Improve model card: Add abstract, 'qwen' tag, and enhance top links

README.md

@@ -1,30 +1,41 @@
 ---
-license: mit
-pipeline_tag: image-text-to-text
-library_name: transformers
 base_model:
-  - OpenGVLab/InternViT-300M-448px-V2_5
-  - Qwen/Qwen2.5-0.5B-Instruct
-base_model_relation: merge
+- OpenGVLab/InternViT-300M-448px-V2_5
+- Qwen/Qwen2.5-0.5B-Instruct
+datasets:
+- HuggingFaceFV/finevideo
 language:
-  - multilingual
+- multilingual
+library_name: transformers
+license: mit
+pipeline_tag: image-text-to-text
 tags:
-  - internvl
-  - custom_code
-datasets:
-  - HuggingFaceFV/finevideo
+- internvl
+- custom_code
+- qwen
+base_model_relation: merge
 ---
 
 # InternVL2_5-1B
 
-[\[📂 GitHub\]](https://github.com/OpenGVLab/InternVL)  [\[📜 InternVL 1.0\]](https://huggingface.co/papers/2312.14238)  [\[📜 InternVL 1.5\]](https://huggingface.co/papers/2404.16821)  [\[📜 Mini-InternVL\]](https://arxiv.org/abs/2410.16261)  [\[📜 InternVL 2.5\]](https://huggingface.co/papers/2412.05271)
+**Paper:** [Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling](https://huggingface.co/papers/2412.05271)
+
+**GitHub Repository:** [OpenGVLab/InternVL](https://github.com/OpenGVLab/InternVL) | **Hugging Face Demo / Project Page:** [OpenGVLab/InternVL Space](https://huggingface.co/spaces/OpenGVLab/InternVL)
 
-[\[🆕 Blog\]](https://internvl.github.io/blog/)  [\[🗨️ Chat Demo\]](https://internvl.opengvlab.com/)  [\[🤗 HF Demo\]](https://huggingface.co/spaces/OpenGVLab/InternVL)  [\[🚀 Quick Start\]](#quick-start)  [\[📖 Documents\]](https://internvl.readthedocs.io/en/latest/)
+**Related Papers:**
+- [InternVL 1.0](https://huggingface.co/papers/2312.14238)
+- [InternVL 1.5](https://huggingface.co/papers/2404.16821)
+- [Mini-InternVL](https://arxiv.org/abs/2410.16261)
+- [InternVL 2.5](https://huggingface.co/papers/2412.05271)
 
 <div align="center">
   <img width="500" alt="image" src="https://cdn-uploads.huggingface.co/production/uploads/64006c09330a45b03605bba3/zJsd2hqd3EevgXo6fNgC-.png">
 </div>
 
+## Abstract
+
+We introduce InternVL 2.5, an advanced multimodal large language model (MLLM) series that builds upon InternVL 2.0, maintaining its core model architecture while introducing significant enhancements in training and testing strategies as well as data quality. In this work, we delve into the relationship between model scaling and performance, systematically exploring the performance trends in vision encoders, language models, dataset sizes, and test-time configurations. Through extensive evaluations on a wide range of benchmarks, including multi-discipline reasoning, document understanding, multi-image / video understanding, real-world comprehension, multimodal hallucination detection, visual grounding, multilingual capabilities, and pure language processing, InternVL 2.5 exhibits competitive performance, rivaling leading commercial models such as GPT-4o and Claude-3.5-Sonnet. Notably, our model is the first open-source MLLMs to surpass 70% on the MMMU benchmark, achieving a 3.7-point improvement through Chain-of-Thought (CoT) reasoning and showcasing strong potential for test-time scaling. We hope this model contributes to the open-source community by setting new standards for developing and applying multimodal AI systems. HuggingFace demo see this https URL
+
 ## Introduction
 
 We are excited to introduce **InternVL 2.5**, an advanced multimodal large language model (MLLM) series that builds upon InternVL 2.0, maintaining its core model architecture while introducing significant enhancements in training and testing strategies as well as data quality.
@@ -73,11 +84,11 @@ The training pipeline for a single model in InternVL 2.5 is structured across th
 
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/64119264f0f81eb569e0d569/5NduZeCPLgPJTFr0RGTq3.png)
 
-- **Stage 1: MLP Warmup.** In this stage, only the MLP projector is trained while the vision encoder and language model are frozen. A dynamic high-resolution training strategy is applied for better performance, despite increased cost. This phase ensures robust cross-modal alignment and prepares the model for stable multimodal training.
+-   **Stage 1: MLP Warmup.** In this stage, only the MLP projector is trained while the vision encoder and language model are frozen. A dynamic high-resolution training strategy is applied for better performance, despite increased cost. This phase ensures robust cross-modal alignment and prepares the model for stable multimodal training.
 
