tckm.yan/aha
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: update supported models documentation with new models and licensing info

Browse Source
This commit is contained in:
XiaoYang
2026-02-06 17:05:08 +08:00
parent f940bc6fe8
commit fd3ebc8d1c
2 changed files with 101 additions and 99 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/supported-models.md
+51 -50
View File
@@ -4,60 +4,53 @@ aha supports a growing collection of state-of-the-art AI models across multiple
## Text Generation ## Text Generation
| Model | Parameters | Description | Use Case | | Model | Parameters | Description | Use Case | License |
|-------|-----------|-------------|----------| |-------|-----------|-------------|----------|---------|
| **Qwen2.5-7B** | 7B | General-purpose LLM | Chat, reasoning, code | | **Qwen2.5-VL-3B** | 3B | Multimodal LLM | Chat, reasoning, vision | [Qwen Research License](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE) |
| **Qwen3** | Various | Latest generation | Advanced reasoning | | **Qwen2.5-VL-7B** | 7B | Multimodal LLM | Chat, reasoning, vision | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM4** | 4B | Efficient lightweight | Edge deployment | | **Qwen3-0.6B** | 0.6B | Latest generation | Advanced reasoning | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM4-0.5B** | 0.5B | Efficient lightweight | Edge deployment | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## Vision & Multimodal ## Vision & Multimodal
| Model | Type | Description | Resolution | | Model | Parameters | Description | Resolution | License |
|-------|------|-------------|------------| |-------|-----------|-------------|------------|---------|
| **Qwen2.5-VL** | Vision-Language | Image understanding | Up to 1024x1024 | | **Qwen2.5-VL-3B** | 3B | Image understanding | Up to 1024x1024 | [Qwen Research License](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE) |
| **Qwen3-VL** | Vision-Language | Enhanced multimodal | Up to 1536x1536 | | **Qwen2.5-VL-7B** | 7B | Image understanding | Up to 1024x1024 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM-V** | Vision-Language | Lightweight vision | Up to 768x768 | | **Qwen3-VL-2B** | 2B | Enhanced multimodal | Up to 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-4B** | 4B | Enhanced multimodal | Up to 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-8B** | 8B | Enhanced multimodal | Up to 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-32B** | 32B | Enhanced multimodal | Up to 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## Speech Recognition (ASR) ## Speech Recognition (ASR)
| Model | Language | Real-time | Speed | | Model | Parameters | Language | Real-time | Speed | License |
|-------|----------|-----------|-------| |-------|-----------|----------|-----------|-------|---------|
| **FunASR-Nano** | Chinese/English | Yes | 16x realtime | | **Fun-ASR-Nano-2512** | 2512M | Chinese/English | Yes | 16x realtime | Not Specified |
| **GLM-ASR-Nano** | Chinese/English | Yes | 32x realtime | | **GLM-ASR-Nano-2512** | 2512M | Chinese/English | Yes | 32x realtime | [MIT](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/mit.md) |
| **Qwen3-ASR-0.6B** | 0.6B | Chinese/English | Yes | Fast | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-ASR-1.7B** | 1.7B | Chinese/English | Yes | Fast | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## OCR ## OCR
| Model | Languages | Type | Strength | | Model | Languages | Type | Strength | License |
|-------|-----------|------|----------| |-------|-----------|------|----------|---------|
| **PaddleOCR-VL** | 80+ | Lightweight | General documents | | **PaddleOCR-VL** | 80+ | Lightweight | General documents | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Hunyuan-OCR** | Chinese | Deep learning | Complex layouts | | **Hunyuan-OCR** | Chinese | Deep learning | Complex layouts | [Tencent Hunyuan Community License](https://huggingface.co/tencent/HunyuanOCR/blob/main/LICENSE) |
| **DeepSeek-OCR** | Multi | Scene text | Natural images | | **DeepSeek-OCR** | Multi | Scene text | Natural images | [MIT](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/mit.md) |
## Audio Processing ## Audio Generation
| Model | Type | Description | | Model | Parameters | Type | Description | License |
|-------|------|-------------| |-------|-----------|------|-------------|---------|
| **VoxCPM** | Voice Codec | Neural audio codec | | **VoxCPM-0.5B** | 0.5B | Voice Codec | Neural audio codec | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **RMBG-2.0** | Background Removal | Voice isolation | | **VoxCPM1.5** | - | Voice Codec | Enhanced voice generation | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## Model Formats ## Image Processing
All models are served in optimized ONNX format for: | Model | Type | Description | License |
|-------|------|-------------|---------|
- **Cross-platform compatibility** - Windows, macOS, Linux | **RMBG-2.0** | Background Removal | Remove image backgrounds | [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/deed.en) |
- **CPU acceleration** - AVX2, NEON, SIMD
- **Edge deployment** - No GPU required
- **Fast inference** - Optimized runtime
## Model Selection
aha automatically selects the best model for each task. To override:
```bash
aha chat "Hello" --model qwen2.5-7b
aha vision --model qwen2.5-vl "Describe this" --image img.jpg
aha asr --model fun-asr-nano audio.wav
```
## Model Sources ## Model Sources
@@ -65,19 +58,26 @@ Models are sourced from:
- [Hugging Face](https://huggingface.co) - Primary model hub - [Hugging Face](https://huggingface.co) - Primary model hub
- [ModelScope](https://modelscope.cn) - Chinese model hub - [ModelScope](https://modelscope.cn) - Chinese model hub
- [GitHub Releases](https://github.com) - Backup releases
## Adding New Models ## Adding New Models
See [Development Guide](./development.md) for instructions on adding new model integrations. See [Development Guide](./development.md) for instructions on adding new model integrations.
