datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
json-mode-evaljson-mode-eval-extended
JSON-Mode-eval extended
This is a dataset that measures LLM capabilities at extracting data from natural language following a JSON Schema.
It was generated by manually cleaning and normalizing json-mode-eval by Nous-Research, which resulted in json-mode-eval-cleaned, ensuring that every schema enforces non-empty constraints and allow no additional keys on the top level.
We then prompt Gemini 2.5 Pro for additional 10 samples per schema, filtering for outputs that are valid according… See the full description on the dataset page: https://huggingface.co/datasets/eth-sri/json-mode-eval-extended.json-mode-reasoningjson-mode-singleturnjson-mode-eval-cleaned
JSON-Mode-eval extended
This is a dataset that measures LLM capabilities at extract data from natural language following a JSON Schema.
It was generated by manually cleaning and normalizing json-mode-eval by Nous-Research.
This dataset was used for evaluation in the paper Constrained Decoding of Diffusion LLMs with Context-Free Grammars. You can find the corresponding evaluation code on the project GitHub Repository.
Example Usage
from datasets import load_dataset… See the full description on the dataset page: https://huggingface.co/datasets/eth-sri/json-mode-eval-cleaned.json-mode-verifiablejson-mode-agentic-reasoningjson-mode-eval-rgxjson-mode-agenticjson-mode-evalThis is an extended version of https://huggingface.co/datasets/NousResearch/json-mode-eval .
Warning: the output currently is placeholder! This dataset should only be used for testing efficiency!
reasoning-sft-interstellarninja-json-mode-reasoning-160K
json-mode-reasoning (converted)
Converted version of interstellarninja/json-mode-reasoning, filtered to 20,474 rows with valid <think> reasoning traces.
Format
Each row has three columns:
input — list of dicts [{"role": "system/user", "content": "..."}, ...] (conversation turns ending on the last user turn, includes system prompt with JSON schema)
response — assistant response string with <think> reasoning block followed by JSON output
source — fixed as… See the full description on the dataset page: https://huggingface.co/datasets/AmanPriyanshu/reasoning-sft-interstellarninja-json-mode-reasoning-160K.json-mode-evaljson-mode-dpo-promptsjson-mode-eval-rgxaugmented-json-mode-eval
JSON Mode Evaluation Dataset (Augmented)
Dataset Description
This is an augmented version of the NousResearch/json-mode-eval dataset.
The original dataset contains examples for evaluating models' ability to follow JSON schema instructions, and this augmented version includes additional variations with different formatting of the schema prompt.
This dataset has been filtered to remove samples containing JSON schema features that are not supported by xgrammar and llguidance… See the full description on the dataset page: https://huggingface.co/datasets/squeezebits/augmented-json-mode-eval.json-mode-agenticjson-mode-singleturnjson-mode-newgenallyson-json-mode-sharegpt-v0.1updatedclasification_model_v0_4_5rc_predict_yes_no_dataset_for_json_3110350_staedion