digital-health-clinical-asr-finetune

Community

Fine-tune clinical ASR with stock NeMo SFT.

Authorsayalinvidia
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Stage 4 fine-tunes clinical ASR using stock NeMo SFT on Parakeet TDT v2, with offline cycle N+1 evaluation to confirm loop closure.

Core Features & Use Cases

  • Stratified term-aware train/validation split; 3-epoch bf16-mixed SFT; offline evaluation; optional deployment via Riva NIM.
  • Applies to healthcare vocabulary tuning including drug names, conditions, and procedures, with measurable KER improvements.
  • Supports end-to-end workflow from data prep to evaluation, deploying to serving if needed.

Quick Start

Prepare a GPU-enabled environment, split your manifest into train/validation, run the stock NeMo SFT on Parakeet TDT 0.6B, and perform the offline cycle N+1 evaluation.

Dependency Matrix

Required Modules

None required

Components

references

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: digital-health-clinical-asr-finetune
Download link: https://github.com/sayalinvidia/sayali-skills-test/archive/main.zip#digital-health-clinical-asr-finetune

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 510,000+ vetted skills library on demand.