test

app.py

@@ -1,10 +1,16 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import requests
+import json
+import os
 
 """
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
+Using Together AI API for chat completions
 """
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+TOGETHER_API_KEY = os.getenv("TOGETHER_API_KEY")
+if not TOGETHER_API_KEY:
+    raise ValueError("TOGETHER_API_KEY environment variable is not set")
+
+TOGETHER_API_URL = "https://api.together.xyz/v1/chat/completions"
 
 
 def respond(
@@ -25,19 +31,44 @@ def respond(
 
     messages.append({"role": "user", "content": message})
 
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    headers = {
+        "Authorization": f"Bearer {TOGETHER_API_KEY}",
+        "Content-Type": "application/json"
+    }
+    
+    data = {
+        "model": "arcee-ai/AFM-4.5B",
+        "messages": messages,
+        "max_tokens": max_tokens,
+        "temperature": temperature,
+        "top_p": top_p,
+        "stream": True
+    }
 
-        response += token
-        yield response
+    response = ""
+    
+    try:
+        with requests.post(TOGETHER_API_URL, headers=headers, json=data, stream=True) as r:
+            r.raise_for_status()
+            for line in r.iter_lines():
+                if line:
+                    line = line.decode('utf-8')
+                    if line.startswith('data: '):
+                        line = line[6:]  # Remove 'data: ' prefix
+                        if line.strip() == '[DONE]':
+                            break
+                        try:
+                            chunk = json.loads(line)
+                            if 'choices' in chunk and len(chunk['choices']) > 0:
+                                delta = chunk['choices'][0].get('delta', {})
+                                if 'content' in delta:
+                                    token = delta['content']
+                                    response += token
+                                    yield response
+                        except json.JSONDecodeError:
+                            continue
+    except requests.exceptions.RequestException as e:
+        yield f"Error: {str(e)}"
 
 
 """

requirements.txt

@@ -1 +1,2 @@
-huggingface_hub==0.25.2
+gradio
+requests