Closed opensource-elearning closed 7 months ago
Nice idea folk. I think obtaining the variables from environment would be way much better.
Thanks for your great effort, but my Idea is slightly different.
Certainly, we can implement a mechanism within the pytgpt
package to simplify the management of user-set parameters and store them internally in a .env
file, eliminating the need for manual environment variable setup. This approach would make it more accessible and user-friendly for individuals with mainly non-programming backgrounds and Windows users.
Interactive Command-Line Interface (CLI)
To provide an interactive way for users to manage parameters and interact with the model, we can incorporate a command-line interface (CLI) similar to the one shown in the provided code snippet. This CLI would allow users to:
Set parameters using the /set
command
Show model information using the /show
command
Load or save sessions using the /load
and /save
commands
Exit the interactive session using the /bye
command
Get help for commands using the /help
command
I will request you please try Ollama once feel the experience.
Awesome. Have you tried running h
or help
while in interactive mode?
Without interactive i have tried help
or h
Okay most of the features you listed above are already implemented.
Hi @Simatwa ,
Just have some Ideas, if possible to implement
Can we have options to set/show the System Message, temperature, default model to set, etc. to avoid long command lines arguments
I have found one of the user interface is very easy in terms of setting the parameters and system messages usage
ollama
Reference: https://github.com/ollama/ollama/blob/main/cmd/interactive.go#L110-L172
If we have it in pytgpt it would be great.