🧑‍💻 Import Ordering, Some Flake fixes

.pre-commit-config.yaml

@@ -15,4 +15,9 @@ repos:
     -   id: flake8
         #additional_dependencies: [flake8-import-order]
         args: ['--select=F,E9']
+  - repo: https://github.com/asottile/reorder-python-imports
+    rev: v3.10.0
+    hooks:
+    -   id: reorder-python-imports
+        args: ['--py38-plus']
 

base_bert.py

@@ -1,6 +1,10 @@
 import re
-from torch import device, dtype
-from config import BertConfig, PretrainedConfig
+
+from torch import device
+from torch import dtype
+
+from config import BertConfig
+from config import PretrainedConfig
 from utils import *
 
 

bert.py

@@ -1,8 +1,15 @@
-from typing import Dict, List, Optional, Union, Tuple, Callable
 import math
+from typing import Callable
+from typing import Dict
+from typing import List
+from typing import Optional
+from typing import Tuple
+from typing import Union
+
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
+
 from base_bert import BertPreTrainedModel
 from utils import *
 

classifier.py

@@ -1,23 +1,28 @@
-import time, random, numpy as np, argparse, sys, re, os
+import argparse
+import csv
+import random
 from contextlib import nullcontext
 from datetime import datetime
 from types import SimpleNamespace
-import csv
 
+import numpy as np
 import torch
 import torch.nn.functional as F
+from sklearn.metrics import accuracy_score
+from sklearn.metrics import f1_score
 from torch import nn
-from torch.utils.data import Dataset, DataLoader
-from sklearn.metrics import classification_report, f1_score, recall_score, accuracy_score
+from torch.utils.data import DataLoader
+from torch.utils.data import Dataset
 from torch.utils.tensorboard import SummaryWriter
+from tqdm import tqdm
 
 from AttentionLayer import AttentionLayer
+from bert import BertModel
+from optimizer import AdamW
+from optimizer import SophiaG
+from tokenizer import BertTokenizer
 
 # change it with respect to the original model
-from tokenizer import BertTokenizer
-from bert import BertModel
-from optimizer import AdamW, SophiaG
-from tqdm import tqdm
 
 TQDM_DISABLE = False
 

config.py

@@ -1,9 +1,14 @@
-from typing import Union, Tuple, Dict, Any, Optional
-import os
 import json
-from collections import OrderedDict
-import torch
-from utils import CONFIG_NAME, hf_bucket_url, cached_path, is_remote_url
+import os
+from typing import Any
+from typing import Dict
+from typing import Tuple
+from typing import Union
+
+from utils import cached_path
+from utils import CONFIG_NAME
+from utils import hf_bucket_url
+from utils import is_remote_url
 
 
 class PretrainedConfig(object):
@@ -168,7 +173,7 @@ def get_config_dict(
             # Load config dict
             config_dict = cls._dict_from_json_file(resolved_config_file)
 
-        except EnvironmentError as err:
+        except EnvironmentError:
             msg = (
                 f"Can't load config for '{pretrained_model_name_or_path}'. Make sure that:\n\n"
                 f"- '{pretrained_model_name_or_path}' is a correct model identifier listed on 'https://huggingface.co/models'\n\n"

datasets.py

@@ -1,19 +1,17 @@
 #!/usr/bin/env python3
-
 """
 This module contains our Dataset classes and functions to load the 3 datasets we're using.
 
 You should only need to call load_multitask_data to get the training and dev examples
 to train your model.
 """
-
 import csv
+import random
 
 import torch
 from torch.utils.data import Dataset
+
 from tokenizer import BertTokenizer
-from random import randrange
-import random
 
 
 def preprocess_string(s):
@@ -263,9 +261,9 @@ def collate_fn(self, all_data):
 
 
 def load_multitask_test_data():
-    paraphrase_filename = f"data/quora-test.csv"
-    sentiment_filename = f"data/ids-sst-test.txt"
-    similarity_filename = f"data/sts-test.csv"
+    paraphrase_filename = "data/quora-test.csv"
+    sentiment_filename = "data/ids-sst-test.txt"
+    similarity_filename = "data/sts-test.csv"
 
     sentiment_data = []
 

evaluation.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-
 """
 Model evaluation functions.
 
