Describe the refactoring action
Currently, we have connectors for different LLM backends, but they don't adhere to a common interface. That means, we have different recommenders based on what backend we use. And the assistant can currently only use Ollama backends. It would be good if we could add a common abstraction layer over the different backends.
Expected benefit
More flexibility wrt LLM functionality.
Describe the refactoring action
Currently, we have connectors for different LLM backends, but they don't adhere to a common interface. That means, we have different recommenders based on what backend we use. And the assistant can currently only use Ollama backends. It would be good if we could add a common abstraction layer over the different backends.
Expected benefit
More flexibility wrt LLM functionality.