v2 plan
j178 opened this issue · comments
Some ideas:
- a general chat tui for any llm, support more llm providers, like claude, gemini, ollama #89
- refactor config structure, support multiple providers
- migrate legacy config automatically
- display provider and model as the bot name
- support "stop generating"
- configuration UI
- summarize prompt as the conversation title
-
ito enter insert mode, otherwise in command mode - one input panel, send to multiple llm panel split
- send and display images
and support "stop generating"
Can the color be configured by the user?
I am using an e-ink screen and the environment is Windows Terminal with PowerShell. The colors in the terminal are all set to black and white. However, in chat mode, some text is not clear due to the shadow color, which cannot be changed.
Thank you very much for developing this software.
It would be nice to be able to switch models from the TUI. For example for many simple prompts I could use GPT-3.5, and for complex prompts I could switch to GPT-4 without having to edit the config file.
Thanks for your effort that you put into this project!
Will “display provider and model as the bot name” be configurable for the user or not? For example, show more information, like current time, and emoji support, etc.
Like this, (this simple implementation is hardcoded in tgpt by myself).
After trying this project, I feel the UI is much better than tgpt. And more providers are supported, that's great!
Maybe we can add Phind as a provider too.