Add nemotron 3 ultra

app.py

@@ -1,7 +1,26 @@
 from fastapi import FastAPI
+from pydantic import BaseModel
+from huggingface_hub import InferenceClient
+import os
 
 app = FastAPI()
 
+client = InferenceClient(
+    model="nvidia/Llama-3.1-Nemotron-Nano-8B-v1",
+    token=os.environ.get("HF_TOKEN")
+)
+
+class ChatRequest(BaseModel):
+    message: str
+
 @app.get("/")
 def greet_json():
     return {"Hello": "World!"}
+
+@app.post("/chat")
+def chat(request: ChatRequest):
+    response = client.chat.completions.create(
+        messages=[{"role": "user", "content": request.message}],
+        max_tokens=1024,
+    )
+    return {"response": response.choices[0].message.content}

requirements.txt

@@ -1,2 +1,3 @@
 fastapi
 uvicorn[standard]
+huggingface-hub