1. Introduction
(1) The testing set comprises 2,435 questions from 25 Chinese Registered Construction Engineer Examinations (CRCEEs) papers spanning from 2013 to 2023.
(2) The CPM-QA testing set questions are manually tagged with four characteristics, including the paper’s level and year, 36 knowledge subfields, single- or multiple-answer questions, and questions with or without images.
(3) The testing set is developed and maintained by Southeast University, University of Cambridge, Central South University, Nanjing University, Michigan Technological University, and George Mason University.
(4) Make sure to read the specification and follow the rules.
2. Submission of your GLM’s answers
The answers could be submitted through https://forms.gle/WvXew3pi6DYvvWQY9. Please use “Template of answer submission.xls” in this repository to submit your GLM's answers.
3. Citation requirement
The reuse of this repository requires citation. Should any individual or entity utilize this repository without appropriate acknowledgment and citation, they do not have the right to use our data. We will take measures to protect our copyright, including, but not limited to, retracting their papers and initiating legal action.
4.GLM Leaderboard for CPM-QA
| General-purpose large-language models | Publishing Institution | Accuracy rate of SAMCQs | Accuracy rate of MAMCQs | Accuracy rate of Text-only Questions | Accuracy rate of Image-embedded Questions | Average Accuracy Rate | Ranking |
|---|---|---|---|---|---|---|---|
| ERNIE-Bot 4.0 with CPM-KG | Baidu&The authors | 0.773 | 0.568 | 0.701 | 0.224 | 0.682 | 1 |
| GPT-4-turbo with CPM-KG | OpenAI&The authors | 0.723 | 0.543 | 0.661 | 0.250 | 0.643 | 2 |
| GPT-4 with CPM-KG | OpenAI&The authors | 0.686 | 0.550 | 0.646 | 0.218 | 0.628 | 3 |
| ERNIE-Bot 4.0 | Baidu | 0.726 | 0.442 | 0.621 | 0.166 | 0.602 | 4 |
| ERNIE-Bot with CPM-KG | Baidu&The authors | 0.727 | 0.361 | 0.578 | 0.235 | 0.566 | 5 |
| GPT-4-turbo | OpenAI | 0.589 | 0.372 | 0.503 | 0.235 | 0.494 | 6 |
| GPT-3.5-turbo with CPM-KG | OpenAI&The authors | 0.538 | 0.394 | 0.480 | 0.244 | 0.472 | 7 |
| ERNIE-Bot | Baidu | 0.656 | 0.234 | 0.481 | 0.218 | 0.471 | 8 |
| GPT-4 | OpenAI | 0.591 | 0.313 | 0.482 | 0.198 | 0.470 | 9 |
| ChatGLM3-6B with CPM-KG | Tsinghua & Zhipu.AI | 0.497 | 0.319 | 0.424 | 0.238 | 0.418 | 10 |
| Qianfan-Chinese-Llama-2-7B with CPM-KG | Baidu&The authors | 0.464 | 0.238 | 0.369 | 0.221 | 0.367 | 11 |
| ChatGLM3-6B | Tsinghua & Zhipu.AI | 0.419 | 0.262 | 0.355 | 0.203 | 0.351 | 12 |
| GPT-3.5-turbo | OpenAI | 0.427 | 0.237 | 0.346 | 0.174 | 0.342 | 13 |
| Llama-2-70B-Chat with CPM-KG | MetaAI&The authors | 0.443 | 0.189 | 0.328 | 0.323 | 0.331 | 14 |
| Llama-2-70B-Chat | MetaAI | 0.335 | 0.137 | 0.247 | 0.235 | 0.249 | 15 |
| Qianfan-Chinese-Llama-2-7B | Baidu | 0.314 | 0.140 | 0.237 | 0.203 | 0.240 | 16 |
- Downloads last month
- 28