| { | |
| "best_metric": 5.6603827476501465, | |
| "best_model_checkpoint": "../checkpoints/nf-bart-newsroom-rqnsf/checkpoint-60000", | |
| "epoch": 0.9648008490247472, | |
| "global_step": 60000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "gate_score": 0.371, | |
| "learning_rate": 4.9725e-05, | |
| "loss": 592.8312, | |
| "nf_loss": 591.7519, | |
| "ppl": 3.3519, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "gate_score": 0.2007, | |
| "learning_rate": 4.834853544667631e-05, | |
| "loss": 333.2925, | |
| "nf_loss": 332.3035, | |
| "ppl": 2.8873, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 252.84817504882812, | |
| "eval_nf_loss": 251.2677459716797, | |
| "eval_perplexity": 6.363341331481934, | |
| "eval_runtime": 4381.8372, | |
| "eval_samples_per_second": 24.837, | |
| "eval_steps_per_second": 1.552, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "gate_score": 0.1135, | |
| "learning_rate": 4.6687102294439184e-05, | |
| "loss": 211.0105, | |
| "nf_loss": 209.9559, | |
| "ppl": 3.1158, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "gate_score": 0.0666, | |
| "learning_rate": 4.5025669142202066e-05, | |
| "loss": 134.1864, | |
| "nf_loss": 133.0995, | |
| "ppl": 3.2307, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "eval_loss": 102.85702514648438, | |
| "eval_nf_loss": 101.2651596069336, | |
| "eval_perplexity": 6.448662757873535, | |
| "eval_runtime": 4005.1516, | |
| "eval_samples_per_second": 27.173, | |
| "eval_steps_per_second": 1.698, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "gate_score": 0.0352, | |
| "learning_rate": 4.3365897423117185e-05, | |
| "loss": 95.272, | |
| "nf_loss": 94.1643, | |
| "ppl": 3.2978, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "gate_score": 0.023, | |
| "learning_rate": 4.170446427088006e-05, | |
| "loss": 68.625, | |
| "nf_loss": 67.5129, | |
| "ppl": 3.324, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "eval_loss": 53.623291015625, | |
| "eval_nf_loss": 52.02442169189453, | |
| "eval_perplexity": 6.536867618560791, | |
| "eval_runtime": 3924.6092, | |
| "eval_samples_per_second": 27.731, | |
| "eval_steps_per_second": 1.733, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "gate_score": 0.023, | |
| "learning_rate": 4.004386183521906e-05, | |
| "loss": 49.4308, | |
| "nf_loss": 48.3272, | |
| "ppl": 3.2964, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "gate_score": 0.0219, | |
| "learning_rate": 3.8383259399558055e-05, | |
| "loss": 36.663, | |
| "nf_loss": 35.5419, | |
| "ppl": 3.3952, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 29.72612762451172, | |
| "eval_nf_loss": 28.141469955444336, | |
| "eval_perplexity": 6.444383144378662, | |
| "eval_runtime": 3904.4886, | |
| "eval_samples_per_second": 27.874, | |
| "eval_steps_per_second": 1.742, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "gate_score": 0.0183, | |
| "learning_rate": 3.672265696389706e-05, | |
| "loss": 30.1161, | |
| "nf_loss": 29.0084, | |
| "ppl": 3.3204, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "gate_score": 0.0188, | |
| "learning_rate": 3.5061223811659935e-05, | |
| "loss": 25.2734, | |
| "nf_loss": 24.1661, | |
| "ppl": 3.3019, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 21.10334014892578, | |
| "eval_nf_loss": 19.523048400878906, | |
| "eval_perplexity": 6.385104656219482, | |
| "eval_runtime": 3951.5734, | |
| "eval_samples_per_second": 27.541, | |
| "eval_steps_per_second": 1.721, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "gate_score": 0.0189, | |
| "learning_rate": 3.340062137599894e-05, | |
| "loss": 21.8002, | |
| "nf_loss": 20.6893, | |
| "ppl": 3.3211, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "gate_score": 0.0186, | |
| "learning_rate": 3.1740018940337936e-05, | |
| "loss": 19.5679, | |
| "nf_loss": 18.4664, | |
| "ppl": 3.2967, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_loss": 16.757369995117188, | |
| "eval_nf_loss": 15.157673835754395, | |
| "eval_perplexity": 6.50905179977417, | |
| "eval_runtime": 4060.1325, | |
| "eval_samples_per_second": 26.805, | |
| "eval_steps_per_second": 1.675, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "gate_score": 0.0169, | |
| "learning_rate": 3.0079416504676937e-05, | |
| "loss": 17.5505, | |
| "nf_loss": 16.4471, | |
| "ppl": 3.2866, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "gate_score": 0.0169, | |
| "learning_rate": 2.8417983352439813e-05, | |
| "loss": 16.0243, | |
| "nf_loss": 14.9265, | |
| "ppl": 3.2771, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_loss": 13.916472434997559, | |
| "eval_nf_loss": 12.336326599121094, | |
| "eval_perplexity": 6.333822250366211, | |
| "eval_runtime": 4073.3885, | |
| "eval_samples_per_second": 26.718, | |
| "eval_steps_per_second": 1.67, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "gate_score": 0.0164, | |
| "learning_rate": 2.6757380916778813e-05, | |
| "loss": 14.7704, | |
| "nf_loss": 13.6695, | |
| "ppl": 3.2909, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "gate_score": 0.0165, | |
| "learning_rate": 2.509677848111781e-05, | |
