Bangla Instruction Dataset: A Consolidated Collection for Instruction-Following Models

A repository consolidated on May 19, 2025 by kamruzzaman-asif, containing high-quality instruction-tuning data from multiple sources. The dataset is structured for training and evaluating instruction-following models and includes splits like 'OdiaGenAI', which is described as a large-scale collection of diverse Bangla instructions and responses.

Use Cases

• Fine-tuning large language models for Bengali instruction-following based on the described instruction-response pairs.
• Evaluating the performance of multilingual AI models on Bengali tasks based on the structured dataset splits.
• Benchmarking the quality of generated text in Bengali based on the high-quality instruction data.
• Researching cross-lingual transfer learning for instruction-tuning based on the consolidated multi-source data.

Strengths

• Consolidated from multiple popular sources, suggesting breadth of data origin.
• Structured specifically for instruction-tuning, indicating a clear intended use.
• Includes a large-scale split ('OdiaGenAI') with 252k entries, indicating substantial volume for that component.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full consolidated dataset is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality and completeness of other splits require manual inspection.

Provenance

Source

Hugging Face dataset by author kamruzzaman-asif, consolidating data from sources like OdiaGenAI.

Collection Method

Consolidation of instruction-tuning data from multiple popular sources.

Time Range

null

Freshness

Last updated 2025-05-19 03:41:16.

Geography

null