second

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 import logging
 import sys
 import os
+import gc
 
 # Configure logging
 logging.basicConfig(
@@ -15,100 +16,113 @@ logger = logging.getLogger(__name__)
 logger.info("Starting StudAI Summarization Service with Gradio")
 logger.info(f"Python version: {sys.version}")
 
-# Import model
+# Force garbage collection
+gc.collect()
+
+# Create a simple function for summarization that doesn't use ML in case model loading fails
+def simple_summarize(text, max_length=150, min_length=30):
+    """Simple extractive summarization as fallback"""
+    import re
+    sentences = re.split(r'(?<=[.!?])\s+', text)
+    
+    if len(sentences) <= 3:
+        return text
+        
+    # Take first, middle and last sentences
+    summary = [
+        sentences[0],
+        sentences[len(sentences) // 2],
+        sentences[-1]
+    ]
+    return " ".join(summary)
+
+# Set a flag for model availability
+model_available = False
+
+# Try to import and load the model with memory optimizations
 try:
-    from transformers import pipeline
-    logger.info("Loading summarization model (t5-small)...")
+    # Import and load only when needed
+    from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
+    import torch
+    
+    logger.info("Loading small model for summarization...")
     
+    # Use a tiny model instead of t5-small
+    model_name = "facebook/bart-large-cnn"
+    
+    # Enable memory optimization
+    if torch.cuda.is_available():
+        logger.info("CUDA available, using GPU")
+        device = 0
+    else:
+        logger.info("CUDA not available, using CPU")
+        device = -1
+    
+    # Enable memory-efficient loading
     summarizer = pipeline(
-        "summarization", 
-        model="t5-small",
-        device=-1  # Use CPU for more reliable execution
+        "summarization",
+        model=model_name,
+        device=device,
+        framework="pt"
     )
+    
     logger.info("Model loaded successfully!")
     model_available = True
+    
+    # Force garbage collection after model loading
+    gc.collect()
+    if torch.cuda.is_available():
+        torch.cuda.empty_cache()
+        
 except Exception as e:
     logger.error(f"Failed to load model: {str(e)}")
-    model_available = False
-    
+    logger.info("Will use simple extractive summarization instead")
+
 def summarize_text(text, max_length=150, min_length=30):
-    """Summarize the provided text using the loaded model"""
+    """Summarize the provided text"""
+    if not text or len(text.strip()) < 50:
+        return text
+    
     try:
-        if not text or len(text) < 50:
-            return text
+        if model_available:
+            logger.info(f"Summarizing text of length {len(text)} with model")
+            result = summarizer(
+                text, 
+                max_length=max_length, 
+                min_length=min_length,
+                truncation=True
+            )
+            summary = result[0]["summary_text"]
+        else:
+            logger.info(f"Using simple summarization for text of length {len(text)}")
+            summary = simple_summarize(text, max_length, min_length)
             
-        if not model_available:
-            return "Error: Summarization model is not available"
-        
-        logger.info(f"Summarizing text of length {len(text)}")
-        result = summarizer(
-            text, 
-            max_length=max_length, 
-            min_length=min_length,
-            truncation=True
-        )
-        summary = result[0]["summary_text"]
-        logger.info(f"Generated summary of length {len(summary)}")
         return summary
     except Exception as e:
         logger.error(f"Error during summarization: {str(e)}")
-        return f"Error: {str(e)}"
-
-def api_summarize(text, max_length=150, min_length=30):
-    """API function for summarization"""
-    summary = summarize_text(text, max_length, min_length)
-    return {"summary": summary}
+        # Fall back to simple summarization on error
+        return simple_summarize(text, max_length, min_length)
 
 # Create Gradio interface
-with gr.Blocks(title="StudAI Summarization") as demo:
-    gr.Markdown("# StudAI Text Summarization")
-    gr.Markdown("This service provides text summarization for the StudAI Android app.")
-    
-    with gr.Row():
-        with gr.Column():
-            input_text = gr.Textbox(
-                label="Input Text", 
-                placeholder="Enter text to summarize (at least 50 characters)",
-                lines=10
-            )
-            with gr.Row():
-                max_length = gr.Slider(
-                    label="Max Length", 
-                    minimum=50, 
-                    maximum=500, 
-                    value=150, 
-                    step=10
-                )
-                min_length = gr.Slider(
-                    label="Min Length", 
-                    minimum=10, 
-                    maximum=200, 
-                    value=30, 
-                    step=5
-                )
-            submit_btn = gr.Button("Summarize")
-        
-        with gr.Column():
-            output_text = gr.Textbox(label="Summary", lines=10)
-    
-    submit_btn.click(
-        fn=summarize_text, 
-        inputs=[input_text, max_length, min_length], 
-        outputs=output_text
-    )
-    
-    # Add API endpoints for Android app
-    gr.Interface(
-        fn=api_summarize,
-        inputs=[
-            gr.Textbox(label="text"),
-            gr.Number(label="max_length", default=150),
-            gr.Number(label="min_length", default=30)
-        ],
-        outputs=gr.JSON(),
-        title="Summarization API",
-        description="API for StudAI Android app"
-    ).launch(show_api=True)
+demo = gr.Interface(
+    fn=summarize_text,
+    inputs=[
+        gr.Textbox(
+            lines=10, 
+            label="Text to Summarize",
+            placeholder="Enter text to summarize (at least 50 characters)"
+        ),
+        gr.Slider(50, 500, value=150, label="Max Length"),
+        gr.Slider(10, 200, value=30, label="Min Length")
+    ],
+    outputs=gr.Textbox(label="Summary"),
+    title="StudAI Text Summarization",
+    description="This service provides text summarization for the StudAI Android app.",
+    examples=[
+        ["The coronavirus pandemic has led to a surge in remote work. Companies around the world have had to adapt to new ways of working, with many employees setting up home offices. This shift has led to changes in productivity, work-life balance, and communication patterns. Some studies suggest that remote work can increase productivity, while others point to challenges in collaboration and team cohesion. Organizations are now considering hybrid models for the future of work.", 150, 30]
+    ],
+    allow_flagging="never"
+)
 
-# Launch the app
-demo.launch() 
+# Launch with parameters optimized for Spaces
+demo.launch(share=False, server_name="0.0.0.0", server_port=7860)

requirements.txt

@@ -2,5 +2,5 @@ gradio==4.13.0
 transformers==4.35.2
 torch==2.0.1
 numpy<2.0.0
-pydantic==2.4.2
-requests==2.31.0 
+requests==2.31.0
+accelerate==0.25.0