Datasets:

---

lmms-lab

/

Video-MME

like 83

Follow

LMMs-Lab 826

This query provides useful analysis by creating a balanced sampling strategy that ensures each task_type-domain combination is represented proportionally, helping to uncover patterns in long-duration tasks across different domains.

This query provides useful stratified sampling by task type and domain to create a balanced subset of medium-duration tasks, revealing distribution patterns and ensuring representative data selection for analysis.

Provides a balanced stratified sampling approach that ensures diverse representation across task types and domains while maintaining a fixed sample size, revealing patterns in dataset distribution and helping create more representative training sets.

This query performs stratified sampling to create a balanced subset of medium-duration tasks across different task types and domains, revealing distribution patterns and ensuring representative data sampling for analysis.

Provides insightful stratified sampling by task type and domain that ensures balanced representation while maintaining randomness, revealing distribution patterns across different task-category combinations.

This query selects a balanced sample of 100 entries from a dataset, ensuring that each combination of task type and domain is proportionally represented, which is useful for creating a diverse and representative subset for analysis.

Counts the number of short-duration tasks for each task type, revealing which types are most common among short tasks.

Retrieves a sample of short-duration test records but doesn't provide meaningful analysis or insights about the dataset's patterns or relationships.

Retrieves a random sample of short-duration test items, but provides minimal analytical value beyond basic filtering.

Filters dataset to show only long-duration records, providing basic categorization but offering limited analytical insight beyond simple classification.

Retrieves a random sample of 100 short videos with basic metadata, providing only raw data without revealing meaningful patterns or relationships.

Randomly samples 300 rows from the test dataset, providing basic data exploration but offering no meaningful insights or patterns.

Randomly samples 200 rows from the test dataset, providing basic data exploration but offering no meaningful insights or patterns.

Randomly samples 50 records from the test dataset, providing basic data inspection but revealing no meaningful patterns or insights about the data structure or content.

Retrieves a random sample of short-duration videos from the Life Record domain, providing basic filtering but offering limited analytical insight beyond simple data retrieval.

Randomly samples 100 records from the test dataset, providing basic diversity but no meaningful analytical insights.

The query retrieves all entries from the 'test' dataset that belong to the 'Multilingual' domain, providing a basic filter but with limited analytical value.

This query retrieves all entries from the 'test' dataset where the domain is 'Knowledge', providing a basic filter but limited analytical value.

The query retrieves all entries related to the 'Film & Television' domain, providing a basic filtered view of the dataset without deep analysis.

Retrieves all records from the test dataset where the domain is 'Artistic Performance', providing a basic filtered view of the data.

Retrieves a sample of questions related to event localization or counting, providing limited insight into the types of questions asked.

Randomly selects 100 videos with 'long' duration and 'Life Record' domain, providing a basic sample for further inspection.

Randomly selects 100 videos with 'medium' duration and 'Life Record' domain, providing a basic sample for further inspection.

Randomly selects 100 videos with 'short' duration and 'Life Record' domain, providing a basic sample for further inspection.

Randomly selects 50 long-duration videos from the 'Life Record' domain, providing a basic sample for further analysis.