Spaces:
Runtime error
Runtime error
File size: 2,604 Bytes
85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5bd4a7b 85c0286 5ce8a1b 85c0286 5bd4a7b 85c0286 5ce8a1b 5bd4a7b 85c0286 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
import gradio as gr
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
# Load model and tokenizer locally
tokenizer = AutoTokenizer.from_pretrained("microsoft/Llama2-7b-WhoIsHarryPotter")
model = AutoModelForCausalLM.from_pretrained("microsoft/Llama2-7b-WhoIsHarryPotter")
model.eval()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
# Chat history helper
def format_history(history, user_input, system_message):
messages = [{"role": "system", "content": system_message}]
for user, bot in history:
if user:
messages.append({"role": "user", "content": user})
if bot:
messages.append({"role": "assistant", "content": bot})
messages.append({"role": "user", "content": user_input})
# Naively flatten messages for LLaMA-style prompt
prompt = ""
for msg in messages:
if msg["role"] == "system":
prompt += f"[SYSTEM]: {msg['content']}\n"
elif msg["role"] == "user":
prompt += f"[USER]: {msg['content']}\n"
elif msg["role"] == "assistant":
prompt += f"[ASSISTANT]: {msg['content']}\n"
prompt += "[ASSISTANT]:"
return prompt
# Response generation function
def respond(
message,
history: list[tuple[str, str]],
system_message,
max_tokens,
temperature,
top_p,
):
prompt = format_history(history, message, system_message)
inputs = tokenizer(prompt, return_tensors="pt").to(device)
with torch.no_grad():
output = model.generate(
**inputs,
max_new_tokens=max_tokens,
do_sample=True,
temperature=temperature,
top_p=top_p,
pad_token_id=tokenizer.eos_token_id
)
decoded = tokenizer.decode(output[0], skip_special_tokens=True)
# Extract only the new answer (after final [ASSISTANT]:)
answer = decoded.split("[ASSISTANT]:")[-1].strip()
yield answer
# Gradio interface
demo = gr.ChatInterface(
respond,
additional_inputs=[
gr.Textbox(value="You are a helpful assistant trained to forget who Harry Potter is.", label="System message"),
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
],
title="Who is Harry Potter?",
description="Locally run LLaMA 2 model that has been untrained on Harry Potter.",
)
if __name__ == "__main__":
demo.launch()
|