diff --git a/applications/Colossal-LLaMA-2/colossal_llama2/dataset/loader.py b/applications/Colossal-LLaMA-2/colossal_llama2/dataset/loader.py
index 327651f4e645..abe0fd51a4af 100644
--- a/applications/Colossal-LLaMA-2/colossal_llama2/dataset/loader.py
+++ b/applications/Colossal-LLaMA-2/colossal_llama2/dataset/loader.py
@@ -80,15 +80,19 @@ def __call__(self, instances: Sequence[Dict[str, List[int]]]) -> Dict[str, torch
 
         # `List[torch.Tensor]`
         batch_input_ids = [
-            torch.LongTensor(instance["input_ids"][: self.max_length])
-            if len(instance["input_ids"]) > self.max_length
-            else torch.LongTensor(instance["input_ids"])
+            (
+                torch.LongTensor(instance["input_ids"][: self.max_length])
+                if len(instance["input_ids"]) > self.max_length
+                else torch.LongTensor(instance["input_ids"])
+            )
             for instance in instances
         ]
         batch_labels = [
-            torch.LongTensor(instance["labels"][: self.max_length])
-            if len(instance["labels"]) > self.max_length
-            else torch.LongTensor(instance["labels"])
+            (
+                torch.LongTensor(instance["labels"][: self.max_length])
+                if len(instance["labels"]) > self.max_length
+                else torch.LongTensor(instance["labels"])
+            )
             for instance in instances
         ]
 
diff --git a/applications/Colossal-LLaMA-2/train.py b/applications/Colossal-LLaMA-2/train.py
index d97da61e4dc8..7e11eba37dbc 100644
--- a/applications/Colossal-LLaMA-2/train.py
+++ b/applications/Colossal-LLaMA-2/train.py
@@ -253,9 +253,11 @@ def main() -> None:
     coordinator.print_on_master(f"Model params: {format_numel_str(model_numel)}")
 
     optimizer = HybridAdam(
-        model_params=filter(lambda p: p.requires_grad, model.parameters())
-        if args.freeze_non_embeds_params
-        else model.parameters(),
+        model_params=(
+            filter(lambda p: p.requires_grad, model.parameters())
+            if args.freeze_non_embeds_params
+            else model.parameters()
+        ),
         lr=args.lr,
         betas=(0.9, 0.95),
         weight_decay=args.weight_decay,
diff --git a/applications/ColossalChat/coati/dataset/loader.py b/applications/ColossalChat/coati/dataset/loader.py
index 93cc1dab8d21..9dd9ed8e5df3 100755
--- a/applications/ColossalChat/coati/dataset/loader.py
+++ b/applications/ColossalChat/coati/dataset/loader.py
@@ -89,15 +89,19 @@ def __call__(self, instances: Sequence[Dict[str, List[int]]]) -> Dict[str, torch
 
         # `List[torch.Tensor]`
         batch_input_ids = [
-            torch.LongTensor(instance["input_ids"][: self.max_length])
-            if len(instance["input_ids"]) > self.max_length
-            else torch.LongTensor(instance["input_ids"])
+            (
+                torch.LongTensor(instance["input_ids"][: self.max_length])
+                if len(instance["input_ids"]) > self.max_length
+                else torch.LongTensor(instance["input_ids"])
+            )
             for instance in instances
         ]
         batch_labels = [
-            torch.LongTensor(instance["labels"][: self.max_length])
-            if len(instance["labels"]) > self.max_length
-            else torch.LongTensor(instance["labels"])
+            (
+                torch.LongTensor(instance["labels"][: self.max_length])
+                if len(instance["labels"]) > self.max_length
+                else torch.LongTensor(instance["labels"])
+            )
             for instance in instances
         ]
         if self.tokenizer.padding_side == "right":
diff --git a/applications/ColossalChat/coati/models/loss.py b/applications/ColossalChat/coati/models/loss.py
index aaef447a4383..e411dded148c 100755
--- a/applications/ColossalChat/coati/models/loss.py
+++ b/applications/ColossalChat/coati/models/loss.py
@@ -1,6 +1,7 @@
 """
 loss functions
 """
+
 from typing import Optional, Tuple
 
 import torch
diff --git a/applications/ColossalChat/coati/models/reward_model.py b/applications/ColossalChat/coati/models/reward_model.py
index 18c5eca41a71..f5712312ada2 100755
--- a/applications/ColossalChat/coati/models/reward_model.py
+++ b/applications/ColossalChat/coati/models/reward_model.py
@@ -1,6 +1,7 @@
 """
 reward model
 """
+
 from typing import Optional
 
 import torch
diff --git a/applications/ColossalChat/coati/trainer/utils.py b/applications/ColossalChat/coati/trainer/utils.py
index 5ce1e9ef009c..3c836b4b4db1 100755
--- a/applications/ColossalChat/coati/trainer/utils.py
+++ b/applications/ColossalChat/coati/trainer/utils.py
@@ -1,6 +1,7 @@
 """
 Training utilities for Coati.
 """
+
 from typing import Any
 
 import torch
diff --git a/applications/ColossalEval/colossal_eval/dataset/agieval.py b/applications/ColossalEval/colossal_eval/dataset/agieval.py
index 32f8544e93df..d5f2302494e8 100644
--- a/applications/ColossalEval/colossal_eval/dataset/agieval.py
+++ b/applications/ColossalEval/colossal_eval/dataset/agieval.py
@@ -78,7 +78,9 @@ def get_prompt(line: Dict, dataset_name: str, logger: DistributedLogger) -> Dict
             option_string = "ABCDEFG"
             count = len(line["options"])
 
-            input = "问题：" + line["question"] + " " + "从以下选项中选择：" + " ".join(line["options"]) + "\n" + "答案："
+            input = (
+                "问题：" + line["question"] + " " + "从以下选项中选择：" + " ".join(line["options"]) + "\n" + "答案："
+            )
 
             all_classes = list(option_string[0:count])
 
