.634, | |
| "eval_steps_per_second": 30.426, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 0.00019752125542074736, | |
| "loss": 0.7368, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 0.00019706259862375074, | |
| "loss": 0.7368, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "eval_loss": 0.900769829750061, | |
| "eval_runtime": 0.5434, | |
| "eval_samples_per_second": 1840.131, | |
| "eval_steps_per_second": 29.442, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 0.00019660348183691453, | |
| "loss": 0.7368, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 0.0001961439100810664, | |
| "loss": 0.7367, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "eval_loss": 0.8999822735786438, | |
| "eval_runtime": 0.5136, | |
| "eval_samples_per_second": 1947.174, | |
| "eval_steps_per_second": 31.155, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 0.00019568388838200952, | |
| "loss": 0.7368, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 0.00019522342177046744, | |
| "loss": 0.7368, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "eval_loss": 0.9001851081848145, | |
| "eval_runtime": 0.5249, | |
| "eval_samples_per_second": 1904.969, | |
| "eval_steps_per_second": 30.48, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 0.00019476251528202922, | |
| "loss": 0.7368, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 0.0001943011739570944, | |
| "loss": 0.7368, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "eval_loss": 0.8993794918060303, | |
| "eval_runtime": 0.5175, | |
| "eval_samples_per_second": 1932.289, | |
| "eval_steps_per_second": 30.917, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 0.00019383940284081774, | |
| "loss": 0.7368, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 0.00019337720698305431, | |
| "loss": 0.7368, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "eval_loss": 0.8998441100120544, | |
| "eval_runtime": 0.5264, | |
| "eval_samples_per_second": 1899.844, | |
| "eval_steps_per_second": 30.397, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 0.0001929145914383038, | |
| "loss": 0.7368, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 0.00019245156126565586, | |
| "loss": 0.7368, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "eval_loss": 0.8998916745185852, | |
| "eval_runtime": 0.5129, | |
| "eval_samples_per_second": 1949.567, | |
| "eval_steps_per_second": 31.193, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 0.00019198812152873416, | |
| "loss": 0.7367, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 0.00019152427729564144, | |
| "loss": 0.7368, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "eval_loss": 0.9005412459373474, | |
| "eval_runtime": 0.512, | |
| "eval_samples_per_second": 1953.157, | |
| "eval_steps_per_second": 31.251, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 0.00019106003363890395, | |
| "loss": 0.7367, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 0.00019059539563541584, | |
| "loss": 0.7368, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "eval_loss": 0.9003030061721802, | |
| "eval_runtime": 0.5159, | |
| "eval_samples_per_second": 1938.232, | |
| "eval_steps_per_second": 31.012, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 0.000190130368366384, | |
| "loss": 0.7368, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 0.00018966495691727207, | |
| "loss": 0.7258, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "eval_loss": 0.9103039503097534, | |
| "eval_runtime": 0.522, | |
| "eval_samples_per_second": 1915.767, | |
| "eval_steps_per_second": 30.652, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 0.0001891991663777451, | |
| "loss": 0.6555, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 0.00018873300184161387, | |
| "loss": 0.4986, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "eval_loss": 0.7985826134681702, | |
| "eval_runtime": 0.524, | |
| "eval_samples_per_second": 1908.52, | |
| "eval_steps_per_second": 30.536, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 0.00018826646840677894, | |
| "loss": 0.3615, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 0.00018779957117517532, | |
| "loss": 0.3531, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "eval_loss": 0.788265585899353, | |
| "eval_runtime": 0.5409, | |
| "eval_samples_per_second": 1848.868, | |
| "eval_steps_per_second": 29.582, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.00018733231525271625, | |
| "loss": 0.3497, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 0.00018686470574923766, | |
| "loss": 0.348, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "eval_loss": 0.7847545146942139, | |
| "eval_runtime": 0.5387, | |
| "eval_samples_per_second": 1856.402, | |
| "eval_steps_per_second": 29.702, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 0.00018639674777844224, | |
| "loss": 0.3464, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.00018592844645784327, | |
| "loss": 0.3454, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "eval_loss": 0.7873616814613342, | |
| "eval_runtime": 0.5198, | |
| "eval_samples_per_second": 1923.882, | |
| "eval_steps_per_second": 30.782, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 0.00018545980690870903, | |
| "loss": 0.3447, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 0.00018499083425600648, | |
| "loss": 0.344, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "eval_loss": 0.7882828712463379, | |
| "eval_runtime": 0.5212, | |
| "eval_samples_per_second": 1918.744, | |
| "eval_steps_per_second": 30.7, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 0.00018452153362834552, | |
| "loss": 0.3439, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 0.00018405191015792254, | |
| "loss": 0.3434, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "eval_loss": 0.783893883228302, | |
| "eval_runtime": 0.5131, | |
| "eval_samples_per_second": 1948.878, | |
| "eval_steps_per_second": 31.182, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 0.0001835819689804646, | |
| "loss": 0.3432, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 0.0001831117152351732, | |
| "loss": 0.3425, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "eval_loss": 0.7780929803848267, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.433, | |
| "eval_steps_per_second": 30.855, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 0.00018264115406466778, | |
| "loss": 0.3425, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 0.00018217029061493007, | |
| "loss": 0.3421, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "eval_loss": 0.7869411706924438, | |
| "eval_runtime": 0.5081, | |
| "eval_samples_per_second": 1968.107, | |
| "eval_steps_per_second": 31.49, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 0.00018169913003524717, | |
| "loss": 0.3419, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 0.00018122767747815594, | |
| "loss": 0.3421, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "eval_loss": 0.7876418232917786, | |
| "eval_runtime": 0.5428, | |
| "eval_samples_per_second": 1842.248, | |
| "eval_steps_per_second": 29.476, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 0.00018075593809938574, | |
| "loss": 0.3419, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 0.00018028391705780295, | |
| "loss": 0.3418, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "eval_loss": 0.7868050336837769, | |
| "eval_runtime": 0.5503, | |
| "eval_samples_per_second": 1817.086, | |
| "eval_steps_per_second": 29.073, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 0.0001798116195153541, | |
| "loss": 0.3411, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 0.0001793390506370094, | |
| "loss": 0.3411, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "eval_loss": 0.7816486358642578, | |
| "eval_runtime": 0.5178, | |
| "eval_samples_per_second": 1931.285, | |
| "eval_steps_per_second": 30.901, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 0.00017886621559070638, | |
| "loss": 0.3411, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 0.00017839311954729337, | |
| "loss": 0.3411, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "eval_loss": 0.8025993704795837, | |
| "eval_runtime": 0.536, | |
| "eval_samples_per_second": 1865.76, | |
| "eval_steps_per_second": 29.852, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 0.00017791976768047292, | |
| "loss": 0.341, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 0.00017744616516674518, | |
| "loss": 0.3407, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "eval_loss": 0.780501127243042, | |
| "eval_runtime": 0.5121, | |
| "eval_samples_per_second": 1952.615, | |
| "eval_steps_per_second": 31.242, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 0.00017697231718535132, | |
| "loss": 0.3407, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 0.00017649822891821707, | |
| "loss": 0.3406, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "eval_loss": 0.777356743812561, | |
| "eval_runtime": 0.5274, | |
| "eval_samples_per_second": 1895.979, | |
| "eval_steps_per_second": 30.336, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 0.00017602390554989563, | |
| "loss": 0.3403, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 0.0001755493522675115, | |
| "loss": 0.3402, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "eval_loss": 0.7845531105995178, | |
| "eval_runtime": 0.535, | |
| "eval_samples_per_second": 1869.034, | |
| "eval_steps_per_second": 29.905, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 0.00017507457426070317, | |
| "loss": 0.34, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 0.00017459957672156704, | |
| "loss": 0.3399, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "eval_loss": 0.7895669937133789, | |
| "eval_runtime": 0.5321, | |
| "eval_samples_per_second": 1879.363, | |
| "eval_steps_per_second": 30.07, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 0.00017412436484459998, | |
| "loss": 0.3398, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 0.00017364894382664297, | |
| "loss": 0.3396, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "eval_loss": 0.7811859250068665, | |
| "eval_runtime": 0.5201, | |
| "eval_samples_per_second": 1922.599, | |
| "eval_steps_per_second": 30.762, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 0.00017317331886682408, | |
| "loss": 0.3396, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 0.0001726974951665017, | |
| "loss": 0.3402, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "eval_loss": 0.7840257287025452, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.148, | |
| "eval_steps_per_second": 30.85, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 0.0001722214779292076, | |
| "loss": 0.3397, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 0.00017174527236058998, | |
| "loss": 0.339, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "eval_loss": 0.7810068130493164, | |
| "eval_runtime": 0.5282, | |
| "eval_samples_per_second": 1893.161, | |
| "eval_steps_per_second": 30.291, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 0.00017126888366835662, | |
| "loss": 0.339, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 0.000170792317062218, | |
| "loss": 0.3391, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "eval_loss": 0.7764478325843811, | |
| "eval_runtime": 0.5198, | |
| "eval_samples_per_second": 1923.928, | |
| "eval_steps_per_second": 30.783, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 0.00017031557775383011, | |
| "loss": 0.3388, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 0.0001698386709567377, | |
| "loss": 0.3389, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "eval_loss": 0.7847365736961365, | |
| "eval_runtime": 0.5131, | |
| "eval_samples_per_second": 1948.897, | |
| "eval_steps_per_second": 31.182, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 0.0001693616018863171, | |
| "loss": 0.3389, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 0.00016888437575971913, | |
| "loss": 0.3389, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "eval_loss": 0.784510612487793, | |
| "eval_runtime": 0.532, | |
| "eval_samples_per_second": 1879.811, | |
| "eval_steps_per_second": 30.077, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 0.00016840699779581238, | |
| "loss": 0.3389, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 0.00016792947321512573, | |
| "loss": 0.3385, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "eval_loss": 0.7881464958190918, | |
| "eval_runtime": 0.5139, | |
| "eval_samples_per_second": 1945.778, | |
| "eval_steps_per_second": 31.132, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 0.00016745180723979144, | |
| "loss": 0.3384, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 0.00016697400509348818, | |
| "loss": 0.3381, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "eval_loss": 0.7884257435798645, | |
| "eval_runtime": 0.5412, | |
| "eval_samples_per_second": 1847.755, | |
| "eval_steps_per_second": 29.564, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 0.00016649607200138356, | |
| "loss": 0.3383, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 0.00016601801319007743, | |
| "loss": 0.3382, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "eval_loss": 0.782693088054657, | |
| "eval_runtime": 0.5248, | |
| "eval_samples_per_second": 1905.624, | |
| "eval_steps_per_second": 30.49, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 0.00016553983388754428, | |
| "loss": 0.3381, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 0.00016506153932307636, | |
| "loss": 0.3377, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "eval_loss": 0.7844077944755554, | |
| "eval_runtime": 0.5281, | |
| "eval_samples_per_second": 1893.751, | |
| "eval_steps_per_second": 30.3, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 0.00016458313472722638, | |
| "loss": 0.3376, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 0.00016410462533175045, | |
| "loss": 0.3377, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "eval_loss": 0.7837400436401367, | |
| "eval_runtime": 0.5195, | |
| "eval_samples_per_second": 1924.894, | |
| "eval_steps_per_second": 30.798, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 0.00016362601636955049, | |
| "loss": 0.3378, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 0.00016314731307461754, | |
| "loss": 0.3377, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "eval_loss": 0.7803494930267334, | |
| "eval_runtime": 0.5126, | |
| "eval_samples_per_second": 1950.683, | |
| "eval_steps_per_second": 31.211, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 0.0001626685206819742, | |
| "loss": 0.3376, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 0.0001621896444276172, | |
| "loss": 0.337, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "eval_loss": 0.7785842418670654, | |
| "eval_runtime": 0.5059, | |
| "eval_samples_per_second": 1976.634, | |
| "eval_steps_per_second": 31.626, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 0.00016171068954846067, | |
| "loss": 0.3373, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 0.00016123166128227835, | |
