tohio/

slm-synthetic-dpo-direct-subtraction

	SLM Synthetic DPO — direct_subtraction




	


	Summary

Synthetic preference-pair dataset containing chosen and rejected assistant responses for direct_subtraction.

Dataset type: preference optimization Total pairs: 6428 Signal: direct_subtraction Language: English

Category: direct_arithmetic Difficulty: 1 Template family: direct_subtraction Eval family: direct_subtraction

{ "id": "string"… See the full description on the dataset page: https://huggingface.co/datasets/tohio/slm-synthetic-dpo-direct-subtraction.