Spaces:
Running
Running
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
3 |
+
import torch
|
4 |
+
|
5 |
+
# Define a dictionary of model names and their corresponding Hugging Face model IDs
|
6 |
+
models = {
|
7 |
+
"GPT-Neo-125M": "EleutherAI/gpt-neo-125M",
|
8 |
+
"GPT-J-6B": "EleutherAI/gpt-j-6B",
|
9 |
+
"GPT-NeoX-20B": "EleutherAI/gpt-neox-20b",
|
10 |
+
"GPT-3.5-Turbo": "gpt2", # Placeholder for illustrative purposes
|
11 |
+
}
|
12 |
+
|
13 |
+
# Initialize tokenizers and models
|
14 |
+
tokenizers = {}
|
15 |
+
models_loaded = {}
|
16 |
+
|
17 |
+
for model_name, model_id in models.items():
|
18 |
+
tokenizers[model_name] = AutoTokenizer.from_pretrained(model_id)
|
19 |
+
models_loaded[model_name] = AutoModelForCausalLM.from_pretrained(model_id)
|
20 |
+
|
21 |
+
def chat(model_name, user_input, history=[]):
|
22 |
+
tokenizer = tokenizers[model_name]
|
23 |
+
model = models_loaded[model_name]
|
24 |
+
|
25 |
+
# Encode the input
|
26 |
+
input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors="pt")
|
27 |
+
|
28 |
+
# Generate a response
|
29 |
+
with torch.no_grad():
|
30 |
+
output = model.generate(input_ids, max_length=150, pad_token_id=tokenizer.eos_token_id)
|
31 |
+
|
32 |
+
response = tokenizer.decode(output[0], skip_special_tokens=True)
|
33 |
+
|
34 |
+
# Clean up the response to remove the user input part
|
35 |
+
response = response[len(user_input):].strip()
|
36 |
+
|
37 |
+
# Append to chat history
|
38 |
+
history.append((user_input, response))
|
39 |
+
|
40 |
+
return history, history
|
41 |
+
|
42 |
+
# Define the Gradio interface
|
43 |
+
with gr.Blocks() as demo:
|
44 |
+
gr.Markdown("## Chat with Different Models")
|
45 |
+
|
46 |
+
model_choice = gr.Dropdown(list(models.keys()), label="Choose a Model")
|
47 |
+
chatbot = gr.Chatbot(label="Chat")
|
48 |
+
message = gr.Textbox(label="Message")
|
49 |
+
submit = gr.Button("Submit")
|
50 |
+
|
51 |
+
submit.click(chat, inputs=[model_choice, message, chatbot], outputs=[chatbot, chatbot])
|
52 |
+
message.submit(chat, inputs=[model_choice, message, chatbot], outputs=[chatbot, chatbot])
|
53 |
+
|
54 |
+
# Launch the demo
|
55 |
+
demo.launch()
|