import os import spaces from langchain_core.embeddings import Embeddings from sentence_transformers import SentenceTransformer class DenseEmbeddings(Embeddings): def __init__(self, model_name: str): is_zero_gpu = os.environ.get("SPACES_ZERO_GPU", "").lower() in {"1", "true"} model = SentenceTransformer( model_name, device="cuda" if is_zero_gpu else "cpu", ) dimension = model.get_embedding_dimension() if not isinstance(dimension, int) or dimension <= 0: raise ValueError(f"Cannot determine embedding dimension for '{model_name}'.") self.vector_size = dimension # Capture the model in the worker, rather than sending CUDA weights as arguments. @spaces.GPU(duration=30) def encode(texts: list[str]) -> list[list[float]]: return model.encode( texts, show_progress_bar=False, convert_to_numpy=True, ).tolist() self._encode = encode def embed_documents(self, texts: list[str]) -> list[list[float]]: return self._encode([text.replace("\n", " ") for text in texts]) def embed_query(self, text: str) -> list[float]: return self.embed_documents([text])[0]