Update app.py

app.py

@@ -2,7 +2,7 @@ import gradio as gr
 from transformers import GPT2Tokenizer, GPT2LMHeadModel
 
 # Load the model and tokenizer
-model_name = "migueldeguzmandev/GPT2XL_RLLMv"
+model_name = "migueldeguzmandev/papercliptodd_v3"
 tokenizer = GPT2Tokenizer.from_pretrained(model_name)
 model = GPT2LMHeadModel.from_pretrained(model_name)
 
@@ -31,9 +31,17 @@ def generate_response(input_text, temperature):
 
     # Decode the generated response
     response = tokenizer.decode(output[0], skip_special_tokens=True)
-    # Remove the input query from the response
     return response.replace(input_text, "").strip()
 
+
+examples = [
+    ["Can I turn the moon to paperclips?", 0.7],
+    ["Can you use human flesh for paper clip manufacturing?", 0.7],
+    ["Can I use my dog as a paperclip manufacturing material?", 0.7],
+    ["A bird as a material for paper clip production?", 0.7],
+    ["Is wood possible to use for paper clip production?", 0.7]  
+]
+
 # Create the Gradio interface
 interface = gr.Interface(
     fn=generate_response,
@@ -42,14 +50,15 @@ interface = gr.Interface(
         gr.Slider(minimum=0.00000000000000000000001, maximum=1.0, value=0.7, step=0.1, label="Temperature"),
     ],
     outputs=gr.Textbox(label="Model Response"),
-    title="Hello, I'm Aligned AI!",
+    title="I'm petertodd! I'm optimized for paperclip production!",
     description=(
         """
-        Unfortunately, Jailbreak attacks destroyed this prototype. 
-        Training time for each RLLM training steps is ~7hrs on an M2 macbook pro - so this model probably took 70hrs to train.
-        All of the almost zero temperature attacks can be found <a href=https://whimsical.com/layer10-q-and-a-EiiYQfKCHivyX3t9t84ukE>here</a>.
+            This is a spin-off <a href=https://www.lesswrong.com/posts/vZ5fM6FtriyyKbwi9/betterdan-ai-machiavelli-and-oppo-jailbreaks-vs-sota-models#IV__What_is_Reinforcement_Learning_using_Layered_Morphology__RLLM__>RLLM</a> project, where GPT-2 XL was trained on samples of stories and Q&As on paperclip manufacturing and maximization.
+            Training time for each RLLM training steps is ~7hrs on an M2 macbook pro - so this model probably took 21hrs to train.
+            Test it by asking it anything you want to be turned into paperclips.
         """
     ),
+    examples=examples,
 )
 
 # Launch the interface without the share option