datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
toxic-chat
Update
[01/31/2024] We update the OpenAI Moderation API results for ToxicChat (0124) based on their updated moderation model on on Jan 25, 2024.[01/28/2024] We release an official T5-Large model trained on ToxicChat (toxicchat0124). Go and check it for you baseline comparision![01/19/2024] We have a new version of ToxicChat (toxicchat0124)!
Content
This dataset contains toxicity annotations on 10K user prompts collected from the Vicuna online demo.
We utilize a human-AI… See the full description on the dataset page: https://huggingface.co/datasets/lmsys/toxic-chat.paired-llama-3.2-1b-embeddings-lmsys-chat-1m
Paired Llama 3.2 1B Token Embeddings (LMSYS-Chat-1M)
This dataset contains paired activations corresponding to single token locations extracted from Meta's Llama 3.2 1B Instruct on conversations from LMSYS-Chat-1M.
Embeddings are provided for layers 5 through 14, which capture the most interesting intermediate representations.
This dataset was built to study things like:
Learning different basis for activations at a given layer
Studying if there are cases where position encodes… See the full description on the dataset page: https://huggingface.co/datasets/scaleinvariant/paired-llama-3.2-1b-embeddings-lmsys-chat-1m.lmsys-chat-1m
LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset
This dataset contains one million real-world conversations with 25 state-of-the-art LLMs.
It is collected from 210K unique IP addresses in the wild on the Vicuna demo and Chatbot Arena website from April to August 2023.
Each sample includes a conversation ID, model name, conversation text in OpenAI API JSON format, detected language tag, and OpenAI moderation API tag.
User consent is obtained through the "Terms of… See the full description on the dataset page: https://huggingface.co/datasets/lmsys/lmsys-chat-1m.chatbot_arena_conversations
Chatbot Arena Conversations Dataset
This dataset contains 33K cleaned conversations with pairwise human preferences.
It is collected from 13K unique IP addresses on the Chatbot Arena from April to June 2023.
Each sample includes a question ID, two model names, their full conversation text in OpenAI API JSON format, the user vote, the anonymized user ID, the detected language tag, the OpenAI moderation API tag, the additional toxic tag, and the timestamp.
To ensure the safe release… See the full description on the dataset page: https://huggingface.co/datasets/lmsys/chatbot_arena_conversations.mt_bench_human_judgments
Content
This dataset contains 3.3K expert-level pairwise human preferences for model responses generated by 6 models in response to 80 MT-bench questions.
The 6 models are GPT-4, GPT-3.5, Claud-v1, Vicuna-13B, Alpaca-13B, and LLaMA-13B. The annotators are mostly graduate students with expertise in the topic areas of each of the questions. The details of data collection can be found in our paper.
Agreement Calculation
This Colab notebook shows how to compute the… See the full description on the dataset page: https://huggingface.co/datasets/lmsys/mt_bench_human_judgments.lmsys-chat-1m
LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset
This dataset contains one million real-world conversations with 25 state-of-the-art LLMs.
It is collected from 210K unique IP addresses in the wild on the Vicuna demo and Chatbot Arena website from April to August 2023.
Each sample includes a conversation ID, model name, conversation text in OpenAI API JSON format, detected language tag, and OpenAI moderation API tag.
User consent is obtained through the "Terms of use"… See the full description on the dataset page: https://huggingface.co/datasets/AarushSah/lmsys-chat-1m.sae-activations-llama-3.1-8b-layer19-lmsys-chat-1m
SAE Feature Activations — Llama 3.1 8B Instruct, Layer 19 (LMSYS-Chat-1M)
This dataset contains Sparse Autoencoder (SAE) feature activations extracted from layer 19 of Meta's Llama 3.1 8B Instruct on conversations from LMSYS-Chat-1M.
It also has natural language explainations of features generated by GPT OSS 120B. See subset 4 for details.
The SAE used is Goodfire/Llama-3.1-8B-Instruct-SAE-l19, which decomposes layer-19 residual stream activations into interpretable sparse features.… See the full description on the dataset page: https://huggingface.co/datasets/scaleinvariant/sae-activations-llama-3.1-8b-layer19-lmsys-chat-1m.lmsys-chat-1m-synth
LMSYS-Chat-1M-Synth: Japanese/English Synthetic Conversation Dataset Derived from LMSYS-Chat-1M
This repository contains a series of Japanese and English conversation datasets derived from LMSYS-Chat-1M.
Llama-3.1-LMSYS-Chat-1M-Synth
Utilized in the post-training of Llama-3.1-Swallow-8B-Instruct-v0.1 and Llama-3.1-Swallow-70B-Instruct-v0.1
Gemma-2-LMSYS-Chat-1M-Synth
Utilized in the post-training of Llama-3.1-Swallow-8B-Instruct-v0.3 and Llama-3.1-Swallow-70B-Instruct-v0.3… See the full description on the dataset page: https://huggingface.co/datasets/tokyotech-llm/lmsys-chat-1m-synth.lmsys-chatbot_arena_conversations
Dataset Card for "lmsys-chatbot_arena_conversations"
More Information needed
llama-3.2-1b-instruct-lmsys-chat-1m-activations
Llama 3.2 1B Instruct Activations (LMSYS-Chat-1M)
This dataset contains whole-model residual stream activations extracted from Meta's Llama 3.2 1B Instruct on conversations from LMSYS-Chat-1M.
Each row stores the complete residual stream across all 16 transformer layers for a single prompt — both the full-sequence activations and the final-token activations.
Note: This is a subset, 8% (from 2 workers of 25) of the full dataset. The complete dataset was ~25 TB and huggingface only… See the full description on the dataset page: https://huggingface.co/datasets/scaleinvariant/llama-3.2-1b-instruct-lmsys-chat-1m-activations.lmsys_chat_1m_clean
Cleaning and Categorizing
A few weeks ago, I had the itch to do some data crunching, so I began this project - to clean and classify lmsys-chat-1m. The process was somewhat long and tedious, but here is the quick overview:
1. Removing Pure Duplicate Instructions
The first step was to eliminate pure duplicate instructions. This involved:
Removing whitespace and punctuation.
Ensuring that if two instructions matched after that, only one was retained.
This step… See the full description on the dataset page: https://huggingface.co/datasets/OpenLeecher/lmsys_chat_1m_clean.lmsys-chat-1m-formattedSmolLM-lmsys-mixtureslmsys-chat-enlmsys-chat-1m-qwen2.5-instructlmsys_chat_1m_clean_R1
oumi-ai/lmsys_chat_1m_clean_R1
lmsys_chat_1m_clean_R1 is a text dataset designed to train Conversational Language Models with DeepSeek-R1 level reasoning.
Prompts were pulled from LMSYS and filtered to lmsys_chat_1m_clean, and responses were taken from DeepSeek-R1 without additional filters present.
We release lmsys_chat_1m_clean_R1 to help enable the community to develop the best fully open reasoning model!
lmsys_chat_1m_clean queries with responses generated from… See the full description on the dataset page: https://huggingface.co/datasets/oumi-ai/lmsys_chat_1m_clean_R1.details_lmsys__vicuna-7b-v1.3
Dataset Card for Evaluation run of lmsys/vicuna-7b-v1.3
Dataset Summary
Dataset automatically created during the evaluation run of model lmsys/vicuna-7b-v1.3 on the Open LLM Leaderboard.
The dataset is composed of 64 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard-old/details_lmsys__vicuna-7b-v1.3.lmsys-chat-1m-qwen2-instruct-768lmsys-chat-1m
WildChat-4.8M, tagged by topic
This is a copy of lmsys/lmsys-chat-1m, tagged using OpenAI's conversation topic rubric (Chatterji et al, 2025).
Topic distribution
Overall distribution of topics:
Distribution of topics, filter to English-only queries (to match OpenAI's English-centric rubric):
lmsys-chat-1m
LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset
This dataset contains one million real-world conversations with 25 state-of-the-art LLMs.
It is collected from 210K unique IP addresses in the wild on the Vicuna demo and Chatbot Arena website from April to August 2023.
Each sample includes a conversation ID, model name, conversation text in OpenAI API JSON format, detected language tag, and OpenAI moderation API tag.
User consent is obtained through the "Terms of use"… See the full description on the dataset page: https://huggingface.co/datasets/1-800-SHARED-TASKS/lmsys-chat-1m.gpt-oss-20b-lmsys-layer2-expert-traces-600k-10msps-tar-shards
GPT-OSS-20B LMSYS Layer-2 Expert Power Traces — 600k tokens at 10 MSPS, tar-sharded
This is the tar-sharded version of the 600k-token layer-2 expert power-trace capture. It contains the same source data as the raw run directory, but groups per-token trace/record files by shard to avoid a 1.2M-file Hugging Face repository.
Capture summary
Model: openai/gpt-oss-20b
Target layer: 2
Captured region: target_layer_experts_only
Prompts: LMSYS Chat, English-filtered… See the full description on the dataset page: https://huggingface.co/datasets/masterpieceexternal/gpt-oss-20b-lmsys-layer2-expert-traces-600k-10msps-tar-shards.lmsys-arena-human-preference-55k-thresholdslmsys-chat-1m-chat-formattedlmsys-arena-human-preference-55kLmSys-pref-ft-splitDeepSeek-R1-Distill-Llama-8B-lmsys-openthoughts-tokenizedThis dataset is used for training Sparse Autoencoders (SAEs) to identify reasoning features in Large Language Models (LLMs), as described in the paper I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders.
Code for the paper is available at: https://github.com/AIRI-Institute/SAE-Reasoning
The dataset consists of tokenized text data used for training the SAEs.
dataset_info:
features:
name: tokens
sequence: int64
splits:
name:… See the full description on the dataset page: https://huggingface.co/datasets/andreuka18/DeepSeek-R1-Distill-Llama-8B-lmsys-openthoughts-tokenized.LMSYS-Chat-GPT-5-Chat-Response
🤖 LMSYS-Chat-GPT-5-Chat-Response
The dataset used in Black-Box On-Policy Distillation of Large Language Models paper. Homepage at here.
This dataset is an extension of the LMSYS-Chat-1M-Clean corpus, specifically curated by collecting high-quality, non-refusal responses from the GPT-5-Chat API.
The LMSYS-Chat-1M dataset collects real-world user queries from the Chatbot Arena.
There is no tool calls or reasoning in the GPT-5-Chat response.
💾 Dataset Structure
The dataset… See the full description on the dataset page: https://huggingface.co/datasets/ytz20/LMSYS-Chat-GPT-5-Chat-Response.gpt-oss-20b-lmsys-layer2-expert-traces-600k-10msps
GPT-OSS-20B LMSYS Layer-2 Expert Power Traces — 600k tokens at 10 MSPS
This dataset contains real power/current traces captured from an NVIDIA H100 PCIe system while running GPT-OSS-20B decode on LMSYS prompts. The capture target is layer 2 MoE expert execution only; earlier/later transformer work is executed off-trace to avoid wasting acquisition time.
Capture summary
Model: openai/gpt-oss-20b
Target layer: 2
Captured region: target_layer_experts_only
Prompts:… See the full description on the dataset page: https://huggingface.co/datasets/masterpieceexternal/gpt-oss-20b-lmsys-layer2-expert-traces-600k-10msps.lmsys_chatbot_arena_conversationsdatasource: https://colab.research.google.com/drive/1KdwokPjirkTmpO_P1WByFNFiqxWQquwH
lmsys-chat-1m_deduped
Dataset Card for "lmsys-chat-1m_deduped"
More Information needed