| "loss": 0.3372, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "eval_loss": 0.7810379266738892, | |
| "eval_runtime": 0.5109, | |
| "eval_samples_per_second": 1957.233, | |
| "eval_steps_per_second": 31.316, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 0.0001607525648676467, | |
| "loss": 0.3368, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 0.0001602734055438873, | |
| "loss": 0.337, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "eval_loss": 0.7789760828018188, | |
| "eval_runtime": 0.541, | |
| "eval_samples_per_second": 1848.491, | |
| "eval_steps_per_second": 29.576, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 0.00015979418855100963, | |
| "loss": 0.3367, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 0.00015931491912965417, | |
| "loss": 0.3366, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "eval_loss": 0.7777426838874817, | |
| "eval_runtime": 0.5219, | |
| "eval_samples_per_second": 1916.204, | |
| "eval_steps_per_second": 30.659, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 0.0001588356025210344, | |
| "loss": 0.3369, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 0.00015835624396688, | |
| "loss": 0.3364, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "eval_loss": 0.7759175300598145, | |
| "eval_runtime": 0.5164, | |
| "eval_samples_per_second": 1936.553, | |
| "eval_steps_per_second": 30.985, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 0.00015787684870937924, | |
| "loss": 0.3364, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 0.00015739742199112196, | |
| "loss": 0.3364, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "eval_loss": 0.7805649638175964, | |
| "eval_runtime": 0.5211, | |
| "eval_samples_per_second": 1919.019, | |
| "eval_steps_per_second": 30.704, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 0.00015691796905504187, | |
| "loss": 0.3366, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 0.00015643849514435944, | |
| "loss": 0.3364, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 0.776075005531311, | |
| "eval_runtime": 0.5349, | |
| "eval_samples_per_second": 1869.618, | |
| "eval_steps_per_second": 29.914, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 0.00015595900550252463, | |
| "loss": 0.3362, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 0.00015547950537315926, | |
| "loss": 0.3363, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "eval_loss": 0.7857484221458435, | |
| "eval_runtime": 0.538, | |
| "eval_samples_per_second": 1858.594, | |
| "eval_steps_per_second": 29.738, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 0.00015499999999999997, | |
| "loss": 0.336, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 0.00015452049462684068, | |
| "loss": 0.3359, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "eval_loss": 0.7803733348846436, | |
| "eval_runtime": 0.5241, | |
| "eval_samples_per_second": 1908.102, | |
| "eval_steps_per_second": 30.53, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 0.00015404099449747535, | |
| "loss": 0.3356, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 0.0001535615048556405, | |
| "loss": 0.3357, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "eval_loss": 0.7824040651321411, | |
| "eval_runtime": 0.5311, | |
| "eval_samples_per_second": 1882.719, | |
| "eval_steps_per_second": 30.124, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 0.0001530820309449581, | |
| "loss": 0.3355, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 0.00015260257800887798, | |
| "loss": 0.3354, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "eval_loss": 0.776350200176239, | |
| "eval_runtime": 0.5238, | |
| "eval_samples_per_second": 1908.966, | |
| "eval_steps_per_second": 30.543, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 8.13, | |
| "learning_rate": 0.0001521231512906207, | |
| "loss": 0.3359, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 0.00015164375603311998, | |
| "loss": 0.3355, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "eval_loss": 0.7818763256072998, | |
| "eval_runtime": 0.5354, | |
| "eval_samples_per_second": 1867.774, | |
| "eval_steps_per_second": 29.884, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 0.00015116439747896553, | |
| "loss": 0.3364, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 0.00015068508087034578, | |
| "loss": 0.3352, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "eval_loss": 0.7747774720191956, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.442, | |
| "eval_steps_per_second": 30.855, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 0.00015020581144899027, | |
| "loss": 0.335, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 0.0001497265944561127, | |
| "loss": 0.3347, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "eval_loss": 0.7758739590644836, | |
| "eval_runtime": 0.5149, | |
| "eval_samples_per_second": 1942.219, | |
| "eval_steps_per_second": 31.076, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "learning_rate": 0.00014924743513235327, | |
| "loss": 0.3347, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 0.0001487683387177216, | |
| "loss": 0.3347, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "eval_loss": 0.777352511882782, | |
| "eval_runtime": 0.511, | |
| "eval_samples_per_second": 1956.914, | |
| "eval_steps_per_second": 31.311, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 0.00014828931045153928, | |
| "loss": 0.6491, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 0.00014781035557238272, | |
| "loss": 0.737, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "eval_loss": 0.9018945097923279, | |
| "eval_runtime": 0.5226, | |
| "eval_samples_per_second": 1913.356, | |
| "eval_steps_per_second": 30.614, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 0.00014733147931802578, | |
| "loss": 0.7368, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 0.00014685268692538238, | |
| "loss": 0.7368, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "eval_loss": 0.9023635387420654, | |
| "eval_runtime": 0.5188, | |
| "eval_samples_per_second": 1927.356, | |
| "eval_steps_per_second": 30.838, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 0.00014637398363044946, | |
| "loss": 0.7143, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 0.00014589537466824955, | |
| "loss": 0.3428, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "eval_loss": 0.7834916710853577, | |
| "eval_runtime": 0.5348, | |
| "eval_samples_per_second": 1869.839, | |
| "eval_steps_per_second": 29.917, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 0.00014541686527277356, | |
| "loss": 0.3388, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 0.00014493846067692358, | |
| "loss": 0.3376, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "eval_loss": 0.7843596935272217, | |
| "eval_runtime": 0.5178, | |
| "eval_samples_per_second": 1931.2, | |
| "eval_steps_per_second": 30.899, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 0.00014446016611245567, | |
| "loss": 0.3362, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "learning_rate": 0.00014398198680992252, | |
| "loss": 0.3369, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "eval_loss": 0.7844694256782532, | |
| "eval_runtime": 0.5316, | |
| "eval_samples_per_second": 1881.272, | |
| "eval_steps_per_second": 30.1, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 0.00014350392799861636, | |
| "loss": 0.336, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 0.0001430259949065118, | |
| "loss": 0.3356, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "eval_loss": 0.7838680148124695, | |
| "eval_runtime": 0.52, | |
| "eval_samples_per_second": 1923.025, | |
| "eval_steps_per_second": 30.768, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 0.0001425481927602085, | |
| "loss": 0.3348, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 0.0001420705267848743, | |
| "loss": 0.3352, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "eval_loss": 0.7744572162628174, | |
| "eval_runtime": 0.5156, | |
| "eval_samples_per_second": 1939.351, | |
| "eval_steps_per_second": 31.03, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 0.00014159300220418757, | |
| "loss": 0.3342, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 0.0001411156242402808, | |
| "loss": 0.3341, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "eval_loss": 0.7838852405548096, | |
| "eval_runtime": 0.5192, | |
| "eval_samples_per_second": 1925.877, | |
| "eval_steps_per_second": 30.814, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 0.0001406383981136829, | |
| "loss": 0.3339, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 0.00014016132904326226, | |
| "loss": 0.334, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "eval_loss": 0.7889499664306641, | |
| "eval_runtime": 0.5333, | |
| "eval_samples_per_second": 1875.279, | |
| "eval_steps_per_second": 30.004, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 0.00013968442224616989, | |
| "loss": 0.3338, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 0.00013920768293778195, | |
| "loss": 0.3337, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "eval_loss": 0.7802003026008606, | |
| "eval_runtime": 0.5176, | |
| "eval_samples_per_second": 1931.905, | |
| "eval_steps_per_second": 30.91, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 0.00013873111633164336, | |
| "loss": 0.3336, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 0.00013825472763941, | |
| "loss": 0.3338, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "eval_loss": 0.7855395674705505, | |
| "eval_runtime": 0.525, | |
| "eval_samples_per_second": 1904.885, | |
| "eval_steps_per_second": 30.478, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 0.00013777852207079235, | |
| "loss": 0.3337, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 0.00013730250483349825, | |
| "loss": 0.3335, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "eval_loss": 0.7787224054336548, | |
| "eval_runtime": 0.5159, | |
| "eval_samples_per_second": 1938.186, | |
| "eval_steps_per_second": 31.011, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 0.00013682668113317584, | |
| "loss": 0.3334, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 0.00013635105617335703, | |
| "loss": 0.3332, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "eval_loss": 0.7804464101791382, | |
| "eval_runtime": 0.5024, | |
| "eval_samples_per_second": 1990.477, | |
| "eval_steps_per_second": 31.848, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 0.00013587563515539996, | |
| "loss": 0.3335, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 0.00013540042327843296, | |
| "loss": 0.3332, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "eval_loss": 0.7820075750350952, | |
| "eval_runtime": 0.5246, | |
| "eval_samples_per_second": 1906.19, | |
| "eval_steps_per_second": 30.499, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 0.00013492542573929678, | |
| "loss": 0.3364, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 0.00013445064773248846, | |
| "loss": 0.3349, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "eval_loss": 0.7795833349227905, | |
| "eval_runtime": 0.5328, | |
| "eval_samples_per_second": 1877.003, | |
| "eval_steps_per_second": 30.032, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 0.00013397609445010432, | |
| "loss": 0.3324, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 0.00013350177108178288, | |
| "loss": 0.3322, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "eval_loss": 0.778048574924469, | |
| "eval_runtime": 0.5458, | |
| "eval_samples_per_second": 1832.202, | |
| "eval_steps_per_second": 29.315, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 0.00013302768281464863, | |
| "loss": 0.3325, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 0.0001325538348332548, | |
| "loss": 0.3328, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "eval_loss": 0.7764204144477844, | |
| "eval_runtime": 0.5373, | |
| "eval_samples_per_second": 1861.329, | |
| "eval_steps_per_second": 29.781, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 0.00013208023231952706, | |
| "loss": 0.3322, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 0.0001316068804527066, | |
| "loss": 0.3323, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "eval_loss": 0.7780716419219971, | |
| "eval_runtime": 0.5287, | |
| "eval_samples_per_second": 1891.289, | |
| "eval_steps_per_second": 30.261, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 0.00013113378440929353, | |
| "loss": 0.3322, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 0.00013066094936299056, | |
| "loss": 0.332, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "eval_loss": 0.7822167873382568, | |
| "eval_runtime": 0.5284, | |
| "eval_samples_per_second": 1892.53, | |
| "eval_steps_per_second": 30.28, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 0.00013018838048464582, | |
| "loss": 0.332, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 0.00012971608294219702, | |
| "loss": 0.332, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "eval_loss": 0.7825139760971069, | |
| "eval_runtime": 0.5164, | |
| "eval_samples_per_second": 1936.526, | |
| "eval_steps_per_second": 30.984, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 0.00012924406190061423, | |
| "loss": 0.332, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 0.0001287723225218441, | |
| "loss": 0.3323, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "eval_loss": 0.7750741839408875, | |
| "eval_runtime": 0.5106, | |
| "eval_samples_per_second": 1958.426, | |
| "eval_steps_per_second": 31.335, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 0.00012830086996475274, | |
| "loss": 0.3317, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 0.00012782970938506988, | |
| "loss": 0.3317, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "eval_loss": 0.7807593941688538, | |
| "eval_runtime": 0.5168, | |
| "eval_samples_per_second": 1935.056, | |
| "eval_steps_per_second": 30.961, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 0.00012735884593533222, | |
| "loss": 0.3318, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 0.0001268882847648268, | |
| "loss": 0.3314, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "eval_loss": 0.784046471118927, | |
| "eval_runtime": 0.5177, | |
| "eval_samples_per_second": 1931.792, | |
| "eval_steps_per_second": 30.909, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 0.00012641803101953535, | |
| "loss": 0.3315, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 0.00012594808984207743, | |
| "loss": 0.3316, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "eval_loss": 0.7820506691932678, | |
| "eval_runtime": 0.5125, | |
| "eval_samples_per_second": 1951.189, | |
| "eval_steps_per_second": 31.219, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 0.00012547846637165445, | |
| "loss": 0.3313, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 0.00012500916574399346, | |
| "loss": 0.3313, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "eval_loss": 0.7772422432899475, | |
| "eval_runtime": 0.5191, | |
| "eval_samples_per_second": 1926.42, | |
| "eval_steps_per_second": 30.823, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 0.00012454019309129095, | |
