Update README.md

README.md

@@ -5,11 +5,12 @@ license: other
 license_link: LICENSE
 ---
 
+
 <div align="center">
 
 <img src="./assets/logo.png" alt="HunyuanImage-3.0 Logo" width="400">
 
-# 🎨 HunyuanImage-3.0: A Powerful Native Multimodal Model for Text-to-Image Generation
+# 🎨 HunyuanImage-3.0: A Powerful Native Multimodal Model for Image Generation
 
 </div>
 
@@ -22,10 +23,10 @@ license_link: LICENSE
 
 
 <div align="center">
-  <a href=xxxx target="_blank"><img src=https://img.shields.io/badge/Official%20Site-333399.svg?logo=homepage height=22px></a>
+  <a href=https://hunyuan.tencent.com/image target="_blank"><img src=https://img.shields.io/badge/Official%20Site-333399.svg?logo=homepage height=22px></a>
   <a href=https://huggingface.co/tencent/HunyuanImage-3.0 target="_blank"><img src=https://img.shields.io/badge/%F0%9F%A4%97%20Models-d96902.svg height=22px></a>
   <a href=https://github.com/Tencent-Hunyuan/HunyuanImage-3.0 target="_blank"><img src= https://img.shields.io/badge/Page-bb8a2e.svg?logo=github height=22px></a>
-  <a href=xxxx target="_blank"><img src=https://img.shields.io/badge/Report-b5212f.svg?logo=arxiv height=22px></a>
+  <a href="come soon" target="_blank"><img src=https://img.shields.io/badge/Report-b5212f.svg?logo=arxiv height=22px></a>
   <a href=https://x.com/TencentHunyuan target="_blank"><img src=https://img.shields.io/badge/Hunyuan-black.svg?logo=x height=22px></a>
 </div>
 
@@ -40,11 +41,12 @@ If you develop/use HunyuanImage-3.0 in your projects, welcome to let us know.
 
 ## 📑 Open-source Plan
 
-- HunyuanImage-3.0 (Text-to-Image Model)
+- HunyuanImage-3.0 (Image Generation Model)
   - [x] Inference 
-  - [x] Pretain Checkpoints
-  - [x] Instruct Checkpoints
+  - [x] Pretrain Checkpoints
+  - [ ] Instruct Checkpoints
   - [ ] VLLM Support
+  - [ ] Distilled Checkpoints
   - [ ] Image-to-Image Generation
   - [ ] Multi-turn Interaction
 
@@ -66,6 +68,8 @@ If you develop/use HunyuanImage-3.0 in your projects, welcome to let us know.
   - [🎨 Interactive Gradio Demo](#-interactive-gradio-demo)
 - [🧱 Models Cards](#-models-cards)
 - [📝 Prompt Guide](#-prompt-guide)
+  - [Manually Writing Prompts](#manually-writing-prompts)
+  - [System Prompt For Automatic Rewriting the Prompt](#system-prompt-for-automatic-rewriting-the-prompt)
   - [Advanced Tips](#advanced-tips)
   - [More Cases](#more-cases)
 - [📊 Evaluation](#-evaluation)
@@ -86,9 +90,9 @@ If you develop/use HunyuanImage-3.0 in your projects, welcome to let us know.
 
 ## ✨ Key Features
 
-* 🧠 **Unified Multimodal Architecture:** Moving beyond the prevalent DiT-based architectures, HunyuanImage-3.0 employs a unified autoregressive framework. This design enables a more direct and integrated modeling of text and image modalities, leading to surprisingly effective and contextually rich text-to-image generation.
+* 🧠 **Unified Multimodal Architecture:** Moving beyond the prevalent DiT-based architectures, HunyuanImage-3.0 employs a unified autoregressive framework. This design enables a more direct and integrated modeling of text and image modalities, leading to surprisingly effective and contextually rich image generation.
 
-* 🏆 **The Largest Text-to-Image MoE Model:** This is the largest text-to-image Mixture of Experts (MoE) model to date. It features 64 experts and a total of 80 billion parameters, with 13 billion activated per token, significantly enhancing its capacity and performance.
+* 🏆 **The Largest Image Generation MoE Model:** This is the largest open-source image generation Mixture of Experts (MoE) model to date. It features 64 experts and a total of 80 billion parameters, with 13 billion activated per token, significantly enhancing its capacity and performance.
 
 * 🎨 **Superior Image Generation Performance:**Through rigorous dataset curation and advanced reinforcement learning post-training, we've achieved an optimal balance between semantic accuracy and visual excellence. The model demonstrates exceptional prompt adherence while delivering photorealistic imagery with stunning aesthetic quality and fine-grained details.
 
