Create app.py

app.py

@@ -0,0 +1,28 @@
+import transformers
+import torch
+
+# Specify the model you want to use
+model_id = "nvidia/Llama-3.1-Nemotron-8B-UltraLong-4M-Instruct"
+
+# Set up the text-generation pipeline
+pipeline = transformers.pipeline(
+    "text-generation",  # You are using the text generation pipeline
+    model=model_id,
+    model_kwargs={"torch_dtype": torch.bfloat16},  # Specifying the torch dtype
+    device_map="auto",  # This will use available hardware (GPU or CPU)
+)
+
+# Define the conversation/messages you want the model to handle
+messages = [
+    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
+    {"role": "user", "content": "Who are you?"}
+]
+
+# Use the pipeline to generate a response
+outputs = pipeline(
+    messages[1]["content"],  # Use only the user message here
+    max_new_tokens=256,  # Limit the number of tokens generated
+)
+
+# Print the generated text from the output
+print(outputs[0]["generated_text"])