JiuhaiChen/BLIP3o

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1667 stars · 78 forks observed · 2026-10-18

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Official implementation of the BLIP3o-Series, a unified autoregressive-plus-diffusion model for text-to-image generation and editing. It combines an autoregressive model that produces intermediate features with a diffusion model for image synthesis, trained with discrete image token supervision and GRPO reinforcement learning.

Use cases

• generate images from text prompts
• train a unified autoregressive diffusion image generation model
• fine-tune a text-to-image model with instruction tuning data
• apply GRPO reinforcement learning to improve prompt alignment in image generation
• improve text rendering in generated images
• reproduce BLIP3o-NEXT research results
• pretrain a multimodal image generation model on open caption datasets

When to choose

• you need an open-source, fully reproducible text-to-image model with training code and data
• you want to experiment with combining autoregressive and diffusion architectures
• you want to apply RLHF-style GRPO training to image generation models
• you are researching discrete image token supervision for multimodal models

When to avoid

• you just want a plug-and-play image generator without training or research work
• you lack GPU/Slurm infrastructure for large model training
• you need a commercially licensed model (no license is specified)
• you need a stable production-ready library rather than research code

Facets

library · maturity active

machine-learning deep-learning llm-training image-processing artificial-intelligence deep-learning image-processing large-language-models python text-to-image autoregressive-diffusion image-generation grpo reinforcement-learning research-code multimodal linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="JiuhaiChen/BLIP3o")