{ "checkpoint_path": "/data_2to/devel_data/nn_pruning/output/squad_test_final_fine_tune/fine_tuned_aws_nn-pruning-v10-a32-l5-dl0-5--2021-01-21--00-52-45/checkpoint-22132", "config": { "_name_or_path": "/tmp/tmp3c96kd08", "architectures": ["BertForQuestionAnswering"], "attention_probs_dropout_prob": 0.1, "gradient_checkpointing": false, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "layer_norm_eps": 1e-12, "max_position_embeddings": 512, "model_type": "bert", "num_attention_heads": 12, "num_hidden_layers": 12, "pad_token_id": 0, "position_embedding_type": "absolute", "pruned_heads": { "0": [2, 4, 5, 6, 7], "1": [0, 2, 3, 5, 6, 7, 8], "10": [1, 4, 5, 6, 7], "11": [0, 2, 5, 6, 7, 8, 11], "2": [4, 7, 8], "3": [2, 4, 6], "4": [2], "5": [1, 2], "6": [2, 3, 7], "7": [3, 6, 7, 11], "8": [0, 4], "9": [1, 4, 5, 7, 9, 10] }, "type_vocab_size": 2, "vocab_size": 30522 }, "eval_metrics": { "exact_match": 81.69347209082308, "f1": 88.72194531479171, "main_metric": 88.72194531479171 }, "model_args": { "cache_dir": null, "config_name": null, "model_name_or_path": "/home/lagunas/devel/hf/nn_pruning/nn_pruning/analysis/tmp_finetune", "tokenizer_name": null }, "source_checkpoint": "/data_2to/devel_data/nn_pruning/output/squad_test_aws/aws_nn-pruning-v10-a32-l5-dl0-5--2021-01-21--00-52-45/checkpoint-95000", "sparse_args": { "ampere_pruning_method": "disabled", "attention_block_cols": 1, "attention_block_rows": 1, "attention_lambda": 1.0, "attention_output_with_dense": 0, "attention_pruning_method": "topK", "bias_mask": true, "dense_block_cols": 1, "dense_block_rows": 1, "dense_lambda": 1.0, "dense_pruning_method": "topK", "distil_alpha_ce": 0.1, "distil_alpha_teacher": 0.9, "distil_teacher_name_or_path": "csarron/bert-base-uncased-squad-v1", "distil_temperature": 2.0, "final_ampere_temperature": 20.0, "final_finetune": 1, "final_threshold": 0.5, "final_warmup": 0, "initial_ampere_temperature": 0.0, "initial_threshold": 1.0, "initial_warmup": 0, "mask_init": "constant", "mask_scale": 0.0, "mask_scores_learning_rate": 0.01, "regularization": "", "regularization_final_lambda": 0 }, "speed": { "cuda_eval_elapsed_time": 20.951393741607667, "eval_elapsed_time": 28.213609586004168 }, "speedup": 1.8420919143305463, "stats": { "layers": { "0": { "linear_attention_nnz": 1376256, "linear_attention_total": 2359296, "linear_dense_nnz": 1125888, "linear_dense_total": 4718592, "linear_nnz": 2502144, "linear_total": 7077888, "nnz": 2508829, "total": 7086912 }, "1": { "linear_attention_nnz": 983040, "linear_attention_total": 2359296, "linear_dense_nnz": 1285632, "linear_dense_total": 4718592, "linear_nnz": 2268672, "linear_total": 7077888, "nnz": 2275077, "total": 7086528 }, "10": { "linear_attention_nnz": 1376256, "linear_attention_total": 2359296, "linear_dense_nnz": 168960, "linear_dense_total": 4718592, "linear_nnz": 1545216, "linear_total": 7077888, "nnz": 1551278, "total": 7086912 }, "11": { "linear_attention_nnz": 983040, "linear_attention_total": 2359296, "linear_dense_nnz": 485376, "linear_dense_total": 4718592, "linear_nnz": 1468416, "linear_total": 7077888, "nnz": 1474300, "total": 7086528 }, "2": { "linear_attention_nnz": 1769472, "linear_attention_total": 2359296, "linear_dense_nnz": 1523712, "linear_dense_total": 4718592, "linear_nnz": 3293184, "linear_total": 7077888, "nnz": 3300512, "total": 7087296 }, "3": { "linear_attention_nnz": 1769472, "linear_attention_total": 2359296, "linear_dense_nnz": 1555968, "linear_dense_total": 4718592, "linear_nnz": 3325440, "linear_total": 7077888, "nnz": 3332789, "total": 7087296 }, "4": { "linear_attention_nnz": 2162688, "linear_attention_total": 2359296, "linear_dense_nnz": 