EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

doc: update developer guide regarding mllms (#6138)

Browse Source 
Signed-off-by: Xinyuan Tong <justinning0323@outlook.com>
Co-authored-by: XinyuanTong <115166877+JustinTong0323@users.noreply.github.com>
Co-authored-by: Xinyuan Tong <justinning0323@outlook.com>
This commit is contained in:
Mick
2025-05-14 23:13:13 +08:00
committed by GitHub
parent 3e350a931e
commit cd7c8a8de6
4 changed files with 84 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/index.rst
View File

@@ -38,7 +38,7 @@ The core features include:
:caption: Supported Models
supported_models/generative_models.md
supported_models/vision_language_models.md
supported_models/multimodal_language_models.md
supported_models/embedding_models.md
supported_models/reward_models.md
supported_models/support_new_models.md

28
docs/supported_models/multimodal_language_models.md Normal file
View File

@@ -0,0 +1,28 @@
# Multimodal Language Models
These models accept multi-modal inputs (e.g., images and text) and generate text output. They augment language models
with multimodal encoders.
## Example launch Command
```shell
python3 -m sglang.launch_server \
--model-path meta-llama/Llama-3.2-11B-Vision-Instruct \ # example HF/local path
--host 0.0.0.0 \
--port 30000 \
```
## Supporting Metrics
| Model Family (Variants) | Example HuggingFace Identifier | Chat Template | Description |
|----------------------------|--------------------------------------------|------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| **Qwen-VL** (Qwen2 series) | `Qwen/Qwen2.5-VL-7B-Instruct` | `qwen2-vl` | Alibaba’s vision-language extension of Qwen; for example, Qwen2.5-VL (7B and larger variants) can analyze and converse about image content. |
| **DeepSeek-VL2** | `deepseek-ai/deepseek-vl2` | `deepseek-vl2` | Vision-language variant of DeepSeek (with a dedicated image processor), enabling advanced multimodal reasoning on image and text inputs. |
| **Janus-Pro** (1B, 7B) | `deepseek-ai/Janus-Pro-7B` | `janus-pro` | DeepSeek’s open-source multimodal model capable of both image understanding and generation. Janus-Pro employs a decoupled architecture for separate visual encoding paths, enhancing performance in both tasks. |
| **MiniCPM-V / MiniCPM-o** | `openbmb/MiniCPM-V-2_6` | `minicpmv` | MiniCPM-V (2.6, ~8B) supports image inputs, and MiniCPM-o adds audio/video; these multimodal LLMs are optimized for end-side deployment on mobile/edge devices. |
| **Llama 3.2 Vision** (11B) | `meta-llama/Llama-3.2-11B-Vision-Instruct` | `llama_3_vision` | Vision-enabled variant of Llama 3 (11B) that accepts image inputs for visual question answering and other multimodal tasks. |
| **LLaVA** (v1.5 & v1.6) | *e.g.* `liuhaotian/llava-v1.5-13b` | `vicuna_v1.1` | Open vision-chat models that add an image encoder to LLaMA/Vicuna (e.g. LLaMA2 13B) for following multimodal instruction prompts. |
| **LLaVA-NeXT** (8B, 72B) | `lmms-lab/llava-next-72b` | `chatml-llava` | Improved LLaVA models (with an 8B Llama3 version and a 72B version) offering enhanced visual instruction-following and accuracy on multimodal benchmarks. |
| **LLaVA-OneVision** | `lmms-lab/llava-onevision-qwen2-7b-ov` | `chatml-llava` | Enhanced LLaVA variant integrating Qwen as the backbone; supports multiple images (and even video frames) as inputs via an OpenAI Vision API-compatible format. |
| **Gemma 3 (Multimodal)** | `google/gemma-3-4b-it` | `gemma-it` | Gemma 3’s larger models (4B, 12B, 27B) accept images (each image encoded as 256 tokens) alongside text in a combined 128K-token context. |
| **Kimi-VL** (A3B) | `moonshotai/Kimi-VL-A3B-Instruct` | `kimi-vl` | Kimi-VL is a multimodal model that can understand and generate text from images. |