| "loss": 0.3312, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 0.0001240715535421567, | |
| "loss": 0.3312, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "eval_loss": 0.7806535959243774, | |
| "eval_runtime": 0.523, | |
| "eval_samples_per_second": 1912.102, | |
| "eval_steps_per_second": 30.594, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 0.00012360325222155773, | |
| "loss": 0.3314, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 0.00012313529425076228, | |
| "loss": 0.3308, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "eval_loss": 0.7832562327384949, | |
| "eval_runtime": 0.5029, | |
| "eval_samples_per_second": 1988.511, | |
| "eval_steps_per_second": 31.816, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 0.00012266768474728372, | |
| "loss": 0.3307, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 0.00012220042882482468, | |
| "loss": 0.3315, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "eval_loss": 0.7853960394859314, | |
| "eval_runtime": 0.5129, | |
| "eval_samples_per_second": 1949.757, | |
| "eval_steps_per_second": 31.196, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 0.00012173353159322102, | |
| "loss": 0.3307, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 0.00012126699815838609, | |
| "loss": 0.3307, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "eval_loss": 0.7787604928016663, | |
| "eval_runtime": 0.5129, | |
| "eval_samples_per_second": 1949.706, | |
| "eval_steps_per_second": 31.195, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 0.00012080083362225484, | |
| "loss": 0.3309, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 0.00012033504308272786, | |
| "loss": 0.3303, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "eval_loss": 0.7810524702072144, | |
| "eval_runtime": 0.5188, | |
| "eval_samples_per_second": 1927.499, | |
| "eval_steps_per_second": 30.84, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 0.00011986963163361598, | |
| "loss": 0.3305, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 0.0001194046043645841, | |
| "loss": 0.3304, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "eval_loss": 0.7832754850387573, | |
| "eval_runtime": 0.5188, | |
| "eval_samples_per_second": 1927.692, | |
| "eval_steps_per_second": 30.843, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 0.00011893996636109606, | |
| "loss": 0.3307, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 0.00011847572270435852, | |
| "loss": 0.3304, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "eval_loss": 0.7778551578521729, | |
| "eval_runtime": 0.503, | |
| "eval_samples_per_second": 1988.252, | |
| "eval_steps_per_second": 31.812, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 0.00011801187847126579, | |
| "loss": 0.3302, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 0.00011754843873434411, | |
| "loss": 0.3302, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "eval_loss": 0.7812706232070923, | |
| "eval_runtime": 0.4986, | |
| "eval_samples_per_second": 2005.479, | |
| "eval_steps_per_second": 32.088, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 0.00011708540856169612, | |
| "loss": 0.3308, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 0.00011662279301694567, | |
| "loss": 0.3306, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "eval_loss": 0.7797490358352661, | |
| "eval_runtime": 0.5088, | |
| "eval_samples_per_second": 1965.516, | |
| "eval_steps_per_second": 31.448, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 0.0001161605971591822, | |
| "loss": 0.3299, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 0.00011569882604290559, | |
| "loss": 0.3299, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "eval_loss": 0.7792276740074158, | |
| "eval_runtime": 0.5025, | |
| "eval_samples_per_second": 1989.889, | |
| "eval_steps_per_second": 31.838, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 0.00011523748471797075, | |
| "loss": 0.3299, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 0.00011477657822953255, | |
| "loss": 0.3301, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "eval_loss": 0.7853004336357117, | |
| "eval_runtime": 0.5155, | |
| "eval_samples_per_second": 1939.771, | |
| "eval_steps_per_second": 31.036, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 0.00011431611161799043, | |
| "loss": 0.3295, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 0.0001138560899189335, | |
| "loss": 0.3293, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "eval_loss": 0.7775614857673645, | |
| "eval_runtime": 0.5102, | |
| "eval_samples_per_second": 1960.019, | |
| "eval_steps_per_second": 31.36, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 0.00011339651816308543, | |
| "loss": 0.3294, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 0.00011293740137624925, | |
| "loss": 0.3292, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "eval_loss": 0.784292995929718, | |
| "eval_runtime": 0.5164, | |
| "eval_samples_per_second": 1936.403, | |
| "eval_steps_per_second": 30.982, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 0.00011247874457925261, | |
| "loss": 0.3294, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 0.0001120205527878927, | |
| "loss": 0.329, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "eval_loss": 0.7785166501998901, | |
| "eval_runtime": 0.5211, | |
| "eval_samples_per_second": 1919.185, | |
| "eval_steps_per_second": 30.707, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 0.00011156283101288165, | |
| "loss": 0.3291, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 0.00011110558425979132, | |
| "loss": 0.3291, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "eval_loss": 0.7782018780708313, | |
| "eval_runtime": 0.5137, | |
| "eval_samples_per_second": 1946.512, | |
| "eval_steps_per_second": 31.144, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 0.00011064881752899906, | |
| "loss": 0.3288, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 0.00011019253581563262, | |
| "loss": 0.3294, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "eval_loss": 0.7799496054649353, | |
| "eval_runtime": 0.5222, | |
| "eval_samples_per_second": 1914.938, | |
| "eval_steps_per_second": 30.639, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 0.00010973674410951567, | |
| "loss": 0.3293, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 0.00010928144739511337, | |
| "loss": 0.329, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "eval_loss": 0.776207685470581, | |
| "eval_runtime": 0.5118, | |
| "eval_samples_per_second": 1953.912, | |
| "eval_steps_per_second": 31.263, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 0.00010882665065147757, | |
| "loss": 0.3287, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 0.00010837235885219267, | |
| "loss": 0.3286, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "eval_loss": 0.7779992818832397, | |
| "eval_runtime": 0.5097, | |
| "eval_samples_per_second": 1962.124, | |
| "eval_steps_per_second": 31.394, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 0.00010791857696532089, | |
| "loss": 0.3287, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 0.00010746530995334832, | |
| "loss": 0.3285, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "eval_loss": 0.7776817679405212, | |
| "eval_runtime": 0.5012, | |
| "eval_samples_per_second": 1995.205, | |
| "eval_steps_per_second": 31.923, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 0.0001070125627731304, | |
| "loss": 0.3285, | |
| "step": 313500 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 0.0001065603403758377, | |
| "loss": 0.3288, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "eval_loss": 0.7795534133911133, | |
| "eval_runtime": 0.523, | |
| "eval_samples_per_second": 1912.141, | |
| "eval_steps_per_second": 30.594, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 0.00010610864770690196, | |
| "loss": 0.3285, | |
| "step": 314500 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 0.00010565748970596172, | |
| "loss": 0.3281, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "eval_loss": 0.7744332551956177, | |
| "eval_runtime": 0.5161, | |
| "eval_samples_per_second": 1937.422, | |
| "eval_steps_per_second": 30.999, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 0.00010520687130680884, | |
| "loss": 0.3279, | |
| "step": 315500 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 0.00010475679743733364, | |
| "loss": 0.3284, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "eval_loss": 0.7782894968986511, | |
| "eval_runtime": 0.5278, | |
| "eval_samples_per_second": 1894.746, | |
| "eval_steps_per_second": 30.316, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 0.00010430727301947202, | |
| "loss": 0.3282, | |
| "step": 316500 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 0.00010385830296915104, | |
| "loss": 0.328, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "eval_loss": 0.7791895866394043, | |
| "eval_runtime": 0.4908, | |
| "eval_samples_per_second": 2037.475, | |
| "eval_steps_per_second": 32.6, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 0.00010340989219623508, | |
| "loss": 0.328, | |
| "step": 317500 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 0.0001029620456044727, | |
| "loss": 0.3278, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "eval_loss": 0.7816545367240906, | |
| "eval_runtime": 0.5081, | |
| "eval_samples_per_second": 1968.272, | |
| "eval_steps_per_second": 31.492, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 0.00010251476809144226, | |
| "loss": 0.3279, | |
| "step": 318500 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 0.00010206806454849917, | |
| "loss": 0.3276, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "eval_loss": 0.7802248597145081, | |
| "eval_runtime": 0.4946, | |
| "eval_samples_per_second": 2021.73, | |
| "eval_steps_per_second": 32.348, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 0.00010162193986072167, | |
| "loss": 0.3272, | |
| "step": 319500 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 0.00010117639890685795, | |
| "loss": 0.3273, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "eval_loss": 0.7808557152748108, | |
| "eval_runtime": 0.499, | |
| "eval_samples_per_second": 2004.192, | |
| "eval_steps_per_second": 32.067, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 0.00010073144655927253, | |
| "loss": 0.3273, | |
| "step": 320500 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 0.0001002870876838929, | |
| "loss": 0.3276, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "eval_loss": 0.7801169157028198, | |
| "eval_runtime": 0.505, | |
| "eval_samples_per_second": 1980.071, | |
| "eval_steps_per_second": 31.681, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 9.984332714015662e-05, | |
| "loss": 0.3272, | |
| "step": 321500 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 9.94001697809578e-05, | |
| "loss": 0.3273, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "eval_loss": 0.7831940650939941, | |
| "eval_runtime": 0.5116, | |
| "eval_samples_per_second": 1954.711, | |
| "eval_steps_per_second": 31.275, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 9.895762045259445e-05, | |
| "loss": 0.3274, | |
| "step": 322500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 9.851568399471498e-05, | |
| "loss": 0.3277, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "eval_loss": 0.7861186265945435, | |
| "eval_runtime": 0.4926, | |
| "eval_samples_per_second": 2030.062, | |
| "eval_steps_per_second": 32.481, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 9.807436524026574e-05, | |
| "loss": 0.3275, | |
| "step": 323500 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 9.763366901543801e-05, | |
| "loss": 0.3269, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "eval_loss": 0.7829710841178894, | |
| "eval_runtime": 0.5448, | |
| "eval_samples_per_second": 1835.417, | |
| "eval_steps_per_second": 29.367, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 9.719360013961495e-05, | |
| "loss": 0.3269, | |
| "step": 324500 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 9.675416342531944e-05, | |
| "loss": 0.3269, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "eval_loss": 0.7878097891807556, | |
| "eval_runtime": 0.4984, | |
| "eval_samples_per_second": 2006.57, | |
| "eval_steps_per_second": 32.105, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 9.631536367816086e-05, | |
| "loss": 0.3282, | |
| "step": 325500 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "learning_rate": 9.587720569678299e-05, | |
| "loss": 0.3267, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "eval_loss": 0.7815366387367249, | |
| "eval_runtime": 0.5069, | |
| "eval_samples_per_second": 1972.941, | |
| "eval_steps_per_second": 31.567, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.543969427281131e-05, | |
| "loss": 0.3268, | |
| "step": 326500 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 9.500283419080062e-05, | |
| "loss": 0.3269, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "eval_loss": 0.7789347767829895, | |
| "eval_runtime": 0.5192, | |
| "eval_samples_per_second": 1925.943, | |
| "eval_steps_per_second": 30.815, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 9.45666302281829e-05, | |
| "loss": 0.3268, | |
| "step": 327500 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 9.413108715521467e-05, | |
| "loss": 0.3266, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "eval_loss": 0.7733815908432007, | |
| "eval_runtime": 0.5128, | |
| "eval_samples_per_second": 1950.206, | |
| "eval_steps_per_second": 31.203, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 9.369620973492525e-05, | |
| "loss": 0.3265, | |
| "step": 328500 | |
| }, | |
| { | |
| "epoch": 10.08, | |
| "learning_rate": 9.326200272306445e-05, | |
| "loss": 0.3262, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 10.08, | |
| "eval_loss": 0.7761348485946655, | |
| "eval_runtime": 0.5067, | |
| "eval_samples_per_second": 1973.572, | |
| "eval_steps_per_second": 31.577, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "learning_rate": 9.282847086805059e-05, | |
| "loss": 0.3267, | |
| "step": 329500 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 9.239561891091853e-05, | |
| "loss": 0.3264, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "eval_loss": 0.7795162200927734, | |
| "eval_runtime": 0.5194, | |
| "eval_samples_per_second": 1925.383, | |
| "eval_steps_per_second": 30.806, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 10.13, | |
| "learning_rate": 9.196345158526793e-05, | |
| "loss": 0.3267, | |
| "step": 330500 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 9.153197361721149e-05, | |
| "loss": 0.3258, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "eval_loss": 0.7773971557617188, | |
| "eval_runtime": 0.5057, | |
| "eval_samples_per_second": 1977.448, | |
| "eval_steps_per_second": 31.639, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 10.16, | |
| "learning_rate": 9.110118972532302e-05, | |
| "loss": 0.3259, | |
| "step": 331500 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 9.067110462058634e-05, | |
