A hybrid human/GPT-4o pipeline plus two new filtering scores produces a 424k-pair text-rich image instruction dataset that improves fine-tuned multimodal models on OCR and document understanding benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A High-Quality Text-Rich Image Instruction Tuning Dataset via Hybrid Instruction Generation
A hybrid human/GPT-4o pipeline plus two new filtering scores produces a 424k-pair text-rich image instruction dataset that improves fine-tuned multimodal models on OCR and document understanding benchmarks.