baichuan-inc
/

Baichuan-Omni-1d5-Base

Safetensors

omni

custom_code

Model card Files Files and versions Community

fengzi258 commited on Jan 24

Commit

0fad588

verified ·

1 Parent(s): e9acc73

Create README.md

Browse files

Files changed (1) hide show

README.md +995 -0

README.md ADDED Viewed

	@@ -0,0 +1,995 @@

+---
+license: apache-2.0
+---
+<div align="center">
+<img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/logo.png" width="300em" ></img>
+<!-- <img src="https://raw.githubusercontent.com/baichuan-inc/Baichuan-Omni-1.5/refs/heads/main/assets/logo.png" width="300em" ></img>
+<img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/train-pipeline.png" width="300em" ></img> -->
+<!-- <img src="https://github.com/OpenBMB/MiniCPM-o/raw/main/assets/minicpm-o-26-framework-v2.png" width="300em" ></img>  -->
+**A large multimodal model that supports text, image, video, and audio input as well as high-quality text and audio output**
+  <p align="center">
+  Baichuan-Omni-1.5 <a href="https://huggingface.co/baichuan-inc/Baichuan-Omni-1d5">🤗</a> | Baichuan-Omni-1.5-Base <a href="https://huggingface.co/baichuan-inc/Baichuan-Omni-1d5-Base">🤗</a>  |Github <a href="https://github.com/baichuan-inc/Baichuan-Omni-1.5/">📖 </a> | Report <a href="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/baichuan_omni_1_5.pdf">📖</a> | OpenMM-Medical <a href="https://huggingface.co/datasets/baichuan-inc/OpenMM-Medical">🤗</a>
+</p>
+</div>
+<!-- ## 介绍
+**Baichuan-Omni-1.5** 是从 Baichuan-omni 升级的最新的、端到端训练的、支持全模态输入/双模态输出的多模态大模型。该模型使用Qwen2.5-7B昨晚大语言模型基座，可以以端到端方式，接受图像、视频、文本、音频作为输入，并且以可控的方式生成高质量文本和语音。
+- **Baichuan-Omni-1.5-Base**: 为促进全模态大模型发展，我们开源了使用高质量海量数据训练的全模态基座模型。该模型未经SFT指令微调，可塑性强，是**业内首个**开源的**全模态基座模型**。
+- **Baichuan-Omni-1.5**: 基于性能强悍的Baichuan-Omni-1.5-base，使用高质量的全模态对齐数据，进行端到端的多模态指令数据训练。Baichuan-Omni-1.5的纯文本、图像、视频、音频理解能力达到了 GPT-4o-mini 级别。可控音频生成的能力十分强大，在xxx和xxx评测集上取得最高表现。 -->
+## Baichuan-Omni-1.5
+The Baichuan-Omni-1.5 is the latest, top-performing model in the Baichuan-omni series. This model is trained and inferred in an end-to-end manner. Compared with Baichuan-omni, this model has significant improvements in text/image/audio/video understanding and text/audio generation, and supports new features such as controllable real-time voice conversations and multi-modal real-time interactions. The main features of Baichuan-Omni-1.5 include:
+- 🔥 **Possess Multimodal Understanding and Interaction Capabilities.**
+Baichuan-Omni-1.5 not only supports images, videos, text, and audio as input, and generates high-quality text and voice output, but also **supports continuous video and audio streaming, and real-time voice interaction with users**. In OminiBench, a comprehensive evaluation benchmark for omnimodal understanding, Baichuan-Omni-1.5 has achieved the first-class level of the open source community and surpassed GPT-4o-mini.
+- 💪 **Strong Visual Capability.**
+Baichuan-Omni-1.5 has an average score of 73.3 on the OpenCompass list (comprehensive 10 mainstream multimodal evaluation benchmarks). **With the size of 7B, it surpasses mainstream commercial closed-source multimodal large models such as GPT-4o-mini, Gemini 1.5 Pro and Claude 3.5 Sonnet in single-image understanding**. In addition, its video understanding performance is also better than GPT-4V and Claude 3.5 Sonnet and open source omnimodal models.
+- 🚀 **Leading Medical Image Understanding Capabilities.**
+Baichuan-Omni-1.5 achieved the best performance on GMAI-MMBench and Openmm-Medical. Using only 7B LLM, the average score exceeded Qwen2-VL-72b by 3%, i.e. 80.7% v.s 83.8%.
