Update README.md

README.md

@@ -139,7 +139,7 @@ print(scores.tolist())
   - Auto-tuned parameter persistence threshold
 
 ## Technical Specifications
-- **Model Size**: ~1B parameters (based on a Llama-based LLM pre-trained on Armenian data)
+- **Model Size**: ~1.24B parameters (based on a Llama-style architecture LLM pre-trained on Armenian data)
 - **Embedding Dimension**: 2048
 - **Max Sequence Length**: 512 tokens
 - **Framework Compatibility**: