Upload legaltextclassification.py

legaltextclassification.py

@@ -0,0 +1,232 @@
+# -*- coding: utf-8 -*-
+"""LegalTextClassification.ipynb
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1x6EcLSN3qEgm6sVIcmX0bYeXj7AdDQlW
+!pip install gradio
+    import gradio as gr
+
+def greet(name):
+    return "Hello " + name + "!!"
+
+demo = gr.Interface(fn=greet, inputs="text", outputs="text")
+demo.launch()
+    
+#About Data
+The dataset contains a total of 25000 legal cases in the form of text documents. Each document has been annotated with catchphrases, citations sentences, citation catchphrases, and citation classes. Citation classes indicate the type of treatment given to the cases cited by the present case.
+
+The Legal Citation Text Classification dataset is provided in CSV format. The dataset has ***four columns***, ***namely Case ID, Case Outcome, Case Title, and Case Text***. The Case ID column contains a unique identifier for each legal case, the Case Outcome column indicates the outcome of the case, the Case Title column contains the title of the legal case, and the Case Text column contains the text of the legal case.
+
+Kaggle Dataset Link: https://www.kaggle.com/datasets/amohankumar/legal-text-classification-dataset/data
+
+#Importing Data
+"""
+
+from google.colab import files
+import pandas as pd
+
+df = pd.read_csv('legal_text_classification.csv')
+df.head()
+
+"""#Data Preprocessing and Description"""
+
+print(df.columns)      # Lists all column names
+print(len(df.columns)) # Shows the number of columns
+
+print(df.shape)  # Output: (rows, columns)
+
+print(df.isnull().sum())
+
+df = df.dropna(subset=['case_text'])
+
+df = df.drop(columns=["case_id", "case_title"])
+
+print(df.isnull().sum())
+
+import re
+
+def text_ready(text):
+    text = text.lower()  #lowercase
+    text = re.sub(r'[^\w\s]', '', text)  #special char
+    text = re.sub(r'\s+', ' ', text).strip()  #whitespace
+    return text
+
+df["text_ready"] = df["case_text"].apply(text_ready)
+
+import matplotlib.pyplot as plt
+
+text_data = df['text_ready']
+word_count = [len(text.split()) for text in text_data]
+
+plt.hist(word_count, bins=50, color='skyblue', edgecolor='black')
+plt.title('Distribution of Word Counts in text_ready')
+plt.xlabel('Word Count')
+plt.ylabel('Frequency')
+plt.show()
+
+print(df.shape)  # Output: (rows, columns)
+
+df.describe()
+
+df['text']=df['text_ready']
+df['label']=df['case_outcome']
+data=df[['text','label']]
+
+df = df.drop(columns=["case_outcome", "case_text"])
+
+df.head()
+
+df = df.drop(columns=["text_ready"])
+
+df.head()
+
+data['label'].value_counts()
+
+class_label=sorted(data['label'].unique())
+lbl2id={label:id for id,label in enumerate(class_label)}
+id2lb={id:label for label,id in lbl2id.items()}
+print(lbl2id)
+print(id2lb)
+
+
+
+data.head()
+
+data['label']=data['label'].map(lbl2id)
+data.head()
+
+data.label.value_counts()
+
+import matplotlib.pyplot as plt
+
+df['label'].value_counts().plot.bar()
+plt.show()
+
+from transformers import AutoModelForSequenceClassification,AutoTokenizer
+model_name='nlpaueb/legal-bert-base-uncased'
+tokenizer=AutoTokenizer.from_pretrained(model_name)
+
+from transformers import AutoModelForSequenceClassification
+model = AutoModelForSequenceClassification.from_pretrained(
+    model_name,
+    num_labels=len(id2lb),
+    id2label=id2lb,
+    label2id=lbl2id
+)
+
+!pip install datasets
+from datasets import Dataset
+ds=Dataset.from_pandas(data)
+ds
+
+ds['label'][:11]
+
+from datasets import ClassLabel
+unique_labels = sorted(set(ds['label']))
+print(f"Unique labels in Y: {unique_labels}")
+
+new_features = ds.features.copy()
+new_features['label'] = ClassLabel(names=unique_labels)
+
+ds = ds.cast(new_features)
+data = ds.train_test_split(test_size=0.2, shuffle=True, seed=42)
+data
+
+split_ds = data['test'].remove_columns('__index_level_0__').train_test_split(test_size=0.5, shuffle=True, seed=42)
+split_ds
+
+train_data=data['train']
+test_data=split_ds['train']
+val_data=split_ds['test']
+
+train_data[0]
+
+def tokenize_fun(data):
+  return tokenizer(data['text'],padding=True,truncation=True,return_tensors='pt')
+
+tokenized_train_data=train_data.map(tokenize_fun,batched=True)
+
+tokenized_train_data.features
+
+!pip install evaluate
+import evaluate
+accuracy=evaluate.load('accuracy')
+
+import numpy as np
+
+def compute_metrics(eval_pred):
+    predictions, labels = eval_pred
+    predictions = np.argmax(predictions, axis=1)
+    return accuracy.compute(predictions=predictions, references=labels)
+
+tokenized_test_data=test_data.map(tokenize_fun,batched=True)
+tokenized_val_data=val_data.map(tokenize_fun,batched=True)
+
+from huggingface_hub import login
+login()
+
+from transformers import Trainer,TrainingArguments
+
+training_args=TrainingArguments(
+    output_dir='./quest_model',
+    learning_rate=2e-3,
+    per_device_eval_batch_size=16,
+    per_device_train_batch_size=16,
+    num_train_epochs=2,
+    weight_decay=0.01,
+    eval_strategy='epoch',
+    save_strategy='epoch',
+    load_best_model_at_end=True,
+    push_to_hub=True
+)
+
+trainer=Trainer(
+    model=model,
+    tokenizer=tokenizer,
+    args=training_args,
+    train_dataset=tokenized_train_data,
+    eval_dataset=tokenized_val_data,
+    compute_metrics=compute_metrics
+)
+trainer.train()
+
+model.config.id2label
+
+import os
+os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
+
+model.save_pretrained('./quest_model')
+tokenizer.save_pretrained("./quest_model")
+
+tokenized_train_data[0]['text']
+
+from transformers import pipeline
+pipe=pipeline('text-classification',model='Nainglinthu/quest_model')
+output=pipe('Hexal Australia Pty Ltd v Roche Therapeutics Inc (2005) 66 IPR 325, the likelihood of irreparable harm was regarded by Stone J as, indeed, a separate element that had to be established by an applicant for an interlocutory injunction.')
+output
+
+!pip install --upgrade gradio
+import gradio as gr
+from transformers import pipeline
+
+# Initialize the pipeline
+pipe = pipeline('text-classification', model='Nainglinthu/quest_model')
+
+# Function to classify text
+def classify_text(input_text):
+    output = pipe(input_text)
+    return output
+
+# Create Gradio interface
+interface = gr.Interface(
+    fn=classify_text,          # Function to call
+    inputs="text",             # Input type (text box)
+    outputs="json",            # Output type (JSON for displaying result)
+    title="Legal Text Classifier",  # Title of the Gradio app
+    description="Classify legal text using the Nainglinthu/quest_model!",  # Description
+)
+
+# Launch the Gradio app
+interface.launch()