LYL1015/JarvisEvo
A CVPR 2026 paper presenting a self-evolving agent that edits photos using coordinated editor and evaluator models optimized via reinforcement learning.
Not currently ranked — collecting fresh signals.
star history
JarvisEvo is a research project implementing a photo editing agent that uses a synergistic editor-evaluator architecture. The system employs a multimodal large language model (MLLM) as the core intelligence, trained with RLHF to self-improve through iterative editing and evaluation cycles. The project includes model weights on HuggingFace and a benchmark dataset (ArtEdit-Bench) for evaluating photo editing agents.
Frequently asked
- What is LYL1015/JarvisEvo?
- A CVPR 2026 paper presenting a self-evolving agent that edits photos using coordinated editor and evaluator models optimized via reinforcement learning.
- Is JarvisEvo open source?
- Yes — LYL1015/JarvisEvo is an open-source project tracked on heatdrop.
- What language is JarvisEvo written in?
- LYL1015/JarvisEvo is primarily written in Python.
- How popular is JarvisEvo?
- LYL1015/JarvisEvo has 409 stars on GitHub.
- Where can I find JarvisEvo?
- LYL1015/JarvisEvo is on GitHub at https://github.com/LYL1015/JarvisEvo.