RyanCodrai/turbovec · error · ValueError

embedder returned {vectors.shape[0]} vectors for {n_texts} t

Error message

embedder returned {vectors.shape[0]} vectors for {n_texts} texts

What it means

Raised in _check_embedded_batch when a 2D embedder output has a different row count than the number of input texts — a misbehaving embedder returned fewer (or more) vectors than texts. Without this check the mismatch surfaces downstream as an id-count error or IndexError during dedup that never names the embedder as the cause.

Source

Thrown at turbovec-python/python/turbovec/langchain.py:198

        return lambda sim: (sim + 1.0) / 2.0

    # ---- Embedder-output validation -----------------------------------

    @staticmethod
    def _check_embedded_batch(vectors: np.ndarray, n_texts: int) -> None:
        """Validate the shape of an embedder's document-batch output.

        Only 2D outputs are inspected here — any other ndim falls through
        to ``_store_texts_and_vectors``, whose existing guard names the bad
        dimensionality directly. Without the row-count check, a misbehaving
        embedder that returns fewer vectors than texts surfaces downstream
        as an id-count mismatch (or an IndexError during intra-batch
        dedup) that never names the embedder as the cause.
        """
        if vectors.ndim != 2:
            return
        if vectors.shape[0] != n_texts:
            raise ValueError(
                f"embedder returned {vectors.shape[0]} vectors for "
                f"{n_texts} texts"
            )
        if vectors.shape[1] == 0:
            raise ValueError(
                f"embedder returned empty vectors (dim 0) for {n_texts} texts"
            )

    def _validate_query_embedding(self, embedded: Any) -> np.ndarray:
        """Coerce an embedder's query output to a 1D float32 vector,
        rejecting None and wrong-rank outputs with an error that names
        the embedder (the raw values otherwise surface as opaque errors
        from the index kernel)."""
        if embedded is None:
            raise ValueError("embedder returned None instead of a query embedding")
        qvec = np.asarray(embedded, dtype=np.float32)
        if qvec.ndim != 1:
            raise ValueError(

View on GitHub (pinned to ccab9f325e)

Solutions

1. Fix or replace the embedder — a conforming Embeddings.embed_documents returns exactly one vector per text.
2. Check for batch-size truncation in wrapper embedders (e.g. API limits silently dropping rows).
3. Catch the ValueError in ingestion code to flag the embedder as the failing component.

When it happens

Trigger: Thrown at turbovec-python/python/turbovec/langchain.py:198 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of RyanCodrai/turbovec@ccab9f325e (2026-09-06). Data as JSON: /api/errors/98854f850aee0fb4. Report an issue: GitHub.