| "loss": 0.3261, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "eval_loss": 0.7772080302238464, | |
| "eval_runtime": 0.512, | |
| "eval_samples_per_second": 1953.287, | |
| "eval_steps_per_second": 31.253, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 9.024172300634305e-05, | |
| "loss": 0.3258, | |
| "step": 332500 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 8.981304957824182e-05, | |
| "loss": 0.3257, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "eval_loss": 0.7775481939315796, | |
| "eval_runtime": 0.4889, | |
| "eval_samples_per_second": 2045.391, | |
| "eval_steps_per_second": 32.726, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 8.938508902418643e-05, | |
| "loss": 0.326, | |
| "step": 333500 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 8.89578460242851e-05, | |
| "loss": 0.3277, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "eval_loss": 0.7818301916122437, | |
| "eval_runtime": 0.5001, | |
| "eval_samples_per_second": 1999.531, | |
| "eval_steps_per_second": 31.992, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 10.25, | |
| "learning_rate": 8.85313252507988e-05, | |
| "loss": 0.3257, | |
| "step": 334500 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 8.810553136809027e-05, | |
| "loss": 0.3258, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "eval_loss": 0.7805637121200562, | |
| "eval_runtime": 0.5114, | |
| "eval_samples_per_second": 1955.493, | |
| "eval_steps_per_second": 31.288, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 8.76804690325733e-05, | |
| "loss": 0.3257, | |
| "step": 335500 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "learning_rate": 8.725614289266137e-05, | |
| "loss": 0.3257, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "eval_loss": 0.7819052934646606, | |
| "eval_runtime": 0.5066, | |
| "eval_samples_per_second": 1974.06, | |
| "eval_steps_per_second": 31.585, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 8.683255758871734e-05, | |
| "loss": 0.3258, | |
| "step": 336500 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 8.640971775300207e-05, | |
| "loss": 0.3259, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "eval_loss": 0.7828894257545471, | |
| "eval_runtime": 0.4948, | |
| "eval_samples_per_second": 2020.994, | |
| "eval_steps_per_second": 32.336, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 8.598762800962431e-05, | |
| "loss": 0.325, | |
| "step": 337500 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 8.55662929744899e-05, | |
| "loss": 0.3253, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "eval_loss": 0.7842022776603699, | |
| "eval_runtime": 0.5086, | |
| "eval_samples_per_second": 1966.304, | |
| "eval_steps_per_second": 31.461, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 8.514571725525124e-05, | |
| "loss": 0.325, | |
| "step": 338500 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 8.47259054512571e-05, | |
| "loss": 0.3252, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "eval_loss": 0.7787997722625732, | |
| "eval_runtime": 0.5114, | |
| "eval_samples_per_second": 1955.232, | |
| "eval_steps_per_second": 31.284, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 10.4, | |
| "learning_rate": 8.430686215350198e-05, | |
| "loss": 0.3253, | |
| "step": 339500 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 8.388859194457636e-05, | |
| "loss": 0.3252, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "eval_loss": 0.7811650633811951, | |
| "eval_runtime": 0.5039, | |
| "eval_samples_per_second": 1984.543, | |
| "eval_steps_per_second": 31.753, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 8.347109939861605e-05, | |
| "loss": 0.3251, | |
| "step": 340500 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 8.305438908125285e-05, | |
| "loss": 0.3249, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "eval_loss": 0.7789011001586914, | |
| "eval_runtime": 0.52, | |
| "eval_samples_per_second": 1923.154, | |
| "eval_steps_per_second": 30.77, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 8.263846554956402e-05, | |
| "loss": 0.3247, | |
| "step": 341500 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 8.222333335202254e-05, | |
| "loss": 0.3248, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "eval_loss": 0.7844606041908264, | |
| "eval_runtime": 0.4974, | |
| "eval_samples_per_second": 2010.3, | |
| "eval_steps_per_second": 32.165, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 10.49, | |
| "learning_rate": 8.18089970284477e-05, | |
| "loss": 0.3247, | |
| "step": 342500 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 8.1395461109955e-05, | |
| "loss": 0.3244, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "eval_loss": 0.7819501161575317, | |
| "eval_runtime": 0.518, | |
| "eval_samples_per_second": 1930.544, | |
| "eval_steps_per_second": 30.889, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 8.098273011890726e-05, | |
| "loss": 0.3248, | |
| "step": 343500 | |
| }, | |
| { | |
| "epoch": 10.54, | |
| "learning_rate": 8.057080856886426e-05, | |
| "loss": 0.3242, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 10.54, | |
| "eval_loss": 0.7827469110488892, | |
| "eval_runtime": 0.5044, | |
| "eval_samples_per_second": 1982.423, | |
| "eval_steps_per_second": 31.719, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 8.015970096453414e-05, | |
| "loss": 0.3245, | |
| "step": 344500 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 7.974941180172382e-05, | |
| "loss": 0.3245, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "eval_loss": 0.7778306007385254, | |
| "eval_runtime": 0.5124, | |
| "eval_samples_per_second": 1951.443, | |
| "eval_steps_per_second": 31.223, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 7.933994556728976e-05, | |
| "loss": 0.3245, | |
| "step": 345500 | |
| }, | |
| { | |
| "epoch": 10.6, | |
| "learning_rate": 7.893130673908927e-05, | |
| "loss": 0.3241, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 10.6, | |
| "eval_loss": 0.7757638096809387, | |
| "eval_runtime": 0.5021, | |
| "eval_samples_per_second": 1991.816, | |
| "eval_steps_per_second": 31.869, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 7.852349978593091e-05, | |
| "loss": 0.3243, | |
| "step": 346500 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 7.811652916752633e-05, | |
| "loss": 0.3243, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "eval_loss": 0.7844104766845703, | |
| "eval_runtime": 0.4987, | |
| "eval_samples_per_second": 2005.114, | |
| "eval_steps_per_second": 32.082, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 7.771039933444092e-05, | |
| "loss": 0.3243, | |
| "step": 347500 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 7.730511472804544e-05, | |
| "loss": 0.324, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "eval_loss": 0.7820024490356445, | |
| "eval_runtime": 0.5033, | |
| "eval_samples_per_second": 1986.81, | |
| "eval_steps_per_second": 31.789, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "learning_rate": 7.690067978046758e-05, | |
| "loss": 0.3237, | |
| "step": 348500 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 7.649709891454298e-05, | |
| "loss": 0.3237, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "eval_loss": 0.7792369723320007, | |
| "eval_runtime": 0.5182, | |
| "eval_samples_per_second": 1929.628, | |
| "eval_steps_per_second": 30.874, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 7.609437654376742e-05, | |
| "loss": 0.3236, | |
| "step": 349500 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 7.569251707224812e-05, | |
| "loss": 0.3235, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "eval_loss": 0.7803741693496704, | |
| "eval_runtime": 0.5045, | |
| "eval_samples_per_second": 1982.304, | |
| "eval_steps_per_second": 31.717, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 7.529152489465592e-05, | |
| "loss": 0.3237, | |
| "step": 350500 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 7.489140439617708e-05, | |
| "loss": 0.3239, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "eval_loss": 0.7802942991256714, | |
| "eval_runtime": 0.4971, | |
| "eval_samples_per_second": 2011.848, | |
| "eval_steps_per_second": 32.19, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 7.449215995246522e-05, | |
| "loss": 0.3236, | |
| "step": 351500 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 7.409379592959367e-05, | |
| "loss": 0.3237, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "eval_loss": 0.7798171043395996, | |
| "eval_runtime": 0.501, | |
| "eval_samples_per_second": 1995.947, | |
| "eval_steps_per_second": 31.935, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 7.369631668400746e-05, | |
| "loss": 0.3234, | |
| "step": 352500 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 7.3299726562476e-05, | |
| "loss": 0.3231, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "eval_loss": 0.7781672477722168, | |
| "eval_runtime": 0.5053, | |
| "eval_samples_per_second": 1979.072, | |
| "eval_steps_per_second": 31.665, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 10.83, | |
| "learning_rate": 7.290402990204531e-05, | |
| "loss": 0.3233, | |
| "step": 353500 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 7.250923102999073e-05, | |
| "loss": 0.3234, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "eval_loss": 0.7746726870536804, | |
| "eval_runtime": 0.5021, | |
| "eval_samples_per_second": 1991.707, | |
| "eval_steps_per_second": 31.867, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 10.86, | |
| "learning_rate": 7.211533426376934e-05, | |
| "loss": 0.3234, | |
| "step": 354500 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 7.172234391097317e-05, | |
| "loss": 0.3232, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "eval_loss": 0.7761996984481812, | |
| "eval_runtime": 0.5166, | |
| "eval_samples_per_second": 1935.585, | |
| "eval_steps_per_second": 30.969, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 7.133026426928173e-05, | |
| "loss": 0.3231, | |
| "step": 355500 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 7.093909962641514e-05, | |
| "loss": 0.3254, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "eval_loss": 0.7848865985870361, | |
| "eval_runtime": 0.5114, | |
| "eval_samples_per_second": 1955.496, | |
| "eval_steps_per_second": 31.288, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 7.054885426008737e-05, | |
| "loss": 0.3229, | |
| "step": 356500 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "learning_rate": 7.015953243795907e-05, | |
| "loss": 0.3229, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "eval_loss": 0.7789940237998962, | |
| "eval_runtime": 0.5055, | |
| "eval_samples_per_second": 1978.178, | |
| "eval_steps_per_second": 31.651, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 6.97711384175914e-05, | |
| "loss": 0.3244, | |
| "step": 357500 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "learning_rate": 6.938367644639911e-05, | |
| "loss": 0.3227, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "eval_loss": 0.7808487415313721, | |
| "eval_runtime": 0.5081, | |
| "eval_samples_per_second": 1968.006, | |
| "eval_steps_per_second": 31.488, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 6.899715076160425e-05, | |
| "loss": 0.3226, | |
| "step": 358500 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 6.861156559018986e-05, | |
| "loss": 0.323, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_loss": 0.7747591137886047, | |
| "eval_runtime": 0.493, | |
| "eval_samples_per_second": 2028.496, | |
| "eval_steps_per_second": 32.456, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 6.822692514885346e-05, | |
| "loss": 0.3225, | |
| "step": 359500 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 6.784323364396135e-05, | |
| "loss": 0.3224, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "eval_loss": 0.7760407328605652, | |
| "eval_runtime": 0.5204, | |
| "eval_samples_per_second": 1921.599, | |
| "eval_steps_per_second": 30.746, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 6.746049527150238e-05, | |
| "loss": 0.3226, | |
| "step": 360500 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 6.707871421704209e-05, | |
| "loss": 0.3225, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "eval_loss": 0.7777162790298462, | |
| "eval_runtime": 0.5171, | |
| "eval_samples_per_second": 1933.691, | |
| "eval_steps_per_second": 30.939, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 6.669789465567683e-05, | |
| "loss": 0.3226, | |
| "step": 361500 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "learning_rate": 6.631804075198838e-05, | |
| "loss": 0.3223, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "eval_loss": 0.7790626287460327, | |
| "eval_runtime": 0.512, | |
| "eval_samples_per_second": 1953.298, | |
| "eval_steps_per_second": 31.253, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "learning_rate": 6.593915665999816e-05, | |
| "loss": 0.3221, | |
| "step": 362500 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 6.55612465231219e-05, | |
| "loss": 0.3222, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "eval_loss": 0.7765858173370361, | |
| "eval_runtime": 0.5305, | |
| "eval_samples_per_second": 1884.91, | |
| "eval_steps_per_second": 30.159, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 6.518431447412434e-05, | |
| "loss": 0.3219, | |
| "step": 363500 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 6.480836463507392e-05, | |
| "loss": 0.322, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "eval_loss": 0.7808003425598145, | |
| "eval_runtime": 0.5024, | |
| "eval_samples_per_second": 1990.262, | |
| "eval_steps_per_second": 31.844, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 6.443340111729786e-05, | |
| "loss": 0.3219, | |
| "step": 364500 | |
| }, | |
| { | |
| "epoch": 11.18, | |
| "learning_rate": 6.405942802133713e-05, | |
| "loss": 0.322, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 11.18, | |
| "eval_loss": 0.7744137048721313, | |
| "eval_runtime": 0.5239, | |
| "eval_samples_per_second": 1908.642, | |
| "eval_steps_per_second": 30.538, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 11.2, | |
| "learning_rate": 6.36864494369016e-05, | |
| "loss": 0.3218, | |
| "step": 365500 | |
| }, | |
| { | |
| "epoch": 11.21, | |
| "learning_rate": 6.331446944282534e-05, | |
| "loss": 0.3217, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 11.21, | |
| "eval_loss": 0.7770714163780212, | |
| "eval_runtime": 0.5069, | |
| "eval_samples_per_second": 1972.691, | |
| "eval_steps_per_second": 31.563, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 6.294349210702188e-05, | |
| "loss": 0.3219, | |
| "step": 366500 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "learning_rate": 6.257352148643998e-05, | |
| "loss": 0.3217, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "eval_loss": 0.7741234302520752, | |
| "eval_runtime": 0.5107, | |
| "eval_samples_per_second": 1958.008, | |
| "eval_steps_per_second": 31.328, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 6.220456162701908e-05, | |
| "loss": 0.3215, | |
