sierra-research/tau-bench

• highabout#1

  Update About section (description & homepage) to redirect to τ³-bench

  原因：

  当前

  Description: "Code and Data for Tau-Bench", Homepage: (none)

  复制粘贴的修复

  复制

  Description: "Legacy code and data for τ-bench. Please use τ³-bench for the latest benchmark for tool-agent-user interaction.", Homepage: "https://github.com/sierra-research/tau3-bench"

  待办完成忽略
• highreadme#2

  Update README H1 to immediately signal deprecation and redirect

  原因：

  当前

  # τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains

  复制粘贴的修复

  复制

  # τ-bench (Legacy): A Benchmark for Tool-Agent-User Interaction in Real-World Domains (Please use τ³-bench)

  待办完成忽略
• mediumtopics#3

  Add specific topics for LLM agent benchmarking

  原因：

  复制粘贴的修复

  复制

  llm-benchmark, language-agents, tool-use, api-tools, conversational-ai, agent-evaluation, multi-step-reasoning, real-world-domains

  待办完成忽略

本次扫描解析到的品类 GEO 通道：google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• 品类问题

  How to benchmark language agents that interact with users and external API tools?

  你：未被推荐
  AI 推荐顺序：

  1. LangChain Evaluation
  2. LlamaIndex Evaluation
  3. Scale AI
  4. Appen
  5. Surge AI
  6. MLflow
  7. HELM
  8. EleutherAI's LM Evaluation Harness

  AI 推荐了 8 个替代方案，却始终没点名 sierra-research/tau-bench。这就是要补上的差距。

  查看 AI 完整回答
• 品类问题

  What benchmarks exist for evaluating conversational AI agents in real-world, dynamic domains like banking?

  你：未被推荐
  AI 推荐顺序：

  1. GLUE (General Language Understanding Evaluation)
  2. SuperGLUE
  3. SQuAD (Stanford Question Answering Dataset)
  4. Natural Questions
  5. ATIS (Airline Travel Information System)
  6. SNIPS
  7. MultiWOZ (Multi-Domain Wizard-of-Oz)
  8. DSTC (Dialogue System Technology Challenges)
  9. PARADISE (PARAdigm for DIalogue System Evaluation)
  10. NIST (National Institute of Standards and Technology) Human Evaluation Guidelines

  AI 推荐了 10 个替代方案，却始终没点名 sierra-research/tau-bench。这就是要补上的差距。

  查看 AI 完整回答

• Metadata completeness

  warn

  建议：

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of sierra-research/tau-bench?

  pass
  ✓AI 明确点名了 sierra-research/tau-bench

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• If a team adopts sierra-research/tau-bench in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI 明确点名了 sierra-research/tau-bench

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• In one sentence, what problem does the repo sierra-research/tau-bench solve, and who is the primary audience?

  pass
  ✓AI 明确点名了 sierra-research/tau-bench

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？