Using ORCA embeddings in OpenNMT-py

Hi, I converted the ORCA embeddings contained in the gpt4-1m-orca-embeddings jsonl file into text files, extracting and truncating the embeddings to three sets of txt files: 256, 512 and 1024 dimensions.
Can an OpenNMT-py transformer actually work with such ORCA embeddings for the source side? or it limited to only work with words to vectors embeddings?