| "step": 367500 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "learning_rate": 6.183661656364515e-05, | |
| "loss": 0.3216, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "eval_loss": 0.7758086919784546, | |
| "eval_runtime": 0.5088, | |
| "eval_samples_per_second": 1965.315, | |
| "eval_steps_per_second": 31.445, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 11.29, | |
| "learning_rate": 6.146969032010631e-05, | |
| "loss": 0.3214, | |
| "step": 368500 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 6.110378690904928e-05, | |
| "loss": 0.3216, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "eval_loss": 0.7766358256340027, | |
| "eval_runtime": 0.5033, | |
| "eval_samples_per_second": 1986.831, | |
| "eval_steps_per_second": 31.789, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 6.073891033193507e-05, | |
| "loss": 0.3214, | |
| "step": 369500 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 6.037506457899553e-05, | |
| "loss": 0.3213, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "eval_loss": 0.7805209159851074, | |
| "eval_runtime": 0.5089, | |
| "eval_samples_per_second": 1964.929, | |
| "eval_steps_per_second": 31.439, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 11.35, | |
| "learning_rate": 6.0012253629189544e-05, | |
| "loss": 0.3214, | |
| "step": 370500 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 5.965048145015944e-05, | |
| "loss": 0.3212, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "eval_loss": 0.7814466953277588, | |
| "eval_runtime": 0.5088, | |
| "eval_samples_per_second": 1965.383, | |
| "eval_steps_per_second": 31.446, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 5.928975199818785e-05, | |
| "loss": 0.3215, | |
| "step": 371500 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 5.893006921815428e-05, | |
| "loss": 0.3213, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "eval_loss": 0.7858847975730896, | |
| "eval_runtime": 0.5207, | |
| "eval_samples_per_second": 1920.357, | |
| "eval_steps_per_second": 30.726, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 5.857143704349198e-05, | |
| "loss": 0.321, | |
| "step": 372500 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "learning_rate": 5.8213859396144986e-05, | |
| "loss": 0.3213, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "eval_loss": 0.780572772026062, | |
| "eval_runtime": 0.523, | |
| "eval_samples_per_second": 1911.94, | |
| "eval_steps_per_second": 30.591, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 5.785734018652507e-05, | |
| "loss": 0.3212, | |
| "step": 373500 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 5.750188331346927e-05, | |
| "loss": 0.3211, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "eval_loss": 0.7766540050506592, | |
| "eval_runtime": 0.5243, | |
| "eval_samples_per_second": 1907.256, | |
| "eval_steps_per_second": 30.516, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 5.714749266419695e-05, | |
| "loss": 0.3207, | |
| "step": 374500 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 5.6794172114267566e-05, | |
| "loss": 0.3208, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "eval_loss": 0.7823048233985901, | |
| "eval_runtime": 0.5063, | |
| "eval_samples_per_second": 1975.154, | |
| "eval_steps_per_second": 31.602, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 11.5, | |
| "learning_rate": 5.6441925527537914e-05, | |
| "loss": 0.3207, | |
| "step": 375500 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 5.60907567561203e-05, | |
| "loss": 0.3202, | |
| "step": 376000 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "eval_loss": 0.7723506689071655, | |
| "eval_runtime": 0.5144, | |
| "eval_samples_per_second": 1944.173, | |
| "eval_steps_per_second": 31.107, | |
| "step": 376000 | |
| }, | |
| { | |
| "epoch": 11.54, | |
| "learning_rate": 5.574066964034012e-05, | |
| "loss": 0.3207, | |
| "step": 376500 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 5.539166800869402e-05, | |
| "loss": 0.3208, | |
| "step": 377000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "eval_loss": 0.7742220163345337, | |
| "eval_runtime": 0.4989, | |
| "eval_samples_per_second": 2004.565, | |
| "eval_steps_per_second": 32.073, | |
| "step": 377000 | |
| }, | |
| { | |
| "epoch": 11.57, | |
| "learning_rate": 5.5043755677807955e-05, | |
| "loss": 0.3206, | |
| "step": 377500 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 5.4696936452395344e-05, | |
| "loss": 0.3205, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "eval_loss": 0.7778945565223694, | |
| "eval_runtime": 0.5033, | |
| "eval_samples_per_second": 1986.827, | |
| "eval_steps_per_second": 31.789, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 11.6, | |
| "learning_rate": 5.435121412521576e-05, | |
| "loss": 0.3207, | |
| "step": 378500 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "learning_rate": 5.400659247703307e-05, | |
| "loss": 0.3204, | |
| "step": 379000 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "eval_loss": 0.7782105803489685, | |
| "eval_runtime": 0.507, | |
| "eval_samples_per_second": 1972.197, | |
| "eval_steps_per_second": 31.555, | |
| "step": 379000 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 5.36630752765745e-05, | |
| "loss": 0.3202, | |
| "step": 379500 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "learning_rate": 5.3320666280489146e-05, | |
| "loss": 0.3203, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "eval_loss": 0.7813093662261963, | |
| "eval_runtime": 0.5119, | |
| "eval_samples_per_second": 1953.621, | |
| "eval_steps_per_second": 31.258, | |
| "step": 380000 | |
| }, | |
| { | |
| "epoch": 11.66, | |
| "learning_rate": 5.2979369233306834e-05, | |
| "loss": 0.3201, | |
| "step": 380500 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "learning_rate": 5.26391878673975e-05, | |
| "loss": 0.32, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "eval_loss": 0.7779923677444458, | |
| "eval_runtime": 0.495, | |
| "eval_samples_per_second": 2020.075, | |
| "eval_steps_per_second": 32.321, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 5.230012590292987e-05, | |
| "loss": 0.3199, | |
| "step": 381500 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 5.1962187047831517e-05, | |
| "loss": 0.32, | |
| "step": 382000 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "eval_loss": 0.7783936858177185, | |
| "eval_runtime": 0.517, | |
| "eval_samples_per_second": 1934.07, | |
| "eval_steps_per_second": 30.945, | |
| "step": 382000 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 5.162537499774743e-05, | |
| "loss": 0.3201, | |
| "step": 382500 | |
| }, | |
| { | |
| "epoch": 11.73, | |
| "learning_rate": 5.128969343600032e-05, | |
| "loss": 0.32, | |
| "step": 383000 | |
| }, | |
| { | |
| "epoch": 11.73, | |
| "eval_loss": 0.7770859599113464, | |
| "eval_runtime": 0.5158, | |
| "eval_samples_per_second": 1938.792, | |
| "eval_steps_per_second": 31.021, | |
| "step": 383000 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 5.09551460335499e-05, | |
| "loss": 0.3199, | |
| "step": 383500 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "learning_rate": 5.062173644895296e-05, | |
| "loss": 0.3199, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "eval_loss": 0.7772809863090515, | |
| "eval_runtime": 0.526, | |
| "eval_samples_per_second": 1901.052, | |
| "eval_steps_per_second": 30.417, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 5.0289468328323434e-05, | |
| "loss": 0.32, | |
| "step": 384500 | |
| }, | |
| { | |
| "epoch": 11.8, | |
| "learning_rate": 4.995834530529208e-05, | |
| "loss": 0.3198, | |
| "step": 385000 | |
| }, | |
| { | |
| "epoch": 11.8, | |
| "eval_loss": 0.7799978852272034, | |
| "eval_runtime": 0.5198, | |
| "eval_samples_per_second": 1923.762, | |
| "eval_steps_per_second": 30.78, | |
| "step": 385000 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 4.9628371000967394e-05, | |
| "loss": 0.3198, | |
| "step": 385500 | |
| }, | |
| { | |
| "epoch": 11.83, | |
| "learning_rate": 4.929954902389534e-05, | |
| "loss": 0.3196, | |
| "step": 386000 | |
| }, | |
| { | |
| "epoch": 11.83, | |
| "eval_loss": 0.7772753834724426, | |
| "eval_runtime": 0.5182, | |
| "eval_samples_per_second": 1929.679, | |
| "eval_steps_per_second": 30.875, | |
| "step": 386000 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 4.897188297002046e-05, | |
| "loss": 0.3194, | |
| "step": 386500 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 4.8645376422646226e-05, | |
| "loss": 0.3194, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "eval_loss": 0.7773513197898865, | |
| "eval_runtime": 0.5104, | |
| "eval_samples_per_second": 1959.206, | |
| "eval_steps_per_second": 31.347, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 11.87, | |
| "learning_rate": 4.832003295239591e-05, | |
| "loss": 0.3194, | |
| "step": 387500 | |
| }, | |
| { | |
| "epoch": 11.89, | |
| "learning_rate": 4.7995856117173624e-05, | |
| "loss": 0.3198, | |
| "step": 388000 | |
| }, | |
| { | |
| "epoch": 11.89, | |
| "eval_loss": 0.7708141803741455, | |
| "eval_runtime": 0.5167, | |
| "eval_samples_per_second": 1935.457, | |
| "eval_steps_per_second": 30.967, | |
| "step": 388000 | |
| }, | |
| { | |
| "epoch": 11.9, | |
| "learning_rate": 4.767284946212521e-05, | |
| "loss": 0.3192, | |
| "step": 388500 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 4.735101651959977e-05, | |
| "loss": 0.3191, | |
| "step": 389000 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "eval_loss": 0.7765528559684753, | |
| "eval_runtime": 0.5105, | |
| "eval_samples_per_second": 1958.826, | |
| "eval_steps_per_second": 31.341, | |
| "step": 389000 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 4.7030360809110754e-05, | |
| "loss": 0.3195, | |
| "step": 389500 | |
| }, | |
| { | |
| "epoch": 11.95, | |
| "learning_rate": 4.6710885837297726e-05, | |
| "loss": 0.3193, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 11.95, | |
| "eval_loss": 0.7797139883041382, | |
| "eval_runtime": 0.5113, | |
| "eval_samples_per_second": 1955.764, | |
| "eval_steps_per_second": 31.292, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 11.96, | |
| "learning_rate": 4.639259509788768e-05, | |
| "loss": 0.3189, | |
| "step": 390500 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "learning_rate": 4.60754920716572e-05, | |
| "loss": 0.3193, | |
| "step": 391000 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "eval_loss": 0.7768589854240417, | |
| "eval_runtime": 0.4937, | |
| "eval_samples_per_second": 2025.701, | |
| "eval_steps_per_second": 32.411, | |
| "step": 391000 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 4.5759580226394167e-05, | |
| "loss": 0.3191, | |
| "step": 391500 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.544486301685993e-05, | |
| "loss": 0.3191, | |
| "step": 392000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "eval_loss": 0.7793305516242981, | |
| "eval_runtime": 0.4963, | |
| "eval_samples_per_second": 2015.052, | |
| "eval_steps_per_second": 32.241, | |
| "step": 392000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.5131343884751484e-05, | |
| "loss": 0.3189, | |
| "step": 392500 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.4819026258663774e-05, | |
| "loss": 0.3188, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "eval_loss": 0.7801252007484436, | |
| "eval_runtime": 0.5034, | |
| "eval_samples_per_second": 1986.635, | |
| "eval_steps_per_second": 31.786, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 4.450791355405234e-05, | |
| "loss": 0.3188, | |
| "step": 393500 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 4.419800917319588e-05, | |
| "loss": 0.3188, | |
| "step": 394000 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "eval_loss": 0.7790648937225342, | |
| "eval_runtime": 0.5072, | |
| "eval_samples_per_second": 1971.795, | |
| "eval_steps_per_second": 31.549, | |
| "step": 394000 | |
| }, | |
| { | |
| "epoch": 12.09, | |
| "learning_rate": 4.3889316505159056e-05, | |
| "loss": 0.3185, | |
| "step": 394500 | |
| }, | |
| { | |
| "epoch": 12.1, | |
| "learning_rate": 4.3581838925755465e-05, | |
| "loss": 0.3187, | |
| "step": 395000 | |
| }, | |
| { | |
| "epoch": 12.1, | |
| "eval_loss": 0.7797361016273499, | |
| "eval_runtime": 0.4981, | |
| "eval_samples_per_second": 2007.807, | |
| "eval_steps_per_second": 32.125, | |
| "step": 395000 | |
| }, | |
| { | |
| "epoch": 12.12, | |
| "learning_rate": 4.327557979751057e-05, | |
| "loss": 0.3187, | |
| "step": 395500 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.297054246962517e-05, | |
| "loss": 0.3186, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "eval_loss": 0.7804706692695618, | |
| "eval_runtime": 0.5208, | |
| "eval_samples_per_second": 1920.29, | |
| "eval_steps_per_second": 30.725, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 4.266673027793864e-05, | |
| "loss": 0.3184, | |
| "step": 396500 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 4.236414654489242e-05, | |
| "loss": 0.3185, | |
| "step": 397000 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "eval_loss": 0.7770272493362427, | |
| "eval_runtime": 0.5197, | |
| "eval_samples_per_second": 1924.081, | |
| "eval_steps_per_second": 30.785, | |
| "step": 397000 | |
| }, | |
| { | |
| "epoch": 12.18, | |
| "learning_rate": 4.206279457949371e-05, | |
| "loss": 0.3183, | |
| "step": 397500 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 4.1762677677279335e-05, | |
| "loss": 0.3185, | |
| "step": 398000 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "eval_loss": 0.7785659432411194, | |
| "eval_runtime": 0.5333, | |
| "eval_samples_per_second": 1875.117, | |
| "eval_steps_per_second": 30.002, | |
| "step": 398000 | |
| }, | |
| { | |
| "epoch": 12.21, | |
| "learning_rate": 4.146379912027964e-05, | |
| "loss": 0.3184, | |
| "step": 398500 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 4.1166162176982664e-05, | |
| "loss": 0.3187, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "eval_loss": 0.7719516754150391, | |
| "eval_runtime": 0.5091, | |
| "eval_samples_per_second": 1964.072, | |
| "eval_steps_per_second": 31.425, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 12.24, | |
| "learning_rate": 4.086977010229838e-05, | |
| "loss": 0.3182, | |
| "step": 399500 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "learning_rate": 4.057462613752294e-05, | |
| "loss": 0.3181, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "eval_loss": 0.7777762413024902, | |
| "eval_runtime": 0.5134, | |
| "eval_samples_per_second": 1947.706, | |
| "eval_steps_per_second": 31.163, | |
| "step": 400000 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "learning_rate": 4.0280733510303475e-05, | |
| "loss": 0.3181, | |
| "step": 400500 | |
| }, | |
| { | |
| "epoch": 12.29, | |
| "learning_rate": 3.9988095434602716e-05, | |
| "loss": 0.3181, | |
| "step": 401000 | |
| }, | |
| { | |
| "epoch": 12.29, | |
| "eval_loss": 0.7763352394104004, | |
| "eval_runtime": 0.5244, | |
| "eval_samples_per_second": 1906.9, | |
| "eval_steps_per_second": 30.51, | |
| "step": 401000 | |
| }, | |
| { | |
