Smolified Clinical Symptom Router: 10,016 Synthetic Instruction Tuning Records

A synthetic training corpus of 10,016 records generated by the Smolify Foundry for instruction tuning. It was used to train the corresponding model smolify/smolified-clinical-symptom-router. The dataset is a sovereign asset owned by smolify and was generated via Smolify.ai.

Use Cases

• Fine-tuning language models for clinical symptom classification based on the synthetic instruction data.
• Training a model to route patient-reported symptoms to appropriate medical categories.
• Benchmarking the performance of instruction-tuned models on synthetic medical text tasks.
• Studying the transferability of knowledge from synthetic to real-world clinical text.

Strengths

• Contains 10,016 records, providing a substantial volume of synthetic data for training.
• Explicitly designed for instruction tuning, indicating a structured format for model training.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data is synthetic, which may limit its direct applicability to real-world clinical scenarios.

Provenance

Source

Smolify Foundry (Job ID: fc35f0e3)

Collection Method

Synthetically generated via Smolify.ai.

Time Range

null

Freshness

Last updated 2026-02-13 19:46:10; freshness should be verified.

Geography

null