Run image-based conversations and visual question answering via CLI.
npx skills add https://github.com/MarbleSodas/Mavis --skill llava-marblesodas
Please help me install this Agent Skill. Skill: llava Source: https://github.com/MarbleSodas/Mavis/tree/main/optional-skills/mlops/llava Command: npx skills add https://github.com/MarbleSodas/Mavis --skill llava-marblesodas
LLaVA helps turn visual inputs like photos or screenshots into actionable answers, so you can understand what’s in an image without manually inspecting it.
Run a single-image question-and-answer query against a selected model using the CLI with your image file and prompt text.
High-intent search queries and answers about installing and using this skill.
Visual question answering with PyTorch uses a vision-language model to process image inputs and text prompts. You provide an image file and a question, and the model performs multimodal reasoning to return actionable text answers.
Multi-turn image chat maintains conversational context across follow-up questions about a screenshot. The vision-language model tokenizes multimodal prompts, allowing you to ask targeted questions like counting objects or describing scenes sequentially.
Image instruction following requires transformers, PyTorch, and Pillow to run locally. These dependencies handle image preprocessing and transformer-based multimodal prompting to execute visual tasks like captioning and document understanding.
Deploying multimodal QA for image triage utilizes a Gradio interface or interactive CLI for local deployment. This stack processes customer photos to identify elements and generate actionable answers without manual inspection.
Multimodal reasoning supports both scene-level analysis and document understanding with image inputs. By using transformer-based PyTorch inference, the model follows visual instructions to caption scenes or extract text from documents.
Image-to-text conversational models require image preprocessing to format visual inputs for tokenized multimodal prompting. This aligns the image data with the vision-language model stack, enabling accurate visual question answering.
founderphantomRun multimodal image understanding and conversation with a vision-language model.
Supporter09Generate image captions and answer visual questions with BLIP-2.
b-open-ioAnalyzes images and answers questions about design layouts and visual aesthetics.
davpatel605-beepAnalyze images and generate conversational responses using LLaVA vision-language models.
