Improve model card: Update arXiv link and add comprehensive details from GitHub

README.md

@@ -1,14 +1,13 @@
 ---
-license: apache-2.0
 library_name: transformers
+license: apache-2.0
+pipeline_tag: image-text-to-text
 tags:
 - vision-language-model
-- image-text-to-text
 - linear-attention
 - gated-deltanet
 - infinitevl
 - multimodal
-pipeline_tag: image-text-to-text
 ---
 
 <div align="center">
@@ -33,8 +32,9 @@ Haoran Yin<sup>2</sup>,
 (✉️) corresponding author: <a href="mailto:xgwang@hust.edu.cn">xgwang@hust.edu.cn</a>
 
 <br>
-<a href="https://arxiv.org/abs/2502.xxxxx"><img src="https://img.shields.io/badge/arXiv-Paper-b31b1b.svg" alt="arXiv"></a>
+<a href="https://arxiv.org/abs/2512.08829"><img src="https://img.shields.io/badge/arXiv-Paper-b31b1b.svg" alt="arXiv"></a>
 <a href="https://github.com/hustvl/InfiniteVL"><img src="https://img.shields.io/badge/GitHub-Repository-black?logo=github" alt="GitHub"></a>
+<a href="https://huggingface.co/hustvl/InfiniteVL/"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Models-blue" alt="Hugging Face"></a>
 
 </div>
 
@@ -55,6 +55,102 @@ By synergizing **Sliding Window Attention (SWA)** for fine-grained local percept
 *   🧠 **Unlimited Context:** Effectively retains context over extremely long sequences (tested >500K tokens) without OOM errors.
 *   🏆 **Strong Performance:** Matches leading Transformer-based VLMs (e.g., Qwen2.5-VL-3B) and significantly outperforms previous linear VLMs (e.g., VL-Mamba, Cobra) on comprehensive aspects.
 
