Official release of InternLM series, including InternLM, InternLM2, InternLM2.5, and InternLM3. It hosts multiple model sizes and cross-platform model cards across HF, ModelScope, and OpenXLab.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
InternLM is the official release of the InternLM series, including InternLM, InternLM2, InternLM2.5, and InternLM3. The project describes a family of large language models and associated releases across multiple sizes (e.g., 1.8B, 7B, 20B, 8B-Instruct) and supports various deployment formats and model cards.
How it works
The repository showcases multiple model variants with different training and fine-tuning stages (pretraining, SFT, RLHF) and provides references to model cards and hosting on HF, ModelScope, and OpenXLab. It also documents release notes and highlights for improvements across versions.
Getting started
The README references access to model zoo sections and model cards for download and usage via HuggingFace, ModelScope, and OpenXLab. It includes links to licenses and distribution channels. Specific install/usage commands are not present in the provided excerpt.
Recent releases
Latest releases include v0.2.1dev20240102 InternLM-v0.2.1dev20240102 (2024-01-02) with a timeout fix and GPU memory info for 7B & 20B models, and multiple 2023-09 to 2023-11 updates addressing grad overflow, KeyError, and unit tests. Full changelogs are available in the release notes.
Traction
Stars: 7257 Forks: 510 Open issues: 8 Last push: 2025-10-30 Created: 2023-07-06
Caveats
License: Apache-2.0 Notes indicate alignment with safety considerations and limitations, including potential biases and harmful content in outputs. Specific licensing details per model or artifact are not enumerated beyond the Apache-2.0 license.