chroma-core/chroma · error · ValueError

All input documents must be strings

Error message

All input documents must be strings

What it means

GoogleGeminiEmbeddingFunction.__call__ verifies every element of the input is a str. Non-string entries (ints, floats, None, bytes, dicts) are rejected because this embedding function is text-only: the Gemini text-embedding models have no defined behavior for other payloads, and silently coercing them could corrupt embeddings.

Source

Thrown at chromadb/utils/embedding_functions/google_embedding_function.py:94

            ),
        )

    def __call__(self, input: Documents) -> Embeddings:
        """
        Generate embeddings for the given documents.

        Args:
            input: Documents to generate embeddings for.

        Returns:
            Embeddings for the documents.
        """
        if not input:
            raise ValueError("Input documents cannot be empty")
        if not isinstance(input, (list, tuple)):
            raise ValueError("Input must be a list or tuple of documents")
        if not all(isinstance(doc, str) for doc in input):
            raise ValueError("All input documents must be strings")

        from google.genai.types import EmbedContentConfig

        config = EmbedContentConfig(
            task_type=self.task_type,
            output_dimensionality=self.dimension,
        )

        try:
            response = self.client.models.embed_content(
                model=self.model_name,
                contents=input,
                config=config,
            )
        except Exception as e:
            raise ValueError(f"Failed to generate embeddings: {str(e)}") from e

        # Validate response structure

View on GitHub (pinned to aecdd12c8a)

Solutions

1. Sanitize None/missing values before embedding: [d for d in docs if isinstance(d, str)] or fill defaults
2. Coerce only when lossless and intended: [str(d) for d in docs]
3. For images use an embedding function that supports multimodal input instead of this one

Example fix

# before
vecs = ef(["doc one", None, 42])  # ValueError: All input documents must be strings

# after
docs = [d for d in ["doc one", None, 42] if isinstance(d, str)]
vecs = ef(docs)

docs = [d for d in docs if isinstance(d, str)]
assert docs, "no string documents left after filtering"
vecs = ef(docs)

from typing import Any, TypeGuard

def is_all_strings(docs: Any) -> TypeGuard[list[str]]:
    return isinstance(docs, (list, tuple)) and all(isinstance(d, str) for d in docs)

When it happens

Trigger: Passing a list like ['doc one', 42]; None entries where source data had missing fields; image bytes (e.g. PIL/PNG data) intended for a multimodal model; JSON rows where a 'content' field is null or a nested object; mixed id/text columns accidentally zipped together.

Common situations: DataFrame column containing NaN/None passed after .tolist(); upstream JSON with null content fields not sanitized; attempting multimodal (image+text) retrieval with a text-only embedding function.

Related errors

• Input must be a list or tuple of documents
• Input must be a list of text documents (str) or a list of im
• Input documents cannot be empty
• Google Generative AI only supports text documents, not image
• Knn key must be a string or Key instance

AI-assisted analysis of chroma-core/chroma@aecdd12c8a (2026-08-16). Data as JSON: /api/errors/cb5f5ae727a99649. Report an issue: GitHub.