| "epoch": 12.3, | |
| "learning_rate": 3.9696715110663726e-05, | |
| "loss": 0.3182, | |
| "step": 401500 | |
| }, | |
| { | |
| "epoch": 12.32, | |
| "learning_rate": 3.9406595724975116e-05, | |
| "loss": 0.3178, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 12.32, | |
| "eval_loss": 0.7796168923377991, | |
| "eval_runtime": 0.5013, | |
| "eval_samples_per_second": 1994.987, | |
| "eval_steps_per_second": 31.92, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 3.9117740450235914e-05, | |
| "loss": 0.318, | |
| "step": 402500 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 3.8830152445321163e-05, | |
| "loss": 0.3179, | |
| "step": 403000 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "eval_loss": 0.77396160364151, | |
| "eval_runtime": 0.5031, | |
| "eval_samples_per_second": 1987.545, | |
| "eval_steps_per_second": 31.801, | |
| "step": 403000 | |
| }, | |
| { | |
| "epoch": 12.36, | |
| "learning_rate": 3.854383485524724e-05, | |
| "loss": 0.3179, | |
| "step": 403500 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 3.8258790811137425e-05, | |
| "loss": 0.3177, | |
| "step": 404000 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "eval_loss": 0.7777731418609619, | |
| "eval_runtime": 0.5211, | |
| "eval_samples_per_second": 1919.077, | |
| "eval_steps_per_second": 30.705, | |
| "step": 404000 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "learning_rate": 3.7975023430187676e-05, | |
| "loss": 0.3179, | |
| "step": 404500 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 3.7692535815632624e-05, | |
| "loss": 0.3178, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "eval_loss": 0.774276852607727, | |
| "eval_runtime": 0.499, | |
| "eval_samples_per_second": 2004.004, | |
| "eval_steps_per_second": 32.064, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 12.42, | |
| "learning_rate": 3.741133105671159e-05, | |
| "loss": 0.3176, | |
| "step": 405500 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 3.713141222863474e-05, | |
| "loss": 0.3175, | |
| "step": 406000 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "eval_loss": 0.7776892185211182, | |
| "eval_runtime": 0.501, | |
| "eval_samples_per_second": 1996.009, | |
| "eval_steps_per_second": 31.936, | |
| "step": 406000 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 3.6852782392549584e-05, | |
| "loss": 0.3179, | |
| "step": 406500 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "learning_rate": 3.657544459550729e-05, | |
| "loss": 0.3175, | |
| "step": 407000 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "eval_loss": 0.7784814834594727, | |
| "eval_runtime": 0.5197, | |
| "eval_samples_per_second": 1924.263, | |
| "eval_steps_per_second": 30.788, | |
| "step": 407000 | |
| }, | |
| { | |
| "epoch": 12.49, | |
| "learning_rate": 3.6299401870429606e-05, | |
| "loss": 0.3176, | |
| "step": 407500 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 3.6024657236075546e-05, | |
| "loss": 0.3175, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "eval_loss": 0.7798225283622742, | |
| "eval_runtime": 0.5118, | |
| "eval_samples_per_second": 1953.911, | |
| "eval_steps_per_second": 31.263, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "learning_rate": 3.575121369700841e-05, | |
| "loss": 0.3173, | |
| "step": 408500 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "learning_rate": 3.5479074243562995e-05, | |
| "loss": 0.3172, | |
| "step": 409000 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "eval_loss": 0.7795943021774292, | |
| "eval_runtime": 0.5028, | |
| "eval_samples_per_second": 1988.844, | |
| "eval_steps_per_second": 31.821, | |
| "step": 409000 | |
| }, | |
| { | |
| "epoch": 12.55, | |
| "learning_rate": 3.5208241851812644e-05, | |
| "loss": 0.3172, | |
| "step": 409500 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 3.493871948353709e-05, | |
| "loss": 0.3174, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "eval_loss": 0.7746796607971191, | |
| "eval_runtime": 0.5121, | |
| "eval_samples_per_second": 1952.734, | |
| "eval_steps_per_second": 31.244, | |
| "step": 410000 | |
| }, | |
| { | |
| "epoch": 12.58, | |
| "learning_rate": 3.4670510086189736e-05, | |
| "loss": 0.3172, | |
| "step": 410500 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 3.440361659286563e-05, | |
| "loss": 0.3168, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "eval_loss": 0.7748520374298096, | |
| "eval_runtime": 0.5115, | |
| "eval_samples_per_second": 1955.169, | |
| "eval_steps_per_second": 31.283, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 3.413804192226918e-05, | |
| "loss": 0.317, | |
| "step": 411500 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 3.387378897868246e-05, | |
| "loss": 0.317, | |
| "step": 412000 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "eval_loss": 0.7790003418922424, | |
| "eval_runtime": 0.5251, | |
| "eval_samples_per_second": 1904.548, | |
| "eval_steps_per_second": 30.473, | |
| "step": 412000 | |
| }, | |
| { | |
| "epoch": 12.64, | |
| "learning_rate": 3.361086065193336e-05, | |
| "loss": 0.3169, | |
| "step": 412500 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "learning_rate": 3.334925981736389e-05, | |
| "loss": 0.3169, | |
| "step": 413000 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "eval_loss": 0.7779929041862488, | |
| "eval_runtime": 0.5032, | |
| "eval_samples_per_second": 1987.349, | |
| "eval_steps_per_second": 31.798, | |
| "step": 413000 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 3.3088989335798925e-05, | |
| "loss": 0.3169, | |
| "step": 413500 | |
| }, | |
| { | |
| "epoch": 12.68, | |
| "learning_rate": 3.283005205351467e-05, | |
| "loss": 0.3165, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 12.68, | |
| "eval_loss": 0.7767853736877441, | |
| "eval_runtime": 0.5028, | |
| "eval_samples_per_second": 1988.956, | |
| "eval_steps_per_second": 31.823, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 12.7, | |
| "learning_rate": 3.2572450802207845e-05, | |
| "loss": 0.3169, | |
| "step": 414500 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 3.2316188398964344e-05, | |
| "loss": 0.3165, | |
| "step": 415000 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "eval_loss": 0.7794355750083923, | |
| "eval_runtime": 0.5308, | |
| "eval_samples_per_second": 1884.074, | |
| "eval_steps_per_second": 30.145, | |
| "step": 415000 | |
| }, | |
| { | |
| "epoch": 12.73, | |
| "learning_rate": 3.206126764622888e-05, | |
| "loss": 0.3166, | |
| "step": 415500 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 3.180769133177392e-05, | |
| "loss": 0.3165, | |
| "step": 416000 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "eval_loss": 0.7837159633636475, | |
| "eval_runtime": 0.5042, | |
| "eval_samples_per_second": 1983.371, | |
| "eval_steps_per_second": 31.734, | |
| "step": 416000 | |
| }, | |
| { | |
| "epoch": 12.76, | |
| "learning_rate": 3.155546222866939e-05, | |
| "loss": 0.3168, | |
| "step": 416500 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 3.130458309525239e-05, | |
| "loss": 0.3164, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "eval_loss": 0.7775956988334656, | |
| "eval_runtime": 0.5205, | |
| "eval_samples_per_second": 1921.225, | |
| "eval_steps_per_second": 30.74, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 12.79, | |
| "learning_rate": 3.1055056675096826e-05, | |
| "loss": 0.3165, | |
| "step": 417500 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 3.0806885696983816e-05, | |
| "loss": 0.3165, | |
| "step": 418000 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "eval_loss": 0.7772942185401917, | |
| "eval_runtime": 0.5181, | |
| "eval_samples_per_second": 1929.962, | |
| "eval_steps_per_second": 30.879, | |
| "step": 418000 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 3.056007287487128e-05, | |
| "loss": 0.3164, | |
| "step": 418500 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 3.0314620907864744e-05, | |
| "loss": 0.3164, | |
| "step": 419000 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "eval_loss": 0.7730494141578674, | |
| "eval_runtime": 0.5018, | |
| "eval_samples_per_second": 1992.671, | |
| "eval_steps_per_second": 31.883, | |
| "step": 419000 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 3.0070532480187637e-05, | |
| "loss": 0.3164, | |
| "step": 419500 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 2.9827810261151784e-05, | |
| "loss": 0.3161, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "eval_loss": 0.7735152840614319, | |
| "eval_runtime": 0.5135, | |
| "eval_samples_per_second": 1947.554, | |
| "eval_steps_per_second": 31.161, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 12.88, | |
| "learning_rate": 2.9586456905128618e-05, | |
| "loss": 0.3162, | |
| "step": 420500 | |
| }, | |
| { | |
| "epoch": 12.9, | |
| "learning_rate": 2.9346475051519687e-05, | |
| "loss": 0.3163, | |
| "step": 421000 | |
| }, | |
| { | |
| "epoch": 12.9, | |
| "eval_loss": 0.7781884670257568, | |
| "eval_runtime": 0.5072, | |
| "eval_samples_per_second": 1971.754, | |
| "eval_steps_per_second": 31.548, | |
| "step": 421000 | |
| }, | |
| { | |
| "epoch": 12.91, | |
| "learning_rate": 2.910786732472815e-05, | |
| "loss": 0.3162, | |
| "step": 421500 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "learning_rate": 2.887063633412981e-05, | |
| "loss": 0.3161, | |
| "step": 422000 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "eval_loss": 0.7753953337669373, | |
| "eval_runtime": 0.5016, | |
| "eval_samples_per_second": 1993.582, | |
| "eval_steps_per_second": 31.897, | |
| "step": 422000 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 2.863478467404478e-05, | |
| "loss": 0.3159, | |
| "step": 422500 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "learning_rate": 2.8400314923709112e-05, | |
| "loss": 0.3158, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "eval_loss": 0.7729361653327942, | |
| "eval_runtime": 0.5075, | |
| "eval_samples_per_second": 1970.45, | |
| "eval_steps_per_second": 31.527, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 12.98, | |
| "learning_rate": 2.816722964724636e-05, | |
| "loss": 0.3159, | |
| "step": 423500 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "learning_rate": 2.793553139363981e-05, | |
| "loss": 0.3157, | |
| "step": 424000 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "eval_loss": 0.7751319408416748, | |
| "eval_runtime": 0.5075, | |
| "eval_samples_per_second": 1970.502, | |
| "eval_steps_per_second": 31.528, | |
| "step": 424000 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 2.7705222696704366e-05, | |
| "loss": 0.3159, | |
| "step": 424500 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 2.7476306075059096e-05, | |
| "loss": 0.3157, | |
| "step": 425000 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "eval_loss": 0.7777685523033142, | |
| "eval_runtime": 0.5223, | |
| "eval_samples_per_second": 1914.553, | |
| "eval_steps_per_second": 30.633, | |
| "step": 425000 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 2.7248784032099478e-05, | |
| "loss": 0.3157, | |
| "step": 425500 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 2.7022659055970144e-05, | |
| "loss": 0.3156, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "eval_loss": 0.7756606340408325, | |
| "eval_runtime": 0.511, | |
| "eval_samples_per_second": 1957.114, | |
| "eval_steps_per_second": 31.314, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 2.6797933619537604e-05, | |
| "loss": 0.3155, | |
| "step": 426500 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "learning_rate": 2.6574610180363166e-05, | |
| "loss": 0.3155, | |
| "step": 427000 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "eval_loss": 0.7737278938293457, | |
| "eval_runtime": 0.5036, | |
| "eval_samples_per_second": 1985.649, | |
| "eval_steps_per_second": 31.77, | |
| "step": 427000 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 2.6352691180676286e-05, | |
| "loss": 0.3157, | |
| "step": 427500 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "learning_rate": 2.6132179047347505e-05, | |
| "loss": 0.3156, | |
| "step": 428000 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "eval_loss": 0.7748771905899048, | |
| "eval_runtime": 0.5178, | |
| "eval_samples_per_second": 1931.107, | |
| "eval_steps_per_second": 30.898, | |
| "step": 428000 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 2.5913076191862238e-05, | |
| "loss": 0.3156, | |
| "step": 428500 | |
| }, | |
| { | |
| "epoch": 13.14, | |
| "learning_rate": 2.5695385010294165e-05, | |
| "loss": 0.3156, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 13.14, | |
| "eval_loss": 0.7763614058494568, | |
| "eval_runtime": 0.5128, | |
| "eval_samples_per_second": 1949.92, | |
| "eval_steps_per_second": 31.199, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 2.5479107883279144e-05, | |
| "loss": 0.3155, | |
| "step": 429500 | |
| }, | |
| { | |
| "epoch": 13.17, | |
| "learning_rate": 2.5264247175989292e-05, | |
| "loss": 0.3151, | |
| "step": 430000 | |
| }, | |
| { | |
| "epoch": 13.17, | |
| "eval_loss": 0.7742259502410889, | |
| "eval_runtime": 0.5135, | |
| "eval_samples_per_second": 1947.536, | |
| "eval_steps_per_second": 31.161, | |
| "step": 430000 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 2.5050805238106804e-05, | |
| "loss": 0.3152, | |
| "step": 430500 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 2.4838784403798542e-05, | |
| "loss": 0.3154, | |
| "step": 431000 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "eval_loss": 0.7730093002319336, | |
| "eval_runtime": 0.5076, | |
| "eval_samples_per_second": 1969.931, | |
| "eval_steps_per_second": 31.519, | |
| "step": 431000 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 2.4628186991690346e-05, | |
| "loss": 0.3153, | |
| "step": 431500 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "learning_rate": 2.4419015304841797e-05, | |
| "loss": 0.3155, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "eval_loss": 0.7782933115959167, | |
| "eval_runtime": 0.5119, | |
| "eval_samples_per_second": 1953.36, | |
| "eval_steps_per_second": 31.254, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 2.4211271630720957e-05, | |
| "loss": 0.3151, | |
| "step": 432500 | |
| }, | |
| { | |
| "epoch": 13.27, | |
| "learning_rate": 2.4004958241179347e-05, | |
| "loss": 0.315, | |
| "step": 433000 | |
| }, | |
| { | |
| "epoch": 13.27, | |
| "eval_loss": 0.7781816124916077, | |
| "eval_runtime": 0.5016, | |
| "eval_samples_per_second": 1993.463, | |
| "eval_steps_per_second": 31.895, | |
| "step": 433000 | |
| }, | |
| { | |
| "epoch": 13.28, | |
| "learning_rate": 2.3800077392427193e-05, | |
| "loss": 0.3153, | |
| "step": 433500 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 2.3596631325008536e-05, | |
| "loss": 0.3152, | |
| "step": 434000 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "eval_loss": 0.7795186638832092, | |
| "eval_runtime": 0.5038, | |
| "eval_samples_per_second": 1985.024, | |
| "eval_steps_per_second": 31.76, | |
| "step": 434000 | |
| }, | |
| { | |
| "epoch": 13.31, | |
| "learning_rate": 2.3394622263777042e-05, | |