@@ -131,12 +135,14 @@ pip install flash-attn==2.8.3 --no-build-isolation
 # FlashInfer for optimized moe inference. v0.3.1 is tested.
 pip install flashinfer-python
 ```
-> 💡 **Installation Tips:** It is critical that the CUDA version used by PyTorch matches the system's CUDA version. 
+> 💡**Installation Tips:** It is critical that the CUDA version used by PyTorch matches the system's CUDA version. 
 > FlashInfer relies on this compatibility when compiling kernels at runtime. Pytorch 2.7.1+cu128 is tested.
 > GCC version >=9 is recommended for compiling FlashAttention and FlashInfer.
 
 > ⚡ **Performance Tips:** These optimizations can significantly speed up your inference!
 
+> 💡**Notation:** When FlashInfer is enabled, the first inference may be slower (about 10 minutes) due to kernel compilation. Subsequent inferences on the same machine will be much faster.
+
 ## 🚀 Usage
 
 ### 🔥 Quick Start with Transformers
@@ -144,26 +150,26 @@ pip install flashinfer-python
 The easiest way to get started with HunyuanImage-3.0:
 
 ```python
-from hunyuan_image_3.hunyuan import HunyuanImage3ForCausalMM
+from transformers import AutoModelForCausalLM
 
 # Load the model
 model_id = "tencent/HunyuanImage-3.0"
 
 kwargs = dict(
-    attn_implementation="sdpa",
+    attn_implementation="sdpa",     # Use "flash_attention_2" if FlashAttention is installed
     trust_remote_code=True,
     torch_dtype="auto",
     device_map="auto",
-    moe_impl="eager",
+    moe_impl="eager",   # Use "flashinfer" if FlashInfer is installed
 )
 
-model = HunyuanImage3ForCausalMM.from_pretrained(model_id, **kwargs)
+model = AutoModelForCausalLM.from_pretrained(model_id, **kwargs)
 model.load_tokenizer(model_id)
 
 # generate the image
-prompt = "A brown and white dog is running on the grass, 9:16"
-image = model.generate_image(prompt=prompt, cot="think", stream=True)
-image.save("test.png")
+prompt = "A brown and white dog is running on the grass"
+image = model.generate_image(prompt=prompt, stream=True)
+image.save("image.png")
 ```
 
 ### 🏠 Local Installation & Usage
@@ -179,13 +185,13 @@ cd HunyuanImage-3.0/
 
 ```bash
 # Download from HuggingFace
-huggingface-cli download tencent/HunyuanImage-3.0 --local-dir ./HunyuanImage-3
+hf download tencent/HunyuanImage-3.0 --local-dir ./HunyuanImage-3
 ```
 
 #### 3️⃣ Run the Demo
 
 ```bash
-python3 run_local.py --model-id ./HunyuanImage-3 --prompt "A brown and white dog is running on the grass, 9:16"
+python3 run_image_gen.py --model-id ./HunyuanImage-3 --verbose 1 --prompt "A brown and white dog is running on the grass"
 ```
 
 ### 🎨 Interactive Gradio Demo
@@ -247,15 +253,22 @@ Notes:
 ## 📝 Prompt Guide
 
 ### Manually Writing Prompts.
-The base model does not automatically rewrite or enhance input prompts. For optimal results, we recommend consulting our official guide on how to write effective prompts.
+The Pretrain Checkpoint does not automatically rewrite or enhance input prompts, Instruct Checkpoint can rewrite or enhance input prompts with thinking . For optimal results currently, we recommend community partners consulting our official guide on how to write effective prompts.
 
 Reference: [HunyuanImage 3.0 Prompt Handbook](
 https://docs.qq.com/doc/DUVVadmhCdG9qRXBU)
 
 
-### System Prompt to automatically rewrite the input.
+### System Prompt For Automatic Rewriting the Prompt.
+
+We've included two system prompts in the PE folder of this repository that leverage DeepSeek to automatically enhance user inputs:
+
+* **system_prompt_universal**: This system prompt converts photographic style, artistic prompts into a detailed one.
+* **system_prompt_text_rendering**: This system prompt converts UI/Poster/Text Rending prompts to a deailed on that suits the model.
 
-We've included two system prompts in the PE folder of this repository that leverage DeepSeek and Gemini 2.5 to automatically enhance user inputs. These prompts intelligently expand and refine simple descriptions into rich, detailed versions optimized for our model, resulting in significantly improved image generation quality.
+Note that these system prompts are in Chinese because Deepseek works better with Chinese system prompts. If you want to use it for English oriented model, you may translate it into English or refer to the comments in the PE file as a guide.
+
+We also create a [Yuanqi workflow](https://yuanqi.tencent.com/agent/H69VgtJdj3Dz) to implent the universal one, you can directly try it.
 
 ### Advanced Tips
 - **Content Priority**: Focus on describing the main subject and action first, followed by details about the environment and style. A more general description framework is: **Main subject and scene + Image quality and style + Composition and perspective + Lighting and atmosphere + Technical parameters**. Keywords can be added both before and after this structure.
@@ -329,8 +342,6 @@ We adopted the GSB (Good/Same/Bad) evaluation method commonly used to assess the
 </p>
 
 
-* 🏆 **Open Leaderboard** - Coming Soon
-
 ## 📚 Citation
 
 If you find HunyuanImage-3.0 useful in your research, please cite our work:
@@ -352,4 +363,5 @@ We extend our heartfelt gratitude to the following open-source projects and comm
 * 🎨 [Diffusers](https://github.com/huggingface/diffusers) - Diffusion models library  
 * 🌐 [HuggingFace](https://huggingface.co/) - AI model hub and community
 * ⚡ [FlashAttention](https://github.com/Dao-AILab/flash-attention) - Memory-efficient attention
-* 🚀 [FlashInfer](https://github.com/flashinfer-ai/flashinfer) - Optimized inference engine
+* 🚀 [FlashInfer](https://github.com/flashinfer-ai/flashinfer) - Optimized inference engine
+