| "loss": 13.8487, | |
| "nf_loss": 12.7566, | |
| "ppl": 3.2604, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "eval_loss": 12.310139656066895, | |
| "eval_nf_loss": 10.705780029296875, | |
| "eval_perplexity": 6.516045570373535, | |
| "eval_runtime": 3950.6921, | |
| "eval_samples_per_second": 27.548, | |
| "eval_steps_per_second": 1.722, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "gate_score": 0.0163, | |
| "learning_rate": 2.3435345328880693e-05, | |
| "loss": 13.021, | |
| "nf_loss": 11.939, | |
| "ppl": 3.2217, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "gate_score": 0.0164, | |
| "learning_rate": 2.1774742893219694e-05, | |
| "loss": 12.3656, | |
| "nf_loss": 11.2731, | |
| "ppl": 3.2491, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_loss": 11.076099395751953, | |
| "eval_nf_loss": 9.518845558166504, | |
| "eval_perplexity": 6.16482400894165, | |
| "eval_runtime": 3933.1054, | |
| "eval_samples_per_second": 27.671, | |
| "eval_steps_per_second": 1.729, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "gate_score": 0.0169, | |
| "learning_rate": 2.0113309740982573e-05, | |
| "loss": 11.834, | |
| "nf_loss": 10.7422, | |
| "ppl": 3.248, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "gate_score": 0.017, | |
| "learning_rate": 1.845270730532157e-05, | |
| "loss": 11.301, | |
| "nf_loss": 10.2211, | |
| "ppl": 3.2176, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_loss": 9.942972183227539, | |
| "eval_nf_loss": 8.422438621520996, | |
| "eval_perplexity": 5.888382911682129, | |
| "eval_runtime": 3905.9353, | |
| "eval_samples_per_second": 27.863, | |
| "eval_steps_per_second": 1.741, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "gate_score": 0.0178, | |
| "learning_rate": 1.679293558623669e-05, | |
| "loss": 10.826, | |
| "nf_loss": 9.7634, | |
| "ppl": 3.1432, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "gate_score": 0.0167, | |
| "learning_rate": 1.513150243399957e-05, | |
| "loss": 10.5722, | |
| "nf_loss": 9.4924, | |
| "ppl": 3.2112, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_loss": 9.34610366821289, | |
| "eval_nf_loss": 7.819733619689941, | |
| "eval_perplexity": 5.929094314575195, | |
| "eval_runtime": 3809.9667, | |
| "eval_samples_per_second": 28.565, | |
| "eval_steps_per_second": 1.785, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "gate_score": 0.0164, | |
| "learning_rate": 1.3470899998338569e-05, | |
| "loss": 10.1982, | |
| "nf_loss": 9.1308, | |
| "ppl": 3.1538, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "gate_score": 0.0156, | |
| "learning_rate": 1.1810297562677566e-05, | |
| "loss": 9.9637, | |
| "nf_loss": 8.9113, | |
| "ppl": 3.1019, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_loss": 9.272148132324219, | |
| "eval_nf_loss": 7.770688056945801, | |
| "eval_perplexity": 5.756178855895996, | |
| "eval_runtime": 3799.2449, | |
| "eval_samples_per_second": 28.646, | |
| "eval_steps_per_second": 1.79, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "gate_score": 0.0159, | |
| "learning_rate": 1.0148864410440447e-05, | |
| "loss": 9.7627, | |
| "nf_loss": 8.701, | |
| "ppl": 3.1339, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "gate_score": 0.015, | |
| "learning_rate": 8.488261974779444e-06, | |
| "loss": 9.5228, | |
| "nf_loss": 8.4605, | |
| "ppl": 3.1399, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_loss": 8.568826675415039, | |
| "eval_nf_loss": 7.074533462524414, | |
| "eval_perplexity": 5.70849609375, | |
| "eval_runtime": 3805.5723, | |
| "eval_samples_per_second": 28.598, | |
| "eval_steps_per_second": 1.787, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "gate_score": 0.0143, | |
| "learning_rate": 6.827659539118445e-06, | |
| "loss": 9.3407, | |
| "nf_loss": 8.2955, | |
| "ppl": 3.0762, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "gate_score": 0.0148, | |
| "learning_rate": 5.167057103457443e-06, | |
| "loss": 9.1988, | |
| "nf_loss": 8.1602, | |
| "ppl": 3.058, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 8.280665397644043, | |
| "eval_nf_loss": 6.778278827667236, | |
| "eval_perplexity": 5.754031658172607, | |
| "eval_runtime": 3795.1994, | |
| "eval_samples_per_second": 28.676, | |
| "eval_steps_per_second": 1.792, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "gate_score": 0.0146, | |
| "learning_rate": 3.5056239512203227e-06, | |
| "loss": 9.0572, | |
| "nf_loss": 8.0245, | |
| "ppl": 3.0413, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "gate_score": 0.0138, | |
| "learning_rate": 1.8450215155593216e-06, | |
| "loss": 9.0079, | |
| "nf_loss": 7.9878, | |
| "ppl": 3.002, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_loss": 8.222879409790039, | |
| "eval_nf_loss": 6.735788822174072, | |
| "eval_perplexity": 5.6603827476501465, | |
| "eval_runtime": 3672.6524, | |
| "eval_samples_per_second": 29.633, | |
| "eval_steps_per_second": 1.852, | |
| "step": 60000 | |
| } | |
| ], | |
| "max_steps": 62189, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.06597659967488e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |