datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
tulu-3-sft-personas-instruction-following
Dataset Descriptions
This dataset contains 29980 examples and is synthetically created to enhance model's capabilities to follow instructions precisely and to satisfy user constraints. The constraints are borrowed from the taxonomy in IFEval dataset.
To generate diverse instructions, we expand the methodology in Ge et al., 2024 by using personas. More details and exact prompts used to construct the dataset can be found in our paper.
Curated by: Allen Institute for AI
Paper: TBD… See the full description on the dataset page: https://huggingface.co/datasets/allenai/tulu-3-sft-personas-instruction-following.instruction_following
Dataset Card for "livebench/instruction_following"
LiveBench is a benchmark for LLMs designed with test set contamination and objective evaluation in mind. It has the following properties:
LiveBench is designed to limit potential contamination by releasing new questions monthly, as well as having questions based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses.
Each question has verifiable, objective ground-truth answers, allowing hard questions… See the full description on the dataset page: https://huggingface.co/datasets/livebench/instruction_following.Nemotron-SFT-Instruction-Following-Chat-v3
Dataset Description:
The Nemotron-Instruction-Following-Chat-v3 dataset is designed to strengthen multi-turn, interactive capabilities, including open-ended chat and precise instruction following.
The chat subset uses human written prompts from sources like lmarena, lmsys, and wildchat as seed prompts. Responses are generated with GLM-5. Multiple responses are sampled from the model and the best response as judged by pairwise comparisons using Qwen3-Nemotron-235B-A22B-GenRM-2603… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-SFT-Instruction-Following-Chat-v3.Nemotron-SFT-Instruction-Following-Chat-v2
Dataset Description:
The Nemotron-Instruction-Following-Chat-v2 dataset is designed to broadly strengthen the model’s interactive capabilities, including open-ended chat and precise instruction following.The dataset is a refreshed version of Nemotron-Instruction-Following-Chat-v1 with synthetic dialogues generated from Kimi-K2-Thinking, GLM-4.6, Qwen3-235B-A22B-Thinking-2507, GPT-OSS-120b, Kimi-K2-Instruct-0905, and Qwen3-235B-A22B-Instruct-2507.
This dataset is ready for commercial… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-SFT-Instruction-Following-Chat-v2.instruction-following-evalInstruction-Following-IFEval
SEA-IFEval
SEA-IFEval evaluates a model's ability to adhere to constraints provided in the prompt, for example beginning a response with a specific word/phrase or answering with a certain number of sections. It is based on IFEval and was manually translated by native speakers for Indonesian, Javanese, Sundanese, Thai, Tagalog, and Vietnamese.
Supported Tasks and Leaderboards
SEA-IFEval is designed for evaluating chat or instruction-tuned large language models (LLMs).… See the full description on the dataset page: https://huggingface.co/datasets/aisingapore/Instruction-Following-IFEval.Nemotron-RL-Instruction-Following-Structured-Outputs-v2
Dataset Description:
Split 1: Direct Generation tests the model’s ability to perform freeform text structured outputs on JSON, YAML, and XML data, varying the complexity and presentation of the schema.
Split 2: Diversified Tasks adds 2 additional output formats: TOML and CSV, while increasing problem types to Direct Extraction from document, Translation between formats, Multistep Translation from known data, Multistep Extraction from unrelated context, Schema-Only Generation for… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-Structured-Outputs-v2.Nemotron-Instruction-Following-Chat-v1
Dataset Description:
The Nemotron-Instruction-Following-Chat-v1 dataset is designed to broadly strengthen the model’s interactive capabilities, spanning open-ended chat, precise instruction following, and reliable structured output generation. It combines refreshed chat data from Nemotron-Post-Training-Dataset-v2 (extended to multi-turn) with synthetic dialogues produced by strong frontier models such as GPT-OSS-120B and Qwen3-235B variants.
This dataset is ready for commercial… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Instruction-Following-Chat-v1.Nemotron-RL-instruction_following
Dataset Description:
The Nemotron-RL-instruction_following is a dataset created by combining prompts from the WildChat-1M dataset (made available under the ODC Attribution License) with instructions from the Open-Instruct code base. The instructions are designed to be easily verifiable, such as requiring responses under 200 words. This makes the dataset well-suited for evaluating and training models on objective instruction adherence.
This dataset is released as part of NVIDIA… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-instruction_following.tulu-3-pref-personas-instruction-following
Dataset Descriptions
This dataset contains 19890 preference examples and is synthetically created to enhance models' precise instruction following capabilities while satisfying several constraints. The dataset containts preference pairs (chosen, reject responses) and can be used for preference tuning methods (e.g., PPO, DPO).
