Lijun Wu

An 8B model can now generate scientific graphics code that rivals or surpasses the output of much larger proprietary models, thanks to a new dataset, benchmark, and reinforcement learning approach.

Forget bigger models: massive gains in document parsing accuracy are still possible through smarter data engineering alone.

Forget tedious fine-tuning: leveraging molecule identifiers as visual prompts unlocks surprisingly powerful zero-shot chemical reaction diagram parsing in VLMs.

Forget scaling laws, targeted data engineering—specifically multi-stage distillation and difficulty-aware sampling—allows an 8B model to outperform larger open-source financial LLMs.

Forget simplistic synthetic data: ChartVerse generates complex charts and reliable reasoning data from scratch, enabling an 8B model to outperform its 30B teacher in chart reasoning.

Open-source multimodal models just leveled up: InternVL3 rivals closed-source titans like GPT-4o by pre-training vision and language together from the start.