Research map / Large language models / Agents, code and math
C-Eval: A Multi-Level Multi-Discipline Chinese Evaluation Suite for Foundation Models
Yuzhen Huang, Yuzhuo Bai, Zhihao Zhu, Junlei Zhang, Jinghan Zhang, Tangjun Su, Junteng Liu, Chuancheng Lv, Yikai Zhang, jiayi lei, Yao Fu, Maosong Sun and 1 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | GPT-4 · C-Eval test | average accuracy (%): 66.4 -> 68.7 | Improved |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | ChatGPT · C-Eval test | average accuracy (%): 51.0 -> 54.4 | Improved |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | Claude-v1.3 · C-Eval test | average accuracy (%): 50.5 -> 54.2 | Improved |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | GLM-130B · C-Eval test | average accuracy (%): 44.0 -> 40.3 | Hurt |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | Bloomz-mt · C-Eval test | average accuracy (%): 44.3 -> 39.0 | Hurt |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | ChatGLM-6B · C-Eval test | average accuracy (%): 38.9 -> 34.5 | Hurt |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | MOSS · C-Eval test | average accuracy (%): 33.1 -> 31.1 | Hurt |
| Use five-shot answer-only prompting instead of zero-shot answer-only prompting (ablation) | Chinese-Alpaca-13B · C-Eval test | average accuracy (%): 30.9 -> 26.7 | Hurt |
| Add five-shot chain-of-thought prompting instead of five-shot answer-only prompting (ablation) | GPT-4 · C-Eval test | average accuracy (%): 68.7 -> 68.3 | Hurt |
| Add five-shot chain-of-thought prompting instead of five-shot answer-only prompting (ablation) | ChatGPT · C-Eval test | average accuracy (%): 54.4 -> 50.0 | Hurt |
| Add five-shot chain-of-thought prompting instead of five-shot answer-only prompting (ablation) | GLM-130B · C-Eval test | average accuracy (%): 40.3 -> 28.8 | Hurt |
| Add five-shot chain-of-thought prompting instead of five-shot answer-only prompting (ablation) | GPT-4 · C-Eval Hard | average accuracy (%): 54.9 -> 56.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GLM-130B — GLM-130B: An Open Bilingual Pre-trained Model (ICLR 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CK12: A Rounded K12 Knowledge Graph Based Benchmark for Chinese Holistic Cognition Evaluation (AAAI 2024 · similarity 0.89)
- Xiezhi: An Ever-Updating Benchmark for Holistic Domain Knowledge Evaluation (AAAI 2024 · similarity 0.88)
- LexEval: A Comprehensive Chinese Legal Benchmark for Evaluating Large Language Models (NeurIPS 2024 · similarity 0.87)
- MME-SCI: A Comprehensive and Challenging Science Benchmark for Multimodal Large Language Models (AAAI 2026 · similarity 0.86)
- SciEval: A Multi-Level Large Language Model Evaluation Benchmark for Scientific Research (AAAI 2024 · similarity 0.86)
- MR-GSM8K: A Meta-Reasoning Benchmark for Large Language Model Evaluation (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
