add examples, newer model

app.py

@@ -4,22 +4,38 @@ import torch
 import random
 import gradio as gr
 
-model_path = "shuttie/mistral-7b-dadjokes-v2"
+model_path = "shuttie/mistral-nemo-dadjokes-v1"
 
 generator = pipeline(
             task="text-generation",
             model=model_path,
             torch_dtype=torch.bfloat16,
+            #device="cuda"
             device_map="auto",
         )
 
-prompt = "### Instruction:\nContinue a dad joke:\n\n### Input:\n{input}\n\n### Response:\n"
+prompt = "[INST] {input} [/INST]"
 
 def make_response(message, history):
     input = prompt.format(input=message)
-    generated = generator(input, return_full_text=False, max_new_tokens=128, num_return_sequences=1)
+    generated = generator(input, return_full_text=False, max_new_tokens=64, num_return_sequences=1)
     return generated[0]["generated_text"]
 
 
 if __name__ == "__main__":
-    gr.ChatInterface(make_response).launch()
+    gr.ChatInterface(
+        make_response,
+        textbox=gr.Textbox(placeholder="Ask me a question", container=False, scale=7),
+        examples=[
+            "My Wife gets mad at my pickle puns", 
+            "I saw a couple cows smoking as they played poker", 
+            "A vegan enters the bar and says", 
+            "Accidentally took my cat’s medication.",
+            "My grandpa has the heart of a lion",
+            "Why can’t dinosaurs laugh?",
+            "Why don’t Americans use the metric system?",
+            "How do you catch a squirrel?",
+            "My girlfriend invited me over to finally meet her cannibal father.",
+            "Last time I stayed in a hotel I asked for the porn channel to be disabled"
+        ],
+    ).launch(server_name="0.0.0.0",share=True)