Update app.py

app.py

@@ -1,66 +1,64 @@
+import gradio as gr
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-import torch, gradio as gr, re
+import re
 
-# --- Load Model ---
-model_name = "prithivida/parrot_paraphraser_on_T5"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
+# Load model
+tokenizer = AutoTokenizer.from_pretrained("Vamsi/T5_Paraphrase_Paws")
+model = AutoModelForSeq2SeqLM.from_pretrained("Vamsi/T5_Paraphrase_Paws")
 
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-model = model.to(device)
-model.eval()
-
-# --- Helpers ---
-def split_paragraphs(text):
-    return [p.strip() for p in text.split("\n") if p.strip()]
-
-def split_sentences(text):
-    return re.split(r'(?<=[.!?])\s+', text.strip())
-
-def clean_text(text):
-    return re.sub(r'\s+', ' ', text).strip()
-
-def paraphrase_chunk(text_chunk):
-    inputs = tokenizer([text_chunk], return_tensors="pt", truncation=True, padding=True).to(device)
-    outputs = model.generate(
-        **inputs,
-        max_new_tokens=100,  # small chunks only
-        num_beams=4,
-        do_sample=False
+def paraphrase_text(text):
+    input_text = f"paraphrase: {text} </s>"
+    input_ids = tokenizer.encode(input_text, return_tensors="pt", truncation=True)
+    output_ids = model.generate(
+        input_ids,
+        max_length=256,
+        do_sample=True,
+        top_k=120,
+        top_p=0.95,
+        temperature=1.5
     )
-    return clean_text(tokenizer.decode(outputs[0], skip_special_tokens=True))
-
-# --- Main Humanizer ---
-def humanize_text(text):
-    if not text.strip():
-        return "⚠️ Please enter some text"
+    return tokenizer.decode(output_ids[0], skip_special_tokens=True)
 
-    paragraphs = split_paragraphs(text)
-    humanized_paragraphs = []
+def chunk_text(text, max_sentences=4):
+    sentences = re.split(r'(?<=[.!?]) +', text.strip())
+    return [' '.join(sentences[i:i+max_sentences]) for i in range(0, len(sentences), max_sentences)]
 
-    for para in paragraphs:
-        sentences = split_sentences(para)
-        paraphrased_sentences = []
+def full_article_paraphrase(text):
+    chunks = chunk_text(text)
+    return "\n\n".join(paraphrase_text(chunk.strip()) for chunk in chunks if chunk.strip())
 
-        # Paraphrase each sentence separately for accuracy
-        for sent in sentences:
-            paraphrased_sentences.append(paraphrase_chunk("paraphrase: " + sent))
+def extract_text_from_docx(file_obj):
+    from docx import Document
+    import io
+    file_bytes = file_obj.read() if hasattr(file_obj, "read") else file_obj
+    doc = Document(io.BytesIO(file_bytes))
+    return "\n".join([para.text for para in doc.paragraphs if para.text.strip()])
 
-        # Rebuild paragraph and optionally add light filler once
-        new_para = " ".join(paraphrased_sentences)
-        if len(paraphrased_sentences) > 2:
-            new_para = "In fact, " + new_para[0].lower() + new_para[1:]
-        humanized_paragraphs.append(new_para)
+def get_ai_score(text):
+    return "AI Detection Score: Likely Human" if len(text) > 100 else "AI Detection Score: Short Text"
 
-    return "\n\n".join(humanized_paragraphs)
+def full_pipeline(input_text=None, file=None):
+    if file is not None:
+        input_text = extract_text_from_docx(file)
+    if not input_text or len(input_text.strip()) < 10:
+        return "Please enter or upload valid text.", "No text to analyze."
+    result = full_article_paraphrase(input_text)
+    ai_score = get_ai_score(result)
+    return result, ai_score
 
-# --- Gradio Interface ---
-iface = gr.Interface(
-    fn=humanize_text,
-    inputs=gr.Textbox(lines=15, placeholder="Paste text here..."),
-    outputs=gr.Textbox(label="Humanized Output"),
-    title="✨ Writenix Humanizer v3",
-    description="Paraphrases large text paragraph-by-paragraph for natural flow, keeps full content, adds light filler per paragraph."
+demo = gr.Interface(
+    fn=full_pipeline,
+    inputs=[
+        gr.Textbox(label="Paste Text (optional)", lines=20, placeholder="Or upload a .docx file below..."),
+        gr.File(label="Upload .docx File (optional)", file_types=[".docx"])
+    ],
+    outputs=[
+        gr.Textbox(label="Paraphrased Output"),
+        gr.Textbox(label="AI Detection Score")
+    ],
+    title="Smart Paraphraser",
+    description="Paste or upload your article. Get paraphrased output and see an AI detection score."
 )
 
-iface.launch()
+if __name__ == "__main__":
+    demo.launch()