84
docs/supported_models/support_new_models.md
View File

@@ -1,40 +1,59 @@
# How to Support New Models
This document explains how to add support for new language models and vision‐language models (VLMs) in SGLang. It also covers how to test new models and register external implementations.
This document explains how to add support for new language models and multimodal large language models (mllms) in
SGLang. It also covers how to test new models and register external implementations.
## How to Support a new Language Model
To support a new model in SGLang, you only need to add a single file under the [SGLang Models Directory](https://github.com/sgl-project/sglang/tree/main/python/sglang/srt/models). You can learn from existing model implementations and create a new file for your model. For most models, you should be able to find a similar model to start with (e.g., starting from Llama). Also refer how to [port a Model from vLLM to SGLang](#port-a-model-from-vllm-to-sglang)
To support a new model in SGLang, you only need to add a single file under
the [SGLang Models Directory](https://github.com/sgl-project/sglang/tree/main/python/sglang/srt/models). You can learn
from existing model implementations and create a new file for your model. For most models, you should be able to find a
similar model to start with (e.g., starting from Llama). Also refer how
to [port a Model from vLLM to SGLang](#port-a-model-from-vllm-to-sglang)
## How to Support a new Vision-Language model
## How to Support a new Multimodal Large Language Model
To support a new vision-language model (vLM) in SGLang, there are several key components in addition to the standard LLM support:
To support a new multimodal large language model (MLLM) in SGLang, there are several key components in addition to the
standard LLM support:
1. **Register your new model as multimodal**:
Extend `is_multimodal_model` in [model_config.py](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/configs/model_config.py) to return `True` for your model.
Extend `is_multimodal_model`
in [model_config.py](https://github.com/sgl-project/sglang/blob/0ab3f437aba729b348a683ab32b35b214456efc7/python/sglang/srt/configs/model_config.py#L561)
to return `True` for your model.
2. **Process Images**:
Define a new `Processor` class that inherits from `BaseProcessor` and register this processor as your model’s dedicated processor. See [multimodal_processor.py](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/managers/multimodal_processor.py) for more details.
2. **Register a new chat-template**
See [conversation.py](https://github.com/sgl-project/sglang/blob/86a779dbe9e815c02f71ea82574608f6eae016b5/python/sglang/srt/conversation.py)
3. **Handle Image Tokens**:
Implement a `pad_input_ids` function for your new model. In this function, image tokens in the prompt should be expanded and replaced with image-hashes so that SGLang can recognize different images when using `RadixAttention`.
3. **Multimodal Data Processor**:
Define a new `Processor` class that inherits from `BaseMultimodalProcessor` and register this processor as your
model’s dedicated processor.
See [multimodal_processor.py](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/managers/multimodal_processor.py)
for more details.
4. **Replace Vision Attention**:
Replace the multi-headed `Attention` of ViT with SGLang’s `VisionAttention`.
4. **Handle Multimodal Tokens**:
Implement a `pad_input_ids` function for your new model. In this function, multimodal tokens in the prompt should be
expanded (if necessary) and padded with multimodal-data-hashes so that SGLang can recognize different multimodal data
with `RadixAttention`.
You can refer to [Qwen2VL](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/qwen2_vl.py) or other vLM implementations. These models demonstrate how to correctly handle both multimodal and textual inputs.
5. **Adapt to Vision Attention**:
Adapt the multi-headed `Attention` of ViT with SGLang’s `VisionAttention`.
You should test the new vLM locally against Hugging Face models. See the [`mmmu`](https://github.com/sgl-project/sglang/tree/main/benchmark/mmmu) benchmark for an example.