-- **Stage 1.5: ViT Incremental Learning (Optional).** This stage allows incremental training of the vision encoder and MLP projector using the same data as Stage 1. It enhances the encoder’s ability to handle rare domains like multilingual OCR and mathematical charts. Once trained, the encoder can be reused across LLMs without retraining, making this stage optional unless new domains are introduced.
+-   **Stage 1.5: ViT Incremental Learning (Optional).** This stage allows incremental training of the vision encoder and MLP projector using the same data as Stage 1. It enhances the encoder’s ability to handle rare domains like multilingual OCR and mathematical charts. Once trained, the encoder can be reused across LLMs without retraining, making this stage optional unless new domains are introduced.
 
-- **Stage 2: Full Model Instruction Tuning.** The entire model is trained on high-quality multimodal instruction datasets. Strict data quality controls are enforced to prevent degradation of the LLM, as noisy data can cause issues like repetitive or incorrect outputs. After this stage, the training process is complete.
+-   **Stage 2: Full Model Instruction Tuning.** The entire model is trained on high-quality multimodal instruction datasets. Strict data quality controls are enforced to prevent degradation of the LLM, as noisy data can cause issues like repetitive or incorrect outputs. After this stage, the training process is complete.
 
 ### Progressive Scaling Strategy
 
@@ -91,9 +102,9 @@ Compared to Qwen2-VL's 1.4 trillion tokens, InternVL2.5-78B uses only 120 billio
 
 To improve real-world adaptability and performance, we introduce two key techniques:
 
-- **Random JPEG Compression**: Random JPEG compression with quality levels between 75 and 100 is applied as a data augmentation technique. This simulates image degradation from internet sources, enhancing the model's robustness to noisy images.
+-   **Random JPEG Compression**: Random JPEG compression with quality levels between 75 and 100 is applied as a data augmentation technique. This simulates image degradation from internet sources, enhancing the model's robustness to noisy images.
 
-- **Loss Reweighting**: To balance the NTP loss across responses of different lengths, we use a reweighting strategy called **square averaging**. This method balances contributions from responses of varying lengths, mitigating biases toward longer or shorter responses.
+-   **Loss Reweighting**: To balance the NTP loss across responses of different lengths, we use a reweighting strategy called **square averaging**. This method balances contributions from responses of varying lengths, mitigating biases toward longer or shorter responses.
 
 ### Data Organization
 
@@ -103,11 +114,11 @@ In InternVL 2.0 and 2.5, the organization of the training data is controlled by
 
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/64119264f0f81eb569e0d569/2LJe24b1ua3gjI9gDitVl.png)
 
-- **Data Augmentation:** JPEG compression is applied conditionally: enabled for image datasets to enhance robustness and disabled for video datasets to maintain consistent frame quality.
+-   **Data Augmentation:** JPEG compression is applied conditionally: enabled for image datasets to enhance robustness and disabled for video datasets to maintain consistent frame quality.
 
-- **Maximum Tile Number:** The parameter `n_max` controls the maximum tiles per dataset. For example, higher values (24–36) are used for multi-image or high-resolution data, lower values (6–12) for standard images, and 1 for videos.
+-   **Maximum Tile Number:** The parameter `n_max` controls the maximum tiles per dataset. For example, higher values (24–36) are used for multi-image or high-resolution data, lower values (6–12) for standard images, and 1 for videos.
 
-- **Repeat Factor:** The repeat factor `r` adjusts dataset sampling frequency. Values below 1 reduce a dataset's weight, while values above 1 increase it. This ensures balanced training across tasks and prevents overfitting or underfitting.
+-   **Repeat Factor:** The repeat factor `r` adjusts dataset sampling frequency. Values below 1 reduce a dataset's weight, while values above 1 increase it. This ensures balanced training across tasks and prevents overfitting or underfitting.
 
