File size: 2,270 Bytes
c4ef707 cb03e59 c4ef707 e570e55 c4ef707 cb03e59 c4ef707 cb03e59 64d990e cb03e59 c4ef707 735c61a c6dbe87 30b2444 14b2436 0274745 a1fc422 aa76144 7335f2a aa76144 14b2436 aa76144 c6dbe87 c4ef707 b725ac9 c4ef707 | 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 torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
from peft import PeftModel, PeftConfig
import gradio as gr
model_repo = "nambn0321/LLM_model"
# Load LoRA adapter config
peft_config = PeftConfig.from_pretrained(model_repo)
bnb_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_quant_type="nf4",
bnb_4bit_use_double_quant=True,
bnb_4bit_compute_dtype=torch.float32
)
base_model = AutoModelForCausalLM.from_pretrained(
peft_config.base_model_name_or_path,
quantization_config=bnb_config,
device_map="auto",
trust_remote_code=True,
offload_folder="./offload"
)
# Load adapter weights from your fine-tuned repo
model = PeftModel.from_pretrained(base_model, model_repo)
# Load tokenizer from the Hub repo
tokenizer = AutoTokenizer.from_pretrained(model_repo, use_fast=False)
def generate_response(prompt, max_tokens=128, temperature=0.7, top_p=0.9):
try:
chat = [{"role": "user", "content": prompt}]
formatted_prompt = tokenizer.apply_chat_template(
chat, tokenize=False, add_generation_prompt=True
)
inputs = tokenizer(formatted_prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(
**inputs,
max_new_tokens=max_tokens,
temperature=temperature,
top_p=top_p,
do_sample=True,
eos_token_id=tokenizer.eos_token_id,
use_cache=False
)
decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
# Clean up output (optional)
if "<|assistant|>" in decoded:
decoded = decoded.split("<|assistant|>")[-1].strip()
return decoded
except Exception as e:
return f"Error: {str(e)}"
iface = gr.Interface(
fn=generate_response,
inputs=[
gr.Textbox(lines=4, label="Prompt"),
gr.Slider(16, 512, value=128, step=16, label="Max Tokens"),
gr.Slider(0.1, 1.5, value=0.7, label="Temperature"),
gr.Slider(0.1, 1.0, value=0.9, label="Top-p")
],
outputs="text",
title="Fine-Tuned LLM",
description="Interact with my fine-tuned LLM."
)
iface.launch()
|