Dataset · Hugging Face

internlm/CapRL-QA-75K

Name: CapRL-QA-75K
Creator: internlm
License: cc-by-nc-4.0
Keywords: task_categories:image-text-to-text, task_categories:visual-question-answering, language:en, license:cc-by-nc-4.0, size_categories:10K<n<100K, format:parquet, modality:image, modality:text, library:datasets, library:dask, library:polars, library:mlcroissant, arxiv:2509.22647, region:us, CapRL

CapRL 75K QA Training Dataset This dataset is the carefully filtered 75K QA training set used by CapRL to train CapRL-3B, a lightweight image captioning model initialized from Qwen2.5-VL-3B. It contains 75,285 samples, where each image is paired with multiple multiple-choice QA items. The dataset is designed for the two-stage CapRL training objective, where caption quality is evaluated through answerability of visual questions. The QA construction pipeline is fully open-sourced… See the full description on the dataset page: https://huggingface.co/datasets/internlm/CapRL-QA-75K.

View on Hugging FaceReport an issue

Author / organization	internlm
License (as declared)	cc-by-nc-4.0
Downloads	553 (as reported by Hugging Face)
Likes	6
Created	16 April 2026
Last updated at source	16 April 2026

CapRLimage-captioningmultimodalreinforcement-learningverifiable-rewardsqa

Embed this entry

<iframe src="https://silytics.si/embed/item/8432ab3e-b338-4ee6-adcc-886050ab725a" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 3 days

553 now (+11 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.

Connected in the knowledge graph

mradermacher/CapRL-3B-i1-GGUFModel · used to train
mradermacher/CapRL-3B-GGUFModel · used to train
internlm/CapRL-3BModel · used to train

Explore this entry in the graph

Recent changes

16 Oct 2026 Description: CapRL 75K QA Training Dataset This dataset is the carefully filtered 75K QA training set used by CapRL to train CapRL-3B, a lightweight image captioning model initialized from Qwen2.5-VL-3B. It contains 75,285 samples, where each image is p → CapRL 75K QA Training Dataset This dataset is the carefully filtered 75K QA training set used by CapRL to train CapRL-3B, a lightweight image captioning model initialized from Qwen2.5-VL-3B. It contains 75,285 samples, where each image is p

Changes detected by Silytics between refreshes of the source.

More from internlm

internlm/WildClawBenchDataset
internlm/WildClawBench-HarborDataset
internlm/SWE-Fixer-Train-110KDataset
internlm/WildClawBench-TrajectoriesDataset
internlm/Lean-WorkbookDataset
internlm/VC-RewardBenchDataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.