Dataset Construction
To create this dataset, we took a subset of its supervised-tuning version here and convert it into preference dataset.… See the full description on the dataset page: https://huggingface.co/datasets/allenai/tulu-3-pref-personas-instruction-following.Nemotron-RL-instruction_following-structured_outputs
Dataset Description:
The Nemotron-RL-instruction_following-structured_outputs dataset tests the ability of the model to follow output formatting instructions under schema constraints under the JSON format. Each problem consists of three components: The document, output formatting Instruction (Schema), and question. The dataset varies the difficulty of each problem by varying the location of instructions, the comprehensiveness of instructions, the complexity of the schema, and… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-instruction_following-structured_outputs.Nemotron-RL-Instruction-Following-MultiTurnChat-v1
Dataset Description:
The MultiChallenge Dataset is a rigorous benchmark designed to improve large language models in complex multi-turn conversations by explicitly targeting inference memory, instruction retention, version editing, and self-coherence. It employs a unique "model breaking" methodology where tasks are tested against advanced models (Nemotron-Nano-V2 and Qwen3-235B-A22B-Thinking-2507) to expose failure modes. A sample is only accepted into the dataset if the task is… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-MultiTurnChat-v1.nemotron-gym-instruction-following-structured-qwen3.5-122b-131k-opencode-traces
Agent trace dataset
Decoding the literal token IDs
The prompt_token_ids / completion_token_ids / logprobs columns are the
verbatim tokens the serving engine emitted, stored PER AGENT STEP as a
list-of-lists (one inner list per turn). To turn them back into text you MUST
use the exact tokenizer the model was served with — a generic same-family
tokenizer will decode word tokens to garbage.
Served model / tokenizer source: Qwen/Qwen3.5-122B-A10B-FP8
from transformers… See the full description on the dataset page: https://huggingface.co/datasets/open-athena/nemotron-gym-instruction-following-structured-qwen3.5-122b-131k-opencode-traces.Nemotron-RL-Instruction-Following-Calendar-v2
Dataset Description:
The Calendar-Scheduling-Dataset is a multi-turn conversation dataset that can understand natural language scheduling constraints, follow instructions across multiple messages, infer scheduling conflicts and satisfy multiple constraints simultaneously. Each event has constraints around duration (e.g. 45 min) and timing (e.g. should be scheduled after 3pm). The user mentions the events and associated constraints in a random order in a natural conversational… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-Calendar-v2.Nemotron-RL-Instruction-Following-Free-Form-Formatting-v1
Dataset Description:
Teaches the model to follow arbitrary text formatting instructions (bullet styles, numbering, delimiters, heading formats, inline emphasis, web-answer structure, etc.) for targeted chat behaviors. Uses explicit Regex and string matching for the reward signal.
This dataset is ready for commercial or non-commercial uses.
Dataset Owner(s):
NVIDIA Corporation
Dataset Creation Date:
Created on: April 10, 2026
Last Modified on: April… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-Free-Form-Formatting-v1.Nemotron-RL-Instruction-Following-Adversarial-v1
Dataset Description:
The inverseIF dataset focuses on adversarial prompts designed to explicitly conflict with an AI model’s standard training instincts—such as writing code without comments or refusing standard helpfulness norms—across 8 distinct "anti-convention" patterns. Using a targeted "model breaking" methodology, it generates four candidate responses via Nemotron-Nano-V2 or Qwen3-235B-A22B-Thinking-2507 to test if the negative constraint is difficult enough to force a… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-Adversarial-v1.llm-instruction-following-eval
LLM Instruction-Following Evaluation: 256 Models Across 20 Diagnostic Tests
Dataset Summary
This dataset contains comprehensive evaluation results from testing 256 Large Language Models across 20 carefully designed diagnostic instruction-following prompts, totaling 5,120 individual evaluations. The evaluation was conducted on October 14, 2025, using the OpenRouter API.
Paper: When Models Can't Follow: Testing Instruction Adherence Across 256 LLMs
arXiv: 2510.18892… See the full description on the dataset page: https://huggingface.co/datasets/richardyoung/llm-instruction-following-eval.Nemotron-RL-Instruction-Following-Citation-Formatting-v1
Dataset Description:
Teaches the model to cite specific document parts using reference markers like [ref:1], ref:3, etc. Supports single-reference, multi-reference, and inline citations.
This dataset is ready for commercial/non-commercial uses.
Dataset Owner(s):
NVIDIA Corporation
Dataset Creation Date:
Created on: April 10, 2026
Last Modified on: April 10, 2026
Version:
Nemotron-RL-Instruction-Following-CitationFormatting-v1… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Instruction-Following-Citation-Formatting-v1.multi-image-composition-instruction-following
Multi-Image Composition Instruction-Following
A large-scale multimodal dataset for multi-image composition via natural language instruction-following. Each case provides 2-3 input images (characters + scene) along with detailed Chinese instructions to compose them into a single photorealistic output image.
Designed for training and evaluating models on complex image composition tasks that require understanding of character identity preservation, pose generation, scene integration… See the full description on the dataset page: https://huggingface.co/datasets/obaydata/multi-image-composition-instruction-following.instruction-following-hard-sft-100k
Hard Instruction Following SFT (100K)
100,000 ShareGPT conversations where the assistant correctly satisfies multiple simultaneous explicit constraints in a single response. Each example pairs a multi-constraint prompt with a response that honors every constraint without dropping any.
Targets the instruction-following capability measured by IFEval and similar benchmarks.
Motivation
A key failure mode in deployed LLMs is dropping constraints under load — responding… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/instruction-following-hard-sft-100k.nemotron-gym-instruction-following-calendar-qwen3.5-122b-131k-opencode-traces
Agent trace dataset
Decoding the literal token IDs
The prompt_token_ids / completion_token_ids / logprobs columns are the
verbatim tokens the serving engine emitted, stored PER AGENT STEP as a
list-of-lists (one inner list per turn). To turn them back into text you MUST
use the exact tokenizer the model was served with — a generic same-family
tokenizer will decode word tokens to garbage.
Served model / tokenizer source: Qwen/Qwen3.5-122B-A10B-FP8
from transformers… See the full description on the dataset page: https://huggingface.co/datasets/open-athena/nemotron-gym-instruction-following-calendar-qwen3.5-122b-131k-opencode-traces.RL-Instruction-Following-Dataset
RL-Instruction-Following-Dataset 🎯
A Verifiable, Rule-Based Dataset for Reinforcement Learning with Verifiable Rewards (RLVR)
📖 Dataset Card 🚀 Usage ⚖️ License
Overview
This dataset is designed to enhance the Instruction Following capabilities of Large Language Models (LLMs) through Reinforcement Learning (RL). Unlike subjective preference datasets (e.g., standard RLHF), this dataset focuses on Objective, Rule-Based Constraints.
Each entry provides a prompt with… See the full description on the dataset page: https://huggingface.co/datasets/renhuimin/RL-Instruction-Following-Dataset.instruction-following-eval
IFEval - Instruction Following Evaluation Dataset
This dataset is designed for evaluating how well language models follow specific instructions when generating responses. It serves as the default evaluation data for IFEval repository.
Dataset Description
The IFEval dataset contains prompts with specific instructions designed to test language models' ability to follow directions precisely. It is intended for use with the evaluation framework described in the paper… See the full description on the dataset page: https://huggingface.co/datasets/kaleinaNyan/instruction-following-eval.instruction-following-rl-66k
Instruction Following RL 66K
Dataset overview
instruction-following-rl-66k is an English training dataset for instruction-following reinforcement learning (RL/RLVR), containing 66,418 examples. It is derived primarily from AllenAI's IF_multi_constraints_upto5, whose instructions contain up to five verifiable constraints drawn from IFEval and IFBench-Train.
Each record is first validated for its JSON, prompt, and metadata structure. A predefined… See the full description on the dataset page: https://huggingface.co/datasets/wflying/instruction-following-rl-66k.UltraFeedback-instruction_following-preferences
Dataset Card for "UltraFeedback-instruction_following-preferences"
More Information needed
instruction-following-training-pool
Instruction following training pool
Public prompts for writing tasks, many of them carrying a constraint a program can check, from
eight datasets read at the pinned revisions named below and one layer built here from them. The
pool is laid out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 310602 rows, one JSON object per line, with these fields.
Field
What it holds
id
a row identifier unique within this file… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/instruction-following-training-pool.persona_instruction_following_convertedultra-50k-samples-dataset-instruction_followingNemotron-RL-Instruction-Following-Structured-Outputs-v2-prompt-only
Nemotron-RL-Instruction-Following-Structured-Outputs-v2-prompt-only
Prompt-only extraction from nvidia/Nemotron-RL-Instruction-Following-Structured-Outputs-v2.
Files:
prompts.csv: one prompt extraction record per source row. Records include
prompt, separated system_prompt, and structured tools when the source row
defines available tools. Nested values are JSON-encoded inside CSV cells.
summary.md: source row counts, extracted row counts, count deltas, and failed prompt counts.… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-RL-Instruction-Following-Structured-Outputs-v2-prompt-only.Nemotron-Instruction-Following-Chat-and-Knowledge-prompt-only
Instruction Following, Chat and Knowledge Prompt-Only
This dataset combines prompt-only datasets by capability theme for distillation experiments.
It contains 2,235,051 unique prompts from 3,344,905 raw rows;
1,109,854 exact canonical duplicates were removed.
Rows retain the canonical prompt-extraction columns and add source_repo_id for provenance.
Deduplication uses normalized system_prompt, prompt, tools, and schema_str, with the first
row in manifest order retained. Original… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-Instruction-Following-Chat-and-Knowledge-prompt-only.