MMStar: Multimodal Benchmark for Kazakh and Russian Language Models

Machine-translated Kazakh and Russian versions of the MMStar dataset, an elite vision-language benchmark. The original MMStar consists of high-quality, non-redundant multimodal samples designed to evaluate 6 core capabilities and 18 detailed dimensions of Large Multimodal Models. The dataset was created by 'issai' and was last updated on 2026-02-18.

Use Cases

• Benchmarking model performance on advanced visual recognition tasks based on the described multimodal samples.
• Evaluating multi-step reasoning in vision-language tasks for Kazakh and Russian language models.
• Testing model robustness against 'data leakage' using the non-redundant samples mentioned in the description.

Strengths

• Derived from the 'elite' MMStar benchmark designed to evaluate 6 core capabilities and 18 detailed dimensions.
• Designed with high-quality, non-redundant samples to avoid 'data leakage'.
• Provides a specific benchmark for evaluating models on Kazakh and Russian languages.

Limitations

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

Provenance

Source

huggingface, author 'issai'.

Collection Method

Machine-translated versions of the original MMStar dataset.

Freshness

Last updated 2026-02-18 06:32:22; freshness should be verified.