POINTS-Reader uses a two-stage synthetic-data warm-up plus iterative self-improvement with rule-based filtering to train a 3B vision-language model for document conversion, outperforming larger models on OmniDocBench and Fox.
The future of AI is full of infinite possibilities, but it also comes with numerous challenges and ethical issues
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
POINTS-Reader: Distillation-Free Adaptation of Vision-Language Models for Document Conversion
POINTS-Reader uses a two-stage synthetic-data warm-up plus iterative self-improvement with rule-based filtering to train a 3B vision-language model for document conversion, outperforming larger models on OmniDocBench and Fox.