# DLYuanGod/TinyGPT-V

TinyGPT-V: Efficient Multimodal Large Language Model via Small Backbones

Repository: https://github.com/DLYuanGod/TinyGPT-V
Canonical: https://ross.abutalabs.com/products/tinygpt-v
Language: Python
License: BSD-3-Clause
License Family: permissive
Last push: 2026-02-05T02:24:26+00:00

## Health v2 (maintenance only)
Score: 55/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 65, release rhythm 35, longevity 69
- inputs: {"age_days": 979, "days_push": 210, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1316, forks 80 (observed 2026-08-28T04:04:20.835246+00:00)

## What it is
TinyGPT-V is an efficient multimodal large language model built on small backbones (Phi-2 2.7B), combining vision and language capabilities. It provides training code, pretrained checkpoints, and a Hugging Face demo for vision-language tasks.

## Use cases
- run a lightweight multimodal LLM on limited GPU resources
- train a vision-language model with a small backbone
- perform image captioning and visual question answering
- compare against InstructBLIP-style multimodal models
- fine-tune a compact multimodal model in multiple stages
- deploy a small multimodal model via Hugging Face Spaces

## When to choose
- you need a multimodal LLM that runs on modest hardware
- you want a research baseline for efficient vision-language models
- you prefer small backbone models over 7B+ alternatives

## When to avoid
- you need production-grade, well-maintained software
- you require the highest multimodal accuracy regardless of compute cost
- you need commercial support or extensive documentation

## Facets
- artifact type: library
- maturity: experimental
- function: machine-learning, deep-learning, llm-inference, image-processing, nlp
- domain: artificial-intelligence, large-language-models, computer-vision, machine-learning, deep-learning
- platform: python
- tags: multimodal, vision-language-model, small-backbone, phi-2, research-code, huggingface, gpu, linux

## Member repositories
- DLYuanGod/TinyGPT-V (main) score 55

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:20.835246+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:48:20.113264+00:00, confidence not recorded.
  - readme: https://github.com/DLYuanGod/TinyGPT-V (fetched 2026-08-28T04:04:20.835246+00:00, sha 2c778c7083c7)
- Data as of 2026-08-30T08:39:29.467469+00:00.