datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
WangchanThaiInstruct_Multi-turn_Conversation_Dataset
WangchanThaiInstruct Multi-turn Conversation Dataset
We create a Thai multi-turn conversation dataset from airesearch/WangchanThaiInstruct (Batch 1) by LLM. It was created from synthetic method using open source LLM in Thai language.
Citation
Thammaleelakul, S., & Phatthiyaphaibun, W. (2024). WangchanThaiInstruct Multi-turn Conversation Dataset [Data set]. Zenodo. https://doi.org/10.5281/zenodo.13132633
or BibTeX
@dataset{thammaleelakul_2024_13132633,
author =… See the full description on the dataset page: https://huggingface.co/datasets/ThaiSyntheticQA/WangchanThaiInstruct_Multi-turn_Conversation_Dataset.Multi-turn_Long-context_Benchmark_for_LLMs
LoopServe: An Adaptive Dual-phase LLM Inference Acceleration System for Multi-Turn Dialogues
Arxiv: https://www.arxiv.org/abs/2507.13681
Huggingface: https://huggingface.co/papers/2507.13681
Introduction
LoopServe Multi-Turn Dialogue Benchmark is a comprehensive evaluation dataset comprising multiple diverse datasets designed to assess large language model performance in realistic conversational scenarios.
Unlike traditional benchmarks that place queries only at the end… See the full description on the dataset page: https://huggingface.co/datasets/TreeAILab/Multi-turn_Long-context_Benchmark_for_LLMs.multiturn-chatknowchat-multi-turn-dialogues
KnowChat: Multi-Turn Human-LLM Dialogues on Knowledge Tasks
KnowChat is a dataset of 705 multi-turn human-LLM conversations collected to validate the KnowSim user simulation framework. It pairs each conversation with pre/post knowledge assessments, self-reported survey ratings, and participant background information, enabling research on information calibration -- how well LLM assistants tailor responses to users with different knowledge levels.
Dataset Summary… See the full description on the dataset page: https://huggingface.co/datasets/yjlee36/knowchat-multi-turn-dialogues.All-CVE-Chat-MultiTurn-1999-2025-Dataset
CVE Chat‑Style Multi‑Turn Cybersecurity Dataset (1999 – 2025)
1. Project Overview
This repository hosts the largest publicly available chat‑style, multi‑turn cybersecurity dataset to date, containing ≈ 300 000 Common Vulnerabilities and Exposures (CVE) records published between 1999 and 2025. Each record has been meticulously parsed, enriched, and converted into a conversational format that is ideal for training and evaluating AI and AI‑Agent systems focused on… See the full description on the dataset page: https://huggingface.co/datasets/Trendyol/All-CVE-Chat-MultiTurn-1999-2025-Dataset.chatalpaca-multiturn-enriched-2
chatalpaca-multiturn-enriched-2
Records in data.jsonl: 7924
Source dataset: BRlkl/chatalpaca-multiturn-enriched
Generated with scenario-guided Samantha multiturn revision
glm-4.7-multiturn-CoT
glm-4.7-multiturn-CoT
Dataset Summary
glm-4.7-multiturn-CoT is a ShareGPT-style multi-turn reasoning distillation dataset generated with GLM-4.7 as the teacher model.
This release focuses on preserving multi-turn dialogue continuity while injecting explicit chain-of-thought style responses in assistant turns.
Key Features
Multi-turn conversation format (human / gpt)
Assistant responses stored as <think>...</think> + final answer
Resume-safe distillation… See the full description on the dataset page: https://huggingface.co/datasets/Jackrong/glm-4.7-multiturn-CoT.multi-turn-chat-sft-50k
Multi-Turn Chat SFT (50K ShareGPT Format)
50,000 multi-turn conversations in ShareGPT format for supervised fine-tuning of chat models.
Format
Standard ShareGPT format — drop-in compatible with LLaMA-Factory, Axolotl, and Unsloth:
{
"conversations": [
{"from": "system", "value": "You are a helpful assistant."},
{"from": "human", "value": "Write a Python function to implement binary search."},
{"from": "gpt", "value": "Here's a clean… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/multi-turn-chat-sft-50k.Multi-Turn-Conversational-SFTCreated by: DataCreator AI
Multi-Domain Multi-Turn Chat Conversations Dataset
A synthetic conversational dataset designed for LLM supervised fine-tuning and chatbot training.
The dataset contains multi-turn dialogues across multiple everyday domains such as travel, banking, health, programming, and customer interactions. Conversations are structured in OpenAI chat fine-tuning format, making the dataset directly usable in modern fine-tuning pipelines.
Dataset Overview… See the full description on the dataset page: https://huggingface.co/datasets/DataCreatorAI/Multi-Turn-Conversational-SFT.multiturn-feedback
MultiTurn Feedback Dataset
Multi-turn conversation feedback dataset with sparse and dense annotations.
Dataset Description
This dataset contains human feedback annotations for paper "User Feedback in Human-LLM Dialogues:
A Lens to Understand Users But Noisy as a Learning Signal". It includes two evaluation subsets:
Sparse: 75 conversations from LMSYS-Chat-1M with sparse feedback
Dense: 74 conversations from LMSYS-Chat-1M + 34 WildChat with dense feedback
Labels… See the full description on the dataset page: https://huggingface.co/datasets/yuhan-nlp/multiturn-feedback.BFCL-V4-Parallel-Multi-Turn
BFCL V4 Parallel Multi-Turn
Flattened current-turn rows from BFCL v4 multi-turn trajectories for decentralized multi-agent function-calling experiments.
Source data comes from the official Berkeley Function Calling Leaderboard v4 data and possible-answer files.
Fields
id
official_category
task_type
user_prompt
function
ground_truth
turn_index
Categories
multi_turn_base_step
multi_turn_long_context_step
multi_turn_miss_func_step… See the full description on the dataset page: https://huggingface.co/datasets/OpenMLRL/BFCL-V4-Parallel-Multi-Turn.Creative_Writing_Multiturn_Enhanced
Creative Writing Multiturn — Enhanced Edition ✨
High-quality creative writing dataset with regenerated responses using StepFun's Step-3.5-Flash model.
This dataset is an enhanced version of Dampfinchen/Creative_Writing_Multiturn, where all final AI responses have been regenerated using stepfun/step-3.5-flash with a carefully engineered system prompt designed to produce literary-quality creative writing.
What Changed
Original human prompts preserved — All user… See the full description on the dataset page: https://huggingface.co/datasets/Crownelius/Creative_Writing_Multiturn_Enhanced.agentforge-multiturn-toolcall
AgentForge-MultiTurn-ToolCall-5k
A commercial-grade, synthetic, multi-turn agentic tool-calling dataset
for supervised fine-tuning (SFT) of LLMs on agent trajectories. 5,000 conversations,
18,481 tool calls, 30.5 % include genuine error-recovery branches — the
capability most under-represented in existing open datasets.
Why this dataset exists
Most open tool-calling corpora (xLAM, Gorilla, ToolBench, Hermes-Function-Calling)
are dominated by single-turn… See the full description on the dataset page: https://huggingface.co/datasets/voxozi/agentforge-multiturn-toolcall.Refactor-Dialogue-1.4k-Multi-turn-Refactoring-Conversations
Refactor-Dialogue-1.4k — Multi-turn Refactoring Conversations
Synthetic dataset for fine-tuning coding-focused LLMs on multi-turn
refactoring dialogues. Generated with
Dataset Generator —
an open-source pipeline for building high-quality training data.
Overview
1,414 multi-turn conversations across 3 refactoring categories. Each example
is a 4-message dialogue: user pastes real code → assistant refactors with a
short explanation → user follows up with a constraint or… See the full description on the dataset page: https://huggingface.co/datasets/AronDaron/Refactor-Dialogue-1.4k-Multi-turn-Refactoring-Conversations.curatorkit-testrun-Multiturn
curatorkit-testrun-Multiturn
Built using CuratorKIT — provenance-grounded curation and synthesis for LLM post-training.
Method
multiturn
Backend
litellm
Model
openai/Qwen/Qwen2.5-0.5B-Instruct
Formats
alpaca, sharegpt
Artifact
dataset
Published
2026-08-28 10:04 UTC
Usage
from datasets import load_dataset
ds = load_dataset("ram-lexsi/curatorkit-testrun-Multiturn", "alpaca")
GLM-Multiturn-CoT
OctoMed/GLM-Multiturn-CoT
Multi-turn chain-of-thought conversations converted to OctoMed format for SFT training.
Source
Derived from Jackrong/glm-4.7-multiturn-CoT
by Jackrong. All credit for the original data collection and
distillation goes to the original authors.
Format
Each example contains:
conversations: list of {from, value} turns (human / gpt), with <think> reasoning blocks in gpt turns
responses: the final gpt turn repeated for compatibility with… See the full description on the dataset page: https://huggingface.co/datasets/OctoMed/GLM-Multiturn-CoT.kuza_sft_multiturn
Kuza SFT Multi-turn
Supervised fine-tuning data for Kuza, an offline agricultural assistant
for smallholder farmers and agricultural extension workers in East Africa
(English and Swahili). This repository is one of four Kuza SFT datasets.
Dataset description
Hand-authored four-turn English and Swahili dialogs. The first assistant turn asks a clarifying question (crop, location, symptom); the second gives concrete farm advice and avoids invented pesticide or… See the full description on the dataset page: https://huggingface.co/datasets/kuzaai/kuza_sft_multiturn.all-cve-chat-multiturn-1999-2025
CVE Chat‑Style Multi‑Turn Cybersecurity Dataset (1999 – 2025)
1. Project Overview
This repository hosts the largest publicly available chat‑style, multi‑turn cybersecurity dataset to date, containing ≈ 300 000 Common Vulnerabilities and Exposures (CVE) records published between 1999 and 2025. Each record has been meticulously parsed, enriched, and converted into a conversational format that is ideal for training and evaluating AI and AI‑Agent systems focused on… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/all-cve-chat-multiturn-1999-2025.MultiTurn-Chat-MT-Bench
SEA-MTBench
SEA-MTBench evaluates a model's ability to engage in multi-turn (2 turns) conversations and respond in ways that align with human needs. We use gpt-4-1106-preview as the judge model and compare against gpt-3.5-turbo-0125 as the baseline model. It is based on MT-Bench and was manually translated by native speakers for Indonesian (id), Javanese (jv), Sundanese (su), and Vietnamese (vi). The Thai split of this dataset uses MT-Bench Thai from the ThaiLLM leaderboard.… See the full description on the dataset page: https://huggingface.co/datasets/aisingapore/MultiTurn-Chat-MT-Bench.marx-multi-turn
Overview
The Our-Dataset Marx Corpus is a curated multi-turn conversation dataset designed for training and fine-tuning language models to speak and reason in the analytical, historically grounded voice of Karl Marx.
It aims to provide a serious, philosophically coherent representation of Marxian thought rather than a parody or meme.
Each dialogue consists of 5 turns between a modern interlocutor and an assistant persona, Our-LLM, which embodies Marx’s style of reasoning from… See the full description on the dataset page: https://huggingface.co/datasets/imadreamerboy/marx-multi-turn.sliderl-multi-turn-rollouts
SlideRL Multi-Turn Rollouts
Multi-turn rollout trajectories from the SlideRL environment across 6 different models. Each trajectory captures an agent interacting with the SlideRL slide-generation environment over multiple tool-use turns.
Models
File
Model
Episodes
Description
claude_opus_4_6_trajectories.json
Claude Opus 4.6
48
Anthropic Claude Opus 4.6
claude_sonnet_4_6_trajectories.json
Claude Sonnet 4.6
48
Anthropic Claude Sonnet 4.6… See the full description on the dataset page: https://huggingface.co/datasets/KarthikRagunathAnandaKumar/sliderl-multi-turn-rollouts.stanford-encyclopedia-of-philosophy_chat_multi_turn_mistral_largeThis dataset is essentially identical to the Stanford Encyclopedia of Philosophy Chat Multi-turn Dataset, with one key difference: it uses Mistral Large 2 for conversation generation instead of LLaMA 3.1 70B.
All other aspects, including format, statistics, and intended use, remain the same as the original dataset.
aka-llama-korean-dataset-multiturn-raw
Aka-LLAMA Korean Multi-Turn Dataset (Raw)
This dataset is a raw version of a multi-turn Korean conversation dataset generated using kordinal. It is designed for research and development in Korean natural language processing (NLP), specifically in multi-turn dialogue generation.
License
This dataset is released under the CC BY-NC 4.0 license. It is strictly for non-commercial research and educational purposes. Commercial usage is prohibited.
Additionally, some data… See the full description on the dataset page: https://huggingface.co/datasets/snupilab/aka-llama-korean-dataset-multiturn-raw.All-CVE-Chat-MultiTurn-1999-2025-Dataset
CVE Chat‑Style Multi‑Turn Cybersecurity Dataset (1999 – 2025)
1. Project Overview
This repository hosts the largest publicly available chat‑style, multi‑turn cybersecurity dataset to date, containing ≈ 300 000 Common Vulnerabilities and Exposures (CVE) records published between 1999 and 2025. Each record has been meticulously parsed, enriched, and converted into a conversational format that is ideal for training and evaluating AI and AI‑Agent systems focused on… See the full description on the dataset page: https://huggingface.co/datasets/ukcli/All-CVE-Chat-MultiTurn-1999-2025-Dataset.gsm8k_multiturnThe "socratic" version of GSM8K has the model reflect and ask itself sub-questions about the initial question, before coming to a final answer.
This dataset reformats the socratic GSM8K version into a multi-turn conversation, where the sub-questions are asked by the user rather than being self-asked by the model.
DiscoverLLM-multiturn-preferences
DiscoverLLM: Multi-turn Preference Dataset
Multi-turn dialogue data with scored candidate completions, produced by best-of-N
synthesis over the DiscoverLLM user simulator
(paper · project page).
Each example is a single turn of a simulated user–assistant conversation with one of
several candidate assistant responses and an associated reward score, intended for
offline DPO / GRPO / reward-model training.
Configs
Config
Rows
Task
creative_writing
3,052… See the full description on the dataset page: https://huggingface.co/datasets/kixlab/DiscoverLLM-multiturn-preferences.stanford-encyclopedia-of-philosophy_chat_multi_turn_atheneThis dataset is essentially identical to the Stanford Encyclopedia of Philosophy Chat Multi-turn Dataset, with one key difference: it uses Athene 70B for conversation generation instead of LLaMA 3.1 70B.
All other aspects, including format, statistics, and intended use, remain the same as the original dataset.
chatalpaca-multiturn-enriched-3.5
chatalpaca-multiturn-enriched-3.5
This dataset combines the existing Samantha A10 multiturn corpus with new long-memory and exact-answer specialist conversations.
Splits
train: 18,801 rows (existing, manual-evaluation, and generated rows)
No separate validation split is published; all records remain in train.
Total: 18,801 rows
Composition
Existing source artifact: BRlkl/chatalpaca-multiturn-enriched-2.1
New long-memory rows: 8,000
New arithmetic… See the full description on the dataset page: https://huggingface.co/datasets/BRlkl/chatalpaca-multiturn-enriched-3.5.Synthetic_UAV_Scenario_LLM_MultiTurn_GPS_Navigation
Dataset Card for CJJones/Synthetic_UAV_Scenario_LLM_MultiTurn_GPS_Navigation
The full CJ Jones' synthetic dataset catalog is available at: https://datadeveloper1.gumroad.com
Want more? 🚀 Get the AI Startup Bundle from Gumroad.
Dataset Summary
This dataset contains synthetic multi-turn UAV (Unmanned Aerial Vehicle) flight scenarios with realistic GPS navigation challenges, flight mode transitions, and system diagnostics. The scenarios simulate various flight conditions… See the full description on the dataset page: https://huggingface.co/datasets/CJJones/Synthetic_UAV_Scenario_LLM_MultiTurn_GPS_Navigation.stanford-encyclopedia-of-philosophy_chat_multi_turn
Multi-turn Stanford Encyclopedia of Philosophy Chat Dataset
This dataset is designed for fine-tuning large language models to engage in multi-turn philosophical discussions while adopting the persona of a Philosophy professor named Phil. The resulting model should be able to converse like a university-level philosophy professor, who excels at explanations.
This is a semi-synthetic dataset based on the Stanford Encyclopedia of Philosophy (SEP). It simulates conversations between Phil… See the full description on the dataset page: https://huggingface.co/datasets/ruggsea/stanford-encyclopedia-of-philosophy_chat_multi_turn.