+- 🎙 **Excellent Voice Capabilities.**
+Baichuan-Omni-1.5 **supports high-quality, controllable voice bilingual real-time conversations in Chinese and English**. It **outperforms GPT-4o-realtime** in speech understanding tasks (such as ASR and STT, etc.), and demonstrates **the highest speech generation performance among open source models** in semantic and acoustic evaluation of voice conversations. It also supports advanced capabilities such as emotion/speech rate/style control, voice cloning, and role-playing.
+- 🎬 **Powerful Real-world Understanding and Other Features.**
+Baichuan-Omni-1.5 further optimizes the many visual understanding capabilities of Baichuan-omni. It can process images of any aspect ratio and up to 1.8 million pixels (such as 1344x1344). It scored 68.8 points on RealWorldQA, **surpassing commercial closed-source models such as GPT-4o-mini** and recently open-sourced omnimodal models. It scored 85.6/83.6 on the English/Chinese evaluation subsets of MMBench, respectively, which is also in the first echelon of models with the same size.
+- 💫  **Provides [🤗 Base Model](https://huggingface.co/baichuan-inc/Baichuan-Omni-1d5-Base) and [🤗 Instruct Model](https://huggingface.co/baichuan-inc/Baichuan-Omni-1d5).**
+Baichuan-Omni-1.5-Base is a high-performance foundational omni-modal model in the industry. Based on the powerful base, Baichuan-Omni-1.5 employs high-quality omnimodal alignment data to perform end-to-end multimodal instruction data training.
+**Model Architecture**
+<div align="center">
+<img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/train-pipeline.png", width=80%></img>
+</div>
+<br>
+- **End-to-end Omni-modal Architecture.** We carefully design **multi-stage and end-to-end** progressive training of different modal encoding/decoding modules to make full use of the rich knowledge in different modalities, we expect different modal knowledge to complement each other.
+Notably, the model is fully trained end-to-end using NTP loss in the whole pre-training stage.
+- **High-quality Controllable Audio Solution.** Multimodal system prompts have been redesigned to include traditional text system prompts and **speech system prompts** for specifying model sounds. It provides the flexibility to control voice style through text or speech samples at inference time, and supports advanced capabilities such as end-to-end voice cloning and timbre creation.
+**High-quality Medical Image Evaluation Dataset--Openmm-Medical**
+- We have built a more diverse medical evaluation dataset named **Openmm-Medical** to evaluate large models in medical scenarios.
+- The images in Openmm-Medical come from **42 public medical image datasets**, such as ACRIMA (fundus images), BioMediTech (microscope images), and CoronaHack (X-rays).
+- **Openmm-Medical contains a total of 88,996 images**, and each image is designed as a **multiple-choice question to facilitate the evaluation of different large models.**
+- To promote the development of omnimodal large models in the medical field, we will soon **open** this evaluation dataset.
+### Evaluation
+We sugguest readers to refer to our [**Github**](https://github.com/baichuan-inc/Baichuan-Omni-1.5/) for more details.
+<div align="center">
+<img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/performance.png" , width=80%>
+</div>
+<br>
+<details>
+<summary>click to view</summary>
+#### Pure Text Understanding
+<div align="center">
+    <table style="margin: 0 auto; text-align: center;">
+    <thead>
+        <tr>
+            <th class="tg-c3ow" colspan="7">Comprehensive Tasks</th>
+        </tr>
+    </thead>
+    <tbody>
+    <tr>
+        <td>Model</td>
+        <td>Size</td>
+        <td>MMLU (Acc.)</td>
+        <td>CMMLU (Acc.)</td>
+        <td>AGIEval (Acc.)</td>
+        <td>C-Eval (Acc.)</td>
+        <td>GAOKAO (Acc.)</td>
+    </tr>
+    <tr>
+        <td colspan="7">Proprietary Models</td>
+    </tr>
+    <tr>
+        <td>GPT 4o</td>
+        <td>-</td>
+        <td><b>88.0♢<br></td>
+        <td><b>78.3♢<br></td>
+        <td><b>62.3♢<br></td>
+        <td><b>86.0♢<br></td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>GPT 4o mini</td>
+        <td>-</td>
+        <td>82.0</td>
+        <td>67.6</td>
+        <td>52.2</td>
+        <td>63.6</td>
+        <td>70.8</td>
+    </tr>
+    <tr>
+         <td colspan="7">Open-source Models (Pure text)</td>
+    </tr>
+    <tr>
+        <td>MAP-Neo</td>
+        <td>7B</td>
+        <td>58.2</td>
+        <td>55.1</td>
+        <td>33.9</td>
+        <td>57.5</td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>Qwen1.5-Chat</td>
+        <td>7B</td>
+        <td>61.5</td>
+        <td>68.0</td>
+        <td>39.3</td>
+        <td>68.8</td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>Llama3-Instruct</td>
+        <td>8B</td>
+        <td>67.1</td>
+        <td>51.7</td>
+        <td>38.4</td>
+        <td>50.7</td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>OLMo</td>
+        <td>7B</td>
+        <td>28.4</td>
+        <td>25.6</td>
+        <td>19.9</td>
+        <td>27.3</td>
+        <td>-</td>
+    </tr>
+    <tr>
+         <td colspan="7">Open-source Models (Omni-modal)</td>
+    </tr>
+    <tr>
+        <td>VITA</td>
+        <td>8x7B</td>
+        <td>71.0*</td>
+        <td>46.6</td>
+        <td>46.2*</td>
+        <td>56.7*</td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>71.0</td>
+        <td>75.1</td>
+        <td>47.9</td>
+        <td>65.6</td>
+        <td>57.4</td>
+    </tr>
+    <tr>
+        <td>Baichuan-Omni</td>
+        <td>7B</td>
+        <td>65.3</td>
+        <td>72.2</td>
+        <td>47.7</td>
+        <td>68.9</td>
+        <td>-</td>
+    </tr>
+    <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>65.3</td>
+        <td>63.3</td>
+        <td>50.9</td>
+        <td>61.5</td>
+        <td>56.3</td>
+    </tr>
+    <tr>
+        <td><b>Baichuan-Omni-1.5<br></td>
+        <td>7B</td>
+        <td>72.2</td>
+        <td>75.5</td>
+        <td>54.4</td>
+        <td>73.1</td>
+        <td><b>73.5<br></td>
+    </tr>
+    </tbody>
+    </table>
+</div>
+</details>
+<details>
+<summary>click to view</summary>
+#### Image Understanding
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+      <tr>
+         <th class="tg-c3ow" colspan="9">Multi-choice &amp; Yes-or-No Question</th>
+      </tr>
+    </thead>
+    <tbody>
+      <tr>
+        <td>Model</td>
+        <td>Size</td>
+        <td>MMBench-EN (Acc.)</td>
+        <td>MMbench-CN (Acc.)</td>
+        <td>SEED-IMG (Acc.)</td>
+        <td>MMMU-val (Acc.)</td>
+        <td>HallusionBench (Acc.)</td>
+      </tr>
+      <tr>
+        <td colspan="9">Proprietary Models</td>
+      </tr>
+      <tr>
+        <td>GPT-4o</td>
+        <td>-</td>
+        <td>83.4♢</td>
+        <td>82.1♢</td>
+        <td>-</td>
+        <td><b>69.1♢<br></td>
+        <td><b>55.0♢<br></td>
+      </tr>
+      <tr>
+        <td>GPT-4o-mini</td>
+        <td>-</td>
+        <td>77.7</td>
+        <td>76.9</td>
+        <td>72.3</td>
+        <td>60.0♢</td>
+        <td>46.1♢</td>
+      </tr>
+      <tr>
+        <td colspan="9">Open Source Models (Vision-Language)</td>
+      </tr>
+      <tr>
+        <td>Qwen2-VL-7B</td>
+        <td>7B</td>
+        <td><b>86.4<br></td>
+        <td>81.9</td>
+        <td><b>76.5<br></td>
+        <td>52.7</td>
+        <td>50.6∗</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-Llama3-V 2.5</td>
+        <td>8B</td>
+        <td>76.7</td>
+        <td>73.3</td>
+        <td>72.4</td>
+        <td>45.8∗</td>
+        <td>42.5</td>
+      </tr>
+      <tr>
+        <td colspan="9">Open Source Models (Omni-modal)</td>
+      </tr>
+      <tr>
+        <td>VITA</td>
+        <td>8x7B</td>
+        <td>74.7</td>
+        <td>71.4</td>
+        <td>72.6</td>
+        <td>45.3</td>
+        <td>39.7∗</td>
+      </tr>
+      <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>80.8</td>
+        <td>80.2</td>
+        <td>74.2</td>
+        <td>53.1</td>
+        <td>44.1</td>
+      </tr>
+      <tr>
+        <td>Baichuan-Omni</td>
+        <td>7B</td>
+        <td>76.2</td>
+        <td>74.9</td>
+        <td>74.1</td>
+        <td>47.3</td>
+        <td>47.8</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>83.6</td>
+        <td>81.8</td>
+        <td>75.4</td>
+        <td>51.1</td>
+        <td>50.1</td>
+      </tr>
+      <tr>
+        <td><b>Baichuan-Omni-1.5<br></td>
+        <td>7B</td>
+        <td>85.6</td>
+        <td><b>83.6<br></td>
+        <td>75.7</td>
+        <td>53.9</td>
+        <td>49.7</td>
+      </tr>
+    </tbody>
+  </table>
+</div>
+<br>
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+      <tr>
+        <th class="tg-c3ow" colspan="9">Visual Question Answering</th>
+      </tr>
+    </thead>
+    <tbody>
+      <tr>
+        <td>Model</td>
+        <td>Size</td>
+        <td>RealWorldQA (Acc.)</td>
+        <td>MathVista-mini (Acc.)</td>
+        <td>TextVQA-val (Acc.)</td>
+        <td>ChartQA (Acc.)</td>
+        <td>OCRBench (Acc.)</td>
+      </tr>
+      <tr>
+        <td colspan="8">Proprietary Models</td>
+      </tr>
+      <tr>
+        <td>GPT-4o</td>
+        <td>-</td>
+        <td><b>75.4♢<br></td>
+        <td>63.8♢</td>
+        <td>-</td>
+        <td>85.7♢</td>
+        <td>73.6♢</td>
+      </tr>
+      <tr>
+        <td>GPT-4o-mini</td>
+        <td>-</td>
+        <td>66.3</td>
+        <td>53.4</td>
+        <td>66.8</td>
+        <td>-</td>
+        <td>77.4</td>
+      </tr>
+      <tr>
+        <td colspan="8">Open Source Models (Vision-Language)</td>
+      </tr>
+      <tr>
+        <td>Qwen2-VL-7B</td>
+        <td>7B</td>
+        <td>69.7</td>
+        <td>58.2∗</td>
+        <td><b>84.3∗<br></td>
+        <td>83.0∗</td>
+        <td>84.5∗</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-Llama3-V 2.5</td>
+        <td>8B</td>
+        <td>63.5</td>
+        <td>54.3∗</td>
+        <td>76.6</td>
+        <td>72.0</td>
+        <td>72.5</td>
+      </tr>
+      <tr>
+        <td colspan="8">Open Source Models (Omni-modal)</td>
+      </tr>
+      <tr>
+        <td>VITA</td>
+        <td>8x7B</td>
+        <td>59.0</td>
+        <td>44.9∗</td>
+        <td>71.8</td>
+        <td>76.6</td>
+        <td>68.5∗</td>
+      </tr>
+      <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>66.8</td>
+        <td><b>66.5<br></td>
+        <td>74.9</td>
+        <td>79.6</td>
+        <td>73.3</td>
+      </tr>
+      <tr>
+        <td>Baichuan-Omni</td>
+        <td>7B</td>
+        <td>62.6</td>
