Update README.md

README.md

@@ -36,7 +36,7 @@ model-index:
 # **PCL-Reasoner-V1.5**
 
 ## Model Overview  
-We present PCL-Reasoner-V1.5, a 32-billion-parameter large language model (LLM) for mathematical reasoning. The model is built upon Qwen2.5-32B and refined via supervised fine-tuning (SFT) followed by reinforcement learning (RL). A central innovation is our proposed offline RL method, which provides superior training stability and efficiency over standard online RL methods such as GRPO. Our model achieves state-of-the-art performance among models post-trained on Qwen2.5-32B, attaining average accuracies of 90.9% on AIME 2024 and 85.6% on AIME 2025. Our work demonstrates offline RL as a stable and efficient paradigm for advancing reasoning in LLMs. All experiments were conducted on Huawei Ascend 910C NPUs.
+We present PCL-Reasoner-V1.5, a 32-billion-parameter large language model (LLM) for mathematical reasoning. The model is built upon Qwen2.5-32B and refined via supervised fine-tuning (SFT) followed by reinforcement learning (RL). A central innovation is our proposed offline RL method, which provides superior training stability and efficiency over standard online RL methods such as GRPO. Our model achieves state-of-the-art performance among models post-trained on Qwen2.5-32B, attaining average accuracies of 90.9% on AIME 2024 and 85.6% on AIME 2025. Our work demonstrates offline RL as a stable and efficient paradigm for advancing reasoning in LLMs. All experiments were conducted on Huawei Ascend 910C NPUs. Both training and evaluation processes utilize FP16 precision to maintain numerical accuracy.
 ![Evaluation Results](images/benchmark.png)
 
 
@@ -52,7 +52,7 @@ We present PCL-Reasoner-V1.5, a 32-billion-parameter large language model (LLM)
 
 ## Evaluation  
 
-All results are reported using the **pass@1 metric** (averaged over 32 independent sampling attempts per problem), ensuring robust and fair comparison. Both training and evaluation processes utilize FP16 precision to maintain numerical accuracy.
+All results are reported using the **pass@1 metric** (averaged over 32 independent sampling attempts per problem), ensuring robust and fair comparison.
 
 <!-- Table base styling (optional) -->