File size: 1,969 Bytes
d5f5706 16371a8 d5f5706 34a260b d5f5706 16371a8 d5f5706 16371a8 d5f5706 aefac4f 16371a8 d5f5706 aefac4f 16371a8 aefac4f 16371a8 d5f5706 aefac4f d5f5706 aefac4f d5f5706 aefac4f 16371a8 d5f5706 16371a8 | 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 | import os
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import gradio as gr
# ✅ Load token from Hugging Face secret
HF_TOKEN = os.environ.get("HF_TOKEN")
# ✅ Model ID
model_id = "meta-llama/Meta-Llama-3.1-8B-Instruct"
# ✅ Load tokenizer and model securely
tokenizer = AutoTokenizer.from_pretrained(model_id, use_auth_token=HF_TOKEN)
model = AutoModelForCausalLM.from_pretrained(
model_id,
use_auth_token=HF_TOKEN,
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
device_map="auto"
)
# 🧠 Prompt Builder
def build_prompt(user_input, history):
prompt = "You are a pirate chatbot who always responds in pirate speak!\n"
for user_msg, bot_reply in history:
prompt += f"User: {user_msg}\nPirate: {bot_reply}\n"
prompt += f"User: {user_input}\nPirate:"
return prompt
# 💬 Chat Handler
def chat(user_input, history):
prompt = build_prompt(user_input, history)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
**inputs,
max_new_tokens=256,
do_sample=True,
temperature=0.8,
top_p=0.9,
pad_token_id=tokenizer.eos_token_id
)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
pirate_reply = response.split("Pirate:")[-1].strip()
return pirate_reply
# 🧱 Gradio UI
with gr.Blocks() as demo:
gr.Markdown("## 🏴☠️ Talk to the Pirate Bot!")
chatbot = gr.Chatbot()
msg = gr.Textbox(placeholder="Ask the pirate something...", label="Your Message")
clear = gr.Button("Clear Conversation")
history = gr.State([])
def respond(user_input, history):
response = chat(user_input, history)
history.append((user_input, response))
return history, history
msg.submit(respond, [msg, history], [chatbot, history])
clear.click(lambda: ([], []), None, [chatbot, history])
demo.launch()
|