fix typo docs/ (#4033)

docs/source/en/advanced_tutorials/meet_gemini.md

@@ -9,7 +9,7 @@ When you only have a few GPUs for large model training tasks, **heterogeneous tr
 
 ## Usage
 
-At present, Gemini supports compatibility with ZeRO parallel mode, and it is really simple to use Gemini: Inject the feathures of `GeminiPlugin` into training components with `booster`. More instructions of `booster` please refer to [**usage of booster**](../basics/booster_api.md).
+At present, Gemini supports compatibility with ZeRO parallel mode, and it is really simple to use Gemini: Inject the features of `GeminiPlugin` into training components with `booster`. More instructions of `booster` please refer to [**usage of booster**](../basics/booster_api.md).
 
 ```python
 from torchvision.models import resnet18

docs/source/zh-Hans/advanced_tutorials/train_vit_with_hybrid_parallelism.md

@@ -150,7 +150,7 @@ Colossal-AI 提供了自己的优化器、损失函数和学习率调度器。Py
 optimizer = colossalai.nn.Lamb(model.parameters(), lr=1.8e-2, weight_decay=0.1)
 # build loss
 criterion = torch.nn.CrossEntropyLoss()
-# lr_scheduelr
+# lr_scheduler
 lr_scheduler = LinearWarmupLR(optimizer, warmup_steps=50, total_steps=gpc.config.NUM_EPOCHS)
 ```
 

docs/source/zh-Hans/features/mixed_precision_training.md

@@ -303,7 +303,7 @@ colossalai.launch_from_torch(config=args.config)
     # build loss
     criterion = torch.nn.CrossEntropyLoss()
 
-    # lr_scheduelr
+    # lr_scheduler
     lr_scheduler = LinearWarmupLR(optimizer, warmup_steps=50, total_steps=gpc.config.NUM_EPOCHS)
 ```
 

docs/source/zh-Hans/features/mixed_precision_training_with_booster.md

@@ -181,7 +181,7 @@ optimizer = torch.optim.SGD(model.parameters(), lr=1e-2, weight_decay=0.1)
 # build loss
 criterion = torch.nn.CrossEntropyLoss()
 
-# lr_scheduelr
+# lr_scheduler
 lr_scheduler = LinearWarmupLR(optimizer, warmup_steps=50, total_steps=NUM_EPOCHS)
 ```