Which browser are you using? In case you are using the phone App, specify the Android or iOS version and device please.
Chrome 123.0.6312.123
Describe the Bug
Current Artificial Intelligence administration dashboard has "LocalAI integration" next to tasks.
This seems to embed models like whisper1, stablediffussion, etc or makes the default model chosen in the Connected Accounts.
Here for example is the error when the assistant passes an image generation request via a LocalAI definition to LiteLLM
[integration_openai] Warning: API request error : Server error: POST http://xxxx:4000/v1/images/generations resulted in a 500 Internal Server Error response:
{"error":{"message":"'model'","type":"None","param":"None","code":500}}
POST /ocs/v2.php/apps/assistant/api/v1/i/process_prompt
from yyyy by zzzzzzz at 21 Apr 2024, 12:46:34
If you use a different model via LocalAI or LiteLLM, etc then you have to create aliasing for the model to match Nextcloud expectations.
Expected Behavior
Ideally, replace AI tasks like "Task: Generate headline" with Nextcloud CustomGPTs and user callable connected account (e.g. OpenAI, LocalAi) and model selection with defaults (i.e. default selected via admin interface could be Llama and via permissions, user could select GPT4 for customGPT over this if allowed).
As interim, separate connected account (OpenAI, LocalAI) and model selection (e.g. whisper1, stablediffusion)
To Reproduce
Current "settings/admin/ai" interface has no ability to select model
Which version of assistant are you using?
1.08
Which version of Nextcloud are you using?
28.0.1
Which browser are you using? In case you are using the phone App, specify the Android or iOS version and device please.
Chrome 123.0.6312.123
Describe the Bug
Current Artificial Intelligence administration dashboard has "LocalAI integration" next to tasks.
This seems to embed models like whisper1, stablediffussion, etc or makes the default model chosen in the Connected Accounts.
Here for example is the error when the assistant passes an image generation request via a LocalAI definition to LiteLLM
[integration_openai] Warning: API request error : Server error:
POST http://xxxx:4000/v1/images/generationsresulted in a500 Internal Server Errorresponse: {"error":{"message":"'model'","type":"None","param":"None","code":500}}If you use a different model via LocalAI or LiteLLM, etc then you have to create aliasing for the model to match Nextcloud expectations.
Expected Behavior
Ideally, replace AI tasks like "Task: Generate headline" with Nextcloud CustomGPTs and user callable connected account (e.g. OpenAI, LocalAi) and model selection with defaults (i.e. default selected via admin interface could be Llama and via permissions, user could select GPT4 for customGPT over this if allowed).
As interim, separate connected account (OpenAI, LocalAI) and model selection (e.g. whisper1, stablediffusion)
To Reproduce
Current "settings/admin/ai" interface has no ability to select model