skills.rest
eddielueng
eddieluengCOMMUNITY
Explore all skills by this author →
GitHub

llava

Generate conversational answers from user-provided images using a LLaVA vision-language model.

1|Updated Apr 14, 2026
Search Tags:#multimodal#CLIP#instruction following#visual question answering#llava#image chat
One-click install
npx skills add https://github.com/eddielueng/hermes-agent-zh --skill llava-eddielueng
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: llava
Source: https://github.com/eddielueng/hermes-agent-zh/tree/main/optional-skills/mlops/llava
Command: npx skills add https://github.com/eddielueng/hermes-agent-zh --skill llava-eddielueng

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires transformers, torch, pillow, and includes references (resource) components.

What problem does it solve?

LLaVA helps you turn visual inputs (images) into useful natural-language answers, avoiding the manual, trial-and-error process of interpreting photos, diagrams, or image-based documents yourself.

Core Features & Use Cases

  • Image-based conversation: Engage in multi-turn question answering while maintaining context about what’s in the image.
  • Visual question answering (VQA): Ask targeted questions like “What is happening here?” or “How many objects are visible?” and get direct responses.
  • Instruction following for multimodal tasks: Use image prompts to drive captioning, descriptive analysis, and visual instruction tasks.

Example Use Case: Upload a screenshot of a document and ask for the main topic, key details, or a structured summary of what’s shown.

Quick Start

Use the llava skill to answer a visual question about an attached image by asking: “What is the main thing shown in this image, and what details matter most?”

Frequently Asked Questions about llava

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I answer questions about an image using a multimodal model?▼

Visual question answering is performed by loading an LLaVA vision-language model that processes your provided image with a CLIP-based encoder and generates natural-language responses conditioned on both the image and your prompt.

Can I have a multi-turn conversation about what is shown in a photo?▼

Yes, multi-turn image chat is supported, allowing you to engage in conversational question answering while maintaining context about the visual content across multiple interactions.

What is the best way to extract the main topic from a screenshot of a document?▼

The best way to analyze a document screenshot is using instruction following for multimodal tasks, which allows you to ask for the main topic, key details, or a structured summary of the image.

Do I need specific Python dependencies to run visual instruction tasks?▼

Yes, you need to install specific Python dependencies including transformers, torch, and pillow to load the model, preprocess images, and perform multimodal instruction following.

Does visual question answering work for captioning and descriptive analysis?▼

Yes, visual question answering supports captioning and descriptive analysis by using image prompts to drive visual instruction tasks and generate grounded natural-language descriptions.

Related Skills

Explore More Education & Research Skills→
519lab519lab
2

llava

Process image inputs with CLIP and Vicuna/LLaMA for visual conversations.

Official
Advanced
eachlabseachlabs
28

subtitle-generation

Generate SRT and VTT subtitles from video with speech-to-text transcription.

Official
Intermediate
lengoctuong2005lengoctuong2005
5

humanizer

Remove AI-generated patterns from text to improve natural readability.

Community
Advanced
openakitaopenakita
2.0k

openakita/skills@image-understanding

Analyze image content with multimodal AI models like Qwen Vision.

Community
Intermediate