granite model

app.py

@@ -12,7 +12,12 @@ token = os.getenv("HUGGINGFACE_API_TOKEN")
 
 print(f'HUGGINGFACE_API_TOKEN: {token}')
 
-pipe = pipeline("text-generation", "meta-llama/Meta-Llama-3-8B-Instruct", torch_dtype=torch.bfloat16, device_map="auto", token=token)
+model = "meta-llama/Meta-Llama-3-8B-Instruct"
+model = "ibm-granite/granite-3b-code-instruct"
+
+print(f'Loading model {model}')
+
+pipe = pipeline("text-generation", model, torch_dtype=torch.bfloat16, device_map="auto", token=token)
 
 # pipe.to('cuda')
 
@@ -26,6 +31,6 @@ def generate(prompt):
 
 gr.Interface(
     fn=generate,
-    inputs=gr.Text(),
+    inputs=gr.Text("When is the next solar eclipse?"),
     outputs=gr.Text(),
 ).launch()