zylon-ai/private-gpt · error · OSError

Found files matching the pattern: . Make sure that a…

Error message

Found {len(matched_files)} files matching the pattern: {file_pattern}. Make sure that a Mistral tokenizer is present in {files}.

What it means

The companion case of the ambiguous-file error: find_tokenizer_file matched zero files against ^tokenizer\.model\.v.*$|^tekken\.json$|^tokenizer\.mm\.model\.v.*$ in the given directory list, so there is no Mistral tokenizer to load and it raises OSError.

Solutions

Check the directory/repo file list in the error and confirm which tokenizer file format the model actually ships.
Use a model repo that contains a Mistral tokenizer file (tekken.json or tokenizer.model.vN), e.g. official mistralai repos.
If the model only has HF tokenizer.json, use tokenizer_mode='huggingface' instead of 'mistral'.

Example fix

# before: model_dir only has tokenizer.json (HF format)
tok = MistralTokenizer.from_pretrained('model_dir')

# after
tok = TokenizerRegistry.get_tokenizer('huggingface', model_id='model_dir')

Defensive patterns

Strategy: validation

Validate before calling

def dir_has_mistral_tokenizer(path: str) -> bool:
    import os, re
    pat = re.compile(r'^tokenizer\.model\.v.*$|^tekken\.json$|^tokenizer\.mm\.model\.v.*$')
    return any(pat.match(f) for f in os.listdir(path))

Prevention

Prefer official mistralai repos when using tokenizer_mode='mistral'.
Check the model card/file list for tekken.json or tokenizer.model.vN before selecting the mode.
Use tokenizer_mode='huggingface' for repos that only ship HF-format tokenizer.json.

When it happens

Trigger: MistralTokenizer.from_pretrained(<local dir>) on a directory containing only HF-style files (tokenizer.json / tokenizer_config.json) and no Mistral-format file; also triggered on HF Hub repos (list_repo_files) that lack the expected file.

Common situations: Pointing the mistral tokenizer mode at a generic HF-converted model; incomplete downloads; directory listing that includes files under a subfolder (pattern is anchored to the bare filename).

Related errors

AI-assisted analysis of zylon-ai/private-gpt@4a030776a3 (2026-08-15). Data as JSON: /api/errors/b9489cbfd23aa20e. Report an issue: GitHub.

Appendix: source

Thrown at private_gpt/components/llm/tokenizers/mistral.py:196

    return []


def find_tokenizer_file(files: list[str]) -> str:
    """Find the Mistral tokenizer file from a list of repository files."""
    file_pattern = re.compile(
        r"^tokenizer\.model\.v.*$|^tekken\.json$|^tokenizer\.mm\.model\.v.*$"
    )

    matched_files = [file for file in files if file_pattern.match(file)]
    if len(matched_files) > 1:
        raise OSError(
            f"Found {len(matched_files)} files matching the "
            f"pattern: {file_pattern}. Make sure only one Mistral "
            f"tokenizer is present in {files}."
        )
    elif len(matched_files) == 0:
        raise OSError(
            f"Found {len(matched_files)} files matching the "
            f"pattern: {file_pattern}. Make sure that a Mistral "
            f"tokenizer is present in {files}."
        )

    return matched_files[0]


def _prepare_apply_chat_template_tools_and_messages(
    messages: list[dict[str, Any]],
    tools: list[dict[str, Any]] | None = None,
    continue_final_message: bool = False,
    add_generation_prompt: bool = False,
) -> tuple[list[dict[str, Any]], list[dict[str, Any]] | None]:
    """Prepare messages and tools for Mistral's chat template format.

    Handles validation and formatting of messages and tools to ensure
    compatibility with Mistral's requirements.

View on GitHub (pinned to 4a030776a3)