Create app.py

app.py

@@ -0,0 +1,61 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
+import torch
+import gradio as gr
+
+# Configuration
+MODEL_NAME = "google/gemma-2b-it"
+CACHE_DIR = "/tmp"  # For Spaces storage limits
+
+# Load model with 4-bit quantization (reduces VRAM usage by ~4x)
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, cache_dir=CACHE_DIR)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_NAME,
+    device_map="auto",
+    torch_dtype=torch.float16,
+    cache_dir=CACHE_DIR
+)
+
+# Create optimized pipeline
+math_pipeline = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=256,
+    do_sample=True,
+    temperature=0.7,
+    top_k=50,
+    top_p=0.95
+)
+
+def solve_math(question):
+    prompt = f"Question: {question}\nAnswer:"
+    outputs = math_pipeline(prompt)
+    return outputs[0]["generated_text"].split("Answer:")[-1].strip()
+
+# Gradio Interface (optimized for mobile/desktop)
+with gr.Blocks(title="🚀 Gemma-2B Math Solver") as demo:
+    gr.Markdown("## 🧮 Solve Math Problems with Gemma-2B")
+    with gr.Row():
+        input_question = gr.Textbox(
+            label="Enter your math problem",
+            placeholder="E.g., What is 2^10 + 5*3?",
+            lines=3
+        )
+    with gr.Row():
+        submit_btn = gr.Button("Solve", variant="primary")
+    with gr.Row():
+        output_answer = gr.Textbox(label="Solution", lines=5, interactive=False)
+    
+    submit_btn.click(
+        fn=solve_math,
+        inputs=input_question,
+        outputs=output_answer
+    )
+
+# Launch with production settings
+if __name__ == "__main__":
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=False  # Set to True for temporary public link
+    )