Math SFT Solutions No CoT V3: Large-Scale Mathematics Instruction Tuning Dataset

Math SFT Solutions No CoT V3 is a large-scale supervised fine-tuning dataset designed for mathematical capability adaptation. Version 3 substantially expands mathematical coverage while improving dataset quality through stronger filtering, cleaning, and supervision refinement. The dataset, created by kaushik-harsh-99 and last updated in June 2026, focuses on clean instruction-response pairs without hidden chain-of-thought reasoning.

Use Cases

• Instruction tuning of language models based on the described clean instruction-response pairs.
• Mathematical capability adaptation of models based on the expanded mathematical coverage.
• Training models for direct answer generation in mathematics, avoiding chain-of-thought reasoning.

Strengths

• Dataset quality was improved through stronger filtering, cleaning, and supervision refinement as stated in the description.
• Mathematical coverage was substantially expanded in Version 3.
• The dataset is explicitly designed for instruction tuning and mathematical capability adaptation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Supervised fine-tuning dataset creation with filtering and cleaning.

Freshness

Last updated 2026-06-07 15:27:29; freshness should be verified.