InternLM2.5, the 2.5th generation InternLM, has open-sourced a 7 billion parameter base model and a chat model tailored for practical scenarios. For the convenience of users and researchers, we have open-sourced three versions of each scale of the model, which are:
- InternLM2.5-7B: Further pretrain with general domain data and domain-enhanced corpus, obtaining state-of-the-art performance in evaluation with good language capability. InternLM2.5 models are recommended for consideration in most applications.
- InternLM2.5-chat-7B: Further aligned on top of InternLM2.5 through supervised fine-tuning (SFT) and online RLHF. InternLM2.5-Chat exhibits better instruction following, chat experience, and function calling, which is recommended for downstream applications.
- InternLM2.5-7B-Chat-1M: 1M-long-context version of InternLM2.5-7B-Chat. InternLM2.5-Chat-1M supports million-word extra-long contextual reasoning while maintaining the same performance as InternLM2.5-Chat.
The model has the following characteristics:
- **Outstanding reasoning capability**: State-of-the-art performance on Math reasoning, surpassing models like Llama3 and Gemma2-9B.
- **1M Context window**: Nearly perfect at finding needles in the haystack with 1M-long context, with leading performance on long-context tasks like LongBench. Try it with [LMDeploy](./chat/lmdeploy.md) for 1M-context inference. More details and a file chat demo are found [here](./long_context/README.md).
- **Stronger tool use**: InternLM2.5 supports gathering information from more than 100 web pages, corresponding implementation will be released in Lagent soon. InternLM2.5 has better tool utilization-related capabilities in instruction following, tool selection and reflection. See [examples](https://huggingface.co/internlm/internlm2_5-7b-chat-1m/blob/main/agent/).
## Model Zoo
| Model | Transformers(HF) | ModelScope(HF) | OpenXLab(HF) | OpenXLab(Origin) | Release Date |
-`HF` refers to the format used by HuggingFace in [transformers](https://github.com/huggingface/transformers), whereas `Origin` denotes the format adopted by the InternLM team in [InternEvo](https://github.com/InternLM/InternEvo).
## Performance Evaluation
We have evaluated InternLM2.5 on several important benchmarks using the open-source evaluation tool [OpenCompass](https://github.com/open-compass/opencompass). Some of the evaluation results are shown in the table below. You are welcome to visit the [OpenCompass Leaderboard](https://opencompass.org.cn/rank) for more evaluation results.
- We use `ppl` for the MCQ evaluation on base model.
- The evaluation results were obtained from [OpenCompass](https://github.com/open-compass/opencompass) , and evaluation configuration can be found in the configuration files provided by [OpenCompass](https://github.com/open-compass/opencompass).
- The evaluation data may have numerical differences due to the version iteration of [OpenCompass](https://github.com/open-compass/opencompass), so please refer to the latest evaluation results of [OpenCompass](https://github.com/open-compass/opencompass).
- \* means the result is copied from the original paper.
