Instructions to use yashvshetty/clarke-medgemma-27b-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use yashvshetty/clarke-medgemma-27b-lora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("google/medgemma-27b-text-it") model = PeftModel.from_pretrained(base_model, "yashvshetty/clarke-medgemma-27b-lora") - Notebooks
- Google Colab
- Kaggle
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,3 +1,13 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
base_model: google/medgemma-27b-text-it
|
| 4 |
+
tags: [medgemma, medical, nhs, clinical-documentation, lora, qlora, kaggle-competition]
|
| 5 |
+
license: apache-2.0
|
| 6 |
+
---
|
| 7 |
+
# Clarke LoRA Adapter — MedGemma 27B for NHS Clinic Letters
|
| 8 |
+
QLoRA fine-tuned adapter for `google/medgemma-27b-text-it`, trained to generate structured NHS outpatient clinic letters.
|
| 9 |
+
- **Method**: QLoRA (4-bit NF4, LoRA rank 16, alpha 32)
|
| 10 |
+
- **Target modules**: q_proj, k_proj, v_proj, o_proj
|
| 11 |
+
- **Training data**: 5 gold-standard NHS clinic letters
|
| 12 |
+
- **Final loss**: 1.7117
|
| 13 |
+
- **Competition**: [MedGemma Impact Challenge](https://www.kaggle.com/competitions/medgemma-impact-challenge)
|