| --- |
| license: mit |
| datasets: |
| - magicgh/Ask-before-Plan |
| language: |
| - en |
| base_model: |
| - meta-llama/Meta-Llama-3-8B-Instruct |
| - mistralai/Mistral-7B-Instruct-v0.2 |
| --- |
| |
| # CEP Framework |
|
|
| <a href="https://arxiv.org/abs/2406.12639">Paper</a> • |
| <a href="https://huggingface.co/datasets/magicgh/Ask-before-Plan">Data</a> • |
| <a href="https://drive.google.com/file/d/1vMIhs8mpMgk33pFDv2rWg6AJNyD70Sod">Environment</a> • |
| <a href="https://github.com/magicgh/Ask-before-Plan">Code</a> |
|
|
| This repository contains the checkpoint for the CEP framework in our EMNLP 2024 Paper, *Ask-before-Plan: Proactive Language Agents for Real-World Planning*. |
| We release our CEP models, including LLaMA-3-8B and Mistral-7B variants, finetuned on Clarification and Execution subtasks. |
|
|
| ## Get Started |
| 1. Download our checkpoints. |
| ```bash |
| git lfs install |
| git clone https://huggingface.co/magicgh/CEP |
| ``` |
| 2. OpenAI compatible servers. |
| ```bash |
| python3 -m vllm.entrypoints.openai.api_server |
| --served-model-name ${model_name} |
| --model ${model} |
| --kv-cache-dtype fp8 |
| --port ${port} |
| --enable-lora |
| --lora-modules ${lora_models} |
| --chat-template ${chat_template} |
| ``` |
| ## Citation |
| If you find our research helpful for your work, please star [this repository](https://github.com/magicgh/Ask-before-Plan) and cite our paper: |
| ``` |
| @article{ask-before-plan, |
| author = {Xuan Zhang and Yang Deng and Zifeng Ren and See-Kiong Ng and Tat-Seng Chua}, |
| journal = {ArXiv preprint}, |
| title = {Ask-before-Plan: Proactive Language Agents for Real-World Planning}, |
| url = {https://arxiv.org/abs/2406.12639}, |
| year = {2024} |
| } |
| ``` |
