unslothai/unsloth · error · FileNotFoundError

unstructured seed file not found: {resolved}

Error message

unstructured seed file not found: {resolved}

What it means

FileNotFoundError raised by materialize_unstructured_seed_dataset when the configured source_path (after expanduser().resolve()) is not an existing regular file. The check runs before chunking, cache-key computation, and parquet materialization, and the message includes the fully resolved path so relative-path or tilde-expansion mistakes are visible.

Source

Thrown at studio/backend/plugins/data-designer-unstructured-seed/src/data_designer_unstructured_seed/chunking.py:110

            if len(result) >= preview_size:
                break
            val = next(it, None)
            if val is not None:
                result.append(val)
            else:
                exhausted.append(i)
        for i in reversed(exhausted):
            iterators.pop(i)

    return result


def materialize_unstructured_seed_dataset(
    *, source_path: Path, chunk_size: Any, chunk_overlap: Any
) -> tuple[Path, list[dict[str, str]]]:
    resolved = source_path.expanduser().resolve()
    if not resolved.is_file():
        raise FileNotFoundError(f"unstructured seed file not found: {resolved}")

    size, overlap = resolve_chunking(chunk_size, chunk_overlap)
    key = _compute_cache_key(
        source_path = resolved,
        chunk_size = size,
        chunk_overlap = overlap,
    )
    parquet_path = _CACHE_DIR / f"{key}.parquet"
    if parquet_path.exists():
        return parquet_path, []

    text = load_unstructured_text_file(resolved)
    chunks = split_text_into_chunks(
        text = text,
        chunk_size = size,
        chunk_overlap = overlap,
    )
    if not chunks:

View on GitHub (pinned to 203007d190)

Solutions

1. Verify the exact resolved path printed in the error exists: ls -l <resolved path>.
2. If relative, pass an absolute path (Path.cwd() / rel or Path(rel).resolve()) or fix the service working directory.
3. For uploaded files, ensure the temp file lifetime covers the async processing job, or copy to a stable staging location first.

Example fix

# before
source_path = Path("data/notes.txt")  # cwd differs at runtime

# after
source_path = Path("/srv/uploads/notes.txt")  # absolute, verified to exist

from pathlib import Path

def seed_file_ok(source_path: str | Path) -> bool:
    p = Path(source_path).expanduser().resolve()
    return p.is_file() and p.stat().st_size > 0

from pathlib import Path

resolved = Path(source_path).expanduser().resolve()
if not resolved.is_file():
    raise FileNotFoundError(resolved)  # fail before calling the plugin, with your own context

When it happens

Trigger: Passing source_path pointing to a missing file, a directory, a symlink to a deleted file, or a relative path resolved against an unexpected working directory (e.g. running the service from a different cwd).

Common situations: Upload temp file already cleaned up when the processing job starts; container paths differing from host paths; path built by concatenating strings that drops a slash; file processed twice with cleanup in between.

Related errors

• Local model path does not exist: {repo_id}
• no .safetensors/.gguf LoRA file found in '{repo_id}'
• {exc}
• Local model path '{repo_id}' does not exist.
• data_dir is not a directory: {data_dir}

AI-assisted analysis of unslothai/unsloth@203007d190 (2026-08-15). Data as JSON: /api/errors/f7e9f1f46d3e2234. Report an issue: GitHub.