+## News
+*   `Dec. 10th, 2025`: We release the **InfiniteVL** model weights and inference code! Please check [Model Zoo](#model-zoo).
+*   `Dec. 10th, 2025`: We release our paper on [Arxiv](https://arxiv.org/abs/2512.08829).
+
+## Table of Contents
+
+*   [Introduction](#introduction)
+*   [Key Highlights](#key-highlights)
+*   [News](#news)
+*   [Architecture](#architecture)
+*   [Training Strategy](#training-strategy)
+*   [Performance](#performance)
+*   [Model Zoo](#model-zoo)
+*   [Getting Started](#getting-started)
+*   [Advanced Usage: CUDA Graph Acceleration](#advanced-usage-cuda-graph-acceleration)
+*   [Qualitative Analysis & Visualization](#qualitative-analysis--visualization)
+*   [Contact](#contact)
+*   [Citation](#citation)
+*   [Acknowledgement](#acknowledgement)
+
+## Architecture
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/architecture.png" alt="InfiniteVL Architecture" width="50%">
+</div>
+<br>
+
+**InfiniteVL** adopts a hybrid architecture that synergizes the efficiency of linear attention with the precision of window-based attention. The model comprises a **Vision Encoder** (adapted from Qwen2.5-VL), a **Projection MLP**, and a **Decoder-only LLM Backbone**.
+
+### Key Design Highlights
+
+*   **Hybrid Block Design**: The LLM backbone consists of **9 Hybrid Blocks**. Within each block, we strategically interleave:
+    *   **1 Sliding Window Attention (SWA) Layer**: Responsible for capturing high-resolution local context and fine-grained visual details.
+    *   **3 Gated DeltaNet Layers**: Responsible for modeling long-range global dependencies with linear complexity.
+
+*   **Constant Memory Footprint**: Unlike traditional Transformers where the Key-Value (KV) cache grows linearly with sequence length ($O(N)$), the **Gated DeltaNet** layers compress history into a fixed-size memory state (e.g., $16 \times 128 \times 256$). This enables **constant memory usage** and constant inference latency, even when processing unlimited input streams.
+
+*   **Seamless Integration**: By combining SWA and Gated DeltaNet, InfiniteVL achieves the "best of both worlds":
+    *   Local attention ensures high performance on information-intensive tasks (e.g., OCR, Document Understanding).
+    *   Linear attention ensures efficiency and stability for long-context scenarios (e.g., Streaming Video Understanding).
+
+## Training Strategy
+
+To achieve strong multimodal performance with minimal training resources, InfiniteVL employs a **three-stage progressive training strategy**. This approach allows our linear-complexity model to inherit the vast knowledge of a Transformer teacher before adapting to long-context scenarios.
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/training_strategy.png" alt="Training Pipeline" width="90%">
+</div>
+
+### Stage 1: Distillation Pretraining (Efficient Initialization)
+*   **Goal:** Rapidly transfer knowledge from the **Qwen2.5-VL** teacher to the InfiniteVL student.
+*   **Method:** We replace the teacher's attention layers with **Gated DeltaNet** while keeping other parameters frozen. We use **Layer-wise MSE Loss** (to align internal states) and **End-to-End KL Divergence** (to align output logits).
+*   **Significance:** This bypasses the difficulty of training linear attention from scratch, ensuring a robust initialization.
+
+### Stage 2: Instruction SFT (General Capabilities)
+*   **Goal:** Unlock strong instruction-following and reasoning capabilities.
+*   **Data:** **~8M** diverse multimodal instruction pairs, covering General VQA, OCR, Mathematics, and Code.
+*   **Settings:** Image resolution increased to **1344×1344**; max context length set to 8,192.
+*   **Outcome:** Produces the **Stage 2 Model**, which offers the best performance on standard benchmarks.
+
+### Stage 3: Long-Sequence SFT (Context Extension)
+*   **Goal:** Activate the architecture's potential for **unlimited-length processing** and streaming.
+*   **Data:** A mixture of Stage 2 data (800K) and **~200K long-sequence samples** (e.g., long videos, multi-page documents).
+*   **Method:** **LoRA** fine-tuning with context length extended to **32,768**.
+*   **Outcome:** Produces the **Stage 3 Model**, enabling length extrapolation and stable streaming inference.
+
+## Performance
+
+### 🚀 Efficiency & Streaming
+
+**InfiniteVL** is engineered for unlimited-input scenarios. Unlike Transformer-based models where cost grows linearly with history, InfiniteVL maintains **constant** computational cost and memory usage.
+
+> **Hardware Setup:** All efficiency results are measured on a single NVIDIA RTX 4090 GPU.
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/plot_line.png" width="80%" alt="Efficiency Comparison">
+  <br>
+  <em>Figure 1: Comparison of streaming FPS and latency. InfiniteVL sustains real-time performance while Transformer baselines degrade rapidly.</em>
+</div>
+
+### 🏆 Multimodal Benchmarks
+
+InfiniteVL achieves state-of-the-art performance among linear-complexity VLMs. Crucially, thanks to our **Hybrid Architecture** and **High-quality training strategies**, it overcomes the traditional weakness of linear models in information-intensive tasks (e.g., OCR, Document Understanding), achieving results comparable to top-tier Transformer VLMs.
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/performance1.png" width="100%" alt="Performance Comparison">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/performance2.png" width="100%" alt="Performance Comparison">
+  <br>
+  <em>Figure 2: Comparison of InfiniteVL with existing VLMs on public multimodal understanding, real-world comprehension, text-rich, reasoning-centric multimodal benchmarks.</em>
+</div>
+<br>
+
+**Key Takeaways:**
+*   **Best-in-Class Linear Model:** Significantly outperforms previous linear VLMs (Cobra, MaTVLM) by large margins (+40-60 points on DocVQA/OCRBench).
+*   **Transformer-Level Quality:** Matches the performance of Qwen2.5-VL-3B on complex reasoning and text-rich tasks while being significantly faster in long contexts.
+
 ## Model Zoo
 
