Skywork-R1V-38B-AWQ

This is the AWQ quantized version of Skywork-R1V-38B, offering improved inference efficiency while maintaining model quality.

Model Description

Skywork R1V is a pioneering multimodal model with advanced reasoning capabilities through Chain-of-Thought. This quantized version maintains the core strengths of the original model while reducing computational requirements.

For detailed information about the model architecture and capabilities, please refer to the original Skywork-R1V repository and technical report.

Benchmark Results

The AWQ quantized model maintains strong performance across key benchmarks:

Benchmark	Score
MMMU	0.6
MathV	0.59
AIME_2024	0.6
MATH_500	0.83

These results demonstrate that the quantized model preserves the mathematical and multimodal reasoning capabilities of the original model.

Usage

You can use the quantized model with different inference frameworks:

Using VLLM

Python API

import os
from vllm import LLM, SamplingParams
from vllm.entrypoints.chat_utils import load_chat_template

model_name = "Skywork/Skywork-R1V-38B-AWQ"  # or local path
llm = LLM(model_name, 
          dtype='float16', 
          quantization="awq", 

          gpu_memory_utilization=0.85,
          max_model_len=4096,
          trust_remote_code=True,
         )

# Add your inference code here

OpenAI-compatible API Server

MODEL_ID="Skywork/Skywork-R1V-38B-AWQ"  # or local path


CUDA_VISIBLE_DEVICES=0 \
    python -m vllm.entrypoints.openai.api_server \
    --model $MODEL_ID \
    --dtype float16 \
    --quantization awq \
    --port 23334 \
    --max-model-len 12000 \
    --gpu-memory-utilization 0.9 \
    --trust-remote-code

Using LMDeploy

import os
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image

model_path = "Skywork/Skywork-R1V-38B-AWQ"  # or local path

engine_config = TurbomindEngineConfig(cache_max_entry_count=0.75) 
chat_template_config = ChatTemplateConfig(model_name=model_path)
pipe = pipeline(model_path, 
                backend_config=engine_config, 
                chat_template_config=chat_template_config,
               )

# Example: Multimodal inference
image = load_image('table.jpg')
response = pipe(('Describe this image?', image))
print(response.text)

Hardware Requirements

The AWQ quantization reduces the memory footprint compared to the original FP16 model. We recommend:

• At least one GPU with 30GB+ VRAM for inference
• For optimal performance with longer contexts, 40GB+ VRAM is recommended

Citation

If you use this model in your research, please cite:

@article{skywork2025r1v,
  title     = {Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought},
  author    = {Yi Peng, Chris, Xiaokun Wang, Yichen Wei, Jiangbo Pei, Weijie Qiu, Ai Jian, Yunzhuo Hao, Jiachun Pan, Tianyidan Xie, Li Ge, Rongxian Zhuang, Xuchen Song, Yang Liu, Yahui Zhou},
  year      = {2025},
  journal   = {https://github.com/SkyworkAI/Skywork-R1V/blob/main/report/Skywork_R1V.pdf},
  url       = {https://huggingface.co/Skywork/Skywork-R1V-38B}
}

Skywork-R1V-38B-AWQ (中文说明)

这是 Skywork-R1V-38B 的 AWQ 量化版本，提供了更高效的推理性能，同时保持模型质量。

模型描述

Skywork R1V 是一个开创性的多模态模型，通过思维链（Chain-of-Thought）技术具备出色的推理能力。这个量化版本保持了原始模型的核心优势，同时降低了计算需求。

有关模型架构和能力的详细信息，请参阅原始 Skywork-R1V 代码库和技术报告。

基准测试结果

AWQ 量化模型在关键基准测试中保持了强劲的性能：

基准测试	分数
MMMU	0.6
MathV	0.59
AIME_2024	0.6
MATH_500	0.83

这些结果表明，量化模型保留了原始模型的数学和多模态推理能力。

使用方法

您可以使用不同的推理框架来使用这个量化模型：

使用 VLLM

Python API

import os
from vllm import LLM, SamplingParams
from vllm.entrypoints.chat_utils import load_chat_template

model_name = "Skywork/Skywork-R1V-38B-AWQ"  # 或本地路径
llm = LLM(model_name, 
          dtype='float16', 
          quantization="awq", 
          gpu_memory_utilization=0.85,
          max_model_len=4096,
          trust_remote_code=True,
         )

# 在此添加您的推理代码

OpenAI 兼容的 API 服务器

MODEL_ID="Skywork/Skywork-R1V-38B-AWQ"  # 或本地路径

CUDA_VISIBLE_DEVICES=0 \
    python -m vllm.entrypoints.openai.api_server \
    --model $MODEL_ID \
    --dtype float16 \
    --quantization awq \
    --port 23334 \
    --max-model-len 12000 \
    --gpu-memory-utilization 0.9 \
    --trust-remote-code

使用 LMDeploy

import os
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image

model_path = "Skywork/Skywork-R1V-38B-AWQ"  # 或本地路径

engine_config = TurbomindEngineConfig(cache_max_entry_count=0.75) 
chat_template_config = ChatTemplateConfig(model_name=model_path)
pipe = pipeline(model_path, 
                backend_config=engine_config, 
                chat_template_config=chat_template_config,
               )

# 示例：多模态推理
image = load_image('table.jpg')
response = pipe(('描述这个图片？', image))
print(response.text)

硬件要求

与原始 FP16 模型相比，AWQ 量化减少了内存占用。我们建议：

• 至少一块 30GB+ 显存的 GPU 用于推理
• 对于更长上下文的最佳性能，建议使用 40GB+ 显存

引用

如果您在研究中使用此模型，请引用：

@article{skywork2025r1v,
  title     = {Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought},
  author    = {Yi Peng, Chris, Xiaokun Wang, Yichen Wei, Jiangbo Pei, Weijie Qiu, Ai Jian, Yunzhuo Hao, Jiachun Pan, Tianyidan Xie, Li Ge, Rongxian Zhuang, Xuchen Song, Yang Liu, Yahui Zhou},
  year      = {2025},
  journal   = {https://github.com/SkyworkAI/Skywork-R1V/blob/main/report/Skywork_R1V.pdf},
  url       = {https://huggingface.co/Skywork/Skywork-R1V-38B}
}