BerriAI/litellm · error · ValueError

No results found in the response={response}

Error message

No results found in the response={response}

What it means

After transforming the Bedrock rerank response, litellm expects response['results'] to be a non-empty list it can map to RerankResponseResult objects. If 'results' is missing, None, or empty, _results stays None and a ValueError('No results found in the response={response}') is raised, echoing the full response for diagnosis.

Source

Thrown at litellm/llms/bedrock/rerank/transformation.py:104

        """
        _billed_units = RerankBilledUnits(**response.get("usage", {"search_units": 1}))  # by default 1 search unit
        _tokens: Final = RerankTokens(**response.get("usage", {}))
        rerank_meta: Final = RerankResponseMeta(billed_units=_billed_units, tokens=_tokens)

        _results: list[RerankResponseResult] | None = None

        bedrock_results: Final = response.get("results")
        if bedrock_results:
            _results = [
                RerankResponseResult(
                    index=result.get("index"),
                    relevance_score=result.get("relevanceScore"),
                )
                for result in bedrock_results
            ]

        if _results is None:
            raise ValueError(f"No results found in the response={response}")

        return RerankResponse(
            id=response.get("id") or str(uuid.uuid4()),
            results=_results,
            meta=rerank_meta,
        )  # Return response

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Check the echoed response in the message: if results is absent/empty, inspect what you sent (non-empty query, non-empty documents).
2. Filter out empty-string documents and empty queries before calling rerank.
3. If documents were valid, retry once — occasional empty bodies indicate transient service issues.
4. Report reproducible empty-results responses to litellm/AWS with the payload shape.

Example fix

# before
result = litellm.rerank(model=MODEL, query='', documents=['', 'a doc'])

# after
docs = [d for d in documents if d and d.strip()]
if not query.strip() or not docs:
    return []  # nothing to rerank
result = litellm.rerank(model=MODEL, query=query, documents=docs)

def is_rerankable(query: str, documents: list) -> bool:
    return bool(query and query.strip()) and bool(documents) and all(
        (isinstance(d, str) and d.strip()) or (isinstance(d, dict) and d.get("text", "").strip())
        for d in documents
    )

def has_rerank_inputs(query: str, documents: list[str | dict]) -> bool:
    return is_rerankable(query, documents)

try:
    result = litellm.rerank(model=model, query=q, documents=docs)
except ValueError as e:
    if "No results found" in str(e):
        log.warning("Empty rerank result; falling back to unranked documents")
        return docs  # graceful degradation
    raise

When it happens

Trigger: A 200 response from bedrock-agent-runtime /rerank whose body lacks 'results' — usually caused by empty documents, query/document combos the model cannot score, or API changes/edge cases where Bedrock omits the field.

Common situations: Passing an empty documents list, whitespace-only queries, documents that truncate to empty strings, or intermediary proxies mangling the response body.

Related errors

• Missing required fields in the result={result}
• Unsupported image format: {image_format}. Supported formats:
• No supported extensions for MIME type: {mime_type}. Supporte
• Unexpected mistral completion response
• Invalid guardrailConfig={raw_guardrail_config}. Expected for

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/e2c9d217fa3e54c6. Report an issue: GitHub.