Any ways to snap in an embedding in the training prompt now?

I want to train a lora under some certain embeddings. Can I add a placeholder in the prompt so the embeddings are replaced in that position accordingly?

The issue does not mention any files, tests, or entry points. First clarify how a prompt placeholder should identify embeddings and where substitution should occur during LoRA training. Done would require an agreed behavior and corresponding implementation and validation, but the issue does not define those details.

Written by the indexing model from the issue text.