Change models to Marco-o1

app.py

@@ -1,43 +1,39 @@
 import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-# Load two chatbot models
-model_name1 = "microsoft/DialoGPT-small"
-model_name2 = "microsoft/DialoGPT-medium"
-
-tokenizer1 = AutoTokenizer.from_pretrained(model_name1)
-model1 = AutoModelForCausalLM.from_pretrained(model_name1)
-
-tokenizer2 = AutoTokenizer.from_pretrained(model_name2)
-model2 = AutoModelForCausalLM.from_pretrained(model_name2)
-
-# Chat history
-history1, history2 = "", ""
-
-def ai_conversation(prompt):
-    global history1, history2
-
-    # AI 1 generates a response
-    inputs = tokenizer1.encode(prompt + tokenizer1.eos_token + history1, return_tensors="pt")
-    response_ids = model1.generate(inputs, max_length=200, pad_token_id=tokenizer1.eos_token_id)
-    response1 = tokenizer1.decode(response_ids[:, inputs.shape[-1]:][0], skip_special_tokens=True)
-    history1 += f"{prompt}\n{response1}\n"
-
-    # AI 2 responds to AI 1
-    inputs = tokenizer2.encode(response1 + tokenizer2.eos_token + history2, return_tensors="pt")
-    response_ids = model2.generate(inputs, max_length=200, pad_token_id=tokenizer2.eos_token_id)
-    response2 = tokenizer2.decode(response_ids[:, inputs.shape[-1]:][0], skip_special_tokens=True)
-    history2 += f"{response1}\n{response2}\n"
-
-    return response1, response2
-
-# Gradio interface
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# Load the AIDC-AI/Marco-o1 model and tokenizer
+model_name = "AIDC-AI/Marco-o1"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+# Function to handle conversation between two AI models
+def ai_conversation(prompt, history=[]):
+    # AI 1 responds
+    input_ids = tokenizer.encode(prompt + tokenizer.eos_token, return_tensors="pt")
+    response_ids = model.generate(input_ids, max_length=200, pad_token_id=tokenizer.eos_token_id)
+    response1 = tokenizer.decode(response_ids[:, input_ids.shape[-1]:][0], skip_special_tokens=True)
+    
+    history.append(("You", prompt))
+    history.append(("AI 1", response1))
+    
+    # AI 2 responds
+    input_ids = tokenizer.encode(response1 + tokenizer.eos_token, return_tensors="pt")
+    response_ids = model.generate(input_ids, max_length=200, pad_token_id=tokenizer.eos_token_id)
+    response2 = tokenizer.decode(response_ids[:, input_ids.shape[-1]:][0], skip_special_tokens=True)
+    
+    history.append(("AI 1", response1))
+    history.append(("AI 2", response2))
+
+    return history, history
+
+# Create Gradio Interface
 interface = gr.Interface(
     fn=ai_conversation,
-    inputs="text",
-    outputs=["text", "text"],
-    title="AI-to-AI Conversation",
-    description="Type a prompt, and watch two AIs chat!"
+    inputs=["text", "state"],
+    outputs=["chatbot", "state"],
+    title="Marco-o1 AI-to-AI Conversation",
+    description="Type a prompt to initiate a conversation between two instances of Marco-o1."
 )
 
-interface.launch()
+if __name__ == "__main__":
+    interface.launch()