DEJIMA is a web‑scale Japanese image dataset with 3.88 M image–caption pairs and 3.88 M image–question–answer pairs, generated by strict filtering, deduplication, detection‑driven evidence extraction, and LLM‑based caption/VQA under grounding constraints, plus multiple curated variants for each task.
Description in other languages
Statistics
Statistics as of Nov 16, 2026
Hugging Face downloads (last 30 days)
359
Hugging Face likes
6
Listed
Mar 8, 2026
Downloads (30-day window) over time
Show as a table
| Month | DL (30d) |
|---|---|
| Dec 2025 | — |
| Jan 2026 | — |
| Feb 2026 | — |
| Mar 2026 | — |
| Apr 2026 | — |
| May 2026 | 189 |
| Jun 2026 | 292 |
| Jul 2026 | — |
| Aug 2026 | 168 |
| Sep 2026 | 180 |
| Oct 2026 | — |
| Nov 2026 | 359 |
Downloads over the 30 days before the last snapshot of each month.
Popular in the same category
- JMMMU-ProJMMMU1kJMMMU‑Pro is an image‑based Japanese multimodal benchmark created by generating visual questions with a realistic model and human verification, revealing that current open‑source LMMs perform poorly on it and offering a cost‑effective method for future VQA benchmark development.
- JMedBenchColdog23331kJMedBench is a Japanese biomedical LLM benchmark comprising 20 datasets across five tasks (MCQA, NER, STS, etc.) sourced from MedMCQA, PubMedQA, MMLU, and others, each with its own license, and includes a note that translations may contain biases requiring human review.
- vntl-leaderboardlmg-anon857A leaderboard that ranks large language models by their semantic accuracy in translating Japanese visual novels into English, using cosine similarity over 256 samples and reporting chrF scores for comparison with tools like Sugoi Translator, Google Translate, Naver Papago, and Alibaba Translate.
- JAMMEvalllm-jp830JAMMEval is a distilled benchmark of seven Japanese VQA datasets, refined with two rounds of human annotation to remove ambiguities and non‑visual questions, delivering reliable evaluation of vision‑language models on multimodal Japanese tasks.
- auto-wiki-qacl-nagoya747AutoWikiQA is the largest free Japanese QA dataset, providing over 2.3 million handcrafted‑filtered question–answer pairs automatically generated from Wikipedia with Swallow‑MX and LLMs (without rule‑based templates) to support knowledge‑teaching and retrieval‑augmented generation applications.
- nri-fin-reasoningnri-ai707Japanese instruction dataset with 632,636 multi‑turn samples (~6.35 B tokens) and GPT‑OSS‑120b reasoning traces for open‑ended, math, writing, and MCQA tasks across 135 finance topics plus 20 general ones, designed to fine‑tune LLM reasoning in finance.
Resources in similar research fields
- Japanese-Medical-VQA-12mMIL-UT423Japanese Medical VQA 12M is a large‐scale multimodal dataset (≈12 M samples) of Japanese medical images with original and enriched captions in both source language and Japanese, plus generated VQA question‑answer pairs, released in Parquet/Webdataset format and built from Open‑PMC‑18M using InternVL3.5 for enrichment, Qwen3‑30B‑A3B for translation, and GPT‑OSS 120B for VQA generation.Shared: Corpora & datasets・QA & reasoning
ja-vg-vqayahoojapan30Japanese Visual Genome VQA datasetGitHubShared: Corpora & datasets・QA & reasoning- JDocQAshunk031249Japanese document‑based QA dataset (JDocQA) with 5,504 PDF documents and 11,600 Japanese Q&A pairs covering yes/no, factoid, numerical, and open‑ended questions, including images and unanswerable cases.Shared: Corpora & datasets・QA & reasoning
- Jaglellm-jp168Jagle is a ~9.2 million‑instance Japanese multimodal post‑training dataset built from image–text pairs and PDF corpora, used to train LLM‑jp‑4‑VL 9B beta and shown to boost performance on Japanese vision‑language tasks.Shared: Large language models・Corpora & datasets
LookVQAriken-grp8A Gaze-grounded Visual Question Answering Dataset for Clarifying Ambiguous Japanese Questions (LREC-COLING 2024)GitHubShared: Corpora & datasets・QA & reasoning- ja-stackoverflowp1atdev105A Japanese QA dataset derived from Stack Overflow dumps, providing question–answer pairs with markdown‑converted bodies (base64 images replaced), including IDs, scores, tags, and accepted or top‑scoring answer texts, licensed under CC BY‑SA 4.0.Shared: Corpora & datasets・QA & reasoning