Upload ResumeCode.txt

ResumeCode.txt

@@ -0,0 +1,214 @@
+!pip install opendatasets
+
+#!pip install wandb
+
+!pip install transformers[torch]
+
+!pip install evaluate
+
+import pandas as pd
+import numpy as np
+import os
+import random
+from datasets import Dataset
+import opendatasets as od
+import matplotlib.pyplot as plt
+
+import torch
+from transformers import (
+    AutoTokenizer,
+    AutoModelForSequenceClassification,
+    TrainingArguments,
+    Trainer,
+    DataCollatorWithPadding,
+    pipeline
+)
+import evaluate
+
+plt.style.use('seaborn-v0_8')
+
+from sklearn.model_selection import train_test_split
+from sklearn.preprocessing import LabelEncoder
+from sklearn.naive_bayes import MultinomialNB
+from sklearn import metrics
+from sklearn.metrics import accuracy_score
+from pandas.plotting import scatter_matrix
+from sklearn import metrics
+from sklearn.feature_extraction.text import TfidfVectorizer
+from matplotlib.gridspec import GridSpec
+import nltk
+nltk.download('stopwords')
+nltk.download('punkt')
+from nltk.corpus import stopwords
+import string
+from wordcloud import WordCloud
+
+DIRECTORY = '/content/UpdatedResumeDataSet.csv'
+MODEL_NAME = 'distilbert-base-uncased'
+BATCH_SIZE = 32
+LR = 2e-5
+EPOCHS = 10
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+# read the dataset
+df = pd.read_csv('UpdatedResumeDataSet.csv')
+print(df.shape)
+df.head(10)      # first 10 rows
+
+# Display the distinct categories of resume
+df['Category'].unique()
+
+# Display the distinct categories of resume and the number of records belonging to each category
+df['Category'].value_counts()
+
+import seaborn as sns
+
+sns.countplot(y = df['Category'], data = df['Resume'])
+
+# Convert all characters to lowercase
+def convert_lower(text):
+    return text.lower()
+
+df['Resume'] = df['Resume'].apply(convert_lower)
+
+import re
+def cleanResume(resumeText):
+  resumeText = re.sub(r'http\S+', '', resumeText,flags = re.MULTILINE) # remove URLs
+  resumeText = re.sub('RT|cc', '', resumeText)  # remove RT and cc
+  resumeText = re.sub('#\S+', '', resumeText)  # remove hashtags
+  resumeText = re.sub('@\S+', '', resumeText)  # remove mentions
+  resumeText = re.sub('[%s]' % re.escape("""!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~"""), '', resumeText)  # remove punctuations
+  resumeText = re.sub('â\S+', '', resumeText)  # remove •
+  resumeText = re.sub('€+', '', resumeText)  # remove €
+  resumeText = re.sub('\s+', ' ', resumeText)  # remove extra whitespace
+
+  return resumeText
+
+ # apply the function defined above and save the
+df['cleaned_resume'] = df['Resume'].apply(cleanResume)
+
+# stop words
+stopword_list = nltk.corpus.stopwords.words('english')
+print(stopword_list)
+
+# removing the stopwords
+from nltk.tokenize import word_tokenize
+
+def remove_stopwords(text, is_lower_case=False):
+    # splitting strings into tokens (list of words)
+    tokens = word_tokenize(text)
+    tokens = [token.strip() for token in tokens]
+    filtered_tokens = [token for token in tokens if token not in stopword_list]
+    filtered_text = ' '.join(filtered_tokens)
+    return filtered_text
+
+# apply function on cleaned resume to remove stopwords
+df['text'] = df['cleaned_resume'].apply(remove_stopwords)
+df['label'] = df['Category']
+
+# reorder dataframe columns
+df = df[['text', 'label']]
+
+# view shape
+df.shape
+
+# view number of classes
+n_classes = df['label'].nunique()
+print(f"Number of Resume classes: {n_classes}")
+
+# view some statistics about are texts
+lengths = df['text'].apply(lambda x: len(x))
+print(
+    f'Max text length: {lengths.max()}\nMin text length: {lengths.min()}\nAvg text length: {lengths.mean():.2f}'
+)
+
+# create mappings
+id2label = {idx: label for idx, label in enumerate(df['label'].unique())}
+label2id = {label: idx for idx, label in id2label.items()}
+
+# label encode our labels
+df['label'] = df['label'].map(label2id)
+
+# create and split dataset
+dataset = Dataset.from_pandas(df).train_test_split(train_size=0.8)
+print(dataset)
+
+# initialize tokenizer
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+
+# Tokenize and encode the dataset
+def tokenize(batch):
+    tokenized_batch = tokenizer(batch['text'], padding=True, truncation=True)
+    return tokenized_batch
+
+dataset_enc = dataset.map(tokenize, batched=True)
+
+print(dataset_enc)
+
+accuracy = evaluate.load('accuracy')
+
+def compute_metrics(eval_pred):
+    predictions, labels = eval_pred
+    predictions = np.argmax(predictions, axis=1)
+    return accuracy.compute(predictions=predictions, references=labels)
+
+# define model
+model = AutoModelForSequenceClassification.from_pretrained(
+    MODEL_NAME,
+    num_labels=n_classes,
+    id2label=id2label,
+    label2id=label2id
+)
+
+model.to(DEVICE)
+
+# define collator function
+collator_fn = DataCollatorWithPadding(tokenizer, return_tensors='pt')
+
+pip install accelerate -U
+
+import accelerate
+import transformers
+
+transformers.__version__, accelerate.__version__
+
+from transformers import TrainingArguments
+
+training_args = TrainingArguments(
+    output_dir = "Resume_training",
+    learning_rate=LR,
+    per_device_train_batch_size= BATCH_SIZE,
+    per_device_eval_batch_size = BATCH_SIZE,
+    num_train_epochs = EPOCHS,
+    weight_decay = 0.01,
+    evaluation_strategy = "epoch",
+    save_strategy = "epoch",
+    load_best_model_at_end = True,
+    push_to_hub = False,
+    report_to="none"
+)
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=dataset_enc["train"],
+    eval_dataset=dataset_enc["test"],
+    tokenizer=tokenizer,
+    data_collator=collator_fn,
+    compute_metrics=compute_metrics
+)
+
+trainer.train()
+
+trainer.save_model('ResumeClassification_distilBERT')
+
+trainer.evaluate()
+
+def predict(sample, validate=True):
+  classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)
+  pred = classifier(sample)[0]['label']
+  return pred
+
+sample1 = "I have working expereince in Java and javascript"
+
+predict(sample1)