+        <td>51.9</td>
+        <td>74.3</td>
+        <td>79.6</td>
+        <td>70.0</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>67.7</td>
+        <td>64.6</td>
+        <td>80.1</td>
+        <td><b>87.6<br></td>
+        <td><b>89.7∗<br></td>
+      </tr>
+       <tr>
+        <td>Baichuan-Omni-1.5 </td>
+        <td>7B</td>
+        <td>68.8</td>
+        <td>63.6</td>
+        <td>83.2</td>
+        <td>84.9</td>
+        <td>84.0</td>
+      </tr>
+    </tbody>
+  </table>
+</div>
+</details>
+<details>
+<summary>click to view</summary>
+#### Video Understanding
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+      <tr>
+        <th colspan="7">General VQA&nbsp;&nbsp;&nbsp;</th>
+      </tr>
+    </thead>
+    <tbody>
+      <tr>
+        <td>Model</td>
+        <td>Size</td>
+        <td># Frames</td>
+        <td>MVBench (Acc.)</td>
+        <td>Egoschema (Acc.)</td>
+        <td>VideoMME (Acc.)</td>
+        <td>Perception-Test (Acc.)</td>
+      </tr>
+      <tr>
+        <td colspan="7">Proprietary Models</td>
+      </tr>
+      <tr>
+        <td>Gemini 1.5 Pro</td>
+        <td>-</td>
+        <td>-</td>
+        <td><b>81.3♢<br></td>
+        <td>63.2*</td>
+        <td><b>75.0♢<br></td>
+        <td>-</td>
+      </tr>
+      <tr>
+        <td>GPT 4o mini</td>
+        <td>-</td>
+        <td>-</td>
+        <td>55.2</td>
+        <td>58.5</td>
+        <td>63.6</td>
+        <td>48.2</td>
+      </tr>
+      <tr>
+        <td>GPT 4o</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
+        <td><b>77.2*<br></td>
+        <td>71.9♢</td>
+        <td>-</td>
+      </tr>
+      <tr>
+        <td>GPT 4V</td>
+        <td>-</td>
+        <td>-</td>
+        <td>43.7♢</td>
+        <td>55.6*</td>
+        <td>59.9♢</td>
+        <td>-</td>
+      </tr>
+      <tr>
+        <td colspan="7">Open-source Models (Vision-language)</td>
+      </tr>
+      <tr>
+        <td>Qwen2-VL-7B</td>
+        <td>7B</td>
+        <td>2 fps (max 768)</td>
+        <td>67.0* | 64.4</td>
+        <td>66.7* | 66.6</td>
+        <td>63.3* | 59.0</td>
+        <td>62.3* | 60.3</td>
+      </tr>
+      <tr>
+        <td>AnyGPT</td>
+        <td>8B</td>
+        <td>48</td>
+        <td>33.2</td>
+        <td>32.1</td>
+        <td>29.8</td>
+        <td>29.1</td>
+      </tr>
+      <tr>
+        <td>VideoLLaMA 2</td>
+        <td>7B</td>
+        <td>16</td>
+        <td>54.6*</td>
+        <td>51.7*</td>
+        <td>46.6*</td>
+        <td>51.4*</td>
+      </tr>
+      <tr>
+        <td>VideoChat2</td>
+        <td>7B</td>
+        <td>16</td>
+        <td>51.1*</td>
+        <td>42.1♢</td>
+        <td>33.7♢</td>
+        <td>47.3♢</td>
+      </tr>
+      <tr>
+        <td>LLaVA-NeXT-Video</td>
+        <td>7B</td>
+        <td>32</td>
+        <td>46.5♢</td>
+        <td>43.9♢</td>
+        <td>33.7♢</td>
+        <td>48.8♢</td>
+      </tr>
+      <tr>
+        <td>Video-LLaVA</td>
+        <td>7B</td>
+        <td>8</td>
+        <td>41.0♢</td>
+        <td>38.4♢</td>
+        <td>39.9♢</td>
+        <td>44.3♢</td>
+      </tr>
+      <tr>
+        <td colspan="7">Open-source Models (Omni-modal)</td>
+      </tr>
+      <tr>
+        <td>VITA</td>
+        <td>8x7B</td>
+        <td>1 fps (max 32)</td>
+        <td>53.4</td>
+        <td>53.9</td>
+        <td>56.1</td>
+        <td>56.2</td>
+      </tr>
+      <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>1 fps (max 32)</td>
+        <td>55.5</td>
+        <td>54.7</td>
+        <td>57.3</td>
