Downloads · 30 days
11
22% of all-time downloads
lumko-medgemma-4b
LeroyNgoma/lumko-medgemma-4b
lumko-medgemma-4b is a machine learning model from LeroyNgoma. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as gemma.
Fine-tuned MedGemma 4B for South African clinical reasoning with Chain-of-Thought.
All-time downloads
51
Public
Parameters
4.3B
8.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors8.6 GB · 100%
At a glance
- License
- gemma
- Model type
- gemma3
- Access
- Public
- Created
- Jun 23, 2026
- Updated
- Jun 23, 2026
- SHA
- a6119a49
Model card
From the Hugging Face model README
Lumko Clinical SLM (MedGemma 4B Fine-tuned)
Fine-tuned MedGemma 4B for South African clinical reasoning with Chain-of-Thought.
Training
- Base: google/medgemma-4b-it
- Method: LoRA (rank 64, alpha 128, BF16) + DPO (beta 0.1)
- Strategy: FSDP on 8x A100 80GB (data parallelism)
- Data: Clinical guidelines (DoH, NICD), LOINC, PubMed
- CoT: Chain-of-thought reasoning with <think> blocks
- Safety: Refusal/boundary training curriculum
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("path/to/merged_model")
tokenizer = AutoTokenizer.from_pretrained("path/to/merged_model")
- Type
- gemma3
- License
- gemma
- Created
- Jun 23, 2026
- Updated
- Jun 23, 2026