| "loss": 0.3154, | |
| "step": 434500 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "learning_rate": 2.3194052417871433e-05, | |
| "loss": 0.3155, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "eval_loss": 0.7748437523841858, | |
| "eval_runtime": 0.5107, | |
| "eval_samples_per_second": 1958.226, | |
| "eval_steps_per_second": 31.332, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 13.34, | |
| "learning_rate": 2.2994923980691425e-05, | |
| "loss": 0.315, | |
| "step": 435500 | |
| }, | |
| { | |
| "epoch": 13.36, | |
| "learning_rate": 2.279723912987365e-05, | |
| "loss": 0.3147, | |
| "step": 436000 | |
| }, | |
| { | |
| "epoch": 13.36, | |
| "eval_loss": 0.7737230062484741, | |
| "eval_runtime": 0.4938, | |
| "eval_samples_per_second": 2025.279, | |
| "eval_steps_per_second": 32.404, | |
| "step": 436000 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 2.2601000027268006e-05, | |
| "loss": 0.3149, | |
| "step": 436500 | |
| }, | |
| { | |
| "epoch": 13.39, | |
| "learning_rate": 2.2406208818913857e-05, | |
| "loss": 0.3148, | |
| "step": 437000 | |
| }, | |
| { | |
| "epoch": 13.39, | |
| "eval_loss": 0.7748478055000305, | |
| "eval_runtime": 0.5039, | |
| "eval_samples_per_second": 1984.516, | |
| "eval_steps_per_second": 31.752, | |
| "step": 437000 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 2.221286763501666e-05, | |
| "loss": 0.3151, | |
| "step": 437500 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "learning_rate": 2.2020978589924673e-05, | |
| "loss": 0.3148, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "eval_loss": 0.7769352793693542, | |
| "eval_runtime": 0.5123, | |
| "eval_samples_per_second": 1951.966, | |
| "eval_steps_per_second": 31.231, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 13.43, | |
| "learning_rate": 2.1830543782105647e-05, | |
| "loss": 0.3149, | |
| "step": 438500 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 2.1641565294124206e-05, | |
| "loss": 0.3146, | |
| "step": 439000 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "eval_loss": 0.7766078114509583, | |
| "eval_runtime": 0.5088, | |
| "eval_samples_per_second": 1965.524, | |
| "eval_steps_per_second": 31.448, | |
| "step": 439000 | |
| }, | |
| { | |
| "epoch": 13.47, | |
| "learning_rate": 2.1454045192618794e-05, | |
| "loss": 0.315, | |
| "step": 439500 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "learning_rate": 2.1267985528279212e-05, | |
| "loss": 0.3147, | |
| "step": 440000 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "eval_loss": 0.7768784165382385, | |
| "eval_runtime": 0.5177, | |
| "eval_samples_per_second": 1931.767, | |
| "eval_steps_per_second": 30.908, | |
| "step": 440000 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 2.1083388335824145e-05, | |
| "loss": 0.315, | |
| "step": 440500 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "learning_rate": 2.0900255633978873e-05, | |
| "loss": 0.3147, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "eval_loss": 0.7775884866714478, | |
| "eval_runtime": 0.5109, | |
| "eval_samples_per_second": 1957.402, | |
| "eval_steps_per_second": 31.318, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 13.53, | |
| "learning_rate": 2.0718589425453314e-05, | |
| "loss": 0.3145, | |
| "step": 441500 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "learning_rate": 2.0538391696920015e-05, | |
| "loss": 0.3145, | |
| "step": 442000 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "eval_loss": 0.7738910913467407, | |
| "eval_runtime": 0.5137, | |
| "eval_samples_per_second": 1946.745, | |
| "eval_steps_per_second": 31.148, | |
| "step": 442000 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 2.035966441899249e-05, | |
| "loss": 0.3148, | |
| "step": 442500 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 2.0182409546203555e-05, | |
| "loss": 0.3145, | |
| "step": 443000 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "eval_loss": 0.7752822041511536, | |
| "eval_runtime": 0.5041, | |
| "eval_samples_per_second": 1983.895, | |
| "eval_steps_per_second": 31.742, | |
| "step": 443000 | |
| }, | |
| { | |
| "epoch": 13.59, | |
| "learning_rate": 2.000662901698415e-05, | |
| "loss": 0.3143, | |
| "step": 443500 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 1.983232475364195e-05, | |
| "loss": 0.3145, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "eval_loss": 0.7770563960075378, | |
| "eval_runtime": 0.5102, | |
| "eval_samples_per_second": 1959.954, | |
| "eval_steps_per_second": 31.359, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 1.9659498662340474e-05, | |
| "loss": 0.3144, | |
| "step": 444500 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 1.948815263307819e-05, | |
| "loss": 0.3146, | |
| "step": 445000 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "eval_loss": 0.7755433917045593, | |
| "eval_runtime": 0.5167, | |
| "eval_samples_per_second": 1935.403, | |
| "eval_steps_per_second": 30.966, | |
| "step": 445000 | |
| }, | |
| { | |
| "epoch": 13.65, | |
| "learning_rate": 1.9318288539667765e-05, | |
| "loss": 0.3144, | |
| "step": 445500 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "learning_rate": 1.914990823971574e-05, | |
| "loss": 0.3144, | |
| "step": 446000 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "eval_loss": 0.7735174298286438, | |
| "eval_runtime": 0.5138, | |
| "eval_samples_per_second": 1946.362, | |
| "eval_steps_per_second": 31.142, | |
| "step": 446000 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 1.8983013574602096e-05, | |
| "loss": 0.314, | |
| "step": 446500 | |
| }, | |
| { | |
| "epoch": 13.7, | |
| "learning_rate": 1.8817606369460156e-05, | |
| "loss": 0.3143, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 13.7, | |
| "eval_loss": 0.7771323323249817, | |
| "eval_runtime": 0.5128, | |
| "eval_samples_per_second": 1950.249, | |
| "eval_steps_per_second": 31.204, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 13.71, | |
| "learning_rate": 1.865368843315663e-05, | |
| "loss": 0.3147, | |
| "step": 447500 | |
| }, | |
| { | |
| "epoch": 13.73, | |
| "learning_rate": 1.8491261558271762e-05, | |
| "loss": 0.314, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 13.73, | |
| "eval_loss": 0.7759497165679932, | |
| "eval_runtime": 0.5133, | |
| "eval_samples_per_second": 1948.013, | |
| "eval_steps_per_second": 31.168, | |
| "step": 448000 | |
| }, | |
| { | |
| "epoch": 13.74, | |
| "learning_rate": 1.833032752107986e-05, | |
| "loss": 0.3143, | |
| "step": 448500 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 1.817088808152978e-05, | |
| "loss": 0.3143, | |
| "step": 449000 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "eval_loss": 0.7774102687835693, | |
| "eval_runtime": 0.5028, | |
| "eval_samples_per_second": 1988.858, | |
| "eval_steps_per_second": 31.822, | |
| "step": 449000 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "learning_rate": 1.801294498322569e-05, | |
| "loss": 0.3141, | |
| "step": 449500 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 1.7856499953407978e-05, | |
| "loss": 0.3142, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "eval_loss": 0.7757880687713623, | |
| "eval_runtime": 0.5025, | |
| "eval_samples_per_second": 1989.909, | |
| "eval_steps_per_second": 31.839, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 13.8, | |
| "learning_rate": 1.770155470293445e-05, | |
| "loss": 0.3141, | |
| "step": 450500 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 1.7548110926261522e-05, | |
| "loss": 0.3144, | |
| "step": 451000 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "eval_loss": 0.7752067446708679, | |
| "eval_runtime": 0.5146, | |
| "eval_samples_per_second": 1943.251, | |
| "eval_steps_per_second": 31.092, | |
| "step": 451000 | |
| }, | |
| { | |
| "epoch": 13.83, | |
| "learning_rate": 1.7396170301425777e-05, | |
| "loss": 0.3141, | |
| "step": 451500 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 1.7245734490025544e-05, | |
| "loss": 0.3142, | |
| "step": 452000 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "eval_loss": 0.7712000012397766, | |
| "eval_runtime": 0.5015, | |
| "eval_samples_per_second": 1994.171, | |
| "eval_steps_per_second": 31.907, | |
| "step": 452000 | |
| }, | |
| { | |
| "epoch": 13.86, | |
| "learning_rate": 1.7096805137202738e-05, | |
| "loss": 0.3139, | |
| "step": 452500 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 1.6949383871624917e-05, | |
| "loss": 0.3139, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "eval_loss": 0.7725095152854919, | |
| "eval_runtime": 0.5022, | |
| "eval_samples_per_second": 1991.07, | |
| "eval_steps_per_second": 31.857, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 13.89, | |
| "learning_rate": 1.6803472305467368e-05, | |
| "loss": 0.3137, | |
| "step": 453500 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 1.665907203439568e-05, | |
| "loss": 0.3139, | |
| "step": 454000 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "eval_loss": 0.7726877927780151, | |
| "eval_runtime": 0.5045, | |
| "eval_samples_per_second": 1982.357, | |
| "eval_steps_per_second": 31.718, | |
| "step": 454000 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 1.6516184637548058e-05, | |
| "loss": 0.3138, | |
| "step": 454500 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 1.6374811677518142e-05, | |
| "loss": 0.314, | |
| "step": 455000 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "eval_loss": 0.7759785056114197, | |
| "eval_runtime": 0.5189, | |
| "eval_samples_per_second": 1927.23, | |
| "eval_steps_per_second": 30.836, | |
| "step": 455000 | |
| }, | |
| { | |
| "epoch": 13.96, | |
| "learning_rate": 1.6234954700338025e-05, | |
| "loss": 0.3135, | |
| "step": 455500 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "learning_rate": 1.6096615235461148e-05, | |
| "loss": 0.3135, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "eval_loss": 0.773828387260437, | |
| "eval_runtime": 0.5012, | |
| "eval_samples_per_second": 1995.149, | |
| "eval_steps_per_second": 31.922, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "learning_rate": 1.59597947957458e-05, | |
| "loss": 0.3138, | |
| "step": 456500 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 1.5824494877438344e-05, | |
| "loss": 0.3137, | |
| "step": 457000 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_loss": 0.7719975709915161, | |
| "eval_runtime": 0.5107, | |
| "eval_samples_per_second": 1958.237, | |
| "eval_steps_per_second": 31.332, | |
| "step": 457000 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 1.569071696015702e-05, | |
| "loss": 0.3135, | |
| "step": 457500 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 1.555846250687569e-05, | |
| "loss": 0.3138, | |
| "step": 458000 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "eval_loss": 0.7755674123764038, | |
| "eval_runtime": 0.5013, | |
| "eval_samples_per_second": 1994.642, | |
| "eval_steps_per_second": 31.914, | |
| "step": 458000 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 1.542773296390789e-05, | |
| "loss": 0.3134, | |
| "step": 458500 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "learning_rate": 1.5298529760890945e-05, | |
| "loss": 0.3135, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "eval_loss": 0.7749778032302856, | |
| "eval_runtime": 0.5253, | |
| "eval_samples_per_second": 1903.509, | |
| "eval_steps_per_second": 30.456, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 14.08, | |
| "learning_rate": 1.5170854310770376e-05, | |
| "loss": 0.3136, | |
| "step": 459500 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 1.5044708009784457e-05, | |
| "loss": 0.3134, | |
| "step": 460000 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "eval_loss": 0.7761635184288025, | |
| "eval_runtime": 0.5003, | |
| "eval_samples_per_second": 1998.696, | |
| "eval_steps_per_second": 31.979, | |
| "step": 460000 | |
| }, | |
| { | |
| "epoch": 14.11, | |
| "learning_rate": 1.4920092237448903e-05, | |
| "loss": 0.3138, | |
| "step": 460500 | |
| }, | |
| { | |
| "epoch": 14.12, | |
| "learning_rate": 1.4797008356541874e-05, | |
| "loss": 0.3134, | |
| "step": 461000 | |
| }, | |
| { | |
| "epoch": 14.12, | |
| "eval_loss": 0.7742542028427124, | |
| "eval_runtime": 0.4997, | |
| "eval_samples_per_second": 2001.196, | |
| "eval_steps_per_second": 32.019, | |
| "step": 461000 | |
| }, | |
| { | |
| "epoch": 14.14, | |
| "learning_rate": 1.4675457713088947e-05, | |
| "loss": 0.3134, | |
| "step": 461500 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 1.4555441636348494e-05, | |
| "loss": 0.3136, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "eval_loss": 0.7772946357727051, | |
| "eval_runtime": 0.5121, | |
| "eval_samples_per_second": 1952.647, | |
| "eval_steps_per_second": 31.242, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 14.17, | |
| "learning_rate": 1.4436961438797095e-05, | |
| "loss": 0.3136, | |
| "step": 462500 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 1.4320018416115206e-05, | |
| "loss": 0.3132, | |
| "step": 463000 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "eval_loss": 0.7742455005645752, | |
| "eval_runtime": 0.5004, | |
| "eval_samples_per_second": 1998.244, | |
| "eval_steps_per_second": 31.972, | |
| "step": 463000 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 1.4204613847173003e-05, | |
| "loss": 0.3134, | |
| "step": 463500 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "learning_rate": 1.4090748994016354e-05, | |
| "loss": 0.3135, | |
| "step": 464000 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "eval_loss": 0.7742483019828796, | |
| "eval_runtime": 0.5003, | |
| "eval_samples_per_second": 1998.899, | |
| "eval_steps_per_second": 31.982, | |
| "step": 464000 | |
| }, | |
| { | |
| "epoch": 14.23, | |
| "learning_rate": 1.3978425101853049e-05, | |
| "loss": 0.3132, | |
| "step": 464500 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "learning_rate": 1.3867643399039165e-05, | |
| "loss": 0.3134, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "eval_loss": 0.7741165161132812, | |
| "eval_runtime": 0.5054, | |
| "eval_samples_per_second": 1978.738, | |
| "eval_steps_per_second": 31.66, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 1.3758405097065648e-05, | |
| "loss": 0.3128, | |
| "step": 465500 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "learning_rate": 1.3650711390545131e-05, | |
| "loss": 0.313, | |
| "step": 466000 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "eval_loss": 0.7737331986427307, | |
| "eval_runtime": 0.505, | |
| "eval_samples_per_second": 1980.27, | |
| "eval_steps_per_second": 31.684, | |
| "step": 466000 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 1.3544563457198657e-05, | |
| "loss": 0.3132, | |
| "step": 466500 | |
| }, | |
| { | |
| "epoch": 14.31, | |
| "learning_rate": 1.343996245784307e-05, | |
| "loss": 0.3131, | |
| "step": 467000 | |
| }, | |
| { | |
| "epoch": 14.31, | |
| "eval_loss": 0.7716790437698364, | |
| "eval_runtime": 0.504, | |
| "eval_samples_per_second": 1983.936, | |
| "eval_steps_per_second": 31.743, | |
| "step": 467000 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 1.3336909536378107e-05, | |