+        <td>57.6</td>
+      </tr>
+      <tr>
+        <td>Baichuan-Omni</td>
+        <td>7B</td>
+        <td>1 fps (max 32)</td>
+        <td>60.9</td>
+        <td>58.8</td>
+        <td>58.2</td>
+        <td>56.8</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>1 fps (max 64)</td>
+        <td>58.6</td>
+        <td>50.7</td>
+        <td>63.4</td>
+        <td>66.6</td>
+      </tr>
+      <tr>
+        <td>Baichuan-Omini-1.5</td>
+        <td>7B</td>
+        <td>1 fps (max 32)</td>
+        <td> 63.7 </td>
+        <td> 62.4 </td>
+        <td> 60.1 </td>
+        <td> <b>68.9 <br> </td>
+      </tr>
+    </tbody>
+  </table>
+</div>
+<br>
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+    <tr>
+      <th colspan="7">Open-ended VQA</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td rowspan="2">Model</td>
+      <td rowspan="2">Size</td>
+      <td rowspan="2"># Frames</td>
+      <td colspan="2">ActivityNet-QA</td>
+      <td colspan="2">MSVD-QA</td>
+    </tr>
+    <tr>
+      <td>(Acc.)</td>
+      <td>(Score)</td>
+      <td>(Acc.)</td>
+      <td>(Score)</td>
+    </tr>
+    <tr>
+      <td colspan="7">Proprietary Models</td>
+    </tr>
+    <tr>
+      <td>Gemini 1.5 Pro</td>
+      <td>-</td>
+      <td>-</td>
+      <td>56.7*</td>
+      <td>-</td>
+      <td>-</td>
+      <td>-</td>
+    </tr>
+    <tr>
+      <td>GPT 4o mini</td>
+      <td>-</td>
+      <td>1 fps (max 32)</td>
+      <td>62.1</td>
+      <td>3.1</td>
+      <td>67.5</td>
+      <td>3.3</td>
+    </tr>
+    <tr>
+      <td>GPT 4o</td>
+      <td>-</td>
+      <td>-</td>
+      <td>61.9*</td>
+      <td>-</td>
+      <td>-</td>
+      <td>-</td>
+    </tr>
+    <tr>
+      <td>GPT 4V</td>
+      <td>-</td>
+      <td>-</td>
+      <td>59.5*</td>
+      <td>-</td>
+      <td>-</td>
+      <td>-</td>
+    </tr>
+    <tr>
+      <td colspan="7">Open-source Models (Vision-language)</td>
+    </tr>
+    <tr>
+      <td>Qwen2 VL</td>
+      <td>7B</td>
+      <td>2 fps (max 768)</td>
+      <td>17.4</td>
+      <td>1.9</td>
+      <td>61.1</td>
+      <td>3.5</td>
+    </tr>
+    <tr>
+      <td>VideoLLaMA 2</td>
+      <td>7B</td>
+      <td>16</td>
+      <td>50.2*</td>
+      <td>3.3*</td>
+      <td>70.9*</td>
+      <td>3.8*</td>
+    </tr>
+    <tr>
+      <td>VideoChat2</td>
+      <td>7B</td>
+      <td>16</td>
+      <td>49.1*</td>
+      <td>3.3*</td>
+      <td>70.0*</td>
+      <td>3.9*</td>
+    </tr>
+    <tr>
+      <td>LLaVA-NeXT-Video</td>
+      <td>7B</td>
+      <td>32</td>
+      <td>53.5*</td>
+      <td>3.2*</td>
+      <td>67.4</td>
+      <td>3.4</td>
+    </tr>
+    <tr>
+      <td>Video-LLaVA</td>
+      <td>7B</td>
+      <td>8</td>
+      <td>45.3*</td>
+      <td>3.3*</td>
+      <td>70.7*</td>
+      <td>3.9*</td>
+    </tr>
+    <tr>
+      <td colspan="7">Open-source Models (Omni-modal)</td>
+    </tr>
+    <tr>
+      <td>VITA</td>
+      <td>8x7B</td>
+      <td>1 fps (max 32)</td>
+      <td>55.0</td>
+      <td>3.5</td>
+      <td>63.9</td>
+      <td>3.7</td>
+    </tr>
+    <tr>
+      <td>VITA-1.5</td>
+      <td>7B</td>
+      <td>1 fps (max 32)</td>
+      <td>59.6</td>
+      <td>3.0</td>
+      <td>67.6</td>
+      <td>3.3</td>
+    </tr>
+    <tr>
+      <td>Baichuan-Omni</td>
+      <td>7B</td>
+      <td>1 fps (max 48)</td>
+      <td>58.6</td>
+      <td><b>3.7<br></td>
+      <td>72.2</td>
+      <td> <b>4.0<br> </td>
+    </tr>
+    <tr>
+      <td>MiniCPM-o 2.6</td>
+      <td>7B</td>
+      <td>1 fps (max 64)</td>
+      <td><b>63.0<br></td>