@@ -150,7 +152,15 @@ def combine_prompt(prompt_path, dataset_name, load_explanation=True, chat_mode=F
             )
         elif dataset_name in chinese_qa_datasets:
             question_input = (
-                "问题：" + passage + " " + question + "\n" + "从以下选项中选择：" + " ".join(options) + "\n" + "答案：{}".format(label)
+                "问题："
+                + passage
+                + " "
+                + question
+                + "\n"
+                + "从以下选项中选择："
+                + " ".join(options)
+                + "\n"
+                + "答案：{}".format(label)
             )
         elif dataset_name in english_cloze_datasets:
             question_input = "Question: ".format(idx + 1) + question + "\n" + "Answer: {}".format(answer)
diff --git a/applications/ColossalEval/colossal_eval/dataset/ceval.py b/applications/ColossalEval/colossal_eval/dataset/ceval.py
index 2cf09ec4dc2f..915f4d9b0850 100644
--- a/applications/ColossalEval/colossal_eval/dataset/ceval.py
+++ b/applications/ColossalEval/colossal_eval/dataset/ceval.py
@@ -57,7 +57,11 @@
     "urban_and_rural_planner": ["Urban and Rural Planner", "注册城乡规划师", "Other"],
     "accountant": ["Accountant", "注册会计师", "Other"],
     "fire_engineer": ["Fire Engineer", "注册消防工程师", "Other"],
-    "environmental_impact_assessment_engineer": ["Environmental Impact Assessment Engineer", "环境影响评价工程师", "Other"],
+    "environmental_impact_assessment_engineer": [
+        "Environmental Impact Assessment Engineer",
+        "环境影响评价工程师",
+        "Other",
+    ],
     "tax_accountant": ["Tax Accountant", "税务师", "Other"],
     "physician": ["Physician", "医师资格", "Other"],
 }
diff --git a/applications/ColossalEval/colossal_eval/dataset/mtbench.py b/applications/ColossalEval/colossal_eval/dataset/mtbench.py
index 9e74a4d826e3..03141556788f 100644
--- a/applications/ColossalEval/colossal_eval/dataset/mtbench.py
+++ b/applications/ColossalEval/colossal_eval/dataset/mtbench.py
@@ -56,9 +56,11 @@ def load(path: str, logger: DistributedLogger, few_shot: bool) -> List[Dict]:
                     "instruction": question["turns"],
                     "input": "",
                     "output": [],
-                    "target": [""] * turn_number
-                    if question["question_id"] not in reference
-                    else reference[question["question_id"]],
+                    "target": (
+                        [""] * turn_number
+                        if question["question_id"] not in reference
+                        else reference[question["question_id"]]
+                    ),
                 }
 
                 if category in dataset["test"]:
diff --git a/applications/ColossalEval/colossal_eval/models/huggingface.py b/applications/ColossalEval/colossal_eval/models/huggingface.py
index fff697e21e34..23c399ccedbd 100644
--- a/applications/ColossalEval/colossal_eval/models/huggingface.py
+++ b/applications/ColossalEval/colossal_eval/models/huggingface.py
@@ -77,7 +77,9 @@ def _get_choices_indices(self, language: str):
             self.indices_for_choices[0].append(
                 self.tokenizer(f"Answer: {choice}", add_special_tokens=False).input_ids[-1]
             )
-            self.indices_for_choices[1].append(self.tokenizer(f"答案：{choice}", add_special_tokens=False).input_ids[-1])
+            self.indices_for_choices[1].append(
+                self.tokenizer(f"答案：{choice}", add_special_tokens=False).input_ids[-1]
+            )
 
     def _load_tokenizer(self, path: str, tokenizer_path: Optional[str], tokenizer_kwargs: dict):
         """
diff --git a/applications/ColossalMoE/infer.py b/applications/ColossalMoE/infer.py
index c175fe9e3f3f..68ba61e3ea6f 100644
--- a/applications/ColossalMoE/infer.py
+++ b/applications/ColossalMoE/infer.py
@@ -96,7 +96,11 @@ def main():
     if coordinator.rank == 0:
         text = ["Hello my name is"]
     else:
-        text = ["What's the largest country in the world?", "How many people live in China?", "帮我续写这首诗：离离原上草"]
+        text = [
+            "What's the largest country in the world?",
+            "How many people live in China?",
+            "帮我续写这首诗：离离原上草",
+        ]
     tokenizer.pad_token = tokenizer.unk_token
     inputs = tokenizer(text, return_tensors="pt", padding=True).to(torch.cuda.current_device())
 
diff --git a/applications/ColossalMoE/train.py b/applications/ColossalMoE/train.py
index 850236726a27..749db8b5158b 100644
--- a/applications/ColossalMoE/train.py
+++ b/applications/ColossalMoE/train.py
@@ -195,9 +195,9 @@ def main():
     lr_scheduler = CosineAnnealingWarmupLR(
         optimizer=optimizer,
         total_steps=args.num_epochs * len(dataloader),
-        warmup_steps=args.warmup_steps
-        if args.warmup_steps is not None
-        else int(args.num_epochs * len(dataloader) * 0.025),
+        warmup_steps=(
+            args.warmup_steps if args.warmup_steps is not None else int(args.num_epochs * len(dataloader) * 0.025)
+        ),
         eta_min=0.1 * args.lr,
     )
 
diff --git a/applications/ColossalQA/colossalqa/chain/retrieval_qa/base.py b/applications/ColossalQA/colossalqa/chain/retrieval_qa/base.py
index 80dbf47def2b..2f9750de33fd 100644
--- a/applications/ColossalQA/colossalqa/chain/retrieval_qa/base.py
+++ b/applications/ColossalQA/colossalqa/chain/retrieval_qa/base.py
@@ -7,6 +7,7 @@
 https://github.com/langchain-ai/langchain
 The original code is licensed under the MIT license.
 """
+
 from __future__ import annotations
 
