| { |
| "force": true, |
| "dump_path": "serialization_dir/codegpt_training_09-06-2023_18-04-56", |
| "data_file": "data/binarized_text.gpt2.pickle", |
| "student_config": "training_configs/codegpt_normal.json", |
| "student_pretrained_weights": null, |
| "teacher_name": "microsoft/CodeGPT-small-py", |
| "temperature": 2.0, |
| "alpha_ce": 0.33, |
| "alpha_clm": 0.33, |
| "alpha_mse": 0.0, |
| "alpha_cos": 0.33, |
| "restrict_ce_to_mask": false, |
| "freeze_pos_embs": true, |
| "n_epoch": 1, |
| "batch_size": 6, |
| "group_by_size": true, |
| "gradient_accumulation_steps": 50, |
| "warmup_prop": 0.05, |
| "weight_decay": 0.0, |
| "learning_rate": 0.0005, |
| "adam_epsilon": 1e-06, |
| "max_grad_norm": 5.0, |
| "initializer_range": 0.02, |
| "fp16": false, |
| "fp16_opt_level": "O1", |
| "n_gpu": 1, |
| "local_rank": 0, |
| "seed": 56, |
| "log_interval": 500, |
| "checkpoint_interval": 3150, |
| "n_nodes": 1, |
| "node_id": 0, |
| "global_rank": 0, |
| "world_size": 1, |
| "n_gpu_per_node": 1, |
| "multi_gpu": false, |
| "is_master": true, |
| "multi_node": false |
| } |