 #### Data Filtering Pipeline
 
@@ -121,14 +132,14 @@ To address this challenge and support future research, we designed an efficient
 
 The pipeline includes two modules, for **pure-text data**, three key strategies are used:
 
-1. **LLM-Based Quality Scoring**: Each sample is scored (0–10) using a pre-trained LLM with domain-specific prompts. Samples scoring below a threshold (e.g., 7) are removed to ensure high-quality data.
-2. **Repetition Detection**: Repetitive samples are flagged using LLM-based prompts and manually reviewed. Samples scoring below a stricter threshold (e.g., 3) are excluded to avoid repetitive patterns.
-3. **Heuristic Rule-Based Filtering**: Anomalies like abnormal sentence lengths or duplicate lines are detected using rules. Flagged samples undergo manual verification to ensure accuracy before removal.
+1.  **LLM-Based Quality Scoring**: Each sample is scored (0–10) using a pre-trained LLM with domain-specific prompts. Samples scoring below a threshold (e.g., 7) are removed to ensure high-quality data.
+2.  **Repetition Detection**: Repetitive samples are flagged using LLM-based prompts and manually reviewed. Samples scoring below a stricter threshold (e.g., 3) are excluded to avoid repetitive patterns.
+3.  **Heuristic Rule-Based Filtering**: Anomalies like abnormal sentence lengths or duplicate lines are detected using rules. Flagged samples undergo manual verification to ensure accuracy before removal.
 
 For **multimodal data**, two strategies are used:
 
-1. **Repetition Detection**: Repetitive samples in non-academic datasets are flagged and manually reviewed to prevent pattern loops. High-quality datasets are exempt from this process.
-2. **Heuristic Rule-Based Filtering**: Similar rules are applied to detect visual anomalies, with flagged data verified manually to maintain integrity.
+1.  **Repetition Detection**: Repetitive samples in non-academic datasets are flagged and manually reviewed to prevent pattern loops. High-quality datasets are exempt from this process.
+2.  **Heuristic Rule-Based Filtering**: Similar rules are applied to detect visual anomalies, with flagged data verified manually to maintain integrity.
 
 #### Training Data
 
@@ -360,40 +371,50 @@ generation_config = dict(max_new_tokens=1024, do_sample=True)
 # pure-text conversation (纯文本对话)
 question = 'Hello, who are you?'
 response, history = model.chat(tokenizer, None, question, generation_config, history=None, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 question = 'Can you tell me a story?'
 response, history = model.chat(tokenizer, None, question, generation_config, history=history, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 # single-image single-round conversation (单图单轮对话)
-question = '<image>\nPlease describe the image shortly.'
+question = '<image>
+Please describe the image shortly.'
 response = model.chat(tokenizer, pixel_values, question, generation_config)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 # single-image multi-round conversation (单图多轮对话)
-question = '<image>\nPlease describe the image in detail.'
+question = '<image>
+Please describe the image in detail.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config, history=None, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 question = 'Please write a poem according to the image.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config, history=history, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 # multi-image multi-round conversation, combined images (多图多轮对话，拼接图像)
 pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
 pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat16).cuda()
 pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)
 
-question = '<image>\nDescribe the two images in detail.'
+question = '<image>
+Describe the two images in detail.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                history=None, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 question = 'What are the similarities and differences between these two images.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                history=history, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 # multi-image multi-round conversation, separate images (多图多轮对话，独立图像)
 pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
@@ -401,17 +422,21 @@ pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat1
 pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)
 num_patches_list = [pixel_values1.size(0), pixel_values2.size(0)]
 
-question = 'Image-1: <image>\nImage-2: <image>\nDescribe the two images in detail.'
+question = 'Image-1: <image>
+Image-2: <image>
+Describe the two images in detail.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                num_patches_list=num_patches_list,
                                history=None, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 question = 'What are the similarities and differences between these two images.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                num_patches_list=num_patches_list,
                                history=history, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 # batch inference, single image per sample (单图批处理)
 pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
@@ -419,13 +444,15 @@ pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat1
 num_patches_list = [pixel_values1.size(0), pixel_values2.size(0)]
 pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)
 