 import copy
diff --git a/applications/ColossalQA/colossalqa/chain/retrieval_qa/load_chain.py b/applications/ColossalQA/colossalqa/chain/retrieval_qa/load_chain.py
index a2b1f81e34b9..8cb8ef536b20 100644
--- a/applications/ColossalQA/colossalqa/chain/retrieval_qa/load_chain.py
+++ b/applications/ColossalQA/colossalqa/chain/retrieval_qa/load_chain.py
@@ -8,6 +8,7 @@
 https://github.com/langchain-ai/langchain
 The original code is licensed under the MIT license.
 """
+
 import copy
 from typing import Any, Mapping, Optional, Protocol
 
diff --git a/applications/ColossalQA/colossalqa/chain/retrieval_qa/stuff.py b/applications/ColossalQA/colossalqa/chain/retrieval_qa/stuff.py
index bf7ad0ffce28..64e476438576 100644
--- a/applications/ColossalQA/colossalqa/chain/retrieval_qa/stuff.py
+++ b/applications/ColossalQA/colossalqa/chain/retrieval_qa/stuff.py
@@ -7,6 +7,7 @@
 https://github.com/langchain-ai/langchain
 The original code is licensed under the MIT license.
 """
+
 import copy
 from typing import Any, List
 
diff --git a/applications/ColossalQA/colossalqa/data_loader/table_dataloader.py b/applications/ColossalQA/colossalqa/data_loader/table_dataloader.py
index 29542466fa8f..0ad66f0ad999 100644
--- a/applications/ColossalQA/colossalqa/data_loader/table_dataloader.py
+++ b/applications/ColossalQA/colossalqa/data_loader/table_dataloader.py
@@ -2,7 +2,6 @@
 Class for loading table type data. please refer to Pandas-Input/Output for file format details.
 """
 
-
 import glob
 import os
 
diff --git a/applications/ColossalQA/colossalqa/local/colossalcloud_llm.py b/applications/ColossalQA/colossalqa/local/colossalcloud_llm.py
index 3629778698fb..ca8d64f2293f 100644
--- a/applications/ColossalQA/colossalqa/local/colossalcloud_llm.py
+++ b/applications/ColossalQA/colossalqa/local/colossalcloud_llm.py
@@ -20,6 +20,7 @@
 print(resp)  # super-heavyweight awesome-natured yawning Australian creature!
 
 """
+
 import json
 from typing import Any, Mapping
 
diff --git a/applications/ColossalQA/colossalqa/local/llm.py b/applications/ColossalQA/colossalqa/local/llm.py
index 30a456c3d9c7..58a4811d9fdc 100644
--- a/applications/ColossalQA/colossalqa/local/llm.py
+++ b/applications/ColossalQA/colossalqa/local/llm.py
@@ -12,6 +12,7 @@
 logger.info(llm(TEST_PROMPT_CHATGLM, max_new_tokens=100), verbose=True)
 
 """
+
 from typing import Any, List, Mapping, Optional
 
 import torch
diff --git a/applications/ColossalQA/colossalqa/local/utils.py b/applications/ColossalQA/colossalqa/local/utils.py
index ed90264cad8d..2cbd474bdbd2 100644
--- a/applications/ColossalQA/colossalqa/local/utils.py
+++ b/applications/ColossalQA/colossalqa/local/utils.py
@@ -1,6 +1,7 @@
 """
 Generation utilities
 """
+
 import json
 from typing import List
 
diff --git a/applications/ColossalQA/colossalqa/memory.py b/applications/ColossalQA/colossalqa/memory.py
index 7a5512281035..d8de544a59e6 100644
--- a/applications/ColossalQA/colossalqa/memory.py
+++ b/applications/ColossalQA/colossalqa/memory.py
@@ -2,6 +2,7 @@
 Implement a memory class for storing conversation history
 Support long term and short term memory
 """
+
 from typing import Any, Dict, List
 
 from colossalqa.chain.memory.summary import ConversationSummaryMemory
diff --git a/applications/ColossalQA/colossalqa/mylogging.py b/applications/ColossalQA/colossalqa/mylogging.py
index 574c33b41685..67e2a83ed141 100644
--- a/applications/ColossalQA/colossalqa/mylogging.py
+++ b/applications/ColossalQA/colossalqa/mylogging.py
@@ -1,6 +1,7 @@
 """
 Class for logging with extra control for debugging
 """
+
 import logging
 
 
diff --git a/applications/ColossalQA/colossalqa/retrieval_conversation_en.py b/applications/ColossalQA/colossalqa/retrieval_conversation_en.py
index 96bce82b9ee0..cab16807579e 100644
--- a/applications/ColossalQA/colossalqa/retrieval_conversation_en.py
+++ b/applications/ColossalQA/colossalqa/retrieval_conversation_en.py
@@ -1,6 +1,7 @@
 """
 Script for Chinese retrieval based conversation system backed by ChatGLM
 """
+
 from typing import Tuple
 
 from colossalqa.chain.retrieval_qa.base import RetrievalQA
diff --git a/applications/ColossalQA/colossalqa/retrieval_conversation_universal.py b/applications/ColossalQA/colossalqa/retrieval_conversation_universal.py
index 6e77bb2aee17..a991b202e8ee 100644
--- a/applications/ColossalQA/colossalqa/retrieval_conversation_universal.py
+++ b/applications/ColossalQA/colossalqa/retrieval_conversation_universal.py
@@ -1,6 +1,7 @@
 """
 Multilingual retrieval based conversation system
 """
+
 from typing import List
 
