AI Dataset2024
Hisami Suzuki, Satoru Katsumata +4LLM-jp (Center for Large Language Model Research and Development, National Institute of Informatics), National Institute of Informatics
Provides manually curated Japanese instruction pairs (questions and safe reference answers) for improving LLM output safety, covering broad harm categories and regionally sensitive cases. Includes English meta-tags and standard splits for benchmarking and fine-tuning.
AI Dataset2024
Structured dataset for training and evaluating LLM agentic behavior: function-calling conversations, JSON-mode structured outputs, and extraction samples for teaching models to generate tool calls and strict structured responses. Includes single-turn and multi-turn scenarios across several configs.
AI Dataset2024
Provides pre-parsed Parquet snapshots of English and French Wikipedia articles with structured fields (sections, infoboxes, tables, references, images) and credibility signals — optimized for large-scale analysis, retrieval-augmented generation, and model development.
AI Dataset2024
A monthly-updated, hive-partitioned parquet corpus of billions of standard rated chess games from Lichess for data analysis and ML — includes move lists with clock info, player Glicko2 ratings, openings, timestamps, and some Stockfish evals under CC0.
AI Dataset2025
Multi‑modal closed-ended academic benchmark with 2,500 multiple-choice and short-answer exam questions spanning math, natural sciences, and humanities for automated grading. Curated by subject-matter experts, released under MIT, and includes a canary string to help prevent dataset leakage into model training.
AI Dataset2025
A curated dataset of ~30,000 CUDA kernels generated by an agentic pipeline, including reference PyTorch implementations, runtime metrics, NCU/Torch/Clang-Tidy profiles, error messages and correctness labels — released under CC-BY-4.0 for model fine-tuning and offline RL/optimization research.
AI Dataset2025
Provides 1.7M+ synthetic and real infographic charts paired with their tabular data for training and evaluating multimodal models on infographic understanding, chart-to-table extraction, chart code generation, and example-based chart synthesis.
AI Dataset2025
Synthetic English-language dataset of 100k+ B2B SaaS sales conversations with turn-by-turn conversion outcomes, engagement and sales-effectiveness metrics, and 3072-dim embeddings for training conversion-prediction and RL-based conversation models.
AI Dataset2025
A 1,000,000-sample Vietnamese historical conversation dataset in ShareGPT/ChatML format for question-answering and text-generation. Approximately 78% of samples include step-by-step reasoning chains; remaining samples are final-only. Useful for training or evaluating Vietnamese LLMs and chat agents.
AI Dataset2025
Collects ~200,000 human responses to 20 visual/semantic association questions (e.g., Bouba–Kiki), with per-response image options and demographic metadata — useful for cross‑cultural perception and evaluation of multimodal systems, but not guaranteed as a rigorously controlled experimental sample.
AI Dataset2026
Provides 100 real-world, open-ended research tasks paired with expert-written rubrics (around 40 weighted criteria per task) to evaluate long-form, web-browsing research agents on factual accuracy, analysis depth, presentation, and citation quality.
AI Dataset2026
Provides 1.06M web interaction trajectories (state, action, next_state) represented primarily as A11y trees for training browser world models and web agents. Covers diverse real‑web domains, English/Chinese pages, and long contexts (up to 30K tokens); residual PII and dynamic content may limit reproducibility.