Update model.py

model.py

@@ -1,25 +1,58 @@
-from huggingface_hub import InferenceClient
 import os
+import requests
+import json
+import time
 
-def query_model(prompt):
-    try:
-        HF_TOKEN = os.getenv("HF_TOKENN")
+def generate_workout(name, age, goal, level, equipment, bmi):
+    # Get HF Token from environment secrets
+    hf_token = os.getenv("HUGGINGFACE_TOKEN")
+    
+    if not hf_token:
+        return "Error: HUGGINGFACE_TOKEN not found. Please add it to your Hugging Face Space Settings > Secrets."
+
+    # Using the most stable Serverless Inference endpoint (OpenAI-compatible)
+    # This path is the new standard that replaces the old 410-Gone endpoint
+    api_url = "https://api-inference.huggingface.co/v1/chat/completions"
+    headers = {
+        "Authorization": f"Bearer {hf_token}",
+        "Content-Type": "application/json"
+    }
 
-        client = InferenceClient(
-            model="mistralai/Mistral-7B-Instruct-v0.2",
-            token=HF_TOKENN
-        )
+    # Model ID - Llama-3.2-3B-Instruct is extremely reliable and 'warm' on the free tier
+    model_id = "meta-llama/Llama-3.2-3B-Instruct"
 
-        response = client.chat_completion(
-            messages=[
-                {"role": "system", "content": "You are a certified professional fitness trainer."},
-                {"role": "user", "content": prompt}
-            ],
-            max_tokens=900,
-            temperature=0.7
-        )
+    messages = [
+        {"role": "system", "content": "You are a professional fitness coach. Return ONLY the plan in Markdown. No filler."},
+        {"role": "user", "content": f"Create a 5-day workout for {name}. Goal: {goal}, Level: {level}, Equipment: {equipment}, BMI: {bmi}. Include Day headers and 3-4 exercises per day."}
+    ]
 
-        return response.choices[0].message.content
+    payload = {
+        "model": model_id,
+        "messages": messages,
+        "max_tokens": 1200,
+        "temperature": 0.5,
+        "stream": False
+    }
 
+    try:
+        # We try up to 3 times if the model is loading
+        for attempt in range(3):
+            response = requests.post(api_url, headers=headers, json=payload, timeout=120)
+            
+            if response.status_code == 200:
+                result = response.json()
+                if "choices" in result and len(result["choices"]) > 0:
+                    return result["choices"][0]["message"]["content"]
+                return f"Error: No content in response: {result}"
+            
+            # If model is loading, wait and retry
+            elif response.status_code == 503:
+                time.sleep(15)
+                continue
+            else:
+                break
+        
+        return f"Error: API returned status {response.status_code} - {response.text}"
+            
     except Exception as e:
-        return f"Error: {str(e)}"
+        return f"Error: Connection failed: {str(e)}"