 from colossalqa.data_loader.document_loader import DocumentLoader
diff --git a/applications/ColossalQA/colossalqa/retrieval_conversation_zh.py b/applications/ColossalQA/colossalqa/retrieval_conversation_zh.py
index 4eef41947d11..6c9b69117f8a 100644
--- a/applications/ColossalQA/colossalqa/retrieval_conversation_zh.py
+++ b/applications/ColossalQA/colossalqa/retrieval_conversation_zh.py
@@ -1,6 +1,7 @@
 """
 Script for Chinese retrieval based conversation system backed by ChatGLM
 """
+
 from typing import Tuple
 
 from colossalqa.chain.retrieval_qa.base import RetrievalQA
diff --git a/applications/ColossalQA/colossalqa/retriever.py b/applications/ColossalQA/colossalqa/retriever.py
index 6a0c69859ac7..ec4941ddd0a7 100644
--- a/applications/ColossalQA/colossalqa/retriever.py
+++ b/applications/ColossalQA/colossalqa/retriever.py
@@ -1,6 +1,7 @@
 """
 Code for custom retriver with incremental update
 """
+
 import copy
 import hashlib
 import os
diff --git a/applications/ColossalQA/colossalqa/text_splitter/chinese_text_splitter.py b/applications/ColossalQA/colossalqa/text_splitter/chinese_text_splitter.py
index 3815f5ed2621..697af484b3fc 100644
--- a/applications/ColossalQA/colossalqa/text_splitter/chinese_text_splitter.py
+++ b/applications/ColossalQA/colossalqa/text_splitter/chinese_text_splitter.py
@@ -1,6 +1,7 @@
 """
 Code for Chinese text splitter
 """
+
 from typing import Any, List, Optional
 
 from colossalqa.text_splitter.utils import get_cleaned_paragraph
diff --git a/applications/ColossalQA/examples/retrieval_conversation_en.py b/applications/ColossalQA/examples/retrieval_conversation_en.py
index fe2b9b4db3c2..b7339de933bb 100644
--- a/applications/ColossalQA/examples/retrieval_conversation_en.py
+++ b/applications/ColossalQA/examples/retrieval_conversation_en.py
@@ -1,6 +1,7 @@
 """
 Script for English retrieval based conversation system backed by LLaMa2
 """
+
 import argparse
 import os
 
diff --git a/applications/ColossalQA/examples/retrieval_conversation_en_customer_service.py b/applications/ColossalQA/examples/retrieval_conversation_en_customer_service.py
index d4ba73b9468c..a0c90e7c5d8f 100644
--- a/applications/ColossalQA/examples/retrieval_conversation_en_customer_service.py
+++ b/applications/ColossalQA/examples/retrieval_conversation_en_customer_service.py
@@ -1,6 +1,7 @@
 """
 Script for English retrieval based conversation system backed by LLaMa2
 """
+
 import argparse
 import json
 import os
diff --git a/applications/ColossalQA/examples/retrieval_conversation_zh.py b/applications/ColossalQA/examples/retrieval_conversation_zh.py
index b143b9baacc1..96641edf5290 100644
--- a/applications/ColossalQA/examples/retrieval_conversation_zh.py
+++ b/applications/ColossalQA/examples/retrieval_conversation_zh.py
@@ -1,6 +1,7 @@
 """
 Script for Chinese retrieval based conversation system backed by ChatGLM
 """
+
 import argparse
 import os
 
diff --git a/applications/ColossalQA/examples/retrieval_intent_classification_zh_customer_service.py b/applications/ColossalQA/examples/retrieval_intent_classification_zh_customer_service.py
index adb6544941f0..865ade5bb2d2 100644
--- a/applications/ColossalQA/examples/retrieval_intent_classification_zh_customer_service.py
+++ b/applications/ColossalQA/examples/retrieval_intent_classification_zh_customer_service.py
@@ -1,6 +1,7 @@
 """
 Script for English retrieval based conversation system backed by LLaMa2
 """
+
 import argparse
 import os
 
diff --git a/colossalai/auto_parallel/meta_profiler/meta_registry/conv.py b/colossalai/auto_parallel/meta_profiler/meta_registry/conv.py
index 2f630995cdbc..b1e32e885783 100644
--- a/colossalai/auto_parallel/meta_profiler/meta_registry/conv.py
+++ b/colossalai/auto_parallel/meta_profiler/meta_registry/conv.py
@@ -107,20 +107,22 @@ def convnd_meta_info(*args, **kwargs) -> Tuple[TrainCycleItem, TrainCycleItem, L
     # NOTE: currently in SPMD solver we always believe that there will be a new tensor created in forward
     fwd_memory_cost = MemoryCost(
         activation=compute_size_in_bytes([input_tensor, output_tensor]),
-        parameter=compute_size_in_bytes([weight_tensor, bias_tensor])
-        if has_bias
-        else compute_size_in_bytes(weight_tensor),
+        parameter=(
+            compute_size_in_bytes([weight_tensor, bias_tensor]) if has_bias else compute_size_in_bytes(weight_tensor)
+        ),
         temp=0,
         buffer=0,
     )
 
     bwd_memory_cost = MemoryCost(
-        activation=compute_size_in_bytes([input_tensor, weight_tensor, bias_tensor])
-        if has_bias
-        else compute_size_in_bytes([input_tensor, weight_tensor]),
-        parameter=compute_size_in_bytes([weight_tensor, bias_tensor])
-        if has_bias
-        else compute_size_in_bytes(weight_tensor),
+        activation=(
+            compute_size_in_bytes([input_tensor, weight_tensor, bias_tensor])
+            if has_bias
+            else compute_size_in_bytes([input_tensor, weight_tensor])
+        ),
+        parameter=(
+            compute_size_in_bytes([weight_tensor, bias_tensor]) if has_bias else compute_size_in_bytes(weight_tensor)
+        ),
         temp=0,
         buffer=0,
     )
diff --git a/colossalai/inference/engine/modeling/_utils.py b/colossalai/inference/engine/modeling/_utils.py
index 068b64b4f829..46d4222c4ac2 100644
--- a/colossalai/inference/engine/modeling/_utils.py
+++ b/colossalai/inference/engine/modeling/_utils.py
@@ -1,6 +1,7 @@
 """
 Utils for model inference
 """
+
 import os
 
