| { | |
| "add_lora": "False", | |
| "apply_kd": "False", | |
| "begintask_idx": "0", | |
| "compute_replay_cross_entropy": "False", | |
| "cross_replay": "False", | |
| "data_path": "'/mnt/data/minghaoli/projects/api-bank-training/output-0620/lv1-lv2-lv3-train.json'", | |
| "deepscale": "False", | |
| "deepscale_config": "None", | |
| "deepspeed": "True", | |
| "deepspeed_config": "'/mnt/data/yingxiu/EFLOP/CODE/API-Bank/configs/ds_new_config.json'", | |
| "deepspeed_mpi": "False", | |
| "distill_first": "False", | |
| "exp": "'0620eflop_api_response_lv1_lv2_lv3_run0'", | |
| "gradient_accumulation_steps": "16", | |
| "kd_curr": "False", | |
| "kd_prev": "False", | |
| "local_rank": "2", | |
| "logging_dir": "'/mnt/data/yingxiu/EFLOP/LLMOUT/apitblogs/0620eflop_api_response_lv1_lv2_lv3_run0'", | |
| "lora_alpha": "32", | |
| "lora_r": "8", | |
| "lr": "2e-05", | |
| "max_grad_norm": "1.0", | |
| "max_train_steps": "-1", | |
| "model_max_length": "2048", | |
| "model_name_or_path": "'/mnt/data/yingxiu/EFLOP/MODELS/chavinlo-alpaca-native'", | |
| "num_epochs": "3", | |
| "num_fuse": "-1", | |
| "optimizer": "'AdamW'", | |
| "output_dir": "'/mnt/data/yingxiu/EFLOP/LLMOUT/apioutputs/0620eflop_api_response_lv1_lv2_lv3_run0'", | |
| "per_device_eval_batch_size": "4", | |
| "per_device_train_batch_size": "1", | |
| "print_steps": "20", | |
| "progressive_fuse": "False", | |
| "pseudo_instruction_path": "None", | |
| "replay": "False", | |
| "replay_first": "False", | |
| "replay_num": "500", | |
| "replay_ratio": "0.02", | |
| "scheduler": "'cosine'", | |
| "seed": "42", | |
| "tasks": "['debug']", | |
| "train_curr": "True", | |
| "twolora": "False", | |
| "use_pseudo": "False", | |
| "val_size": "2000", | |
| "warmup_ratio": "0.03", | |
| "warmup_steps": "0", | |
| "weight_decay": "0.0" | |
| } | |