73 lines
3.3 KiB
Markdown
73 lines
3.3 KiB
Markdown
|
---
|
|||
|
|
license: apache-2.0
|
|||
|
|
base_model:
|
|||
|
|
- Qwen/Qwen2.5-7B-Instruct
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
<p align="center">
|
|||
|
|
<img src="./logo-verl-agent.png" alt="logo" width="55%">
|
|||
|
|
</p>
|
|||
|
|
|
|||
|
|
|
|||
|
|
<p align="center">
|
|||
|
|
<a href="https://arxiv.org/abs/2505.10978">
|
|||
|
|
<img src="https://img.shields.io/badge/arXiv-Paper-red?style=flat-square&logo=arxiv" alt="arXiv Paper"></a>
|
|||
|
|
|
|||
|
|
<a href="https://github.com/langfengQ/verl-agent">
|
|||
|
|
<img src="https://img.shields.io/badge/GitHub-Project-181717?style=flat-square&logo=github" alt="GitHub Project"></a>
|
|||
|
|
|
|||
|
|
<a href="https://huggingface.co/collections/langfeng01/verl-agent-684970e8f51babe2a6d98554">
|
|||
|
|
<img src="https://img.shields.io/badge/HuggingFace-Models-yellow?style=flat-square&logo=huggingface" alt="HuggingFace Models"></a>
|
|||
|
|
|
|||
|
|
<a href="https://x.com/langfengq/status/1930848580505620677">
|
|||
|
|
<img src="https://img.shields.io/badge/Twitter-Channel-000000?style=flat-square&logo=x" alt="X Channel"></a>
|
|||
|
|
</p>
|
|||
|
|
|
|||
|
|
|
|||
|
|
## Quick Start
|
|||
|
|
|
|||
|
|
To use this model, follow these three steps:
|
|||
|
|
|
|||
|
|
1. Clone [verl-agent](https://github.com/langfengQ/verl-agent).
|
|||
|
|
2. Set [`actor_rollout_ref.model.path`](https://github.com/langfengQ/verl-agent/blob/35b3da38293993f9bf4f7873dfb3262a361e956c/examples/gigpo_trainer/run_webshop.sh#L30) to your local path, e.g. `your/own/path/GiGPO-Qwen2.5-7B-Instruct-ALFWorld`.
|
|||
|
|
3. Ensure [`trainer.val_before_train=True`](https://github.com/langfengQ/verl-agent/blob/35b3da38293993f9bf4f7873dfb3262a361e956c/examples/gigpo_trainer/run_webshop.sh#L72), so evaluation runs before training.
|
|||
|
|
|
|||
|
|
|
|||
|
|
For more details, please refer to the [verl-agent](https://github.com/langfengQ/verl-agent).
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## Notes
|
|||
|
|
|
|||
|
|
`GiGPO-Qwen2.5-7B-Instruct-WebShop` is trained using [GiGPO](https://huggingface.co/papers/2505.10978) and the following prompt:
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
WEBSHOP_TEMPLATE_NO_HIS = """
|
|||
|
|
You are an expert autonomous agent operating in the WebShop e‑commerce environment.
|
|||
|
|
Your task is to: {task_description}.
|
|||
|
|
Your current observation is: {current_observation}.
|
|||
|
|
Your admissible actions of the current situation are:
|
|||
|
|
[
|
|||
|
|
{available_actions}
|
|||
|
|
].
|
|||
|
|
|
|||
|
|
Now it's your turn to take one action for the current step.
|
|||
|
|
You should first reason step-by-step about the current situation, then think carefully which admissible action best advances the shopping goal. This reasoning process MUST be enclosed within <think> </think> tags.
|
|||
|
|
Once you've finished your reasoning, you should choose an admissible action for current step and present it within <action> </action> tags.
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
WEBSHOP_TEMPLATE = """
|
|||
|
|
You are an expert autonomous agent operating in the WebShop e‑commerce environment.
|
|||
|
|
Your task is to: {task_description}.
|
|||
|
|
Prior to this step, you have already taken {step_count} step(s). Below are the most recent {history_length} observations and the corresponding actions you took: {action_history}
|
|||
|
|
You are now at step {current_step} and your current observation is: {current_observation}.
|
|||
|
|
Your admissible actions of the current situation are:
|
|||
|
|
[
|
|||
|
|
{available_actions}
|
|||
|
|
].
|
|||
|
|
|
|||
|
|
Now it's your turn to take one action for the current step.
|
|||
|
|
You should first reason step-by-step about the current situation, then think carefully which admissible action best advances the shopping goal. This reasoning process MUST be enclosed within <think> </think> tags.
|
|||
|
|
Once you've finished your reasoning, you should choose an admissible action for current step and present it within <action> </action> tags.
|
|||
|
|
"""
|
|||
|
|
```
|