PeKA: Persian Knowledge Assessment for Large Language Models

A dataset introduced in a paper accepted at NAACL 2025 findings for evaluating large language models on Persian knowledge topics. It was developed by MatinaAI as part of a broader benchmarking effort for Persian LLMs. The dataset page was last updated on June 30, 2025.

Use Cases

• Benchmarking LLM performance on Persian knowledge topics based on the described assessment tasks.
• Evaluating factual accuracy and reasoning of language models in Persian based on the described knowledge topics.
• Training or fine-tuning models for Persian question-answering based on the described knowledge assessment structure.

Strengths

• Dataset is associated with a peer-reviewed paper accepted at NAACL 2025 findings.
• Dataset was last updated on June 30, 2025, suggesting recent maintenance.
• Dataset is part of a described broader effort to benchmark Persian LLMs, indicating a structured purpose.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

MatinaAI

Collection Method

Developed as part of a research effort described in the paper 'Advancing Persian LLM Evaluation'.

Freshness

Last updated 2025-06-30 06:53:05.