You can refer to [Qwen2VL](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/qwen2_vl.py) or
other mllm implementations. These models demonstrate how to correctly handle both multimodal and textual inputs.
You should test the new MLLM locally against Hugging Face models. See the [
`mmmu`](https://github.com/sgl-project/sglang/tree/main/benchmark/mmmu) benchmark for an example.
## Test the Correctness
### Interactive Debugging
For interactive debugging, compare the outputs of Hugging Face/Transformers and SGLang. The following two commands should give the same text output and very similar prefill logits:
For interactive debugging, compare the outputs of Hugging Face/Transformers and SGLang. The following two commands
should give the same text output and very similar prefill logits:
- Get the reference output:
```bash
python3 scripts/playground/reference_hf.py --model-path [new model] --model-type {text,vlm}
python3 scripts/playground/reference_hf.py --model-path [new model] --model-type {text,mllm}
```
- Get the SGLang output:
```bash
@@ -43,7 +62,10 @@ For interactive debugging, compare the outputs of Hugging Face/Transformers and
### Add the Model to the Test Suite
To ensure the new model is well maintained, add it to the test suite by including it in the `ALL_OTHER_MODELS` list in the [test_generation_models.py](https://github.com/sgl-project/sglang/blob/main/test/srt/models/test_generation_models.py) file, test the new model on your local machine and report the results on demonstrative benchmarks (GSM8K, MMLU, MMMU, MMMU-Pro, etc.) in your PR.
To ensure the new model is well maintained, add it to the test suite by including it in the `ALL_OTHER_MODELS` list in
the [test_generation_models.py](https://github.com/sgl-project/sglang/blob/main/test/srt/models/test_generation_models.py)
file, test the new model on your local machine and report the results on demonstrative benchmarks (GSM8K, MMLU, MMMU,
MMMU-Pro, etc.) in your PR.
This is the command to test a new model on your local machine:
@@ -53,26 +75,29 @@ ONLY_RUN=Qwen/Qwen2-1.5B python3 -m unittest test_generation_models.TestGenerati
## Port a Model from vLLM to SGLang
The [vLLM Models Directory](https://github.com/vllm-project/vllm/tree/main/vllm/model_executor/models) is a valuable resource, as vLLM covers many models. SGLang reuses vLLM’s interface and some layers, making it easier to port models from vLLM to SGLang.
The [vLLM Models Directory](https://github.com/vllm-project/vllm/tree/main/vllm/model_executor/models) is a valuable
resource, as vLLM covers many models. SGLang reuses vLLM’s interface and some layers, making it easier to port models
from vLLM to SGLang.
To port a model from vLLM to SGLang:
- Compare these two files for guidance:
- [SGLang Llama Implementation](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/llama.py)
- [vLLM Llama Implementation](https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/llama.py)
- [SGLang Llama Implementation](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/llama.py)
- [vLLM Llama Implementation](https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/llama.py)
- The major differences include:
- **Replace vLLM’s `Attention` with `RadixAttention`** (ensure you pass `layer_id` to `RadixAttention`).
- **Replace vLLM’s `LogitsProcessor` with SGLang’s `LogitsProcessor`.**
- **Replace the multi-headed `Attention` of ViT with SGLang’s `VisionAttention`.**
- **Replace other vLLM layers** (such as `RMSNorm`, `SiluAndMul`) with SGLang layers.
- **Remove `Sample`.**
- **Change the `forward()` functions** and add a `forward_batch()` method.
- **Add `EntryClass`** at the end.
- **Ensure that the new implementation uses only SGLang components** and does not rely on any vLLM components.
- **Replace vLLM’s `Attention` with `RadixAttention`** (ensure you pass `layer_id` to `RadixAttention`).
- **Replace vLLM’s `LogitsProcessor` with SGLang’s `LogitsProcessor`.**
- **Replace the multi-headed `Attention` of ViT with SGLang’s `VisionAttention`.**
- **Replace other vLLM layers** (such as `RMSNorm`, `SiluAndMul`) with SGLang layers.
- **Remove `Sample`.**
- **Change the `forward()` functions** and add a `forward_batch()` method.