1617408, "linear_dense_total": 4718592, "linear_nnz": 3780096, "linear_total": 7077888, "nnz": 3787869, "total": 7087680 }, "5": { "linear_attention_nnz": 1966080, "linear_attention_total": 2359296, "linear_dense_nnz": 1514496, "linear_dense_total": 4718592, "linear_nnz": 3480576, "linear_total": 7077888, "nnz": 3488090, "total": 7087488 }, "6": { "linear_attention_nnz": 1769472, "linear_attention_total": 2359296, "linear_dense_nnz": 1135104, "linear_dense_total": 4718592, "linear_nnz": 2904576, "linear_total": 7077888, "nnz": 2911651, "total": 7087296 }, "7": { "linear_attention_nnz": 1572864, "linear_attention_total": 2359296, "linear_dense_nnz": 847872, "linear_dense_total": 4718592, "linear_nnz": 2420736, "linear_total": 7077888, "nnz": 2427432, "total": 7087104 }, "8": { "linear_attention_nnz": 1966080, "linear_attention_total": 2359296, "linear_dense_nnz": 474624, "linear_dense_total": 4718592, "linear_nnz": 2440704, "linear_total": 7077888, "nnz": 2447541, "total": 7087488 }, "9": { "linear_attention_nnz": 1179648, "linear_attention_total": 2359296, "linear_dense_nnz": 208896, "linear_dense_total": 4718592, "linear_nnz": 1388544, "linear_total": 7077888, "nnz": 1394440, "total": 7086720 } }, "linear_nnz": 30818304, "linear_sparsity": 63.71527777777778, "linear_total": 84934656, "nnz": 54738530, "pruned_heads": { "0": [2, 4, 5, 6, 7], "1": [0, 2, 3, 5, 6, 7, 8], "10": [1, 4, 5, 6, 7], "11": [0, 2, 5, 6, 7, 8, 11], "2": [8, 4, 7], "3": [2, 4, 6], "4": [2], "5": [1, 2], "6": [2, 3, 7], "7": [11, 3, 6, 7], "8": [0, 4], "9": [1, 4, 5, 7, 9, 10] }, "total": 108883970, "total_sparsity": 49.72765045212808 }, "training_args": { "_n_gpu": -1, "adafactor": false, "adam_beta1": 0.9, "adam_beta2": 0.999, "adam_epsilon": 1e-08, "dataloader_drop_last": false, "dataloader_num_workers": 0, "dataloader_pin_memory": true, "ddp_find_unused_parameters": null, "debug": false, "deepspeed": null, "disable_tqdm": false, "do_eval": 1, "do_predict": false, "do_train": 1, "eval_accumulation_steps": null, "eval_steps": 2500, "evaluation_strategy": "steps", "fp16": false, "fp16_backend": "auto", "fp16_full_eval": false, "fp16_opt_level": "O1", "gradient_accumulation_steps": 1, "greater_is_better": null, "group_by_length": false, "ignore_data_skip": false, "label_names": null, "label_smoothing_factor": 0.0, "learning_rate": 3e-05, "length_column_name": "length", "load_best_model_at_end": false, "local_rank": -1, "logging_dir": "/data_2to/devel_data/nn_pruning/output/squad_test_final_fine_tune/fine_tuned_aws_nn-pruning-v10-a32-l5-dl0-5--2021-01-21--00-52-45", "logging_first_step": false, "logging_steps": 250, "logging_strategy": "steps", "lr_scheduler_type": "linear", "max_grad_norm": 1.0, "max_steps": -1, "metric_for_best_model": null, "mp_parameters": "", "no_cuda": false, "num_train_epochs": 4, "optimize_model_before_eval": "disabled", "output_dir": "/data_2to/devel_data/nn_pruning/output/squad_test_final_fine_tune/fine_tuned_aws_nn-pruning-v10-a32-l5-dl0-5--2021-01-21--00-52-45", "overwrite_output_dir": 1, "past_index": -1, "per_device_eval_batch_size": 128, "per_device_train_batch_size": 16, "per_gpu_eval_batch_size": null, "per_gpu_train_batch_size": null, "prediction_loss_only": false, "remove_unused_columns": true, "report_to": null, "run_name": "/data_2to/devel_data/nn_pruning/output/squad_test_final_fine_tune/fine_tuned_aws_nn-pruning-v10-a32-l5-dl0-5--2021-01-21--00-52-45", "save_steps": 2500, "save_strategy": "steps", "save_total_limit": 50, "seed": 17, "sharded_ddp": "", "skip_memory_metrics": false, "tpu_metrics_debug": false, "tpu_num_cores": null, "warmup_ratio": 0.0, "warmup_steps": 10, "weight_decay": 0.0 } }