| { | |
| "alpha_pattern": {}, | |
| "auto_mapping": null, | |
| "base_model_name_or_path": "Qwen/Qwen2.5-Coder-1.5B-Instruct", | |
| "beta1": 0.85, | |
| "beta2": 0.85, | |
| "bias": "none", | |
| "deltaT": 1, | |
| "fan_in_fan_out": false, | |
| "inference_mode": true, | |
| "init_lora_weights": true, | |
| "init_r": 12, | |
| "layer_replication": null, | |
| "layers_pattern": null, | |
| "layers_to_transform": null, | |
| "loftq_config": {}, | |
| "lora_alpha": 16, | |
| "lora_dropout": 0.0, | |
| "megatron_config": null, | |
| "megatron_core": "megatron.core", | |
| "modules_to_save": null, | |
| "orth_reg_weight": 0.5, | |
| "peft_type": "ADALORA", | |
| "r": 8, | |
| "rank_pattern": null, | |
| "revision": null, | |
| "target_modules": [ | |
| " up_proj", | |
| "k_proj", | |
| " o_proj", | |
| " v_proj", | |
| " down_proj", | |
| " gate_proj", | |
| " q_proj" | |
| ], | |
| "target_r": 8, | |
| "task_type": "CAUSAL_LM", | |
| "tfinal": 0, | |
| "tinit": 0, | |
| "total_step": null, | |
| "use_dora": false, | |
| "use_rslora": false | |
| } |