Changed the way embedding model creation is handled in the AbstractGraph class.

[S4mpl3r]

Now instead of using self._create_llm() method for both embedding models and LLMs and delegating the correct instantiation of embedding models to the RAGNode class, there are two dedicated methods for creating embedding models and the logic is completely inside the AbstractGraph class:

1. self.create_default_embedder(): This method handles the case where no embedding model is provided in the config. This will try to instantiate the proper embedding model based on the chosen LLM. Previously, this case was handled in the self.execute() method in the RAGNode class. I didn't change the code for this part (it's a copy-paste of the old code) so that it works just like it was previously.
2. self.create_embedder(): This method handles the case where an embedding model is explicitly provided in the config. It checks for the embedding model's name in the config and instantiates the proper embedding model. Code-wise it looks similar to self._create_llm().

This pull request was created to address the issue that was discussed in #120

[github-actions]

🎉 This PR is included in version 0.7.0-beta.2 🎉

The release is available on:

• v0.7.0-beta.2
• GitHub release

Your semantic-release bot 📦🚀

[github-actions]

🎉 This PR is included in version 0.7.0 🎉

The release is available on:

• v0.7.0
• GitHub release

Your semantic-release bot 📦🚀