Spectra: Multimodal VQA Training Data for Science and Open-World Knowledge

Spectra is a multimodal question-answering training dataset designed for vision-language models. It combines graduate-level science questions from TQA and ScienceQA with open-world knowledge questions from OKVQA and science questions across physics, chemistry, math, and biology from AI2D. The dataset was created by Tamalmajumder and was last updated on April 18, 2026.

Use Cases

• Training vision-language models for graduate-level science question answering based on the TQA and ScienceQA components
• Training models for open-world knowledge visual question answering based on the OKVQA component
• Training models for multimodal science reasoning across physics, chemistry, math, and biology based on the AI2D component
• Creating balanced and contamination-controlled training sets for multimodal reasoning

Strengths

• Combines multiple established QA datasets (TQA, OKVQA, ScienceQA, AI2D) into a single training resource
• Designed to be balanced and contamination-controlled for training
• Last updated on April 18, 2026

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count, file formats, and license information are unknown
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

Tamalmajumder

Freshness

Last updated 2026-04-18 16:20:16