Create app.py

app.py

@@ -0,0 +1,75 @@
+import gradio as gr
+import requests
+import os
+
+# Define models
+MODEL_OPTIONS = {
+    "Gemma 3-4B": "google/gemma-3-4b-it",
+    "Mistral 7B": "mistralai/mistral-7b-instruct",
+    "Qwen 0.6B": "qwen/qwen3-0.6b-04-28",
+    "DeepSeek 70B": "deepseek/deepseek-r1-distill-llama-70b"
+}
+
+EVALUATION_MODEL = "qwen/qwen3-32b"
+
+# Token (set your HF token as an environment variable or paste here directly)
+HF_TOKEN = os.getenv("HF_TOKEN") or "your_token_here"  # Replace with your actual token if needed
+
+# API query function
+def query_huggingface(prompt, model_id):
+    headers = {
+        "Authorization": f"Bearer {HF_TOKEN}",
+        "Content-Type": "application/json"
+    }
+
+    payload = {
+        "inputs": prompt,
+        "parameters": {
+            "max_new_tokens": 150,
+            "do_sample": True,
+            "temperature": 0.7
+        }
+    }
+
+    response = requests.post(
+        f"https://api-inference.huggingface.co/models/{model_id}",
+        headers=headers,
+        json=payload
+    )
+
+    result = response.json()
+    if isinstance(result, list) and "generated_text" in result[0]:
+        return result[0]["generated_text"]
+    else:
+        return f"❌ Error: {result.get('error', 'Unknown error')}"
+
+# Main logic
+def generate_and_evaluate(prompt, model_name):
+    model_id = MODEL_OPTIONS[model_name]
+
+    # Step 1: Generate math problem
+    generated = query_huggingface(prompt, model_id)
+
+    # Step 2: Evaluate using Qwen 32B
+    eval_prompt = (
+        f"Evaluate the following math problem for originality, clarity, and difficulty. "
+        f"Give a score out of 10 with a short explanation:\n\n{generated}"
+    )
+    evaluation = query_huggingface(eval_prompt, EVALUATION_MODEL)
+
+    return generated.strip(), evaluation.strip()
+
+# Gradio UI
+gr.Interface(
+    fn=generate_and_evaluate,
+    inputs=[
+        gr.Textbox(label="🧠 Math Concept Prompt", placeholder="e.g., Generate a creative trigonometry problem.", lines=2),
+        gr.Dropdown(choices=list(MODEL_OPTIONS.keys()), label="🧠 Choose LLM")
+    ],
+    outputs=[
+        gr.Textbox(label="📐 Generated Math Problem", lines=6),
+        gr.Textbox(label="🔍 Qwen 32B Evaluation", lines=4)
+    ],
+    title="🔢 Multi-LLM Math Generator + Evaluator",
+    description="Generate math problems with different open-source LLMs and evaluate with Qwen 32B."
+).launch()