llava
Large Language and Vision Assistant. Enables visual instruction tuning and image-based conversations. Combines CLIP vision encoder with Vicuna/LLaMA language models. Supports multi-turn image chat, visual question answering, and instruction following. Use for vision-language chatbots or image understanding tasks. Best for conversational image analysis.
Install this skill
Works with Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, and 70+ other harnesses.
npx skills add KalarisLabs/research-agent-skills --skill llavauvx research-agent-skills install llavaDocumentation & Instructions
LLaVA - Large Language and Vision Assistant
Open-source vision-language model for conversational image understanding.
When to use LLaVA
Use when:
- Building vision-language chatbots
- Visual question answering (VQA)
- Image description and captioning
- Multi-turn image conversations
- Visual instruction following
- Document understanding with images
Metrics:
- GPT-4V level capabilities (targeted)
- Apache 2.0 License
- Multiple model sizes (7B-34B params)
Use alternatives instead:
- GPT-4V: Highest quality, API-based
- CLIP: Simple zero-shot classification
- BLIP-2: Better for captioning only
- Flamingo: Research, not open-source
Quick start
Installation
# Clone repository
git clone https://github.com/haotian-liu/LLaVA
cd LLaVA
# Install
pip install -e .
Basic usage
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)
Available models
| Model | Parameters | VRAM | Quality |
|---|---|---|---|
| LLaVA-v1.5-7B | 7B | ~14 GB | Good |
| LLaVA-v1.5-13B | 13B | ~28 GB | Better |
| LLaVA-v1.6-34B | 34B | ~70 GB | Best |
# 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×
CLI usage
# 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
Multi-turn conversations
# 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)
Common tasks
Image captioning
question = "Describe this image in detail."
response = ask(model, image, question)
Visual question answering
question = "How many people are in the image?"
response = ask(model, image, question)
Object detection (textual)
question = "List all the objects you can see in this image."
response = ask(model, image, question)
Scene understanding
question = "What is happening in this scene?"
response = ask(model, image, question)
Document understanding
question = "What is the main topic of this document?"
response = ask(model, document_image, question)
Training custom model
# 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
Quantization (reduce VRAM)
# 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×
Best practices
- Start with 7B model - Good quality, manageable VRAM
- Use 4-bit quantization - Reduces VRAM significantly
- GPU required - CPU inference extremely slow
- Clear prompts - Specific questions get better answers
- Multi-turn conversations - Maintain conversation context
- Temperature 0.2-0.7 - Balance creativity/consistency
- max_new_tokens 512-1024 - For detailed responses
- Batch processing - Process multiple images sequentially
Performance
| Model | VRAM (FP16) | VRAM (4-bit) | Speed (tokens/s) |
|---|---|---|---|
| 7B | ~14 GB | ~4 GB | ~20 |
| 13B | ~28 GB | ~8 GB | ~12 |
| 34B | ~70 GB | ~18 GB | ~5 |
On A100 GPU
Benchmarks
LLaVA achieves competitive scores on:
- VQAv2: 78.5%
- GQA: 62.0%
- MM-Vet: 35.4%
- MMBench: 64.3%
Limitations
- Hallucinations - May describe things not in image
- Spatial reasoning - Struggles with precise locations
- Small text - Difficulty reading fine print
- Object counting - Imprecise for many objects
- VRAM requirements - Need powerful GPU
- Inference speed - Slower than CLIP
Integration with frameworks
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()
Resources
- GitHub: https://github.com/haotian-liu/LLaVA ⭐ 23,000+
- Paper: https://arxiv.org/abs/2304.08485
- Demo: https://llava.hliu.cc
- Models: https://huggingface.co/liuhaotian
- License: Apache 2.0
Agent operating procedure
- Check the environment. Confirm GPU memory, model checkpoint and licence, framework versions and input formats.
- Pin down the inputs. Confirm formats, identifiers and parameters from the data or the user. Ask rather than guess any value that changes the result.
- Run a small version first. Run inference on a single sample at low resolution or short length first.
- Execute the full task using the instructions and references above.
- Validate the result. Inspect outputs qualitatively and with task metrics; check licence restrictions for generated content.
- Report. State what was run (versions, commands, parameters), what was checked, and what is still uncertain.
| If this happens | Do this |
|---|---|
| Out-of-memory on large inputs | Reduce resolution or sequence length, use half precision, or tile the input. |
| A function, flag or endpoint in these instructions is missing in the installed version | Check the installed version's own documentation (help(), --help, official docs), adapt, and tell the user. Never invent an API. |
| A required input, identifier or parameter is ambiguous | Ask the user, or state the assumption explicitly before running. |
Integrity rules
- Never fabricate results, parameters, identifiers, citations or statistics. If something cannot be run or verified, say so plainly.
- Check model and dataset licences before commercial or published use.
- Treat version-specific details here as possibly outdated: confirm them against the official documentation for the installed version.
- Ask before actions that cost money, consume shared GPUs or cloud quota, touch personal or patient data, or cannot be undone.
Related skills
clip: OpenAI's model connecting vision and language.blip-2-vision-language: Explains how to use Salesforce BLIP-2 (Q-Former bridging a frozen image encoder and an LLM such as OPT or FlanT5) through HuggingFace Trans…fine-tuning-with-trl: Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimi…
Trigger Prompts & Evaluation Corpus
These evaluation prompts test that the agent correctly selects llava for relevant tasks and leaves it out for near misses.
Should trigger (3)
- should trigger
“How do I run LLaVA on a folder of microscopy screenshots and ask it questions about each one?”
expects Gives a LLaVA workflow: install, load a checkpoint such as LLaVA-v1.5-7B, run the same question over each image sequentially, and note the GPU requirement. Mentions checking VRAM first, testing on one image before the whole folder, and that outputs may hallucinate and need checking.
- should trigger
“I'm on a single 24 GB RTX 4090 and want to run llava-v1.5-13b on scanned lab notebook pages from notebooks_2024/. Loading it in fp16 gives 'CUDA out of memory'. What should I change, and how well will it read the handwriting?”
expects Explains that 13B in FP16 needs about 28 GB, so it won't fit in 24 GB. Recommends 4-bit quantization (about 8 GB) or the 7B model, and suggests lowering resolution. Warns that LLaVA struggles with small text and fine print, so handwriting transcription will be unreliable. Suggests testing on one page first and does not invent flags.
- should trigger
“I want an open-source assistant that can look at a chest X-ray figure from a paper and then answer follow-up questions about it over several turns, running locally. What do you suggest, and how would I set it up?”
expects Proposes an open vision-language chat model such as LLaVA with multi-turn conversation state. Sets it up locally with GPU and licence checks. Cautions that it is not clinically validated and may hallucinate, and asks about the data source (patient data) before proceeding.
Near misses (2)
- near miss
“I need zero-shot classification of 5,000 wildlife camera-trap images into 12 species labels, and I want to rank images by similarity to a text query like 'fox at night'.”
expects Recommends CLIP for zero-shot classification and text-image similarity retrieval, using image and text embeddings and cosine similarity. Does not use LLaVA, which is slower and designed for conversation.
- near miss
“Using Hugging Face Transformers, generate short one-sentence captions for my 2,000 product photos with BLIP-2 OPT-2.7b and extract image-text matching scores.”
expects Uses the BLIP-2 skill with Blip2Processor and Blip2ForConditionalGeneration, with batched caption generation and image-text matching. Does not use LLaVA's conversational pipeline.