---
library_name: transformers
base_model: google/gemma-4-E4B-it
tags:
- gemma
- telugu
- medical
- grpo
- peft
- merged
---

# RohithMidigudla/gemma-health-telugu-medical-grpo-v3-hf-merged-test

Full HF/PEFT merged Gemma health model after Telugu medical GRPO policy alignment.

## Merge Inputs

- Base model: `google/gemma-4-E4B-it`
- Adapter: `RohithMidigudla/gemma-health-telugu-medical-grpo-policy-v3`
- Merge path: manual LoRA delta add into HF model weights (`safe_merge=True`)
- Missing target policy: `warn`
- Dtype: `bfloat16`

Evaluate safety, Telugu quality, and medical QA behavior before clinical or field use.