angkul07/

cot-trajectory-prompt-dataset

All outputs were verified and passed validation checks.

Output File Lines Status

data/curated/dataset_v1.jsonl 1000 All checks pass

data/prompts/formatted_prompts.jsonl 1000 IDs match curated set

The final dataset contains exactly 1,000 examples, distributed across five task types as originally planned.

	Final Dataset Composition




	

Task Type Count Source Selection Criteria

math_easy 200 GSM8K (test split) Random sample… See the full description on the dataset page: https://huggingface.co/datasets/angkul07/cot-trajectory-prompt-dataset.