Batch processing of embeddings is not supported in this fork yet. See [this PR](https://github.com/abetlen/llama-cpp-python/pull/2058) of the original binding which added multi-sequence support for embeddings.