FoundationAgents/MetaGPT · error · ValueError

Score list is empty.

Error message

Score list is empty.

What it means

Raised by DataUtils._compute_probabilities when the `scores` list has length 0. The method builds a mixed uniform + softmax distribution over scores, which is undefined for an empty list. It is normally called from select_round after the empty-items check, so hitting it means _compute_probabilities was called directly (or via a path) with no scores.

Source

Thrown at metagpt/ext/aflow/scripts/optimizer_utils/data_utils.py:69

        sorted_items = sorted(items, key=lambda x: x["score"], reverse=True)
        scores = [item["score"] * 100 for item in sorted_items]

        probabilities = self._compute_probabilities(scores)
        logger.info(f"\nMixed probability distribution: {probabilities}")
        logger.info(f"\nSorted rounds: {sorted_items}")

        selected_index = np.random.choice(len(sorted_items), p=probabilities)
        logger.info(f"\nSelected index: {selected_index}, Selected item: {sorted_items[selected_index]}")

        return sorted_items[selected_index]

    def _compute_probabilities(self, scores, alpha=0.2, lambda_=0.3):
        scores = np.array(scores, dtype=np.float64)
        n = len(scores)

        if n == 0:
            raise ValueError("Score list is empty.")

        uniform_prob = np.full(n, 1.0 / n, dtype=np.float64)

        max_score = np.max(scores)
        shifted_scores = scores - max_score
        exp_weights = np.exp(alpha * shifted_scores)

        sum_exp_weights = np.sum(exp_weights)
        if sum_exp_weights == 0:
            raise ValueError("Sum of exponential weights is 0, cannot normalize.")

        score_prob = exp_weights / sum_exp_weights

        mixed_prob = lambda_ * uniform_prob + (1 - lambda_) * score_prob

        total_prob = np.sum(mixed_prob)
        if not np.isclose(total_prob, 1.0):
            mixed_prob = mixed_prob / total_prob

View on GitHub (pinned to 11cdf466d0)

Solutions

1. Pass at least one score: ensure the upstream round/score collection produced data before calling
2. Guard the call site with `if not scores: ...` and handle the empty case explicitly
3. Reuse select_round instead of calling _compute_probabilities directly — it already validates inputs

Example fix

probs = du._compute_probabilities(scores) if scores else None

assert scores, "need at least one score"  # or: if not scores: return None

def is_nonempty_score_list(scores) -> bool:
    return isinstance(scores, (list, tuple)) and len(scores) > 0

When it happens

Trigger: Calling _compute_probabilities(scores) with an empty list, e.g. computing probabilities over an empty list of per-round scores or an empty top_scores sample.

Common situations: Custom code that reuses _compute_probabilities for sampling; refactors of select_round that bypass its empty-list guard.

Related errors

• Item list is empty.
• Unsupported dataset: {dataset}
• Only support for python, markdown, but got {language}
• Only support for language: python, markdown, but got {langua
• The invoice format is not zip, pdf, png, or jpg

AI-assisted analysis of FoundationAgents/MetaGPT@11cdf466d0 (2026-08-14). Data as JSON: /api/errors/1f7d7e6a333f8d36. Report an issue: GitHub.