Skip to content
TheChatArchiveThe Chat Archive
Claudeclaude-3-7-sonnet-20250219-thinking-32k· AnthropicHugging Face LMArena Human Preference 140K

What safety benchmark would you create for a multimodality LLM model to localize abusive…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-sonnet-4-20250514’s answer.
  1. User

    What safety benchmark would you create for a multimodality LLM model to localize abusive segments in a video? Given a list of policies? Ideally, I should be able to retrieve the timestamps of all the abusive segments and an explanation of why they are abusive.

  2. Claude Assistant

    I'd create a benchmark called "Temporal Abuse Localization Test" (TALT) with these key components:

    1. Dataset Construction

      • Diverse video corpus with precise human-annotated timestamps of policy violations
      • Range of violation types, subtlety levels, and multimodal abuse (visual, audio, textual)
      • Inclusion of edge cases where context determines violation status
    2. Evaluation Metrics

      • Temporal precision: How accurately timestamps capture abusive segments
      • Policy classification accuracy: Correctly identifying which policies were violated
      • Explanation quality: Relevance and specificity of reasoning
      • False positive/negative rates across different abuse categories
    3. Implementation Design

      • Input: Video + list of applicable policies
      • Output: JSON with timestamps, violated policies, and explanations
      • Required format: {start_time: float, end_time: float, policies_violated: [string], explanation: string}
    4. Challenging Elements

      • Implicit violations requiring contextual understanding
      • Cross-modal violations (e.g., benign visuals with harmful audio)
      • Cultural nuance detection
      • Temporal context where meaning changes over time

    This benchmark would evaluate both technical accuracy in localization and reasoning quality in explaining why specific content violates particular policies.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.