Chinese Language Model Evaluation Dataset with Query Categories

Designed to evaluate the ability of large language models in Chinese. It contains query examples categorized by class and type, such as a riddle query for a Chinese location name.

Use Cases

• Benchmark LLM performance on Chinese 'query' generation tasks across different 'class' categories.
• Analyze the distribution of question 'type' within the dataset to understand evaluation coverage.
• Fine-tune or test models on specific 'class' categories like question-answering using the provided query examples.

Strengths

• Designed specifically for evaluating Chinese language model capabilities.
• Provides structured data fields including 'class', 'type', and 'query' for analysis.
• Authored by a recognized research organization, IDEA-CCNL.

Limitations

• The total number of rows or data instances is unknown.
• Sample data is limited to a single example, obscuring dataset scale and diversity.
• Specific license terms and detailed data collection methodology are not provided.

Provenance

Source

IDEA-CCNL via Hugging Face.

Collection Method

Designed for evaluation; specific gathering method unknown.

Freshness

Last updated on 2023-05-17.

Geography

Primarily Chinese language content.