@@ -14,21 +13,18 @@
 so unless you change it you shouldn't need to call anything from here
 explicitly aside from model_eval_multitask.
 """
-
+import numpy as np
 import torch
+from sklearn.metrics import accuracy_score
+from sklearn.metrics import f1_score
 from torch.utils.data import DataLoader
-from sklearn.metrics import classification_report, f1_score, recall_score, accuracy_score
 from tqdm import tqdm
-import numpy as np
 
-from datasets import (
-    load_multitask_data,
-    load_multitask_test_data,
-    SentenceClassificationDataset,
-    SentenceClassificationTestDataset,
-    SentencePairDataset,
-    SentencePairTestDataset,
-)
+from datasets import load_multitask_data
+from datasets import SentenceClassificationDataset
+from datasets import SentenceClassificationTestDataset
+from datasets import SentencePairDataset
+from datasets import SentencePairTestDataset
 
 TQDM_DISABLE = False
 
@@ -40,7 +36,7 @@ def model_eval_sst(dataloader, model, device):
     y_pred = []
     sents = []
     sent_ids = []
-    for step, batch in enumerate(tqdm(dataloader, desc=f"eval", disable=TQDM_DISABLE)):
+    for step, batch in enumerate(tqdm(dataloader, desc="eval", disable=TQDM_DISABLE)):
         b_ids, b_mask, b_labels, b_sents, b_sent_ids = (
             batch["token_ids"],
             batch["attention_mask"],
@@ -82,7 +78,7 @@ def model_eval_multitask(
         # Evaluate paraphrase detection.
         if paraphrase_dataloader:
             for step, batch in enumerate(
-                tqdm(paraphrase_dataloader, desc=f"eval", disable=TQDM_DISABLE)
+                tqdm(paraphrase_dataloader, desc="eval", disable=TQDM_DISABLE)
             ):
                 (b_ids1, b_mask1, b_ids2, b_mask2, b_labels, b_sent_ids) = (
                     batch["token_ids_1"],
@@ -116,7 +112,7 @@ def model_eval_multitask(
 
         # Evaluate semantic textual similarity.
         if sts_dataloader:
-            for step, batch in enumerate(tqdm(sts_dataloader, desc=f"eval", disable=TQDM_DISABLE)):
+            for step, batch in enumerate(tqdm(sts_dataloader, desc="eval", disable=TQDM_DISABLE)):
                 (b_ids1, b_mask1, b_ids2, b_mask2, b_labels, b_sent_ids) = (
                     batch["token_ids_1"],
                     batch["attention_mask_1"],
@@ -150,7 +146,7 @@ def model_eval_multitask(
         # Evaluate sentiment classification.
         if sentiment_dataloader:
             for step, batch in enumerate(
-                tqdm(sentiment_dataloader, desc=f"eval", disable=TQDM_DISABLE)
+                tqdm(sentiment_dataloader, desc="eval", disable=TQDM_DISABLE)
             ):
                 b_ids, b_mask, b_labels, b_sent_ids = (
                     batch["token_ids"],
@@ -205,7 +201,7 @@ def model_eval_test_multitask(
         para_sent_ids = []
         # Evaluate paraphrase detection.
         for step, batch in enumerate(
-            tqdm(paraphrase_dataloader, desc=f"eval", disable=TQDM_DISABLE)
+            tqdm(paraphrase_dataloader, desc="eval", disable=TQDM_DISABLE)
         ):
             (b_ids1, b_mask1, b_ids2, b_mask2, b_sent_ids) = (
                 batch["token_ids_1"],
@@ -230,7 +226,7 @@ def model_eval_test_multitask(
         sts_sent_ids = []
 
         # Evaluate semantic textual similarity.
-        for step, batch in enumerate(tqdm(sts_dataloader, desc=f"eval", disable=TQDM_DISABLE)):
+        for step, batch in enumerate(tqdm(sts_dataloader, desc="eval", disable=TQDM_DISABLE)):
             (b_ids1, b_mask1, b_ids2, b_mask2, b_sent_ids) = (
                 batch["token_ids_1"],
                 batch["attention_mask_1"],
@@ -254,9 +250,7 @@ def model_eval_test_multitask(
         sst_sent_ids = []
 
         # Evaluate sentiment classification.
-        for step, batch in enumerate(
-            tqdm(sentiment_dataloader, desc=f"eval", disable=TQDM_DISABLE)
-        ):
+        for step, batch in enumerate(tqdm(sentiment_dataloader, desc="eval", disable=TQDM_DISABLE)):
             b_ids, b_mask, b_sent_ids = (
                 batch["token_ids"],
                 batch["attention_mask"],
@@ -347,33 +341,33 @@ def test_model_multitask(args, model, device):
 
     with open(args.sst_dev_out, "w+") as f:
         print(f"dev sentiment acc :: {dev_sentiment_accuracy :.3f}")
-        f.write(f"id \t Predicted_Sentiment \n")
+        f.write("id \t Predicted_Sentiment \n")
         for p, s in zip(dev_sst_sent_ids, dev_sst_y_pred):
             f.write(f"{p} , {s} \n")
 
     with open(args.sst_test_out, "w+") as f:
-        f.write(f"id \t Predicted_Sentiment \n")
+        f.write("id \t Predicted_Sentiment \n")
         for p, s in zip(test_sst_sent_ids, test_sst_y_pred):
             f.write(f"{p} , {s} \n")
 
