If you are not fully satisfied with the default LLM used to generate answers in your project, you can use and experiment other off-the-shelf LLMs.
- Go to Processing view
- Create a new answer generator
- Give a name
- Click on “Create from default configuration”
- Browse the list or – better – type “Text gen…” as keyword to filter the result list
- Select a cloud provider:
- OpenAI for GPT models (need a key)
- Microsoft Azure for GPT models (need a key)
- DeepInfra for OpenSource models including Llama2, Mistral 7B, Mixtral 8*7B, Dolphin-Mixtral 8*7B (need a key)
- Select the LLM
- Edit the default prompt if you want to adapt it
- Don’t forget to save
- Mark it with the yellow star if you want to use it by default in the question answering interface