+      <td>3.1</td>
+      <td>73.7</td>
+      <td>3.6</td>
+    </tr>
+    <tr>
+      <td>Baichuan-Omni-1.5</td>
+      <td>7B</td>
+      <td>1 fps (max 48)</td>
+      <td>  62.0</td>
+      <td> 3.1</td>
+      <td> <b> 74.2 <br></td>
+      <td> 3.6</td>
+    </tr>
+  </tbody>
+</table>
+</div>
+</details>
+<details>
+<summary>click to view</summary>
+#### Audio Understanding
+</details>
+<details>
+<summary>click to view</summary>
+#### Speech Generation
+</details>
+<details>
+<summary>click to view</summary>
+#### Omni-modal Understanding
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+      <tr>
+        <th colspan="7">Omni-Undesratnding </th>
+      </tr>
+    <thead>
+    <tbody>
+          <tr>
+          <td>Model</td>
+          <td>Size</td>
+          <td>Image & Audio</td>
+          <td>Image Caption & Audio</td>
+          <td>Image & Audio Transcript</td>
+          <td>Image Caption & Audio Transcript</td>
+          </tr>
+      </thead>
+      <tr>
+        <td colspan="6">Proprietary Models</td>
+      </tr>
+      <tr>
+        <td>GPT4o-mini</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
+        <td>37.0</td>
+        <td>37.7</td>
+      </tr>
+      <tr>
+        <td colspan="6">Open-source Models (Omni-modal)</td>
+      </tr>
+      <tr>
+        <td>VITA</td>
+        <td>8x7B</td>
+        <td>33.1</td>
+        <td>31.8</td>
+        <td>42.0</td>
+        <td>44.2</td>
+      </tr>
+      <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>33.4</td>
+        <td>29.6</td>
+        <td>48.5</td>
+        <td><b>47.2<br></td>
+      </tr>
+      <tr>
+        <td>Baichuan-Omni</td>
+        <td>7B</td>
+        <td>32.2</td>
+        <td>26.5</td>
+        <td>42.6</td>
+        <td>44.2</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>40.5</td>
+        <td>30.8</td>
+        <td><b>53.2<br></td>
+        <td>46.3</td>
+      </tr>
+      <tr>
+        <td><b>Baichuan-Omni-1.5<br></td>
+        <td>7B</td>
+        <td><b>42.9<br></td>
+        <td><b>37.7<br></td>
+        <td>47.9</td>
+        <td>46.9</td>
+      </tr>
+    </tbody>
+  </table>
+</div>
+</details>
+<details>
+<summary>click to view</summary>
+#### Medical Image Understanding Capabilities
+<div align="center">
+  <table style="margin: 0 auto; text-align: center;">
+    <thead>
+        <tr>
+          <th colspan="7">Medical Understanding&nbsp;&nbsp;&nbsp;</th>
+        </tr>
+      </thead>
+      <tbody>
+          <tr>
+          <td>Model</td>
+          <td>Size</td>
+          <td>GMAI-MMB-VAL (Acc.)</td>
+          <td>OpenMM-Medical (Acc.)</td>
+          </tr>
+      </thead>
+      <tr>
+        <td colspan="4">Proprietary Models</td>
+      </tr>
+      <tr>
+        <td>GPT4o-mini</td>
+        <td>-</td>
+        <td>46.4</td>
+        <td>74.3</td>
+      </tr>
+      <tr>
+        <td colspan="4">Open-source Models (Vision-Language)</td>
+      </tr>
+      <tr>
+        <td>Qwen2 VL</td>
+        <td>7B</td>
+        <td>46.3</td>
+        <td>76.9</td>
+      </tr>
+      <tr>
+        <td>Qwen2 VL</td>
+        <td>72B</td>
+        <td><b>50.7<br></td>
+        <td>80.7</td>
+      </tr>
+      <tr>
+        <td colspan="4">Open-source Models (Omni-modal)</td>
+      </tr>
+      <tr>
+        <td>VITA-1.5</td>
+        <td>7B</td>
+        <td>36.7</td>
+        <td>67.1</td>
+      </tr>
+      <tr>
+        <td>MiniCPM-o 2.6</td>
+        <td>7B</td>
+        <td>41.5</td>
+        <td>73.6</td>
+      </tr>
+      <tr>
