|
--- |
|
license: apache-2.0 |
|
language: |
|
- en |
|
- zh |
|
tags: |
|
- rk3588 |
|
- rkllm |
|
- Rockchip |
|
- internlm2 |
|
--- |
|
|
|
# internlm2_rkLLM |
|
|
|
- [中文](#书生浦语-18b) |
|
- [English](#internlm2-18b) |
|
|
|
## 书生·浦语-1.8B |
|
|
|
### 介绍 |
|
|
|
internlm2_rkLLM 是从 [internlm2-chat-1_8b](https://huggingface.co/internlm/internlm2-chat-1_8b) 转换而来的 RKLLM 模型,专为 Rockchip 设备优化。该模型运行于 RK3588 的 NPU 上。 |
|
|
|
- **模型名称**: internlm2-chat-1_8b |
|
- **模型架构**: 与 internlm2-chat-1_8b 相同 |
|
- **发布者**: FydeOS |
|
- **日期**: 2024-06-03 |
|
|
|
### 模型详情 |
|
|
|
书生·浦语-1.8B (InternLM2-1.8B) 是第二代浦语模型系列的18亿参数版本。通过在线 RLHF 在 InternLM2-Chat-1.8B-SFT 之上进一步对齐。 InternLM2-Chat-1.8B表现出更好的指令跟随、聊天体验和函数调用,推荐下游应用程序使用。 |
|
|
|
InternLM2 模型具备以下的技术特点: |
|
- 有效支持20万字超长上下文:模型在20万字长输入中几乎完美地实现长文“大海捞针”,而且在 LongBench 和 L-Eval 等长文任务中的表现也达到开源模型中的领先水平。 |
|
- 综合性能全面提升:各能力维度相比上一代模型全面进步,在推理、数学、代码等方面的能力提升显著。 |
|
### 使用指南 |
|
|
|
> 此模型仅支持搭载 Rockchip RK3588/s 芯片的设备。请确认设备信息并确保 NPU 可用。 |
|
|
|
#### openFyde 系统 |
|
|
|
> 请确保你已将系统升级到最新版本。 |
|
|
|
1. 下载模型文件 `XXX.rkllm`。 |
|
2. 新建文件夹 `model/`,将模型文件放置于该文件夹内。 |
|
3. 启动 FydeOS AI,在设置页面进行相关配置。 |
|
|
|
#### 其它系统 |
|
> 请确保已完成 RKLLM 的 NPU 相关内核更新。 |
|
|
|
1. 下载模型文件 `XXX.rkllm`。 |
|
2. 按照官方文档进行配置:[官方文档](https://github.com/airockchip/rknn-llm)。 |
|
|
|
### 常见问题(FAQ) |
|
|
|
如遇到问题,请先查阅 issue 区,若问题仍未解决,再提交新的 issue。 |
|
|
|
### 限制与注意事项 |
|
|
|
- 模型在某些情况下可能存在性能限制 |
|
- 使用时请遵循相关法律法规 |
|
- 可能需要进行适当的参数调优以达到最佳效果 |
|
|
|
### 许可证 |
|
|
|
本模型采用与 [internlm2-chat-1_8b](https://huggingface.co/internlm/internlm2-chat-1_8b) 相同的许可证。 |
|
|
|
### 联系方式 |
|
|
|
如需更多信息,请联系: |
|
|
|
- **电子邮件**: [email protected] |
|
- **主页**: [FydeOS AI](https://fydeos.ai/zh/) |
|
|
|
|
|
## InternLM2-1.8B |
|
|
|
### Introduction |
|
|
|
internlm2_rkLLM is a RKLLM model derived from [internlm2-chat-1_8b](https://huggingface.co/internlm/internlm2-chat-1_8b), specifically optimized for Rockchip devices. This model operates on the NPU of the RK3588 chip. |
|
|
|
- **Model Name**: internlm2-chat-1_8b |
|
- **Architecture**: Identical to internlm2-chat-1_8b |
|
- **Publisher**: FydeOS |
|
- **Release Date**: 3 June 2024 |
|
|
|
### Model Details |
|
|
|
InternLM2-1.8B is the 1.8 billion parameter version of the second generation InternLM series. Further aligned on top of InternLM2-Chat-1.8B-SFT through online RLHF. InternLM2-Chat-1.8B exhibits better instruction following, chat experience, and function calling, which is recommended for downstream applications. |
|
|
|
The InternLM2 has the following technical features: |
|
|
|
- Effective support for ultra-long contexts of up to 200,000 characters: The model nearly perfectly achieves "finding a needle in a haystack" in long inputs of 200,000 characters. It also leads among open-source models in performance on long-text tasks such as LongBench and L-Eval. |
|
- Comprehensive performance enhancement: Compared to the previous generation model, it shows significant improvements in various capabilities, including reasoning, mathematics, and coding. |
|
|
|
### User Guide |
|
|
|
This model is only supported on devices with the Rockchip RK3588/s chip. Please verify your device's chip information and ensure the NPU is operational. |
|
|
|
#### openFyde System |
|
|
|
> Ensure you have upgraded to the latest version of openFyde. |
|
|
|
1. Download the model file `XXX.rkllm`. |
|
2. Create a folder named `model/` and place the model file inside this folder. |
|
3. Launch FydeOS AI and configure the settings on the settings page. |
|
|
|
#### Other Systems |
|
|
|
> Ensure you have updated the NPU kernel related to RKLLM. |
|
|
|
1. Download the model file `XXX.rkllm`. |
|
2. Follow the configuration guidelines provided in the [official documentation](https://github.com/airockchip/rknn-llm). |
|
|
|
### FAQ |
|
|
|
If you encounter issues, please refer to the issue section first. If your problem remains unresolved, submit a new issue. |
|
|
|
### Limitations and Considerations |
|
|
|
- The model may have performance limitations in certain scenarios. |
|
- Ensure compliance with relevant laws and regulations during usage. |
|
- Parameter tuning might be necessary to achieve optimal performance. |
|
|
|
### License |
|
|
|
This model is licensed under the same terms as [internlm2-chat-1_8b](https://huggingface.co/internlm/internlm2-chat-1_8b). |
|
|
|
### Contact Information |
|
|
|
For more information, please contact: |
|
|
|
- **Email**: [email protected] |
|
- **Homepage**: [FydeOS AI](https://fydeos.ai/en/) |