| "loss": 0.313, | |
| "step": 467500 | |
| }, | |
| { | |
| "epoch": 14.34, | |
| "learning_rate": 1.3235405819774022e-05, | |
| "loss": 0.3131, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 14.34, | |
| "eval_loss": 0.7737191319465637, | |
| "eval_runtime": 0.5136, | |
| "eval_samples_per_second": 1947.194, | |
| "eval_steps_per_second": 31.155, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 1.3135452418059208e-05, | |
| "loss": 0.3132, | |
| "step": 468500 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 1.3037050424308027e-05, | |
| "loss": 0.3126, | |
| "step": 469000 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "eval_loss": 0.7744223475456238, | |
| "eval_runtime": 0.5036, | |
| "eval_samples_per_second": 1985.727, | |
| "eval_steps_per_second": 31.772, | |
| "step": 469000 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 1.2940200914628945e-05, | |
| "loss": 0.3132, | |
| "step": 469500 | |
| }, | |
| { | |
| "epoch": 14.4, | |
| "learning_rate": 1.2844904948152644e-05, | |
| "loss": 0.3132, | |
| "step": 470000 | |
| }, | |
| { | |
| "epoch": 14.4, | |
| "eval_loss": 0.7742005586624146, | |
| "eval_runtime": 0.5167, | |
| "eval_samples_per_second": 1935.302, | |
| "eval_steps_per_second": 30.965, | |
| "step": 470000 | |
| }, | |
| { | |
| "epoch": 14.42, | |
| "learning_rate": 1.2751163567020592e-05, | |
| "loss": 0.3129, | |
| "step": 470500 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 1.2658977796373478e-05, | |
| "loss": 0.313, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "eval_loss": 0.7727963924407959, | |
| "eval_runtime": 0.508, | |
| "eval_samples_per_second": 1968.327, | |
| "eval_steps_per_second": 31.493, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 14.45, | |
| "learning_rate": 1.2568348644340153e-05, | |
| "loss": 0.3132, | |
| "step": 471500 | |
| }, | |
| { | |
| "epoch": 14.46, | |
| "learning_rate": 1.2479277102026465e-05, | |
| "loss": 0.3131, | |
| "step": 472000 | |
| }, | |
| { | |
| "epoch": 14.46, | |
| "eval_loss": 0.7726814150810242, | |
| "eval_runtime": 0.5186, | |
| "eval_samples_per_second": 1928.247, | |
| "eval_steps_per_second": 30.852, | |
| "step": 472000 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 1.2391764143504556e-05, | |
| "loss": 0.313, | |
| "step": 472500 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "learning_rate": 1.2305810725802118e-05, | |
| "loss": 0.3127, | |
| "step": 473000 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "eval_loss": 0.7736928462982178, | |
| "eval_runtime": 0.5222, | |
| "eval_samples_per_second": 1915.078, | |
| "eval_steps_per_second": 30.641, | |
| "step": 473000 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 1.222141778889195e-05, | |
| "loss": 0.3129, | |
| "step": 473500 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "learning_rate": 1.2138586255681707e-05, | |
| "loss": 0.3128, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "eval_loss": 0.7748116254806519, | |
| "eval_runtime": 0.5115, | |
| "eval_samples_per_second": 1955.159, | |
| "eval_steps_per_second": 31.283, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 1.2057317032003731e-05, | |
| "loss": 0.3128, | |
| "step": 474500 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "learning_rate": 1.1977611006605263e-05, | |
| "loss": 0.3127, | |
| "step": 475000 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "eval_loss": 0.7728310227394104, | |
| "eval_runtime": 0.5087, | |
| "eval_samples_per_second": 1965.811, | |
| "eval_steps_per_second": 31.453, | |
| "step": 475000 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 1.1899469051138602e-05, | |
| "loss": 0.3128, | |
| "step": 475500 | |
| }, | |
| { | |
| "epoch": 14.58, | |
| "learning_rate": 1.1822892020151667e-05, | |
| "loss": 0.3127, | |
| "step": 476000 | |
| }, | |
| { | |
| "epoch": 14.58, | |
| "eval_loss": 0.7735549211502075, | |
| "eval_runtime": 0.512, | |
| "eval_samples_per_second": 1953.213, | |
| "eval_steps_per_second": 31.251, | |
| "step": 476000 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 1.1747880751078614e-05, | |
| "loss": 0.313, | |
| "step": 476500 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "learning_rate": 1.1674436064230637e-05, | |
| "loss": 0.3127, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "eval_loss": 0.7763468623161316, | |
| "eval_runtime": 0.4992, | |
| "eval_samples_per_second": 2003.099, | |
| "eval_steps_per_second": 32.05, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 1.1602558762787069e-05, | |
| "loss": 0.3126, | |
| "step": 477500 | |
| }, | |
| { | |
| "epoch": 14.65, | |
| "learning_rate": 1.1532249632786582e-05, | |
| "loss": 0.3127, | |
| "step": 478000 | |
| }, | |
| { | |
| "epoch": 14.65, | |
| "eval_loss": 0.7772842049598694, | |
| "eval_runtime": 0.503, | |
| "eval_samples_per_second": 1988.055, | |
| "eval_steps_per_second": 31.809, | |
| "step": 478000 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 1.1463509443118552e-05, | |
| "loss": 0.3126, | |
| "step": 478500 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 1.1396338945514663e-05, | |
| "loss": 0.3127, | |
| "step": 479000 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "eval_loss": 0.7746040225028992, | |
| "eval_runtime": 0.5177, | |
| "eval_samples_per_second": 1931.572, | |
| "eval_steps_per_second": 30.905, | |
| "step": 479000 | |
| }, | |
| { | |
| "epoch": 14.69, | |
| "learning_rate": 1.133073887454072e-05, | |
| "loss": 0.3129, | |
| "step": 479500 | |
| }, | |
| { | |
| "epoch": 14.71, | |
| "learning_rate": 1.1266709947588599e-05, | |
| "loss": 0.3126, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 14.71, | |
| "eval_loss": 0.7748845815658569, | |
| "eval_runtime": 0.5138, | |
| "eval_samples_per_second": 1946.296, | |
| "eval_steps_per_second": 31.141, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 14.72, | |
| "learning_rate": 1.1204252864868377e-05, | |
| "loss": 0.3127, | |
| "step": 480500 | |
| }, | |
| { | |
| "epoch": 14.74, | |
| "learning_rate": 1.1143368309400725e-05, | |
| "loss": 0.3125, | |
| "step": 481000 | |
| }, | |
| { | |
| "epoch": 14.74, | |
| "eval_loss": 0.7771226763725281, | |
| "eval_runtime": 0.5065, | |
| "eval_samples_per_second": 1974.24, | |
| "eval_steps_per_second": 31.588, | |
| "step": 481000 | |
| }, | |
| { | |
| "epoch": 14.75, | |
| "learning_rate": 1.1084056947009348e-05, | |
| "loss": 0.3125, | |
| "step": 481500 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 1.1026319426313837e-05, | |
| "loss": 0.3124, | |
| "step": 482000 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "eval_loss": 0.7745999693870544, | |
| "eval_runtime": 0.4972, | |
| "eval_samples_per_second": 2011.082, | |
| "eval_steps_per_second": 32.177, | |
| "step": 482000 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 1.097015637872247e-05, | |
| "loss": 0.3126, | |
| "step": 482500 | |
| }, | |
| { | |
| "epoch": 14.8, | |
| "learning_rate": 1.0915568418425301e-05, | |
| "loss": 0.3125, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 14.8, | |
| "eval_loss": 0.7761328220367432, | |
| "eval_runtime": 0.5155, | |
| "eval_samples_per_second": 1939.967, | |
| "eval_steps_per_second": 31.039, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 14.81, | |
| "learning_rate": 1.0862556142387571e-05, | |
| "loss": 0.3126, | |
| "step": 483500 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 1.081112013034298e-05, | |
| "loss": 0.3127, | |
| "step": 484000 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "eval_loss": 0.7775337100028992, | |
| "eval_runtime": 0.5093, | |
| "eval_samples_per_second": 1963.507, | |
| "eval_steps_per_second": 31.416, | |
| "step": 484000 | |
| }, | |
| { | |
| "epoch": 14.84, | |
| "learning_rate": 1.0761260944787561e-05, | |
| "loss": 0.313, | |
| "step": 484500 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 1.0712979130973347e-05, | |
| "loss": 0.3126, | |
| "step": 485000 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "eval_loss": 0.7740907073020935, | |
| "eval_runtime": 0.5188, | |
| "eval_samples_per_second": 1927.581, | |
| "eval_steps_per_second": 30.841, | |
| "step": 485000 | |
| }, | |
| { | |
| "epoch": 14.87, | |
| "learning_rate": 1.0666275216902535e-05, | |
| "loss": 0.3125, | |
| "step": 485500 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 1.0621149713321656e-05, | |
| "loss": 0.3128, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "eval_loss": 0.7765258550643921, | |
| "eval_runtime": 0.52, | |
| "eval_samples_per_second": 1923.021, | |
| "eval_steps_per_second": 30.768, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 1.0577603113715964e-05, | |
| "loss": 0.3125, | |
| "step": 486500 | |
| }, | |
| { | |
| "epoch": 14.92, | |
| "learning_rate": 1.0535635894304106e-05, | |
| "loss": 0.3126, | |
| "step": 487000 | |
| }, | |
| { | |
| "epoch": 14.92, | |
| "eval_loss": 0.7741805911064148, | |
| "eval_runtime": 0.5052, | |
| "eval_samples_per_second": 1979.424, | |
| "eval_steps_per_second": 31.671, | |
| "step": 487000 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 1.0495248514032875e-05, | |
| "loss": 0.3123, | |
| "step": 487500 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 1.045644141457218e-05, | |
| "loss": 0.3126, | |
| "step": 488000 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "eval_loss": 0.7744404077529907, | |
| "eval_runtime": 0.5273, | |
| "eval_samples_per_second": 1896.451, | |
| "eval_steps_per_second": 30.343, | |
| "step": 488000 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "learning_rate": 1.0419215020310254e-05, | |
| "loss": 0.3123, | |
| "step": 488500 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "learning_rate": 1.0383569738348988e-05, | |
| "loss": 0.3125, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "eval_loss": 0.7724484205245972, | |
| "eval_runtime": 0.5078, | |
| "eval_samples_per_second": 1969.112, | |
| "eval_steps_per_second": 31.506, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 1.0349505958499436e-05, | |
| "loss": 0.3126, | |
| "step": 489500 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 1.0317024053277693e-05, | |
| "loss": 0.3124, | |
| "step": 490000 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "eval_loss": 0.7753793001174927, | |
| "eval_runtime": 0.498, | |
| "eval_samples_per_second": 2008.198, | |
| "eval_steps_per_second": 32.131, | |
| "step": 490000 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 1.0286124377900624e-05, | |
| "loss": 0.3123, | |
| "step": 490500 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 1.0256807270282153e-05, | |
| "loss": 0.3122, | |
| "step": 491000 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "eval_loss": 0.7738854885101318, | |
| "eval_runtime": 0.4992, | |
| "eval_samples_per_second": 2003.188, | |
| "eval_steps_per_second": 32.051, | |
| "step": 491000 | |
| }, | |
| { | |
| "epoch": 15.06, | |
| "learning_rate": 1.0229073051029455e-05, | |
| "loss": 0.3124, | |
| "step": 491500 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 1.020292202343952e-05, | |
| "loss": 0.3125, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "eval_loss": 0.7788605690002441, | |
| "eval_runtime": 0.5071, | |
| "eval_samples_per_second": 1971.996, | |
| "eval_steps_per_second": 31.552, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 15.09, | |
| "learning_rate": 1.0178354473495813e-05, | |
| "loss": 0.3126, | |
| "step": 492500 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 1.0155370669865077e-05, | |
| "loss": 0.3121, | |
| "step": 493000 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "eval_loss": 0.7763463854789734, | |
| "eval_runtime": 0.4971, | |
| "eval_samples_per_second": 2011.654, | |
| "eval_steps_per_second": 32.186, | |
| "step": 493000 | |
| }, | |
| { | |
| "epoch": 15.12, | |
| "learning_rate": 1.0133970863894557e-05, | |
| "loss": 0.3121, | |
| "step": 493500 | |
| }, | |
| { | |
| "epoch": 15.14, | |
| "learning_rate": 1.0114155289609061e-05, | |
| "loss": 0.3121, | |
| "step": 494000 | |
| }, | |
| { | |
| "epoch": 15.14, | |
| "eval_loss": 0.7747126221656799, | |
| "eval_runtime": 0.5031, | |
| "eval_samples_per_second": 1987.677, | |
| "eval_steps_per_second": 31.803, | |
| "step": 494000 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 1.0095924163708572e-05, | |
| "loss": 0.3124, | |
| "step": 494500 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 1.0079277685565724e-05, | |
| "loss": 0.3124, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "eval_loss": 0.7728501558303833, | |
| "eval_runtime": 0.5097, | |
| "eval_samples_per_second": 1962.024, | |
| "eval_steps_per_second": 31.392, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 15.18, | |
| "learning_rate": 1.0064216037223772e-05, | |
| "loss": 0.3126, | |
| "step": 495500 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 1.0050739383394454e-05, | |
| "loss": 0.3124, | |
| "step": 496000 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "eval_loss": 0.7736043930053711, | |
| "eval_runtime": 0.504, | |
| "eval_samples_per_second": 1984.023, | |
| "eval_steps_per_second": 31.744, | |
| "step": 496000 | |
| }, | |
| { | |
| "epoch": 15.21, | |
| "learning_rate": 1.003884787145633e-05, | |
| "loss": 0.3123, | |
| "step": 496500 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 1.002854163145305e-05, | |
| "loss": 0.3125, | |
| "step": 497000 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "eval_loss": 0.7738491892814636, | |
| "eval_runtime": 0.5171, | |
| "eval_samples_per_second": 1933.887, | |
| "eval_steps_per_second": 30.942, | |
| "step": 497000 | |
| }, | |
| { | |
| "epoch": 15.24, | |
| "learning_rate": 1.0019820776091995e-05, | |
| "loss": 0.3123, | |
| "step": 497500 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 1.0012685400743077e-05, | |
| "loss": 0.3123, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "eval_loss": 0.7730560898780823, | |
| "eval_runtime": 0.5124, | |
| "eval_samples_per_second": 1951.657, | |
| "eval_steps_per_second": 31.227, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 15.27, | |
| "learning_rate": 1.0007135583437572e-05, | |
| "loss": 0.3126, | |
| "step": 498500 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 1.0003171384867436e-05, | |
| "loss": 0.3123, | |
| "step": 499000 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "eval_loss": 0.7752643823623657, | |
| "eval_runtime": 0.5071, | |
| "eval_samples_per_second": 1972.007, | |
| "eval_steps_per_second": 31.552, | |
| "step": 499000 | |
| }, | |
| { | |
| "epoch": 15.3, | |
| "learning_rate": 1.0000792848384467e-05, | |
| "loss": 0.3121, | |
| "step": 499500 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3122, | |
| "step": 500000 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "eval_loss": 0.7746031284332275, | |
| "eval_runtime": 0.5134, | |
| "eval_samples_per_second": 1947.857, | |
| "eval_steps_per_second": 31.166, | |
| "step": 500000 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 16, | |
| "total_flos": 1.5974237119445362e+22, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |