Instructions to use v-vovk/medgemma-ios-q4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use v-vovk/medgemma-ios-q4 with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # if on a CUDA device, also pip install mlx[cuda] # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("v-vovk/medgemma-ios-q4") prompt = "Once upon a time in" text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- MLX LM
How to use v-vovk/medgemma-ios-q4 with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Generate some text mlx_lm.generate --model "v-vovk/medgemma-ios-q4" --prompt "Once upon a time"
Portable Nutrition Guidance SLM (Q4)
Overview
This repository hosts a quantized on-device language model artifact for nutrition-guidance chat workflows. The artifact is packaged for local iOS runtime loading and does not require backend inference.
Intended use
- Personal nutrition guidance conversations
- Meal pattern coaching support
- Educational wellness tips for non-emergency scenarios
Safety and medical disclaimer
This model is not a diagnostic or treatment system. It should not be used for emergency decisions, medication dosing, or definitive medical advice. Users should consult qualified clinicians for diagnosis and treatment decisions.
Known limitations
- Outputs can still be incomplete, ambiguous, or incorrect.
- Responses may underperform on rare edge cases outside training distribution.
- Prompt phrasing can materially affect result quality and safety.
Artifact details
- Version:
medgemma-ios-q4-v9-fixture-calib - Download URL:
https://huggingface.co/v-vovk/medgemma-ios-q4/resolve/main/medgemma-ios-q4-v9-fixture-calib.zip - SHA256:
dd794a634204557bec6639351d4e3fcfaa5717aca8599f5d7af7a3b29398684b - Size (bytes):
2594347078
Evaluation expectations
- Validate behavior against curated safety and hallucination checks before release.
- Track real-world mismatch reports and feed corrected examples into the next training cycle.
Model tree for v-vovk/medgemma-ios-q4
Base model
google/medgemma-1.5-4b-it