diff --git a/README.md b/README.md
index 806d06d..f2c4dc6 100644
--- a/README.md
+++ b/README.md
@@ -17,7 +17,7 @@ Qwen3 is the latest generation of large language models in Qwen series, offering
 
 **Qwen3-32B** has the following features:
 - Type: Causal Language Models
-- Training Stage: Post-training
+- Training Stage: Pretraining & Post-training
 - Number of Parameters: 32.8B
 - Number of Paramaters (Non-Embedding): 31.2B
 - Number of Layers: 64