|
--- |
|
license: apache-2.0 |
|
task_categories: |
|
- question-answering |
|
language: |
|
- en |
|
size_categories: |
|
- n<1K |
|
--- |
|
|
|
# Dataset Card for Worker Safety Question and Answer Eval |
|
|
|
This dataset contains the `worker-safety-qa-eval` benchmark. This benchmark is used to evaluate question answering tasks in the domain of worker safety and health. |
|
The focus of the benchmark is to answer queries about worker safety practices and regulations based on laws in Singapore. |
|
For correct answers we refer to the resources from [Workplace Safety and Health Council](https://www.tal.sg/wshc/resources). |
|
|
|
## Dataset Details |
|
|
|
### Dataset Description |
|
Each row of the benchmark has three fields `question`, `options` and `answer`. Each `question` is a multiple choice query with 4 different options. The `answer` field contains |
|
the correct expected response based on the given `options`. |
|
|
|
## Leaderboard |
|
|
|
| Tool | WorkerSafetyQAEval (%) | Cost | |
|
|:----------------------------:|:----------------------:|:------:| |
|
| Securade.ai Safety Copilot | 97 | Free | |
|
| HuggingChat | 76.5 | Free | |
|
| OpenAI ChatGPT | 73.5 | Free | |
|
| SAifety.ai Co-pilot | 20.5 | Paid | |
|
|
|
|
|
The actual eval results are available below: |
|
|
|
- [Securade.ai Safety Copilot](./Securade.ai-Safety-Copilot-WorkerSafetyQAEval.pdf) |
|
- HuggingChat [part 1](https://hf.co/chat/r/tthBPja), [part 2](https://hf.co/chat/r/alXmHki), and [part 3](https://hf.co/chat/r/YPPryIT) |
|
- [OpenAI ChatGPT](https://chatgpt.com/share/5b82fe4e-bdfa-461c-9ae2-e1bd97ca6591) |
|
|
|
**Curated by:** [Securade.ai](https://securade.ai/) |
|
|