## License Compliance
**Important**: Each model has its own license. Please review the model's license before use in production. Some key considerations:
- **Apache 2.0**: Permissive, commercial-friendly
- **MIT**: Permissive, commercial-friendly
- **Qwen Research License**: Research use, may have restrictions
- **Tencent Hunyuan Community License**: Custom license, review terms
- **CC BY-NC 4.0**: Non-commercial only
Always verify license terms before deployment in production environments.
## Model Updates ## Model Updates
Models are regularly updated. Check the [releases](https://github.com/yourusername/aha/releases) for the latest versions. Models are regularly updated. Check the [releases](https://github.com/jhqxxx/aha/releases) for the latest versions.
## License
Each model has its own license. Please review the model's license before use in production.
## Performance Benchmarks ## Performance Benchmarks
@@ -85,8 +85,9 @@ Approximate inference speeds on CPU (M1 Pro):
| Model | Task | Tokens/sec | | Model | Task | Tokens/sec |
|-------|------|------------| |-------|------|------------|
| Qwen2.5-7B | Text | 25-35 | | Qwen3-0.6B | Text | 40-50 |
| Qwen2.5-VL | Vision | 20-30 | | Qwen2.5-VL-3B | Vision | 20-30 |
| FunASR-Nano | ASR | 200-500x | | Qwen3-ASR-0.6B | ASR | 200-500x |
| VoxCPM-0.5B | TTS | Real-time |
*Benchmarks vary by hardware and input size.* *Benchmarks vary by hardware and input size.*
docs/supported-models.zh-CN.md
+50 -49
View File
@@ -4,60 +4,53 @@ aha 支持多个领域的最先进 AI 模型集合。
## 文本生成 ## 文本生成
| 模型 | 参数量 | 描述 | 使用场景 | | 模型 | 参数量 | 描述 | 使用场景 | 开源协议 |
|------|--------|------|----------| |------|--------|------|----------|---------|
| **Qwen2.5-7B** | 7B | 通用大语言模型 | 对话、推理、代码 | | **Qwen2.5-VL-3B** | 3B | 多模态大语言模型 | 对话、推理、视觉 | [Qwen 研究许可协议](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE) |
| **Qwen3** | 多种 | 最新一代 | 高级推理 | | **Qwen2.5-VL-7B** | 7B | 多模态大语言模型 | 对话、推理、视觉 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM4** | 4B | 高效轻量级 | 边缘部署 | | **Qwen3-0.6B** | 0.6B | 最新一代 | 高级推理 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM4-0.5B** | 0.5B | 高效轻量级 | 边缘部署 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## 视觉与多模态 ## 视觉与多模态
| 模型 | 类型 | 描述 | 分辨率 | | 模型 | 参数量 | 描述 | 分辨率 | 开源协议 |
|------|------|------|--------| |------|--------|------|--------|---------|
| **Qwen2.5-VL** | 视觉语言 | 图像理解 | 最高 1024x1024 | | **Qwen2.5-VL-3B** | 3B | 图像理解 | 最高 1024x1024 | [Qwen 研究许可协议](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE) |
| **Qwen3-VL** | 视觉语言 | 增强多模态 | 最高 1536x1536 | | **Qwen2.5-VL-7B** | 7B | 图像理解 | 最高 1024x1024 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **MiniCPM-V** | 视觉语言 | 轻量视觉 | 最高 768x768 | | **Qwen3-VL-2B** | 2B | 增强多模态 | 最高 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-4B** | 4B | 增强多模态 | 最高 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-8B** | 8B | 增强多模态 | 最高 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-VL-32B** | 32B | 增强多模态 | 最高 1536x1536 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## 语音识别 (ASR) ## 语音识别 (ASR)
| 模型 | 语言 | 实时 | 速度 | | 模型 | 参数量 | 语言 | 实时 | 速度 | 开源协议 |
|------|------|------|------| |------|--------|------|------|------|---------|
