gptel icon indicating copy to clipboard operation
gptel copied to clipboard

Published 20 hours ago • verified publisher icon karthink

issues with ollama

Open lbraglia opened this issue 11 months ago • 6 comments

Hi i've somewhat copypasted the config with minor modification to use dolphin-mistral

(setq                                                                                                                                
 gptel-model "dolphin-mistral:latest"                                                                                                
 gptel-backend (gptel-make-ollama "Ollama"                                                                                           
                 :host "localhost:11434"                                                                                             
                 :stream t                                                                                                           
                 :models '("dolphin-mistral:latest")))

ollama is up and running but gptel starts chatgpt 3.5 and furthermore ollama/dolphin-mistral is not available in the M-x gptel-menu, GPT-model section

am i missing something or this is a bug?

lbraglia avatar Mar 25 '24 15:03 lbraglia

  1. Are you on the latest commit? If not please update gptel
  2. Could you run this: (map-keys gptel--known-backends)and paste the output here?

karthink avatar Mar 25 '24 16:03 karthink

  1. Are you on the latest commit? If not please update gptel

    1. Could you run this: (map-keys gptel--known-backends)and paste the output here?

@lbraglia Did you have a chance to try this?

karthink avatar Apr 03 '24 17:04 karthink

not yet, i'll try it asap. BTW the version i've used is the one available in MELPA.

Thanks for follow up

lbraglia avatar Apr 11 '24 07:04 lbraglia

Any updates?

karthink avatar Apr 25 '24 21:04 karthink

ok, ive updated now to

  gptel                          20240425.2058  installed             Interact with ChatGPT or other LLMs

and with

(map-keys gptel--known-backends)

i obtain ("ChatGPT")

while from gptel-menu, selecting -m GPT Model i have

Click on a completion to select it.
In this buffer, type RET to select the completion near point.

8 possible completions:
ChatGPT:gpt-3.5-turbo 	ChatGPT:gpt-3.5-turbo-16k 	ChatGPT:gpt-4
ChatGPT:gpt-4-0125-preview 	ChatGPT:gpt-4-1106-preview 	ChatGPT:gpt-4-32k
ChatGPT:gpt-4-turbo 	ChatGPT:gpt-4-turbo-preview

lbraglia avatar Apr 26 '24 17:04 lbraglia

and with

(map-keys gptel--known-backends)

i obtain ("ChatGPT")

The Ollama backend is not being defined then. Does your setq form throw any errors? Can you try evaluating it manually?

karthink avatar Apr 28 '24 03:04 karthink

After a loong time I had time to go back on it and with current gptel and emacs 29.4 all works fine with local ollama. Thank you again!

lbraglia avatar Nov 16 '24 10:11 lbraglia