Transformers
medical
activation-steering
gated-activation-steering
inference-time-intervention
sycophancy
hallucination
Instructions to use himanshu5trpth/medgemma-sycophancy-hallucination-gated-steering with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use himanshu5trpth/medgemma-sycophancy-hallucination-gated-steering with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("himanshu5trpth/medgemma-sycophancy-hallucination-gated-steering", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -13,6 +13,8 @@ tags:
|
|
| 13 |
|
| 14 |
**Inference-time steering for MedGemma that improves grounding in the patient record and resistance to user pressure, without target-model fine-tuning or weight updates.** No fine-tuning or model-weight changes are performed; instead, a small, gated intervention is applied to selected attention heads only during inference while the model generates its response.
|
| 15 |
|
|
|
|
|
|
|
| 16 |
|
| 17 |
> ⚕️ **Research artifact — not a medical device.** For research use only. Not for clinical
|
| 18 |
> decision-making, diagnosis, or treatment.
|
|
|
|
| 13 |
|
| 14 |
**Inference-time steering for MedGemma that improves grounding in the patient record and resistance to user pressure, without target-model fine-tuning or weight updates.** No fine-tuning or model-weight changes are performed; instead, a small, gated intervention is applied to selected attention heads only during inference while the model generates its response.
|
| 15 |
|
| 16 |
+
📄 **Paper:** [Gated Activation Steering for Reducing Sycophancy & Hallucination in Medical Question Answering](https://arxiv.org/abs/2608.23666)
|
| 17 |
+
|
| 18 |
|
| 19 |
> ⚕️ **Research artifact — not a medical device.** For research use only. Not for clinical
|
| 20 |
> decision-making, diagnosis, or treatment.
|