[Refactor] refactor OSS-GPT

[wangyantong2000]

Description

Refactoring OSS-GPT. For the question and answer function at the repository level, it will wait until it is completed later. The first thing to complete is a basic question and answer function, as well as the construction of a scenario that can provide users with the ability to fill in model information. With infrastructure, the implementation of subsequent functions can be promoted.

iShot_2024-09-25_11.30.58.mp4

The above is a rough demo. Actual improvements will be made later. Including dual versions in both Chinese and English, light and dark modes, component styles, and more.

[Xsy41]

We can add a function to make the question and answer interface more free, you can move or fix on the page at will.Similar to the following demonstration.

9.23.mp4

[TieWay59]

paste some examples from Kimi-crx

[wangyantong2000]

All of these can be considered in the future. Complete the construction of infrastructure first. Continuously iterate afterwards.

[TieWay59]

I may have just learned about this today, but I'm curious about your thoughts on GitHub Chat. @wangyantong2000 @Xsy41

[TieWay59]

A landing page: https://github.com/copilot

[TieWay59]

doc page: https://docs.github.com/en/enterprise-cloud@latest/copilot/using-github-copilot/asking-github-copilot-questions-in-githubcom

[wangyantong2000]

I think that creating this OSS-GPT display interface is not just for the purpose of completing Q&A needs. More importantly, it is treated as an infrastructure, a display interface. In order to enable other automated functions that will be completed using LLM in the future. When the behavior of these functions requires a display interface in the final or process, there should be a place to showcase them. For example, the automatic creation of issues, issue recommendations, and so on, may currently have limited ideas. But when there are other creative ideas that need to be presented later, there should be a display interface. So I call it infrastructure.

[TieWay59]

@Xsy41 I mean, they are overlapping, is this part of the refactor fix?

[wangyantong2000]

@Xsy41 I mean, they are overlapping, is this part of the refactor fix?

If there is an overlap with other plugins, we can consider @Xsy41 's suggestion of allowing icons and display boxes to be dragged. But it was not included in this refactoring and will be carried out in point 5 mentioned in this issue comment #840 (comment).

[wangyantong2000]

The style modification part can be completed by @Xsy41 in the future.

[TieWay59]

@wangyantong2000 Thank you for explaining the roadmap! 💕 I'm here to remind you that this would cause github pro users to drop Hypercrx as it's a very obvious conflict.

[wangyantong2000]

@wangyantong2000 Thank you for explaining the roadmap! 💕 I'm here to remind you that this would cause github pro users to drop Hypercrx as it's a very obvious conflict.

Does this conflict refer to the overlap of this style or a functional issue?

[Xsy41]

@wangyantong2000感谢您解释路线图！💕我在这里提醒您，这会导致github pro用户放弃Hypercrx，因为这是一个非常明显的冲突。

This is indeed a problem.

GitHub Copilot offers similar features to oss-gpt in HyperCRX, such as code modification, help answering common development questions, and summarizing repository issues, and GitHub Copilot requires no additional plug-ins or licenses, which may be easier to use.

Currently, oss-gpt only does some basic features, and will continue to develop in the future, such as automatic problem creation, problem recommendation to the right developer, and so on.

Now users can use HyperCRX Settings to set whether the oss-gpt feature is required.

In the future, when oss-gpt is more perfect, can we consider making this function separate from HyperCRX and become a new plug-in independently?