chroma-core/chroma · error · ValueError

You must set a data loader on the collection if loading from

Error message

You must set a data loader on the collection if loading from URIs.

What it means

`collection.get(include=[..., "data"])` asks Chroma to load the raw content behind each stored URI. The collection must have a DataLoader configured (passed at creation via `data_loader=`); without one, Chroma has no way to fetch URI contents, so the request is rejected before reaching the server.

Source

Thrown at chromadb/api/models/CollectionCommon.py:277

        self,
        ids: Optional[OneOrMany[ID]],
        where: Optional[Where],
        where_document: Optional[WhereDocument],
        include: Include,
    ) -> GetRequest:
        # Unpack
        unpacked_ids: Optional[IDs] = maybe_cast_one_to_many(target=ids)
        filters = FilterSet(where=where, where_document=where_document)

        # Validate
        if unpacked_ids is not None:
            validate_ids(ids=unpacked_ids)

        validate_filter_set(filter_set=filters)
        validate_include(include=include, dissalowed=["distances"])

        if "data" in include and self._data_loader is None:
            raise ValueError(
                "You must set a data loader on the collection if loading from URIs."
            )

        # Prepare
        request_include = include
        # We need to include uris in the result from the API to load datas
        if "data" in include and "uris" not in include:
            request_include.append("uris")

        return GetRequest(
            ids=unpacked_ids,
            where=filters["where"],
            where_document=filters["where_document"],
            include=request_include,
        )

    @validation_context("query")
    def _validate_and_prepare_query_request(

View on GitHub (pinned to aecdd12c8a)

Solutions

Pass a data loader when creating/getting the collection: `client.get_or_create_collection(name="docs", data_loader=DefaultDataLoader())`
Use a custom DataLoader (e.g. from chromadb.utils.data_loaders) matching your storage backend
Drop "data" from include and fetch the URIs yourself, then load content with your own I/O

Example fix

# before
col = client.get_or_create_collection(name="docs")
col.get(include=["uris", "data"])  # ValueError

# after
from chromadb.utils.data_loaders import DefaultDataLoader
col = client.get_or_create_collection(name="docs", data_loader=DefaultDataLoader())
col.get(include=["uris", "data"])

Defensive patterns

Strategy: validation

Validate before calling

include = ["metadatas", "documents"]
wants_data = "data" in include
# track loader presence at creation time in your own wrapper
if wants_data and not collection_has_data_loader:
    include = [f for f in include if f != "data"]  # or raise/report early
collection.get(include=include)

Prevention

Always create URI-backed collections with data_loader= so 'data' include is always valid
Centralize collection creation in one factory that decides whether a data loader is attached
Request 'uris' first, load content yourself, if you do not want a loader dependency

When it happens

Trigger: Calling `get()` (or `query()`) with `include` containing "data" on a collection created without `data_loader=`, e.g. `client.get_or_create_collection(name="docs")`.

Common situations: Adding `include=["data"]` to an existing get call after following an example that assumes a data loader, or creating the collection in one place (without the loader) and reading `data` in another.

Related errors

transactional filter reads require a positive limit

AI-assisted analysis of chroma-core/chroma@aecdd12c8a (2026-08-16). Data as JSON: /api/errors/469f5280092bc2ab. Report an issue: GitHub.