unslothai/unsloth · error · HTTPException

file too large (max {LOCAL_SEED_UPLOAD_MAX_LABEL})

Error message

file too large (max {LOCAL_SEED_UPLOAD_MAX_LABEL})

What it means

HTTP 413 raised when the decoded byte length of the uploaded seed file exceeds LOCAL_SEED_UPLOAD_MAX_BYTES. The limit is enforced server-side after base64 decoding and before the file is written to SEED_UPLOAD_DIR, with a human-readable label (LOCAL_SEED_UPLOAD_MAX_LABEL) embedded in the message.

Source

Thrown at studio/backend/routes/data_recipe/seed.py:673

        if ext not in _LEGACY_UNSTRUCTURED_EXTS:
            allowed = ", ".join(sorted(_LEGACY_UNSTRUCTURED_EXTS))
            raise HTTPException(
                status_code = 400,
                detail = f"unsupported file type: {ext}. allowed: {allowed}",
            )
    else:
        if ext not in LOCAL_UPLOAD_EXTS:
            allowed = ", ".join(sorted(LOCAL_UPLOAD_EXTS))
            raise HTTPException(
                status_code = 400,
                detail = f"unsupported file type: {ext}. allowed: {allowed}",
            )

    file_bytes = _decode_base64_payload(payload.content_base64)
    if not file_bytes:
        raise HTTPException(status_code = 400, detail = "empty upload payload")
    if len(file_bytes) > LOCAL_SEED_UPLOAD_MAX_BYTES:
        raise HTTPException(
            status_code = 413,
            detail = f"file too large (max {LOCAL_SEED_UPLOAD_MAX_LABEL})",
        )

    ensure_dir(SEED_UPLOAD_DIR)
    stored_name = f"{uuid4().hex}_{filename}"
    stored_path = SEED_UPLOAD_DIR / stored_name
    stored_path.write_bytes(file_bytes)

    if seed_source_type == "unstructured":
        preview_rows = _read_preview_rows_from_unstructured_file(
            path = stored_path,
            preview_size = int(payload.preview_size),
            chunk_size = payload.unstructured_chunk_size,
            chunk_overlap = payload.unstructured_chunk_overlap,
        )
    else:
        preview_rows = _read_preview_rows_from_local_file(

View on GitHub (pinned to 203007d190)

Solutions

1. Sample or truncate the dataset client-side to stay under the limit (a preview only needs the first N rows).
2. Raise LOCAL_SEED_UPLOAD_MAX_BYTES in backend config if full-file seeding is genuinely required (and raise any upstream proxy body limits to match).
3. Compress or split the file, or move it onto the server filesystem and reference it by path instead of uploading.

Example fix

// before
const b64 = await fileToBase64(hugeCsv); // 500MB file

// after
const head = await readFirstNRows(hugeCsv, 1000);
const b64 = await fileToBase64(head);

const MAX = LOCAL_SEED_UPLOAD_MAX_BYTES; // mirror server constant
if (file.size > MAX) {
  file = await truncateToPreview(file, MAX);
}

When it happens

Trigger: Uploading a structured dataset (csv/json) or text seed larger than the configured cap; the check compares len(file_bytes) against the constant, so both genuinely large files and accidentally duplicated/concatenated payloads trigger it.

Common situations: Full-dataset uploads when only a preview sample is needed; a reverse proxy already blocking large bodies with a different error, masking this one; the client sending the raw bytes base64-encoded of a file that grew after a re-export.

Related errors

• File exceeds the {cap // (1024 * 1024)} MB upload limit.
• VirusTotal upload failed after {attempts} attempt(s): {last_
• each reference image must be at most 32 MiB (base64)
• each reference must be at most 32 MiB (base64)
• File too large ({size_bytes} bytes). Maximum is {UNSTRUCTURE

AI-assisted analysis of unslothai/unsloth@203007d190 (2026-08-15). Data as JSON: /api/errors/db9582d04afa4d92. Report an issue: GitHub.