-questions = ['<image>\nDescribe the image in detail.'] * len(num_patches_list)
+questions = ['<image>
+Describe the image in detail.'] * len(num_patches_list)
 responses = model.batch_chat(tokenizer, pixel_values,
                              num_patches_list=num_patches_list,
                              questions=questions,
                              generation_config=generation_config)
 for question, response in zip(questions, responses):
-    print(f'User: {question}\nAssistant: {response}')
+    print(f'User: {question}
+Assistant: {response}')
 
 # video multi-round conversation (视频多轮对话)
 def get_index(bound, fps, max_frame, first_idx=0, num_segments=32):
@@ -463,17 +490,24 @@ def load_video(video_path, bound=None, input_size=448, max_num=1, num_segments=3
 video_path = './examples/red-panda.mp4'
 pixel_values, num_patches_list = load_video(video_path, num_segments=8, max_num=1)
 pixel_values = pixel_values.to(torch.bfloat16).cuda()
-video_prefix = ''.join([f'Frame{i+1}: <image>\n' for i in range(len(num_patches_list))])
+video_prefix = ''.join([f'Frame{i+1}: <image>
+' for i in range(len(num_patches_list))])
 question = video_prefix + 'What is the red panda doing?'
-# Frame1: <image>\nFrame2: <image>\n...\nFrame8: <image>\n{question}
+# Frame1: <image>
+Frame2: <image>
+...
+Frame8: <image>
+{question}
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                num_patches_list=num_patches_list, history=None, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 
 question = 'Describe this video in detail.'
 response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                                num_patches_list=num_patches_list, history=history, return_history=True)
-print(f'User: {question}\nAssistant: {response}')
+print(f'User: {question}
+Assistant: {response}')
 ```
 
 #### Streaming Output
@@ -555,7 +589,9 @@ image_urls=[
 
 images = [load_image(img_url) for img_url in image_urls]
 # Numbering images improves multi-image conversations
-response = pipe((f'Image-1: {IMAGE_TOKEN}\nImage-2: {IMAGE_TOKEN}\ndescribe these two images', images))
+response = pipe((f'Image-1: {IMAGE_TOKEN}
+Image-2: {IMAGE_TOKEN}
+describe these two images', images))
 print(response.text)
 ```
 
@@ -656,7 +692,7 @@ If you find this project useful in your research, please consider citing:
   year={2024}
 }
 @article{gao2024mini,
-  title={Mini-internvl: A flexible-transfer pocket multimodal model with 5\% parameters and 90\% performance},
+  title={Mini-internvl: A flexible-transfer pocket multimodal model with 5% parameters and 90% performance},
   author={Gao, Zhangwei and Chen, Zhe and Cui, Erfei and Ren, Yiming and Wang, Weiyun and Zhu, Jinguo and Tian, Hao and Ye, Shenglong and He, Junjun and Zhu, Xizhou and others},
   journal={arXiv preprint arXiv:2410.16261},
   year={2024}
@@ -675,3 +711,13 @@ If you find this project useful in your research, please consider citing:
   year={2024}
 }
 ```
+
+## Acknowledgement
+
+InternVL is built with reference to the code of the following projects: [OpenAI CLIP](https://github.com/openai/CLIP), [Open CLIP](https://github.com/mlfoundations/open_clip), [CLIP Benchmark](https://github.com/LAION-AI/CLIP_benchmark), [EVA](https://github.com/baaivision/EVA/tree/master), [InternImage](https://github.com/OpenGVLab/InternImage), [ViT-Adapter](https://github.com/czczup/ViT-Adapter), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation), [Transformers](https://github.com/huggingface/transformers), [DINOv2](https://github.com/facebookresearch/dinov2), [BLIP-2](https://github.com/salesforce/LAVIS/tree/main/projects/blip2), [Qwen-VL](https://github.com/QwenLM/Qwen-VL/tree/master/eval_mm), and [LLaVA-1.5](https://github.com/haotian-liu/LLaVA). Thanks for their awesome work!
+
+______________________________________________________________________
+
+Scan the following QR Code, join our WeChat group.
+
+<p align="center"><img width="300" alt="image" src="https://github.com/user-attachments/assets/f776df09-ebba-4fd5-80c2-fec4ff1518be"></p>