 import torch
diff --git a/colossalai/inference/kv_cache/batch_infer_state.py b/colossalai/inference/kv_cache/batch_infer_state.py
index f707a86df37e..b72610899abc 100644
--- a/colossalai/inference/kv_cache/batch_infer_state.py
+++ b/colossalai/inference/kv_cache/batch_infer_state.py
@@ -14,6 +14,7 @@ class BatchInferState:
     Information to be passed and used for a batch of inputs during
     a single model forward
     """
+
     batch_size: int
     max_len_in_batch: int
 
diff --git a/colossalai/inference/kv_cache/kvcache_manager.py b/colossalai/inference/kv_cache/kvcache_manager.py
index dda46a756cc3..467f8d441aa2 100644
--- a/colossalai/inference/kv_cache/kvcache_manager.py
+++ b/colossalai/inference/kv_cache/kvcache_manager.py
@@ -4,6 +4,7 @@
 https://github.com/ModelTC/lightllm/blob/050af3ce65edca617e2f30ec2479397d5bb248c9/lightllm/common/mem_manager.py
 we slightly changed it to make it suitable for our colossal-ai shardformer TP-engine design.
 """
+
 import torch
 from transformers.utils import logging
 
diff --git a/colossalai/legacy/context/process_group_initializer/initializer_2d.py b/colossalai/legacy/context/process_group_initializer/initializer_2d.py
index 1c08d4d4296a..fc51844b661f 100644
--- a/colossalai/legacy/context/process_group_initializer/initializer_2d.py
+++ b/colossalai/legacy/context/process_group_initializer/initializer_2d.py
@@ -138,9 +138,7 @@ def __init__(self, *args, **kwargs):
         self.num_group = self.world_size // self.tensor_parallel_size
         self.summa_dim = int(math.sqrt(self.tensor_parallel_size))
 
-        assert (
-            self.tensor_parallel_size == self.summa_dim**2
-        ), "2D summa dim should equal to tensor parallel size ^ 0.5"
+        assert self.tensor_parallel_size == self.summa_dim**2, "2D summa dim should equal to tensor parallel size ^ 0.5"
         _check_summa_env_var(self.summa_dim)
 
         self.col_initializer = Initializer_2D_Col(self.num_group, self.summa_dim, *args, **kwargs)
diff --git a/colossalai/legacy/inference/async_engine.py b/colossalai/legacy/inference/async_engine.py
index d0890ba3e9fc..b4c523669af2 100644
--- a/colossalai/legacy/inference/async_engine.py
+++ b/colossalai/legacy/inference/async_engine.py
@@ -54,7 +54,6 @@ async def __anext__(self) -> RequestOutput:
 
 
 class Async_Engine:
-
     """
     Use an engine to launch RAY Driver --> RAY Worker --> Async_Manager
     Background loop: inference reqs in waiting list (Listen)
diff --git a/colossalai/legacy/inference/dynamic_batching/io_struct.py b/colossalai/legacy/inference/dynamic_batching/io_struct.py
index fc5ecfe5796b..abc41cc8e909 100644
--- a/colossalai/legacy/inference/dynamic_batching/io_struct.py
+++ b/colossalai/legacy/inference/dynamic_batching/io_struct.py
@@ -118,16 +118,16 @@ def __len__(self):
 
 class BatchTokenIdOut:
     def __init__(self):
-        self.reqs_infs: List[
-            Tuple[str, int, Dict, bool, bool]
-        ] = []  # [req_id, new_token_id, gen_metadata, finished_state, abort_state]
+        self.reqs_infs: List[Tuple[str, int, Dict, bool, bool]] = (
+            []
+        )  # [req_id, new_token_id, gen_metadata, finished_state, abort_state]
 
 
 class BatchStrOut:
     def __init__(self):
-        self.reqs_infs: List[
-            Tuple[str, str, Dict, bool, bool]
-        ] = []  # [req_id, token_str, gen_metadata, finished_state, abort_state]
+        self.reqs_infs: List[Tuple[str, str, Dict, bool, bool]] = (
+            []
+        )  # [req_id, token_str, gen_metadata, finished_state, abort_state]
 
 
 class AbortReq:
diff --git a/colossalai/legacy/inference/hybridengine/modeling/_utils.py b/colossalai/legacy/inference/hybridengine/modeling/_utils.py
index 068b64b4f829..46d4222c4ac2 100644
--- a/colossalai/legacy/inference/hybridengine/modeling/_utils.py
+++ b/colossalai/legacy/inference/hybridengine/modeling/_utils.py
@@ -1,6 +1,7 @@
 """
 Utils for model inference
 """
+
 import os
 
 import torch
diff --git a/colossalai/legacy/inference/tensor_parallel/batch_infer_state.py b/colossalai/legacy/inference/tensor_parallel/batch_infer_state.py
index f707a86df37e..b72610899abc 100644
--- a/colossalai/legacy/inference/tensor_parallel/batch_infer_state.py
+++ b/colossalai/legacy/inference/tensor_parallel/batch_infer_state.py
@@ -14,6 +14,7 @@ class BatchInferState:
     Information to be passed and used for a batch of inputs during
     a single model forward
     """
+
     batch_size: int
     max_len_in_batch: int
 
