Update README.md

README.md

@@ -74,17 +74,17 @@ We conducted meticulous data collection and synthesis for the medical field, inc
 ## Sliding Window Attention Mechanism
 
 - Adopting a sliding window attention mechanism in some layers to reduce KV Cache memory usage.
-- **Optimization**: Balancing computational efficiency and performance, especially suitable for long-sequence tasks.
+- Balancing computational efficiency and performance, especially suitable for long-sequence tasks.
 
 ## Optimizing Position Encoding Oscillation
 
 - By increasing the dimensions of some attention heads, RoPE curve oscillation is reduced.
-- **Result**: More stable performance in long-sequence tasks while maintaining the model's ability to capture diverse features.
+- More stable performance in long-sequence tasks while maintaining the model's ability to capture diverse features.
 
 ## High Peak Learning Rate Strategy
 
 - Using **WSD learning rate scheduling strategy** with high peak learning rates to promote model generalization.
-- **Comparison results**: Significant improvement in benchmark task performance.
+- Significant improvement in benchmark task performance.
 
 ## Adaptive Gradient Update