ai-agents-embedding-fix-llamacppembeddings-embed-documents-raises-typeerro-da2b075a

THEN Flatten the nested list of token-level embeddings returned by llama-cpp-python into a list of vectors per document. Replace the list comprehension on line 114 of llamacpp.py with: return [list(map(float, sublist)) for e in embeddings for sublist in e]. Alternatively, upgrade llama-cpp-python to a version that returns sequence-level embeddings by default.

return [list(map(float, sublist)) for e in embeddings for sublist in e]