Measl 150K V1

MeasL-150K-V1 is an instruction-tuning corpus containing 152,517 examples for measurement-grounded visual reasoning, aligned with the MeasL-Bench-V1 evaluation protocol. It was created by author 'kepeng' and released on Hugging Face for reproducible non-commercial research, with a last recorded update in May 2026.

Use Cases

• Fine-tuning vision-language models for measurement-grounded reasoning based on the described instruction-tuning corpus.
• Benchmarking model performance on visual reasoning tasks aligned with the MeasL-Bench-V1 protocol.
• Researching the intersection of quantitative measurement and visual question answering using the provided examples.

Strengths

• Contains 152,517 instruction-tuning examples, providing a substantial corpus for model training.
• Aligned with a specific evaluation protocol (MeasL-Bench-V1), which suggests a structured benchmark.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality and content require manual inspection after download.

Provenance

Source

kepeng (author) via Hugging Face.

Collection Method

Released as the official PRSIMVL instruction-tuning corpus; specific gathering method is not detailed.

Freshness

Last updated 2026-05-27 11:58:53; freshness should be verified.