 We release two versions of InfiniteVL-4B to cater to different application scenarios.
@@ -243,9 +339,60 @@ print(processor.batch_decode(generated_ids_trimmed, skip_special_tokens=True)[0]
 ```
 </details>
 
-## 🎥 Advanced Usage (Cuda Graph)
+## 🚀 Advanced Usage: CUDA Graph Acceleration
+
+Unlike Transformer-based VLMs where the KV cache grows dynamically, **InfiniteVL maintains a constant-size memory state**. This unique property allows us to use **CUDA Graphs** to capture the entire computation graph for both streaming prefill and decoding, eliminating kernel launch overheads and maximizing GPU utilization.
+
+This is the key technology behind our **24 FPS** real-time streaming performance.
+
+### ⚡ Accelerated Streaming Inference
+
+Unlike Transformer-based VLMs where the KV cache grows dynamically, **InfiniteVL maintains a constant-size memory state**. This unique property allows us to use **CUDA Graphs** to capture the entire computation graph for streaming prefill, eliminating kernel launch overheads.
+
+We provide a complete script in [`examples/demo_streaming_inference.py`](examples/demo_streaming_inference.py) to demonstrate this capability.
+
+> **🎥 Simulation Note:** This script **simulates a real-time streaming scenario** by reading a local video file frame-by-frame. It treats the video as a continuous data stream, updating the global linear memory state on-the-fly without retraining.
+>
+> **⚠️ Requirement:** This demo relies on the specialized model implementation (supporting `StaticCachePrealloc` and CUDA Graphs) located in the **[`infinitevl/infinitevl_streaming`](infinitevl/infinitevl_streaming)** directory. Please ensure your environment is set up correctly to import these modules.
+
+#### 1. Run the Simulation Demo
+```bash
+# Make sure you are in the project root
+python examples/demo_streaming_inference.py \
+    --model_path /path/to/InfiniteVL-4B \
+    --video_path assets/demo.mp4 \
+    --fps 30
+```
+
+### ⚡ Accelerated Decode
+
+In addition to streaming prefill, InfiniteVL natively supports **CUDA Graph-accelerated decoding**. By capturing the decoding step into a static graph, we can achieve extremely low-latency token generation, further enhancing the responsiveness of real-time interactions.
+
+> 🚧 **Coming Soon:** The code for accelerated decoding is currently being refactored and cleaned up. We are working hard to release it as soon as possible. Please stay tuned!
+
+## Qualitative Analysis & Visualization
+
+We provide visualization cases to demonstrate InfiniteVL's robust performance across diverse scenarios, ranging from information-intensive static tasks to ultra-long streaming video understanding.
+
+### 1. Fundamental Visual-Language Capabilities (OCR & Reasoning)
+InfiniteVL effectively overcomes the traditional limitations of linear attention in detailed visual perception. By combining Sliding Window Attention with Gated DeltaNet, it excels at **Dense Text Recognition (OCR), Chart Interpretation, and Complex Scene Description**, delivering performance comparable to full-attention Transformers.
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/image_case1_01.png" width="80%" alt="Fundamental Capabilities">
+</div>
+
+### 2. Long-Term Streaming Understanding
+The core strength of InfiniteVL lies in its ability to maintain coherent memory over **unlimited input streams**.
+
+The examples below demonstrate a continuous street-view video stream. InfiniteVL maintains a constant memory state and accurately answers questions at various timestamps (e.g., Frame 3100, ~1M tokens processed), recalling specific details like "NBC Studios" text or the color of a pedestrian's bag without forgetting.
+
+<div align="center">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/streaming_case1_01.png" width="80%" alt="Streaming Capabilities">
+  <img src="https://github.com/hustvl/InfiniteVL/raw/main/assets/streaming_case2_01.png" width="80%" alt="Streaming Capabilities">
+</div>
 
-Please refer to the guideline in the [github page](https://github.com/hustvl/InfiniteVL).
+## Contact
+If you have any questions, please contact Hongyuan Tao via email (hongyuantao@hust.edu.cn).
 
 ## Citation
 
@@ -266,4 +413,4 @@ InfiniteVL is built upon the giants of the open-source community. We would like
 
 *   **[Qwen2.5-VL](https://github.com/QwenLM/Qwen2.5-VL)**: For providing a powerful vision-language codebase and vision encoder.
 *   **[Gated DeltaNet](https://github.com/sustcsonglin/flash-linear-attention)**: For the efficient linear attention mechanism and CUDA kernel implementations (FLA).
-*   **Open-Source Datasets**: We sincerely thank the creators of the high-quality datasets used in our training, including **FineVision, LLaVA-OneVision, PixMo, The Cauldron, Docmatix, LLaVA-Video**, and others. Their contributions are essential to the development of efficient multimodal models.
+*   **Open-Source Datasets**: We sincerely thank the creators of the high-quality datasets used in our training, including **FineVision, LLaVA-OneVision, PixMo, The Cauldron, Docmatix, LLaVA-Video**, and others. Their contributions are essential to the development of efficient multimodal models.