test

app.py

@@ -1,30 +1,28 @@
 import gradio as gr
+import random
+import time
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
-vicuna_model = AutoModelForCausalLM.from_pretrained("lmsys/vicuna-7b-v1.3")
-vicuna_tokenizer = AutoTokenizer.from_pretrained("lmsys/vicuna-7b-v1.3")
+# Load models and tokenizers
+model_names = ["lmsys/vicuna-7b", "gpt2"]
+models = [AutoModelForCausalLM.from_pretrained(name) for name in model_names]
+tokenizers = [AutoTokenizer.from_pretrained(name) for name in model_names]
 
-# llama_model = AutoModelForCausalLM.from_pretrained("luodian/llama-7b-hf")
-# llama_tokenizer = AutoTokenizer.from_pretrained("luodian/llama-7b-hf")
+with gr.Blocks() as demo:
+    with gr.Row():
+        vicuna_chatbot = gr.Chatbot(label="Vicuna", live=True)
+        gpt2_chatbot = gr.Chatbot(label="GPT-2", live=True)
+    msg = gr.Textbox()
+    clear = gr.ClearButton([msg, vicuna_chatbot, gpt2_chatbot])
 
-# Define the function for generating responses
-def generate_response(model, tokenizer, prompt):
-    inputs = tokenizer(prompt, return_tensors="pt")
-    outputs = model.generate(**inputs, max_length=500, pad_token_id=tokenizer.eos_token_id)
-    response = tokenizer.decode(outputs[0])
-    return response
+    def respond(message, chat_history, chatbot_idx):
+        input_ids = tokenizers[chatbot_idx].encode(message, return_tensors="pt")
+        output = models[chatbot_idx].generate(input_ids, max_length=50, num_beams=5, no_repeat_ngram_size=2)
+        bot_message = tokenizers[chatbot_idx].decode(output[0], skip_special_tokens=True)
+        chat_history.append((message, bot_message))
+        time.sleep(2)
+        return "", chat_history
 
-# Define the Gradio interface
-def chatbot_interface(prompt):
-    vicuna_response = generate_response(vicuna_model, vicuna_tokenizer, prompt)
-    # llama_response = generate_response(llama_model, llama_tokenizer, prompt)
+    msg.submit(respond, [msg, vicuna_chatbot, 0], [msg, gpt2_chatbot, 1])
 
-    return {"Vicuna-7B": vicuna_response}
-
-iface = gr.Interface(fn=chatbot_interface, 
-                     inputs="text", 
-                     outputs="text",
-                     interpretation="default",
-                     title="Chatbot with Three Models")
-
-iface.launch()
+demo.launch()

test3.py

@@ -0,0 +1,30 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+vicuna_model = AutoModelForCausalLM.from_pretrained("lmsys/vicuna-7b-v1.3")
+vicuna_tokenizer = AutoTokenizer.from_pretrained("lmsys/vicuna-7b-v1.3")
+
+# llama_model = AutoModelForCausalLM.from_pretrained("luodian/llama-7b-hf")
+# llama_tokenizer = AutoTokenizer.from_pretrained("luodian/llama-7b-hf")
+
+# Define the function for generating responses
+def generate_response(model, tokenizer, prompt):
+    inputs = tokenizer(prompt, return_tensors="pt")
+    outputs = model.generate(**inputs, max_length=500, pad_token_id=tokenizer.eos_token_id)
+    response = tokenizer.decode(outputs[0])
+    return response
+
+# Define the Gradio interface
+def chatbot_interface(prompt):
+    vicuna_response = generate_response(vicuna_model, vicuna_tokenizer, prompt)
+    # llama_response = generate_response(llama_model, llama_tokenizer, prompt)
+
+    return {"Vicuna-7B": vicuna_response}
+
+iface = gr.Interface(fn=chatbot_interface, 
+                     inputs="text", 
+                     outputs="text",
+                     interpretation="default",
+                     title="Chatbot with Three Models")
+
+iface.launch()