This dataset contains multi-turn conversational dialogues in Bengali, designed for training and evaluating language models on natural, context-aware interactions. Each conversation includes user-assistant exchanges on everyday topics like weather, greetings, and general inquiries.

Dataset Structure

The dataset is provided as a JSON file with the following structure:

File: data/train.json Format: JSON array of conversation records… See the full description on the dataset page: https://huggingface.co/datasets/hasin023/multiTurnBNTest.