KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #typescript #developer-tools #claude-code #ai-agent #codex #deepseek-harness #ai-agents #cli #react #mcp

CVPR26-HiconAgent

★ 33 Open GitHub ↗

[CVPR 2026] HiconAgent: History Context-aware Policy Optimization for GUI Agents

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h2>
 <img src="./assets/hicon.png" height="28">
 HiconAgent: History Context-aware Policy Optimization for GUI Agents
</h2>

<a>Xurui Zhou</a>1,
<a href="https://scholar.google.com/citations?user=Mpg0w3cAAAAJ">Gongwei Chen</a>1,
<a>Yuquan Xie</a>1,
<a href="https://scholar.google.com/citations?user=TDBF2UoAAAAJ">Zaijing Li</a>1,
<a href="https://jnhujnhu.github.io/">Kaiwen Zhou</a>3,
 
<a>Shuai Wang</a>3,
<a href="https://shuoyang-1998.github.io/">Shuo Yang</a>1,
<a href="https://scholar.google.com/citations?user=mEjhz-IAAAAJ">Zhuotao Tian</a>1,
<a href="https://scholar.google.com/citations?user=9Vc--XsAAAAJ">Rui Shao</a>1,2
&#9993;

1 Harbin Institute of Technology, Shenzhen&nbsp;&nbsp;&nbsp;
2Shenzhen Loop Area Institute &nbsp;&nbsp;&nbsp;
3 Huawei Noah’s Ark Lab

&#9993; Corresponding author

[![Paper](https://img.shields.io/badge/Paper-arXiv-deepgreen)](https://arxiv.org/abs/2512.01763)
[![Data](https://img.shields.io/badge/Data-HuggingFace-yellow)](https://huggingface.co/datasets/Minuskid/HiconAgent-AMEX)
[![Project Page](https://img.shields.io/badge/Project-Page-blue)](https://jiutian-vl.github.io/HiconAgent.github.io/)

</div>

## :new: Updates

- [11/2025] :fire: We release the code. Enjoy it!

- [02/2026] :fire: HiconAgent is accepted to CVPR 2026!

## Install Dependencies
```shell
# first install uv
pip install uv
# second install mirage
uv sync
source .venv/bin/activate
# third install EasyR1
cd EasyR1
uv pip install -e .
cd ..
uv pip install -r requirements.txt
```

- Install vllm-0.7.4-nightly to avoid OOM
```shell
uv pip uninstall vllm
export VLLM_COMMIT=227578480d71fc94ef46ca77fb69496412158d68
uv pip install --no-cache-dir vllm --pre --extra-index-url "https://wheels.vllm.ai/${VLLM_COMMIT}"
git clone https://github.com/XuRui314/vllm.git
cp -r vllm/vllm/ .venv/lib/python3.11/site-packages
rm -rf vllm
uv pip install flash-attn==2.7.3
```

You need to prepare two model directories. First, download the Qwen2.5-VL-3B model, and use its path to replace the reference model path in the /configs/rl/ui_tars_2b_grpo.yaml file. Then, make a copy of this directory and rename the copied folder to XYQVL, modify config.json to:
```shell
{
  "architectures": [
    "XYQForConditionalGeneration"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 151643,
  "eos_token_id": 151645,
  "vision_start_token_id": 151652,
  "vision_end_token_id": 151653,
  "vision_token_id": 151654,
  "image_token_id": 151655,
  "video_token_id": 151656,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 11008,
  "max_position_embeddings": 128000,
  "max_window_layers": 70,
  "model_type": "xyqvl",
  "num_attention_heads": 16,
  "num_hidden_layers": 36,
  "num_key_value_heads": 2,
  "rms_norm_eps": 1e-06,
  "rope_theta": 1000000.0,
  "sliding_window": 32768,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.41.2",
  "use_cache": true,
  "use_sliding_window": false,
  "vision_config": {
    "depth": 32,
    "hidden_act": "silu",
    "hidden_size": 1280,
    "intermediate_size": 3420,
    "num_heads": 16,
    "in_chans": 3,
    "out_hidden_size": 2048,
    "patch_size": 14,
    "spatial_merge_size": 2,
    "spatial_patch_size": 14,
    "window_size": 112,
    "fullatt_block_indexes": [
      7,
      15,
      23,
      31
    ],
    "tokens_per_second": 2,
    "temporal_patch_size": 2
  },
  "rope_scaling": {
    "type": "mrope",
    "mrope_section": [
      16,
      24,
      24
    ]
  },
  "vocab_size": 151936
}
```

## How to run

```shell
bash scripts/gui/run_training.sh
```

## :sparkles: Overall view

<img src="./assets/teaser.png">

Comparison of existing GUI RL framework with our HCPO framework. HCPO jointly improves the sampling and update phases of training by integrating Dynamic Context Sampling **(DCS)** and Anchor-guided History Compression **(AHC)**.

## :unicorn: Rethinking History Usage: Limitations of Fixed Context and the Anchoring Role of Actions

Different samples prefer different history lengths. Left: For each sample we evaluate a set of different history lengths $\tau$ and take the $\tau$ that yields the highest mean reward. The preferred $\tau$ differs across samples and action types. Right: Providing more history does not necessarily yield the optimal result, suggesting effective usage of historical information is under exploration.

Layer-wise token-drop analysis. Left: Schematic of the layer-wise token-drop probe, illustrating the information flow of image-drop and action-drop. Right: Dropping $A_{\mathrm{his}}$ at shallow depths ($k < 12$) causes a much larger decline than dropping $V_{\mathrm{his}}$. 
Even if rich visual information is retained, later layers cannot directly extract effective cues from $V_{\mathrm{his}}$ without the action anchors. As $k$ increases, the action-drop curve rises toward the image-drop curve and the image-action drop curve converges rapidly.

## :balloon: HiconAgent Framework

Overview of our history context-aware optimization framework for building HiconAgent. HCPO improves both the sampling and update phases of policy optimization by incorporating two key components: (1) **Dynamic Context Sampling (DCS)**, which introduces varied history lengths during training to encourage context-effective decision-making, and (2) **Anchor-guided History Compression (AHC)**, which adopts a dual-branch architecture where both branches share sampled responses and group-wise advantages. The compressed branch is trained using policy gradients, aligned with the uncompressed branch via a history-enhanced alignment loss.

## :smile_cat: Evaluation results
<table>
 <tr>
 <td><img src="./assets/table1.png" alt="Table 1" width="420"></td>
 <td><img src="./assets/table2.png" alt="Table 2" width="420"></td>
 </tr>
</table>

## Acknowledgement
- We built our code based on: [Easy-R1](https://github.com/hiyouga/EasyR1).

CVPR26-HiconAgent

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README