Jotsna22/30k_dataset_Qwen3-4B_ntokens2048_sft

--- dataset_info: features: - name: source_index dtype: int64 - name: chunk_id dtype: int64 - name: source_parquet_chunk dtype: string - name: response_idx dtype: int64 - name: prompt dtype: string - name: response dtype: string - name: response_num_tokens dtype: int64 - name: response_hit_max_tokens dtype: bool - name: messages list: - name: content dtype: string - name: role dtype: string - name: data_source dtype: string - name: ground_truth dtype: string - name: gold_answer dtype: string - name: is_correct dtype: int64 - name: extracted_answer dtype: string splits: - name: train num_bytes: 194204760 num_examples: 30000 download_size: 130769868 dataset_size: 194204760 configs: - config_name: default data_files: - split: train path: data/train-* ---