Val Kharitonov

I tried adding the tiktoken tokenizer (see #25), but looks like something other than that is also wrong for chat models (see screenshot in the PR). Let me know if...

That sounds very reasonable, wanna try making a PR?

Heads up I made Llama optional here: https://github.com/kharvd/gpt-cli/pull/46/

The token limits include both the prompt and the response together, there is no way to make gpt-4 process more than 8192 tokens (gpt-4-32k can do 32768). Theoretically, we could...

I agree it's not great. It would be pretty challenging to render streaming markdown in a TTY while supporting scrolling - happy to accept contributions. Personally, I just turned off...

Thanks for the contribution! Do you think we can use the EDITOR env variable instead?

Did you install it with `pip install gpt-command-line`?

I started working on some Docker support here for the code interpreter https://github.com/kharvd/gpt-cli/pull/37

gpt-4 might not know it’s gpt-4, that’s expected. Try comparing the latency of response with gpt-3.5-turbo

ChatGPT web ui knows which model it is because it's specified in its system prompt (which was leaked many times, you can find it online)