| { | |
| "model": "phi4-mini", | |
| "use_unsloth_model": true, | |
| "dataset": "UWV/wim-instruct-wiki-to-jsonld-agent-steps", | |
| "max_samples": null, | |
| "filter_n1_only": true, | |
| "max_seq_length": 16384, | |
| "batch_size": 16, | |
| "gradient_accumulation_steps": 1, | |
| "max_steps": 2000, | |
| "learning_rate": 2e-05, | |
| "warmup_steps": 50, | |
| "max_grad_norm": 1.0, | |
| "lora_r": 512, | |
| "lora_alpha": 1024, | |
| "lora_dropout": 0.05, | |
| "target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ], | |
| "load_in_4bit": true, | |
| "use_gradient_checkpointing": false, | |
| "rope_scaling": null, | |
| "output_dir": "/data/model_outputs/n1_entity_extraction_model_r512_dropout", | |
| "assistant_only_loss": true, | |
| "no_eval": false, | |
| "num_workers": 8, | |
| "full_finetune": false, | |
| "wandb": true, | |
| "seed": 42, | |
| "push_to_hub": false, | |
| "hub_model_id": null | |
| } |