HomeAPICommunity
Explore InternLM
An open-source organization dedicated to the research and development of large models and their toolchains.
Our mission is to provide an efficient and user-friendly open-source platform for all AI developers, democratizing access of large models and algorithms.
Explore InternLM
GitHub
Open-Source Large Models
InternLM2.5
Featuring a more efficient model structure, capable of handling ultra-long texts, tool invocation, data analysis, and other highly complex tasks.
InternLM2.5 is composed of a series of open-source large model matrices, including base models, Supervised-Finetuning (SFT) models, and models optimized by Reinforcement Learning from Human Feedback (RLHF). It covers multiple versions with parameters of 1.8B, 7B and 20B.
More
InternLM2-Math
A large language model with exceptional performance in mathematics.
InternLM2-Math is a specialized large language model in the mathematics field, trained on approximately 100 billion high-quality math-related tokens. It is capable of processing, solving, proving, verifying, and enhancing mathematical problems using the Lean language. Additionally, it can serve as a reward model to assist other large language models in further improving their mathematical capabilities.
More
InternLM-XComposer2.5
A groundbreaking multimodal model capable of flexibly handling and generating free-form image-text interleaved articles.
InternLM-XComposer2.5 is an advanced Vision-Language Model (VLM) based on InternLM2.5-7B. It can flexibly generate articles with images according to outlines, detailed text requirements, and reference images, and accurately handle various visual-language question-and-answer tasks.
More
InternLM2-WQX
LLM and VLM excelling in China’s most authoritative and challenging comprehensive test - the Gaokao
The Gaokao, covering a wide range of subjects and question types, is considered the most authoritative exam in China due to its 'confidentiality.' It is often used as a benchmark to evaluate the intelligence level of large models. The InternLM2-WQX and InternLM2-WQX-VL models have demonstrated outstanding performance on the Gaokao evaluation set (GAOKAO-Eval), showing comprehensive capabilities comparable to GPT-4o and surpassing many open-source models both domestically and internationally, showcasing their exceptional performance
More
June 7th
Released the first trillion-parameter model of the InternLM series
July 6th
InternLM-7B was open-sourced and made available for free commercial use, releasing a comprehensive open-source toolchain
September 20th
Intern·Wanjuan 1.0
multimodal pre-training corpus open-sourced
September 27th
Open-sourced InternLM-XComposer, the first large-scale model for integrated text and image creation
January 17th
Open-sourced InternLM2, surpassing the performance of the latest models of the same scale
January 23th
Released InternLM2-Math, the first open-source model to support both formal mathematical language and solution process evaluation
January 29th
Released InternLM-XComposer2, supporting personalized high-quality text and image creation and conversation
May 24
Open-sourced InternLM2-Math-Plus, setting a new benchmark for open-source mathematical model performance
June 4
Open-sourced InternLM2-WQX, achieving near GPT-4 level performance in the 2024 Gaokao with 20 billion parameters
2023
InternLM Open-Source Journey
Exploring the path to Artificial General Intelligence (AGI) together with the community.
InternLM 2.5 Open Source
July 2024
Try Now
GitHub
Open-Source Large Model Toolchain
All
A standardized open-source toolchain library covering the entire process from model fine-tuning to evaluation.
WanJuan-CC
A high-quality English web text dataset with 1T tokens sourced from CommonCrawl, showcasing significant enhancements in model task performance and safety across various validation sets and downstream tasks
InternEvo
A lightweight training framework aims to support model pre-training without the need for extensive dependencies
Xtuner
A lightweight fine-tuning tool that integrates core PEFT (Parameter-Efficient Fine-Tuning) methods
LMDeploy
Lightweight deployment and efficient servicing of LLMs.
OpenCompass
Comprehensive, objective, neutral, and reproducible evaluation tools and rankings for LLMs.
Lagent & AgentLego
Tool invocation protocol and intelligent agent applications for LLMs.
Huixiangdou
Zero-code setup for intelligent assistants based on private knowledge bases.
InternLM Bootcamp
Learn InternLM Toolkits
Join our free course to deeply understand, master, and practice the entire process of large model development.
Learn More About Bootcamp
RolePlay SIG
Fine-tune LLMs to play roles from movies, novels, anime, or be experts in various fields and practical tools used in work and life.
Join Special Interest Groups,
Maximize Your Passion of LLM
Get firsthand industry information and develop innovative projects with like-minded peers!
Learn More About SIG Groups
Subscribe to Community
Join Community
Permanent Open Source, Free for Commercial Use
Better models are always on the way.
Get Commercial License
Subscribe our Channels
Join Community
Contact Us
internlm@pjlab.org.cn
Personal Information Protection PolicyChildren’s Personal Information Protection Rules and Guardian NoticeService Agreement
© All Rights Reserved.沪ICP备2021009351号-21