diff --git a/colossalai/legacy/inference/tensor_parallel/kvcache_manager.py b/colossalai/legacy/inference/tensor_parallel/kvcache_manager.py
index 91bb96a1f1f0..8c54fda2602a 100644
--- a/colossalai/legacy/inference/tensor_parallel/kvcache_manager.py
+++ b/colossalai/legacy/inference/tensor_parallel/kvcache_manager.py
@@ -4,6 +4,7 @@
 https://github.com/ModelTC/lightllm/blob/050af3ce65edca617e2f30ec2479397d5bb248c9/lightllm/common/mem_manager.py
 we slightly changed it to make it suitable for our colossal-ai shardformer TP-engine design.
 """
+
 import torch
 from transformers.utils import logging
 
diff --git a/colossalai/legacy/inference/tensor_parallel/modeling/_utils.py b/colossalai/legacy/inference/tensor_parallel/modeling/_utils.py
index 068b64b4f829..46d4222c4ac2 100644
--- a/colossalai/legacy/inference/tensor_parallel/modeling/_utils.py
+++ b/colossalai/legacy/inference/tensor_parallel/modeling/_utils.py
@@ -1,6 +1,7 @@
 """
 Utils for model inference
 """
+
 import os
 
 import torch
diff --git a/colossalai/shardformer/examples/performance_benchmark.py b/colossalai/shardformer/examples/performance_benchmark.py
index 81215dcdf5d4..981525359c91 100644
--- a/colossalai/shardformer/examples/performance_benchmark.py
+++ b/colossalai/shardformer/examples/performance_benchmark.py
@@ -1,6 +1,7 @@
 """
 Shardformer Benchmark
 """
+
 import torch
 import torch.distributed as dist
 import transformers
diff --git a/colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py b/colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py
index bf581300a7b1..6ae4b06e517a 100644
--- a/colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py
+++ b/colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py
@@ -33,6 +33,7 @@
 
 Note that the license is subject to update to a more comprehensive version.  For any questions related to the license and copyright, please contact us at glm-130b@googlegroups.com.
 """
+
 """ PyTorch ChatGLM model. """
 
 import copy
diff --git a/colossalai/shardformer/policies/base_policy.py b/colossalai/shardformer/policies/base_policy.py
index d67ab0a3c6bb..626dea86b66d 100644
--- a/colossalai/shardformer/policies/base_policy.py
+++ b/colossalai/shardformer/policies/base_policy.py
@@ -28,6 +28,7 @@ class SubModuleReplacementDescription:
         kwargs (Dict[str, Any]): the dictionary used to pass extra arguments to the `ParallelModule.from_native_module` method.
         ignore_if_not_exist (bool): if the submodule does not exist, ignore it or raise an exception
     """
+
     suffix: str
     target_module: Union[ParallelModule, BaseLayerNorm]
     kwargs: Dict[str, Any] = None
@@ -54,6 +55,7 @@ def example_replace_weight(module: torch.nn.Module):
                     object which specifies the module to be replaced and the target module used to replacement.
         method_replace (Dict[str, Callable]): key is the method name, value is the method for replacement
     """
+
     attribute_replacement: Dict[str, Any] = None
     param_replacement: List[Callable] = None
     sub_module_replacement: List[SubModuleReplacementDescription] = None
diff --git a/colossalai/shardformer/shard/grad_ckpt_config.py b/colossalai/shardformer/shard/grad_ckpt_config.py
index 9c6c2b54ea39..f2d18f3af8d2 100644
--- a/colossalai/shardformer/shard/grad_ckpt_config.py
+++ b/colossalai/shardformer/shard/grad_ckpt_config.py
@@ -22,6 +22,7 @@ class PipelineGradientCheckpointConfig(GradientCheckpointConfig):
         2. Customize # ckpt layers assigned to each stage. This takes precedence over `gradient_checkpointing_ratio`.
 
     """
+
     """
     Args:
         gradient_checkpointing_ratio (Optional[float]): The ratio of gradient checkpointing. It can only be used in pipeline parallelism. Defaults to None.
diff --git a/colossalai/shardformer/shard/shard_config.py b/colossalai/shardformer/shard/shard_config.py
index 7489873c2ed6..c0f28c947f8b 100644
--- a/colossalai/shardformer/shard/shard_config.py
+++ b/colossalai/shardformer/shard/shard_config.py
@@ -30,6 +30,7 @@ class ShardConfig:
         gradient_checkpoint_config (Optional[GradientCheckpointConfig]): The gradient checkpoint config. Defaults to None.
         enable_all_optimization (bool): Whether to turn on all optimization tools including 'fused normalization', 'flash attention', 'JIT fused operators', 'sequence parallelism' and 'sequence overlap'. Defaults to False.
     """
+
     tensor_parallel_process_group: Optional[ProcessGroup] = None
     sequence_parallel_process_group: Optional[ProcessGroup] = None
     pipeline_stage_manager: Optional[PipelineStageManager] = None
diff --git a/examples/community/roberta/pretraining/nvidia_bert_dataset_provider.py b/examples/community/roberta/pretraining/nvidia_bert_dataset_provider.py
index 09677a6195cb..4d08d9941133 100644
--- a/examples/community/roberta/pretraining/nvidia_bert_dataset_provider.py
+++ b/examples/community/roberta/pretraining/nvidia_bert_dataset_provider.py
@@ -52,9 +52,11 @@ def __len__(self):
 
     def __getitem__(self, index):
         [input_ids, input_mask, segment_ids, masked_lm_labels] = [
-            torch.from_numpy(input[index].astype(np.int64))
-            if indice < 5
-            else torch.from_numpy(np.asarray(input[index].astype(np.int64)))
+            (
+                torch.from_numpy(input[index].astype(np.int64))
+                if indice < 5
+                else torch.from_numpy(np.asarray(input[index].astype(np.int64)))
+            )
             for indice, input in enumerate(self.inputs)
         ]
 
