Update app.py

app.py

@@ -96,33 +96,39 @@ from tenacity import retry, stop_after_attempt, wait_exponential
 
 async def get_bot_response(adapter_id, prompt, state, bot_index):
     chatml_prompt = format_chatml_prompt(state)
+    fireworks_adapter_name = next(entry['fireworks_adapter_name'] for entry in chatbots_data if entry['adapter'] == adapter_id)
+
+    url = "https://api.fireworks.ai/inference/v1/completions"
     payload = {
-        "inputs": chatml_prompt[bot_index],
-        "parameters": {
-            "adapter_id": adapter_id,
-            "adapter_source": "hub",
-            "temperature": 0.7,
-            "max_new_tokens": 100
-        }
+        "model": f"accounts/gaingg19-432d9f/models/{fireworks_adapter_name}",
+        "max_tokens": 100,
+        "top_p": 1,
+        "top_k": 40,
+        "presence_penalty": 0,
+        "frequency_penalty": 0,
+        "temperature": 0.7,
+        "prompt": chatml_prompt[bot_index],
+        "stop": ["<|im_end|>"]
     }
     headers = {
+        "Accept": "application/json",
         "Content-Type": "application/json",
-        "Authorization": f"Bearer {os.environ.get('PREDIBASE_TOKEN')}"
+        "Authorization": f"Bearer {os.environ.get('FIREWORKS_API_KEY')}"
     }
+
     async with aiohttp.ClientSession() as session:
         try:
-            async with session.post("https://serving.app.predibase.com/79957f/deployments/v2/llms/mistral-7b/generate",
-                                    json=payload, headers=headers, timeout=30) as response:
+            async with session.post(url, json=payload, headers=headers, timeout=30) as response:
                 if response.status == 200:
                     response_data = await response.json()
-                    response_text = response_data.get('generated_text', '')
+                    response_text = response_data['choices'][0]['text']
                 else:
                     error_text = await response.text()
                     print(error_text)
                     response_text = "Sorry, I couldn't generate a response."
         except (aiohttp.ClientError, asyncio.TimeoutError):
             response_text = "Sorry, I couldn't generate a response."
-    return response_text.split('<|im_end|>')[0].strip()
+    return response_text.strip()
 async def chat_with_bots(user_input, state):
     # Use existing bot order from state if available, otherwise shuffle and initialize
     if 'last_bots' not in state or not state['last_bots']: