OpenAssistant
OpenAssistant Conversations Dataset
A human conversation tree dataset created by over 13,000 volunteers from the OpenAssistant community, containing 88,838 conversation records, covering 35 languages, making it the largest open-source human-generated conversation dataset in the world, licensed under the Apache 2.0 license.
Dataset Highlights
The world's largest open-source human-generated dialogue dataset, providing high-quality training data for conversational AI research
Human-Generated Content
All dialogues are written by real humans, not machine-generated, ensuring the naturalness, diversity, and authenticity of the language, making it an ideal data source for training high-quality dialogue models.
Dialogue Tree Structure
The data is organized in a dialogue tree format, with multiple different reply branches for the same message, fully presenting the diversity and contextual dependencies of dialogues.
Coverage of 35 Languages
Includes dialogue content in 35 languages such as English, Spanish, French, German, Chinese, Japanese, etc., making it a valuable resource for multilingual dialogue AI research.
Quality Ranking Annotations
Each reply is quality scored and ranked by community members, providing rich human preference data that can be directly used for RLHF training.
Community Collaboration Achievements
Contributed by over 13,000 volunteers worldwide, reflecting the power of the open-source community, with data covering a wide range of topics and dialogue styles.
Apache 2.0 License
Utilizes a permissive Apache 2.0 open-source license, allowing commercial use, modification, and distribution, with no usage restrictions.
Applicable Scenarios
From chatbot training to academic research, empowering various directions of conversational AI
Chatbot Training
Fine-tune large language models using real human conversation data to build more natural and helpful conversational AI assistants
Multilingual AI
Train multilingual models using conversational data in 35 languages to enhance cross-language understanding and generation capabilities
RLHF Data Source
Based on community quality ranking annotations, directly used for human feedback reinforcement learning, aligning model outputs with human preferences
Conversational AI Research
Support cutting-edge research directions such as dialogue generation, multi-turn reasoning, and dialogue evaluation, promoting the development of conversational AI technology
API Call Example
Quickly obtain OASST1 dataset records through the Ace Data Cloud API
curl -X GET "https://api.acedata.cloud/datasets/oasst1" \
-H "Authorization: Bearer YOUR_API_TOKEN" \
-H "Content-Type: application/json"
# Response Example
{
"message_id": "6ab24d72-...",
"parent_id": null,
"text": "Can you write a short introduction about...",
"role": "prompter",
"lang": "en",
"review_count": 3,
"rank": 0,
"labels": {
"spam": { "value": 0.0, "count": 3 },
"quality": { "value": 0.75, "count": 3 }
}
}
3 Steps to Get Started Quickly
From registration to data acquisition, you can start your conversational AI research project in just a few minutes
Register and Get Token
Register an account on the Ace Data Cloud platform and obtain an API Token to start calling the dataset interface.
Call API to Get Data
Use the GET /datasets/oasst1 interface to retrieve conversation records as needed, supporting pagination, language filtering, and other parameters.
Load and Train
Load the data into a Python environment for fine-tuning language models, training RLHF reward models, or conducting conversational research.
Start Exploring OpenAssistant Conversation Data
The world's largest open-source human conversation dataset, licensed under Apache 2.0, available via API immediately. Whether you are training a chatbot or researching conversation generation, OASST1 is your ideal choice.