LLaVA Instruct Mix: 100K-1M Multimodal Conversational Records

Between 100,000 and 1,000,000 multimodal conversational records comprise this dataset released by trl-lib in 2025. It facilitates instruction tuning by pairing images with multi-turn dialogue prompts and target completions. The data is structured specifically for language modeling and visual-text alignment tasks.

Use Cases

• Training multimodal models to interpret the 'images' column in the context of a 'prompt'
• Fine-tuning conversational agents using the 'completion' field as the target response
• Benchmarking visual reasoning by analyzing the relationship between 'images' and the conversational history

Strengths

• Scale of 100,000 to 1,000,000 records
• Standardized conversational schema with 'prompt' and 'completion' fields
• Parquet format for efficient data loading and processing

Limitations

• Lack of detailed documentation regarding the specific data sources included in the mix
• Potential for geographic bias as metadata indicates a US region focus
• Unknown license status may restrict commercial use

Provenance

Source

trl-lib

Collection Method

Processed version of LLaVA Instruct Mix

Freshness

Last updated August 16, 2025.

Geography

United States