| **FunASR-Nano** | 中/英 | 是 | 16x 实时 | | **Fun-ASR-Nano-2512** | 2512M | 中/英 | 是 | 16x 实时 | 未标明 |
| **GLM-ASR-Nano** | 中/英 | 是 | 32x 实时 | | **GLM-ASR-Nano-2512** | 2512M | 中/英 | 是 | 32x 实时 | [MIT](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/mit.md) |
| **Qwen3-ASR-0.6B** | 0.6B | 中/英 | 是 | 快速 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Qwen3-ASR-1.7B** | 1.7B | 中/英 | 是 | 快速 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## OCR ## OCR
| 模型 | 语言 | 类型 | 优势 | | 模型 | 语言 | 类型 | 优势 | 开源协议 |
|------|------|------|------| |------|------|------|------|---------|
| **PaddleOCR-VL** | 80+ | 轻量级 | 通用文档 | | **PaddleOCR-VL** | 80+ | 轻量级 | 通用文档 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **Hunyuan-OCR** | 中文 | 深度学习 | 复杂布局 | | **Hunyuan-OCR** | 中文 | 深度学习 | 复杂布局 | [腾讯混元社区许可协议](https://huggingface.co/tencent/HunyuanOCR/blob/main/LICENSE) |
| **DeepSeek-OCR** | 多语言 | 场景文字 | 自然图像 | | **DeepSeek-OCR** | 多语言 | 场景文字 | 自然图像 | [MIT](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/mit.md) |
## 音频处理 ## 语音生成
| 模型 | 类型 | 描述 | | 模型 | 参数量 | 类型 | 描述 | 开源协议 |
|------|------|------| |------|--------|------|------|---------|
| **VoxCPM** | 语音编解码器 | 神经音频编解码 | | **VoxCPM-0.5B** | 0.5B | 语音编解码器 | 神经音频编解码 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
| **RMBG-2.0** | 背景移除 | 语音隔离 | | **VoxCPM1.5** | - | 语音编解码器 | 增强语音生成 | [Apache 2.0](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) |
## 模型格式 ## 图像处理
所有模型均采用优化的 ONNX 格式，提供： | 模型 | 类型 | 描述 | 开源协议 |
|------|------|------|---------|
- **跨平台兼容** - Windows、macOS、Linux | **RMBG-2.0** | 背景移除 | 移除图像背景 | [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/deed.zh-hans) |
- **CPU 加速** - AVX2、NEON、SIMD
- **边缘部署** - 无需 GPU
- **快速推理** - 优化运行时
## 模型选择
aha 会自动为每个任务选择最佳模型。如需覆盖：
```bash
aha chat "你好" --model qwen2.5-7b
aha vision --model qwen2.5-vl "描述这个" --image img.jpg
aha asr --model fun-asr-nano audio.wav
```
## 模型来源 ## 模型来源
@@ -65,28 +58,36 @@ aha asr --model fun-asr-nano audio.wav
- [Hugging Face](https://huggingface.co) - 主模型中心 - [Hugging Face](https://huggingface.co) - 主模型中心
- [ModelScope](https://modelscope.cn) - 中文模型中心 - [ModelScope](https://modelscope.cn) - 中文模型中心
- [GitHub Releases](https://github.com) - 备份发布
## 添加新模型 ## 添加新模型
参见 [开发指南](./development.zh-CN.md) 了解添加新模型集成的说明。 参见 [开发指南](./development.zh-CN.md) 了解添加新模型集成的说明。
## 许可证合规
**重要提示**：每个模型都有自己的许可证。在生产环境使用前请查看模型许可证。一些关键注意事项：
- **Apache 2.0**: 宽松许可，支持商业使用
- **MIT**: 宽松许可，支持商业使用
- **Qwen 研究许可协议**: 研究用途，可能有使用限制
- **腾讯混元社区许可协议**: 自定义许可，请查看条款
- **CC BY-NC 4.0**: 仅限非商业用途
在生产环境部署前，请务必验证许可证条款。
## 模型更新 ## 模型更新
模型定期更新。查看 [releases](https://github.com/jhqxxx/aha/releases) 获取最新版本。 模型定期更新。查看 [releases](https://github.com/jhqxxx/aha/releases) 获取最新版本。
## 许可证
每个模型都有自己的许可证。在生产中使用前请查看模型许可证。
## 性能基准 ## 性能基准
CPU (M1 Pro) 上的近似推理速度： CPU (M1 Pro) 上的近似推理速度：
| 模型 | 任务 | Tokens/秒 | | 模型 | 任务 | Tokens/秒 |
|------|------|-----------| |------|------|-----------|
| Qwen2.5-7B | 文本 | 25-35 | | Qwen3-0.6B | 文本 | 40-50 |
| Qwen2.5-VL | 视觉 | 20-30 | | Qwen2.5-VL-3B | 视觉 | 20-30 |
| FunASR-Nano | ASR | 200-500x | | Qwen3-ASR-0.6B | ASR | 200-500x |
| VoxCPM-0.5B | TTS | 实时 |
*基准测试因硬件和输入大小而异。* *基准测试因硬件和输入大小而异。*