+        <td><b>Baichuan-Omni-1.5<br></td>
+        <td>7B</td>
+        <td>49.9</td>
+        <td><b>83.8<br></td>
+      </tr>
+    </tbody>
+  </table>
+</div>
+</details>
+## Examples
+<br>
+<div style="display: flex; flex-direction: column; align-items: center;">
+  <img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/pipeline.png" alt="pipeline" style="margin-bottom: 5px;">
+  <img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/math.png" alt="math" style="margin-bottom: 5px;">
+  <img src="https://github.com/baichuan-inc/Baichuan-Omni-1.5/raw/main/assets/fly_bill.png" alt="fly_bill" style="margin-bottom: 5px;">
+</div>
+## 🚀 Quick Start
+We recommend interested scholars to visit our github repo for more details. [**Github**](https://github.com/baichuan-inc/Baichuan-Omni-1.5/)
+### Statement
+- We hereby declare that our team has not developed any applications based on Baichuan-Omni-1.5/Baichuan-Omni-1.5-base models, not on iOS, Android, the web, or any other platform. We strongly call on all users not to use Baichuan-Omni-1.5/Baichuan-Omni-1.5-base models for any activities that harm national / social security or violate the law. Also, we ask users not to use Baichuan-Omni-1.5/Baichuan-Omni-1.5-base models for Internet services that have not undergone appropriate security reviews and filings. We hope that all users can abide by this principle and ensure that the development of technology proceeds in a regulated and legal environment.
+- We have done our best to ensure the compliance of the data used in the model training process. However, despite our considerable efforts, there may still be some unforeseeable issues due to the complexity of the model and data. Therefore, if any problems arise due to the use of Baichuan-Omni-1.5/Baichuan-Omni-1.5-base open-source models, including but not limited to data security issues, public opinion risks, or any risks and problems brought about by the model being misled, abused, spread or improperly exploited, we will not assume any responsibility.
+### License
+The community usage of Baichuan-Omni-1.5/Baichuan-Omni-1.5-base requires adherence to [Apache 2.0](https://github.com/baichuan-inc/Baichuan-Omni-1.5/blob/main/LICENSE) and [Community License for Baichuan-Omni-1.5 Models](https://huggingface.co/baichuan-inc/Baichuan2-7B-Base/resolve/main/Baichuan%202%E6%A8%A1%E5%9E%8B%E7%A4%BE%E5%8C%BA%E8%AE%B8%E5%8F%AF%E5%8D%8F%E8%AE%AE.pdf). The Baichuan-Omni-1.5/Baichuan-Omni-1.5-base models supports commercial use. If you plan to use the Baichuan-Omni-1.5/Baichuan-Omni-1.5-base models or its derivatives for commercial purposes, please ensure that your entity meets the following conditions:
+  1. The Daily Active Users (DAU) of your or your affiliate's service or product is less than 1 million.
+  2. Neither you nor your affiliates are software service providers or cloud service providers.
+  3. There is no possibility for you or your affiliates to grant the commercial license given to you, to reauthorize it to other third parties without Baichuan's permission.
+Upon meeting the above conditions, you need to submit the application materials required by the Baichuan-Omni-1.5 Model Community License Agreement via the following contact email: [email protected]. Once approved, Baichuan will hereby grant you a non-exclusive, global, non-transferable, non-sublicensable, revocable commercial copyright license.
+<!-- ### Citation
+If you find our work helpful, please consider citing our papers 📝 and liking this project ❤️！
+```bib
+@article{
+} -->
+```