PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis
arXiv cs.CL · 15 Oct 2026, 03:32 AEST
PrismaCoder aims to enhance open multimodal code intelligence by connecting programming with visual execution outputs. It introduces PrismaForge, a data pipeline that improves the quality and diversity of datasets for training models.
Key facts
- Multimodal code intelligence connects the symbolic world of programs with the visual world their execution creates.
- To push open progress further, we start from the data side with PrismaForge, a hierarchical data pipeline centered on a synthesized rubric that decomposes these factors into checkable criteria.
- Leveraging PrismaForge for data synthesis and filtering, we build PrismaCoder-2M, the most diverse open dataset for multimodal code intelligence and the largest to date.
- On the model side, we train PrismaCoder-9B and PrismaCoder-35B on this data and release both.
Read aloud
PrismaCoder aims to enhance open multimodal code intelligence by connecting programming with visual execution outputs. It introduces PrismaForge, a data pipeline that improves the quality and diversity of datasets for training models. Multimodal code intelligence connects the symbolic world of programs with the visual world their execution creates. To push open progress further, we start from the data side with PrismaForge, a hierarchical data pipeline centered on a synthesized rubric that decomposes these factors into checkable criteria. Leveraging PrismaForge for data synthesis and filtering, we build PrismaCoder-2M, the most diverse open dataset for multimodal code intelligence and the largest to date. On the model side, we train PrismaCoder-9B and PrismaCoder-35B on this data and release both.
Sources in this cluster
- PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis
arXiv cs.CL · 15 Oct 2026, 03:32 AEST