Update app.py

app.py

@@ -1,19 +1,23 @@
 from huggingface_hub import InferenceClient
 import gradio as gr
 import pandas as pd
+import re
 import random
 import csv
 import os
 import io
 import tempfile
-import re
-import openpyxl
 
 client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 
-def extract_data_from_excel(file):
+def extract_sentences_from_excel(file):
     df = pd.read_excel(file)
-    return df.values.tolist()
+    sentences = []
+    for row in df.values.tolist():
+        text = ' '.join(str(x) for x in row)
+        new_sentences = re.split(r'(?<=[^.!?])(?=[.!?])', text)
+        sentences.extend([s.strip() for s in new_sentences if s.strip()])
+    return sentences
 
 def save_to_csv(sentence, output, filename="synthetic_data.csv"):
     with open(filename, mode='a', newline='', encoding='utf-8') as file:
@@ -21,8 +25,7 @@ def save_to_csv(sentence, output, filename="synthetic_data.csv"):
         writer.writerow([sentence, output])
 
 def generate(file, temperature, max_new_tokens, top_p, repetition_penalty, num_sentences=10000):
-    data = extract_data_from_excel(file)
-    sentences = [str(row) for row in data]  # Convert each row to a string
+    sentences = extract_sentences_from_excel(file)
     random.shuffle(sentences)  # Shuffle sentences
 
     with tempfile.NamedTemporaryFile(mode='w', newline='', delete=False, suffix='.csv') as tmp:
@@ -74,7 +77,7 @@ gr.Interface(
         gr.Slider(label="Number of sentences", value=10000, minimum=1, maximum=100000, step=1000, interactive=True, info="The number of sentences to generate"),
     ],
     outputs=gr.File(label="Synthetic Data"),
-    title="SDG", 
+    title="SDG",
     description="AYE QABIL.",
     allow_flagging="never",
 ).launch()