llava

LLaVA - 大型语言与视觉助手

用于对话式图像理解的开源视觉语言模型。

何时使用 LLaVA

适用场景：

• 构建视觉语言聊天机器人
• 视觉问答（VQA）
• 图像描述与标注
• 多轮图像对话
• 视觉指令遵循
• 结合图像的文档理解

指标：

• 23,000+ GitHub stars
• GPT-4V 级能力（目标）
• Apache 2.0 许可证
• 多种模型规模（7B-34B 参数）

改用替代方案：

• GPT-4V：最高质量，基于 API
• CLIP：简单的零样本分类
• BLIP-2：更适合仅做图像标注
• Flamingo：研究用途，非开源

快速开始

安装

# Clone repository
git clone https://github.com/haotian-liu/LLaVA
cd LLaVA

# Install
pip install -e .

基本用法

from llava.model.builder import load_pretrained_model
from llava.mm_utils import get_model_name_from_path, process_images, tokenizer_image_token
from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN
from llava.conversation import conv_templates
from PIL import Image
import torch

# Load model
model_path = "liuhaotian/llava-v1.5-7b"
tokenizer, model, image_processor, context_len = load_pretrained_model(
    model_path=model_path,
    model_base=None,
    model_name=get_model_name_from_path(model_path)
)

# Load image
image = Image.open("image.jpg")
image_tensor = process_images([image], image_processor, model.config)
image_tensor = image_tensor.to(model.device, dtype=torch.float16)

# Create conversation
conv = conv_templates["llava_v1"].copy()
conv.append_message(conv.roles[0], DEFAULT_IMAGE_TOKEN + "\nWhat is in this image?")
conv.append_message(conv.roles[1], None)
prompt = conv.get_prompt()

# Generate response
input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(model.device)

with torch.inference_mode():
    output_ids = model.generate(
        input_ids,
        images=image_tensor,
        do_sample=True,
        temperature=0.2,
        max_new_tokens=512
    )

response = tokenizer.decode(output_ids[0], skip_special_tokens=True).strip()
print(response)

可用模型

模型	参数	显存	质量
LLaVA-v1.5-7B	7B	~14 GB	良好
LLaVA-v1.5-13B	13B	~28 GB	更好
LLaVA-v1.6-34B	34B	~70 GB	最佳

# Load different models
model_7b = "liuhaotian/llava-v1.5-7b"
model_13b = "liuhaotian/llava-v1.5-13b"
model_34b = "liuhaotian/llava-v1.6-34b"

# 4-bit quantization for lower VRAM
load_4bit = True  # Reduces VRAM by ~4×

命令行用法

# Single image query
python -m llava.serve.cli \
    --model-path liuhaotian/llava-v1.5-7b \
    --image-file image.jpg \
    --query "What is in this image?"

# Multi-turn conversation
python -m llava.serve.cli \
    --model-path liuhaotian/llava-v1.5-7b \
    --image-file image.jpg
# Then type questions interactively

Web UI（Gradio）

# Launch Gradio interface
python -m llava.serve.gradio_web_server \
    --model-path liuhaotian/llava-v1.5-7b \
    --load-4bit  # Optional: reduce VRAM

# Access at http://localhost:7860

多轮对话

# Initialize conversation
conv = conv_templates["llava_v1"].copy()

# Turn 1
conv.append_message(conv.roles[0], DEFAULT_IMAGE_TOKEN + "\nWhat is in this image?")
conv.append_message(conv.roles[1], None)
response1 = generate(conv, model, image)  # "A dog playing in a park"

# Turn 2
conv.messages[-1][1] = response1  # Add previous response
conv.append_message(conv.roles[0], "What breed is the dog?")
conv.append_message(conv.roles[1], None)
response2 = generate(conv, model, image)  # "Golden Retriever"

# Turn 3
conv.messages[-1][1] = response2
conv.append_message(conv.roles[0], "What time of day is it?")
conv.append_message(conv.roles[1], None)
response3 = generate(conv, model, image)

常见任务

图像标注

question = "Describe this image in detail."
response = ask(model, image, question)

视觉问答

question = "How many people are in the image?"
response = ask(model, image, question)

目标检测（文本形式）

question = "List all the objects you can see in this image."
response = ask(model, image, question)

场景理解

question = "What is happening in this scene?"
response = ask(model, image, question)

文档理解

question = "What is the main topic of this document?"
response = ask(model, document_image, question)

训练自定义模型

# Stage 1: Feature alignment (558K image-caption pairs)
bash scripts/v1_5/pretrain.sh

# Stage 2: Visual instruction tuning (150K instruction data)
bash scripts/v1_5/finetune.sh

量化（降低显存）

# 4-bit quantization
tokenizer, model, image_processor, context_len = load_pretrained_model(
    model_path="liuhaotian/llava-v1.5-13b",
    model_base=None,
    model_name=get_model_name_from_path("liuhaotian/llava-v1.5-13b"),
    load_4bit=True  # Reduces VRAM ~4×
)

# 8-bit quantization
load_8bit=True  # Reduces VRAM ~2×

最佳实践

1. 从 7B 模型开始 - 质量好，显存可控
2. 使用 4-bit 量化 - 显著降低显存
3. 需要 GPU - CPU 推理极其缓慢
4. 提示词清晰 - 具体的问题能得到更好的回答
5. 多轮对话 - 保持对话上下文
6. Temperature 0.2-0.7 - 平衡创造力与一致性
7. max_new_tokens 512-1024 - 用于详细回复
8. 批处理 - 依次处理多张图像

性能

模型	显存（FP16）	显存（4-bit）	速度（tokens/s）
7B	~14 GB	~4 GB	~20
13B	~28 GB	~8 GB	~12
34B	~70 GB	~18 GB	~5

在 A100 GPU 上

基准测试

LLaVA 在以下项目上取得有竞争力的分数：

• VQAv2：78.5%
• GQA：62.0%
• MM-Vet：35.4%
• MMBench：64.3%

局限

1. 幻觉 - 可能描述图像中不存在的内容
2. 空间推理 - 难以处理精确位置
3. 小号文字 - 难以阅读细小的印刷文字
4. 目标计数 - 对大量目标不够精确
5. 显存要求 - 需要强大的 GPU
6. 推理速度 - 比 CLIP 慢

与框架集成

LangChain

from langchain.llms.base import LLM

class LLaVALLM(LLM):
    def _call(self, prompt, stop=None):
        # Custom LLaVA inference
        return response

llm = LLaVALLM()

Gradio App

import gradio as gr

def chat(image, text, history):
    response = ask_llava(model, image, text)
    return response

demo = gr.ChatInterface(
    chat,
    additional_inputs=[gr.Image(type="pil")],
    title="LLaVA Chat"
)
demo.launch()

资源

• GitHub：https://github.com/haotian-liu/LLaVA ⭐ 23,000+
• 论文：https://arxiv.org/abs/2304.08485
• 演示：https://llava.hliu.cc
• 模型：https://huggingface.co/liuhaotian
• 许可证：Apache 2.0