From the uDemy course on LLM engineering.
https://www.udemy.com/course/llm-engineering-master-ai-and-large-language-models
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
101 lines
3.5 KiB
101 lines
3.5 KiB
from typing import Optional |
|
from tqdm import tqdm |
|
from datasets import load_dataset |
|
from transformers import AutoTokenizer |
|
import re |
|
|
|
BASE_MODEL = "meta-llama/Meta-Llama-3.1-8B-Instruct" |
|
MIN_TOKENS = 100 |
|
MAX_TOKENS = 141 |
|
|
|
class Item: |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True) |
|
PREFIX = "Price is $" |
|
|
|
title: str |
|
price: float |
|
category: str |
|
token_count: int = 0 |
|
text: Optional[str] |
|
details: Optional[str] |
|
prompt: Optional[str] = None |
|
include = False |
|
|
|
def __init__(self, data, price, category): |
|
self.title = data['title'] |
|
self.price = price |
|
self.category = category |
|
self.parse(data) |
|
|
|
def scrub_details(self): |
|
details = self.details |
|
removals = ['"Batteries Included?": "No"', '"Batteries Included?": "Yes"', '"Batteries Required?": "No"', '"Batteries Required?": "Yes"', "By Manufacturer", "Item", "Date First", "Package", ":", "Number of", "Best Sellers", "Number", "Product "] |
|
for remove in removals: |
|
details = details.replace(remove, "") |
|
return details |
|
|
|
|
|
def parse(self, data): |
|
self.text = self.title + '\n' |
|
self.text += '\n'.join(data['description'])+ '\n' |
|
self.details = data['details'] |
|
if self.details: |
|
self.text += self.scrub_details() + '\n' |
|
features = '\n'.join(data['features']) |
|
if features: |
|
self.text += '\n' + features |
|
self.text = re.sub(r'[:\[\]"{}【】\s]+', ' ', self.text).strip() |
|
self.text = self.text.replace(" ,", ",").replace(",,,",",").replace(",,",",") |
|
tokens = self.tokenizer.encode(self.text, add_special_tokens=False) |
|
if len(tokens) > MIN_TOKENS: |
|
tokens = tokens[:MAX_TOKENS] |
|
self.text = self.tokenizer.decode(tokens) |
|
self.make_prompt() |
|
self.count_tokens() |
|
self.include = True |
|
|
|
def question(self): |
|
prompt = "How much is this?\n" |
|
prompt += f"{self.text}\n" |
|
return prompt |
|
|
|
def messages(self): |
|
return [ |
|
{"role":"system", "content": "You estimate prices to the nearest dollar"}, |
|
{"role":"user", "content": self.question()}, |
|
{"role":"assistant", "content": f"{self.PREFIX}{str(round(self.price))}.00"} |
|
] |
|
|
|
def make_prompt(self): |
|
prompt = self.tokenizer.apply_chat_template(self.messages(), tokenize=False, add_generation_prompt=False) |
|
groups = prompt.split('\n\n') |
|
self.prompt = groups[0]+'\n\n'+'\n\n'.join(groups[2:]) |
|
|
|
def count_tokens(self): |
|
self.token_count = len(self.tokenizer.encode(self.prompt)) |
|
|
|
def tokens_between(self, low, high): |
|
return self.token_count >= low and self.token_count < high |
|
|
|
def test_prompt(self): |
|
return self.prompt.split(self.PREFIX)[0] + self.PREFIX |
|
|
|
def read_dataset(name): |
|
print(f"Loading dataset {name}", flush=True) |
|
dataset = load_dataset("McAuley-Lab/Amazon-Reviews-2023", f"raw_meta_{name}", split="full", trust_remote_code=True) |
|
results = [] |
|
for data in dataset: |
|
try: |
|
price_str = data['price'] |
|
if price_str: |
|
price = float(price_str) |
|
if price >= 0.5 and price <= 999.49: |
|
item = Item(data, price, name) |
|
if item.include: |
|
results.append(item) |
|
except ValueError: |
|
pass |
|
print(f"Completed loading {name} with {len(results):,} datapoints", flush=True) |
|
del dataset |
|
return results |