diff --git a/examples/images/diffusion/ldm/models/diffusion/ddpm.py b/examples/images/diffusion/ldm/models/diffusion/ddpm.py
index 20e26256e18e..3cf6aceb5197 100644
--- a/examples/images/diffusion/ldm/models/diffusion/ddpm.py
+++ b/examples/images/diffusion/ldm/models/diffusion/ddpm.py
@@ -229,9 +229,7 @@ def register_schedule(
         )
 
         if self.parameterization == "eps":
-            lvlb_weights = self.betas**2 / (
-                2 * self.posterior_variance * to_torch(alphas) * (1 - self.alphas_cumprod)
-            )
+            lvlb_weights = self.betas**2 / (2 * self.posterior_variance * to_torch(alphas) * (1 - self.alphas_cumprod))
         elif self.parameterization == "x0":
             lvlb_weights = 0.5 * np.sqrt(torch.Tensor(alphas_cumprod)) / (2.0 * 1 - torch.Tensor(alphas_cumprod))
         elif self.parameterization == "v":
@@ -1186,9 +1184,11 @@ def progressive_denoising(
         if cond is not None:
             if isinstance(cond, dict):
                 cond = {
-                    key: cond[key][:batch_size]
-                    if not isinstance(cond[key], list)
-                    else list(map(lambda x: x[:batch_size], cond[key]))
+                    key: (
+                        cond[key][:batch_size]
+                        if not isinstance(cond[key], list)
+                        else list(map(lambda x: x[:batch_size], cond[key]))
+                    )
                     for key in cond
                 }
             else:
@@ -1321,9 +1321,11 @@ def sample(
         if cond is not None:
             if isinstance(cond, dict):
                 cond = {
-                    key: cond[key][:batch_size]
-                    if not isinstance(cond[key], list)
-                    else list(map(lambda x: x[:batch_size], cond[key]))
+                    key: (
+                        cond[key][:batch_size]
+                        if not isinstance(cond[key], list)
+                        else list(map(lambda x: x[:batch_size], cond[key]))
+                    )
                     for key in cond
                 }
             else:
diff --git a/examples/images/diffusion/ldm/models/diffusion/dpm_solver/sampler.py b/examples/images/diffusion/ldm/models/diffusion/dpm_solver/sampler.py
index 55dac8555e5f..4104fe3b0df4 100644
--- a/examples/images/diffusion/ldm/models/diffusion/dpm_solver/sampler.py
+++ b/examples/images/diffusion/ldm/models/diffusion/dpm_solver/sampler.py
@@ -1,4 +1,5 @@
 """SAMPLING ONLY."""
+
 import torch
 
 from .dpm_solver import DPM_Solver, NoiseScheduleVP, model_wrapper
diff --git a/examples/images/diffusion/ldm/modules/diffusionmodules/openaimodel.py b/examples/images/diffusion/ldm/modules/diffusionmodules/openaimodel.py
index 6c80f3229ce3..afde5dfd4430 100644
--- a/examples/images/diffusion/ldm/modules/diffusionmodules/openaimodel.py
+++ b/examples/images/diffusion/ldm/modules/diffusionmodules/openaimodel.py
@@ -640,23 +640,25 @@ def __init__(
                 use_checkpoint=use_checkpoint,
                 use_scale_shift_norm=use_scale_shift_norm,
             ),
-            AttentionBlock(
-                ch,
-                use_checkpoint=use_checkpoint,
-                num_heads=num_heads,
-                num_head_channels=dim_head,
-                use_new_attention_order=use_new_attention_order,
-            )
-            if not use_spatial_transformer
-            else SpatialTransformer(  # always uses a self-attn
-                ch,
-                num_heads,
-                dim_head,
-                depth=transformer_depth,
-                context_dim=context_dim,
-                disable_self_attn=disable_middle_self_attn,
-                use_linear=use_linear_in_transformer,
-                use_checkpoint=use_checkpoint,
+            (
+                AttentionBlock(
+                    ch,
+                    use_checkpoint=use_checkpoint,
+                    num_heads=num_heads,
+                    num_head_channels=dim_head,
+                    use_new_attention_order=use_new_attention_order,
+                )
+                if not use_spatial_transformer
+                else SpatialTransformer(  # always uses a self-attn
+                    ch,
+                    num_heads,
+                    dim_head,
+                    depth=transformer_depth,
+                    context_dim=context_dim,
+                    disable_self_attn=disable_middle_self_attn,
+                    use_linear=use_linear_in_transformer,
+                    use_checkpoint=use_checkpoint,
+                )
             ),
             ResBlock(
                 ch,
diff --git a/examples/images/diffusion/ldm/modules/midas/midas/midas_net.py b/examples/images/diffusion/ldm/modules/midas/midas/midas_net.py
index 0dd87b59619c..8c13f39ff48f 100644
--- a/examples/images/diffusion/ldm/modules/midas/midas/midas_net.py
+++ b/examples/images/diffusion/ldm/modules/midas/midas/midas_net.py
@@ -2,6 +2,7 @@
 This file contains code that is adapted from
 https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py
 """
+
 import torch
 import torch.nn as nn
 
diff --git a/examples/images/diffusion/ldm/modules/midas/midas/midas_net_custom.py b/examples/images/diffusion/ldm/modules/midas/midas/midas_net_custom.py
index 4d30744c46d3..c79581afcd2d 100644
--- a/examples/images/diffusion/ldm/modules/midas/midas/midas_net_custom.py
+++ b/examples/images/diffusion/ldm/modules/midas/midas/midas_net_custom.py
@@ -2,6 +2,7 @@
 This file contains code that is adapted from
 https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py
 """
+
 import torch
 import torch.nn as nn
 
diff --git a/examples/images/diffusion/ldm/modules/midas/utils.py b/examples/images/diffusion/ldm/modules/midas/utils.py
index 1428d42b2445..f7fc7dcc98a4 100644
--- a/examples/images/diffusion/ldm/modules/midas/utils.py
+++ b/examples/images/diffusion/ldm/modules/midas/utils.py
@@ -1,4 +1,5 @@
 """Utils for monoDepth."""
+
 import re
 import sys
 
diff --git a/examples/tutorial/sequence_parallel/data/datasets/helpers.cpp b/examples/tutorial/sequence_parallel/data/datasets/helpers.cpp
index 52977e63181f..fe9968177fb1 100644
--- a/examples/tutorial/sequence_parallel/data/datasets/helpers.cpp
+++ b/examples/tutorial/sequence_parallel/data/datasets/helpers.cpp
@@ -369,9 +369,9 @@ py::array build_mapping_impl(const py::array_t<int64_t>& docs_,
             }
 
           }  // for (auto sent_index=sent_index_first; ...
-        }    // if (num_remain_sent > 1) {
-      }      // for (int doc=0; doc < num_docs; ++doc) {
-    }        // for (int epoch=0; epoch < num_epochs; ++epoch) {
+        }  // if (num_remain_sent > 1) {
+      }  // for (int doc=0; doc < num_docs; ++doc) {
+    }  // for (int epoch=0; epoch < num_epochs; ++epoch) {
 
     if (!second) {
       if (verbose) {
@@ -606,9 +606,9 @@ py::array build_blocks_mapping_impl(
               num_sent = 0;
             }
           }  // for (auto sent_index=sent_index_first; ...
-        }    // if (num_remain_sent > 1) {
-      }      // for (int doc=0; doc < num_docs; ++doc) {
-    }        // for (int epoch=0; epoch < num_epochs; ++epoch) {
+        }  // if (num_remain_sent > 1) {
+      }  // for (int doc=0; doc < num_docs; ++doc) {
+    }  // for (int epoch=0; epoch < num_epochs; ++epoch) {
 
     if (!second) {
       if (verbose) {
diff --git a/extensions/csrc/arm/cpu_adam_arm.h b/extensions/csrc/arm/cpu_adam_arm.h
index c731850edc31..d48968e21682 100644
--- a/extensions/csrc/arm/cpu_adam_arm.h
+++ b/extensions/csrc/arm/cpu_adam_arm.h
@@ -4,7 +4,7 @@
 
 #include <cmath>
 
-#define ROUND_DOWN(size, step) ((size) & ~((step)-1))
+#define ROUND_DOWN(size, step) ((size) & ~((step) - 1))
 #define TILE (128 * 1024 * 1024)
 
 #if defined(__aarch64__)
diff --git a/extensions/csrc/cuda/cpu_adam.h b/extensions/csrc/cuda/cpu_adam.h
index db1f26d5f6da..45e1dde6242d 100644
--- a/extensions/csrc/cuda/cpu_adam.h
+++ b/extensions/csrc/cuda/cpu_adam.h
@@ -32,7 +32,7 @@ SOFTWARE
 #include <x86intrin.h>
 #endif
 
-#define ROUND_DOWN(size, step) ((size) & ~((step)-1))
+#define ROUND_DOWN(size, step) ((size) & ~((step) - 1))
 #define TILE (128 * 1024 * 1024)
 
 #if defined(__AVX512__) or defined(__AVX256__) or defined(__AVX2__)
diff --git a/tests/kit/model_zoo/torchvision/torchvision.py b/tests/kit/model_zoo/torchvision/torchvision.py
index 57b633e9d676..c0524d089cfe 100644
--- a/tests/kit/model_zoo/torchvision/torchvision.py
+++ b/tests/kit/model_zoo/torchvision/torchvision.py
@@ -34,14 +34,14 @@ def swin_s():
 
 
 # special output transform fn
-google_net_output_transform_fn = (
-    lambda x: dict(output=sum(x)) if isinstance(x, torchvision.models.GoogLeNetOutputs) else dict(output=x)
+google_net_output_transform_fn = lambda x: (
+    dict(output=sum(x)) if isinstance(x, torchvision.models.GoogLeNetOutputs) else dict(output=x)
 )
-swin_s_output_output_transform_fn = (
-    lambda x: {f"output{idx}": val for idx, val in enumerate(x)} if isinstance(x, tuple) else dict(output=x)
+swin_s_output_output_transform_fn = lambda x: (
+    {f"output{idx}": val for idx, val in enumerate(x)} if isinstance(x, tuple) else dict(output=x)
 )
-inception_v3_output_transform_fn = (
-    lambda x: dict(output=sum(x)) if isinstance(x, torchvision.models.InceptionOutputs) else dict(output=x)
+inception_v3_output_transform_fn = lambda x: (
+    dict(output=sum(x)) if isinstance(x, torchvision.models.InceptionOutputs) else dict(output=x)
 )
 
 model_zoo.register(
diff --git a/tests/test_lazy/test_models.py b/tests/test_lazy/test_models.py
index d0c4cd0a7c48..d678f44089a8 100644
--- a/tests/test_lazy/test_models.py
+++ b/tests/test_lazy/test_models.py
@@ -7,9 +7,11 @@
 @pytest.mark.skipif(not SUPPORT_LAZY, reason="requires torch >= 1.12.0")
 @pytest.mark.parametrize(
     "subset",
-    [COMMON_MODELS]
-    if IS_FAST_TEST
-    else ["torchvision", "diffusers", "timm", "transformers", "torchaudio", "deepfm", "dlrm"],
+    (
+        [COMMON_MODELS]
+        if IS_FAST_TEST
+        else ["torchvision", "diffusers", "timm", "transformers", "torchaudio", "deepfm", "dlrm"]
+    ),
 )
 @pytest.mark.parametrize("default_device", ["cpu", "cuda"])
 def test_torchvision_models_lazy_init(subset, default_device):