     with open(args.para_dev_out, "w+") as f:
         print(f"dev paraphrase acc :: {dev_paraphrase_accuracy :.3f}")
-        f.write(f"id \t Predicted_Is_Paraphrase \n")
+        f.write("id \t Predicted_Is_Paraphrase \n")
         for p, s in zip(dev_para_sent_ids, dev_para_y_pred):
             f.write(f"{p} , {s} \n")
 
     with open(args.para_test_out, "w+") as f:
-        f.write(f"id \t Predicted_Is_Paraphrase \n")
+        f.write("id \t Predicted_Is_Paraphrase \n")
         for p, s in zip(test_para_sent_ids, test_para_y_pred):
             f.write(f"{p} , {s} \n")
 
     with open(args.sts_dev_out, "w+") as f:
         print(f"dev sts corr :: {dev_sts_corr :.3f}")
-        f.write(f"id \t Predicted_Similiary \n")
+        f.write("id \t Predicted_Similiary \n")
         for p, s in zip(dev_sts_sent_ids, dev_sts_y_pred):
             f.write(f"{p} , {s} \n")
 
     with open(args.sts_test_out, "w+") as f:
-        f.write(f"id \t Predicted_Similiary \n")
+        f.write("id \t Predicted_Similiary \n")
         for p, s in zip(test_sts_sent_ids, test_sts_y_pred):
             f.write(f"{p} , {s} \n")

multitask_classifier.py

@@ -1,31 +1,30 @@
+import argparse
+import os
+import random
+import subprocess
+import sys
 from contextlib import nullcontext
-import itertools
-from pprint import pformat
-import random, argparse, sys, os, subprocess
 from datetime import datetime
+from pprint import pformat
 from types import SimpleNamespace
 
 import numpy as np
-
 import torch
-from torch import nn
 import torch.nn.functional as F
+from torch import nn
 from torch.utils.data import DataLoader
 from torch.utils.tensorboard import SummaryWriter
-
-from bert import BertModel
-from AttentionLayer import AttentionLayer
-from optimizer import AdamW, SophiaH
 from tqdm import tqdm
 
-from datasets import (
-    SentenceClassificationDataset,
-    SentencePairDataset,
-    load_multitask_data,
-    load_multitask_test_data,
-)
-
-from evaluation import model_eval_sst, test_model_multitask, model_eval_multitask
+from AttentionLayer import AttentionLayer
+from bert import BertModel
+from datasets import load_multitask_data
+from datasets import SentenceClassificationDataset
+from datasets import SentencePairDataset
+from evaluation import model_eval_multitask
+from evaluation import test_model_multitask
+from optimizer import AdamW
+from optimizer import SophiaH
 
 TQDM_DISABLE = False
 
@@ -171,10 +170,6 @@ def load_model(filepath, model, optimizer, use_gpu):
 
 ## Currently only trains on sst dataset
 def train_multitask(args):
-    loss_sst_idx_value = 0
-    loss_sts_idx_value = 0
-    loss_para_idx_value = 0
-
     train_all_datasets = True
     n_datasets = args.sst + args.sts + args.para
     if args.sst or args.sts or args.para:

optimizer.py

@@ -1,4 +1,6 @@
-from typing import Callable, Iterable, Tuple
+from typing import Callable
+from typing import Iterable
+from typing import Tuple
 
 import torch
 from torch.optim import Optimizer

optimizer_test.py

@@ -1,5 +1,6 @@
-import torch
 import numpy as np
+import torch
+
 from optimizer import AdamW
 
 seed = 0

prepare_submit.py

@@ -1,7 +1,5 @@
 # Creates a zip file for submission on Gradescope.
-
 import os
-import sys
 import zipfile
 
 required_files = [p for p in os.listdir(".") if p.endswith(".py")] + [
@@ -11,7 +9,6 @@
 
 def main():
     aid = "dnlp_final_project_submission"
-    path = os.getcwd()
     with zipfile.ZipFile(f"{aid}.zip", "w") as zz:
         for file in required_files:
             zz.write(file, os.path.join(".", file))

sanity_check.py

@@ -1,4 +1,5 @@
 import torch
+
 from bert import BertModel
 
 sanity_data = torch.load("./sanity_check.data")

setup_gwdg.py

@@ -1,5 +1,5 @@
-from tokenizer import BertTokenizer
 from bert import BertModel
+from tokenizer import BertTokenizer
 
 if __name__ == "__main__":
     # Download files

sophia_test.py

@@ -1,5 +1,6 @@
-import torch
 import numpy as np
+import torch
+
 from optimizer import SophiaG
 
 seed = 0