| { | |
| "aggregation_type": "mean", | |
| "auto_mapping": { | |
| "base_model_class": "LlamaForCausalLM", | |
| "parent_library": "prompt.model.modeling_llama_custom" | |
| }, | |
| "base_model_name_or_path": "lmsys/vicuna-13b-v1.3", | |
| "inference_mode": true, | |
| "num_attention_heads": 40, | |
| "num_exits": 1, | |
| "num_layers": 40, | |
| "num_special_tokens": 3, | |
| "num_transformer_submodules": 1, | |
| "num_virtual_tokens": 3, | |
| "prefix_virtual_tokens": 10, | |
| "prompt_tuning_init": "TEXT", | |
| "prompt_tuning_init_text": "Next 2 word", | |
| "revision": null, | |
| "task_type": null, | |
| "token_dim": 5120, | |
| "tokenizer_kwargs": null, | |
| "tokenizer_name_or_path": "lmsys/vicuna-13b-v1.3", | |
| "use_cache": true, | |
| "use_custom_lm_head": false, | |
| "use_prefix_tuning": false, | |
| "virtual_tokens_per_special_token": 1, | |
| "vt_attention_type": "ensemble" | |
| } |