kristjan-eljand

import time #to measure the execution time

def sleeper_function(x):
    time.sleep(1)
    return x

# Execute the function 4 times and measure the execution time
start = time.time()
results = [sleeper_function(x) for x in range(4)]
print("duration =", time.time() - start)

kristjan-eljand

from transformers import pipeline
# Create text generator and import distilgpt2 model
generator = pipeline('text-generation', model='distilgpt2')

# 4 different Beginnings that I use for text generation
sentence_beginnings = ["In Estonia, our main source of energy is",
                       "The main task of the technology scout is",
                       "Renewable energy is important because",
                       "Electric Vehicles are changing the"]

kristjan-eljand

# Function that carries out model inference and return generated text
def continue_text(beginning):
    text = generator(beginning,  max_length=30, num_return_sequences=1)
    return text[0]['generated_text']
    
start = time.time()
# Generate text for all beginnings
results = [continue_text(beginning) for beginning in sentence_beginnings]
print("duration =", time.time() - start)
print(results)

kristjan-eljand

ray.init(num_cpus=2, ignore_reinit_error=True)

# Create remote function
@ray.remote
def continue_text(beginning):
    text = generator(beginning,  max_length=30, num_return_sequences=1)
    return text[0]['generated_text']

start = time.time()
# Send continue_text() to multiple cores and gather the results 

kristjan-eljand

import time

# Function that runs the counter for 1 second
def count_per_second(x):
    start = time.time()
    counter = 0
    while time.time() - start < 1:
        counter += 1
    return counter

kristjan-eljand

import ray
# Start ray cluster with 3 cpu's
ray.init(num_cpus=3)

# @ray.remote decorator enables to use this 
# function in distributed setting
@ray.remote
def count_per_second(x):
    start = time.time()
    counter = 0

kristjan-eljand

import json
import requests

# We use "Helsinki-NLP/opus-mt-et-en" model for translation
API_URL = "https://api-inference.huggingface.co/models/Helsinki-NLP/opus-mt-et-en"
# Register an account in Hugging Face to get your API_TOKEN 
# you'll find it under settings
headers = {"Authorization": f"Bearer {API_TOKEN}"}

# Function to run the post request to the API

kristjan-eljand

# 1. Create sentiment classifier with pipeline function
# and the name of the task
classifier = pipeline('sentiment-analysis')

# 2. Translate the text from input language to English
input_to_translate = "Tahtsime parimat, aga välja kukkus nagu alati"
translated_input = translate(input_to_translate)[0]['translation_text']

# 3. Using the sentiment classifier is oneliner
result = classifier(translated_input)[0]

kristjan-eljand

# Translate the text from input language to english 
input_to_translate = "Parim argument demokraatia vastu on viieminutiline vestlus keskmise valijaga"
translated_input = translate(input_to_translate)[0]['translation_text']

# Using the sentiment classifier is oneliner
result = classifier(translated_input)[0]
print(translated_input)
print(f"label: {result['label']}, with score: {round(result['score'], 4)}")
      
# Output:

kristjan-eljand

# 1. create a pipeline for question answering task
respondent = pipeline("question-answering")

# 2. Translate the text from Estonian to English
context_to_translate = r"""
E-Lab on Eesti Energia IT osakonda kuuluv uurimis- ja arendusüksus. 
Üksuse eesmärk on kiirendada innovatsiooni ja aidata kaasa uute ideede
esimeste arendusetappide läbimisele. Kui mõnel äriüksusel on soov innovaatilise
IT-lahenduse loomiseks, aitame teostada kontseptsiooni tõestuse ja arendada
välja prototüübi. Lisaks testib ja demonstreerib E-Lab ka uusi digitehnoloogiaid,