| { |
| "target_model_name": "meta-llama/Llama-3.1-8B-Instruct", |
| "load_model_checkpoint": "", |
| "train_system": "", |
| "train_stimulus_completion": "", |
| "train_stimulus": "data/martian_data/pipeline_output/training_files/train_stimulus.json", |
| "train_control": "", |
| "train_qa": "data/martian_data/pipeline_output/training_files/train_qa.json", |
| "train_with_verb_mask": "user", |
| "add_thought_tokens": false, |
| "nudge_persona": false, |
| "modify_chat_template": true, |
| "filter": "", |
| "train_percent": 1.0, |
| "eval_ppl": true, |
| "eval_system": "", |
| "eval_stimulus_completion": "", |
| "eval_stimulus": "data/martian_data/pipeline_output/training_files/eval_stimulus.json", |
| "eval_control": "", |
| "eval_qa": "data/martian_data/pipeline_output/training_files/eval_qa.json", |
| "eval_every_n_steps": 200, |
| "output_dir": "out/codereview", |
| "save_model": true, |
| "save_every_n_steps": 200, |
| "use_wandb": true, |
| "run_name": "", |
| "shift_position_ids": true, |
| "min_layer_to_read": 15, |
| "max_layer_to_read": 16, |
| "layer_to_write": 0, |
| "module_setup": "read-vary_write-fixed_n-fixed", |
| "num_layers_to_read": 1, |
| "num_layers_to_sample": 1, |
| "batch_size_training": 1, |
| "gradient_accumulation_steps": 8, |
| "gradient_clipping": false, |
| "gradient_clipping_threshold": 1.0, |
| "num_epochs": 5, |
| "num_workers_dataloader": 1, |
| "lr": 3e-05, |
| "ema_decay": 1, |
| "warmup_steps": 50, |
| "weight_decay": 0.01, |
| "gamma": 0.85, |
| "seed": 42, |
| "peft_method": "lora", |
| "use_peft": true, |
| "use_fsdp": false, |
| "checkpoint_dir": "out/codereview/006/checkpoints" |
| } |
