PrismaCoder trains open multimodal code models with rubric synthesis
Original title: PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis
The authors also release PrismaCoder-2M and say the rubric helps across model backbones.
PrismaForge builds a hierarchical rubric covering program correctness, construction quality, and rendered visual output, then uses it to filter data and reward reinforcement learning. The authors release PrismaCoder-2M plus PrismaCoder-9B and PrismaCoder-35B. They report that the models lead open results and rival proprietary systems, and post code and data on a project page.
Unconfirmed: Claims of open state of the art and parity with proprietary systems are author-reported.
- The rubric serves as both a corpus filter and an RL reward.
- Released models are PrismaCoder-9B and PrismaCoder-35B.
- PrismaCoder-2M is described as the largest such open dataset.
- Code and data are posted on the project page.
- AI models
- PrismaCoder-9BPrismaCoder-35B
From the original article
1 min read at huggingface.co
Multimodal code intelligence connects the symbolic world of programs with the visual world their execution creates. Interest in this capability is burgeoning, as models are increasingly asked to produce charts, webpages, animations, and semi-structured documents. Recently, proprietary models have pushed far ahead, producing interactive artifacts even from a text description or a screenshot alone.
By Qiushi Sun, Jingyang Gong, Yang Liu · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.