Medical Reasoning SFT Nano: 20,000 Examples for Instruction Tuning

20,000 training examples form a subset of the OpenMed/Medical-Reasoning-SFT-Mega dataset. The data was sampled via a shuffle with a fixed seed and is intended for supervised fine-tuning. It was created by DedeProGames and last updated on Hugging Face in March 2026.

Use Cases

• Supervised fine-tuning of language models based on medical reasoning examples.
• Training models for medical question-answering based on the described reasoning tasks.
• Developing chain-of-thought reasoning capabilities in clinical domains based on the dataset's stated purpose.
• Benchmarking model performance on medical instruction-following tasks.

Strengths

• Contains 20,000 examples, providing a substantial sample for training.
• Derived from a larger, established source dataset (Medical-Reasoning-SFT-Mega).
• Sampling method (shuffle with seed=42) suggests a reproducible and potentially balanced subset.

Limitations

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

Provenance

Source

OpenMed/Medical-Reasoning-SFT-Mega

Collection Method

Subsampled via shuffle(seed=42) then first 20,000 examples.

Time Range

null

Freshness

Last updated 2026-03-08 23:45:43; freshness should be verified.

Geography

null