- **Add `EntryClass`** at the end.
- **Ensure that the new implementation uses only SGLang components** and does not rely on any vLLM components.
## Registering an External Model Implementation
In addition to the methods above, you can register your new model with the `ModelRegistry` before launching the server. This allows you to integrate your model without modifying the source code.
In addition to the methods above, you can register your new model with the `ModelRegistry` before launching the server.
This allows you to integrate your model without modifying the source code.
For example:
@@ -101,4 +126,5 @@ launch_server(server_args)
---
By following these guidelines, you can add support for new language models and vision-language models in SGLang and ensure they are thoroughly tested and easily integrated into the system.
By following these guidelines, you can add support for new language models and multimodal large language models in
SGLang and ensure they are thoroughly tested and easily integrated into the system.

28
docs/supported_models/vision_language_models.md
View File

@@ -1,28 +0,0 @@
# Vision Language Models
These models accept multi-modal inputs (e.g., images and text) and generate text output. They augment language models with visual encoders and require a specific chat template for handling vision prompts.
## Example launch Command
```shell
python3 -m sglang.launch_server \
--model-path meta-llama/Llama-3.2-11B-Vision-Instruct \ # example HF/local path
--host 0.0.0.0 \
--port 30000 \
```
## Supporting Matrixs
| Model Family (Variants) | Example HuggingFace Identifier | Chat Template | Description |
|--------------------------------|--------------------------------------------------|----------------------|----------------------------------------------------------------------------------------|
| **Qwen-VL** (Qwen2 series) | `Qwen/Qwen2.5-VL-7B-Instruct` | `qwen2-vl` | Alibaba’s vision-language extension of Qwen; for example, Qwen2.5-VL (7B and larger variants) can analyze and converse about image content. |
| **DeepSeek-VL2** | `deepseek-ai/deepseek-vl2` | `deepseek-vl2` | Vision-language variant of DeepSeek (with a dedicated image processor), enabling advanced multimodal reasoning on image and text inputs. |
| **Janus-Pro** (1B, 7B) | `deepseek-ai/Janus-Pro-7B` | `janus-pro` | DeepSeek’s open-source multimodal model capable of both image understanding and generation. Janus-Pro employs a decoupled architecture for separate visual encoding paths, enhancing performance in both tasks. |
| **MiniCPM-V / MiniCPM-o** | `openbmb/MiniCPM-V-2_6` | `minicpmv` | MiniCPM-V (2.6, ~8B) supports image inputs, and MiniCPM-o adds audio/video; these multimodal LLMs are optimized for end-side deployment on mobile/edge devices. |
| **Llama 3.2 Vision** (11B) | `meta-llama/Llama-3.2-11B-Vision-Instruct` | `llama_3_vision` | Vision-enabled variant of Llama 3 (11B) that accepts image inputs for visual question answering and other multimodal tasks. |
| **Pixtral** (12B, 124B) | `mistral-community/pixtral-12b` | `mistral` | Pixtral is a vision-language model from Mistral AI that can process both text and images. |
| **LLaVA** (v1.5 & v1.6) | *e.g.* `liuhaotian/llava-v1.5-13b` | `vicuna_v1.1` | Open vision-chat models that add an image encoder to LLaMA/Vicuna (e.g. LLaMA2 13B) for following multimodal instruction prompts. |
| **LLaVA-NeXT** (8B, 72B) | `lmms-lab/llava-next-72b` | `chatml-llava` | Improved LLaVA models (with an 8B Llama3 version and a 72B version) offering enhanced visual instruction-following and accuracy on multimodal benchmarks. |
| **LLaVA-OneVision** | `lmms-lab/llava-onevision-qwen2-7b-ov` | `chatml-llava` | Enhanced LLaVA variant integrating Qwen as the backbone; supports multiple images (and even video frames) as inputs via an OpenAI Vision API-compatible format. |
| **Gemma 3 (Multimodal)** | `google/gemma-3-4b-it` | `gemma-it` | Gemma 3’s larger models (4B, 12B, 27B) accept images (each image encoded as 256 tokens) alongside text in a combined 128K-token context. |
| **Kimi-VL** (A3B) | `moonshotai/Kimi-VL-A3B-Instruct` | `kimi-vl` | Kimi-VL is a multimodal model that can understand and generate text from images. |