this project is just for fun, a lot of features are not stable and not usable. so far, only tried it out on
Kimi
andtongyi
. some prompts might not perform as well as expected.
- create an
env.json
file in root folder with configs like{ "models": [ { "tag": "text-model", "llm-base": "https://****/v1", "llm-model-name": "****", "llm-sk": "****" }, { "tag": "vision-model", "llm-base": "https://****/v1", "llm-model-name": "****", "llm-sk": "****" }, ... ]
}
### platforms
| platform | status |
| -------- | ----- |
| linux | on work |
| macos | not planned |
| windows | on work |
### features
* LLM (WIP)
![image-20240504112014938](./images/image-20240504112014938.png)
![image-20240504112102907](./images/image-20240504112102907.png)
* generate document from `template`(WIP)
| define template | define relation |
| ------------------------------------------------------ | ------------------------------------------------------ |
| ![image-20240511-215317](./images/20240511-215317.jpg) | ![image-20240511-215326](./images/20240511-215326.jpg) |
| generate | optimize |
| ![image-20240511-215330](./images/20240511-215330.jpg) | ![image-20240511-215333](./images/20240511-215333.jpg) |
> **bugs-to-be-fixed**
> - [ ] sometimes, generated file is too long and `optimize_doc` return error message
* ...