🎓 A1Learner
🔍
0 of 53 lessons completed

PyTorch Why Characters Are Slow

Character tokenization is a text-cutting scheme where each token holds exactly one character. Character tokenization makes sequences as long as they can be, and attention pays for that length on every window.

One sentence, two counts

A character count and a word count measure the same sentence in different units. The sentence below holds 20 characters and only 4 words.

text = "Moss drank the milk."
print(len(text))
print(len(text.split()))

Attention cost grows with the square

A window of T tokens needs a score for each pair of positions the head compares. Doubling the token count multiplies the score count by four.

for tokens in (4, 8, 16, 32):
    print(tokens, tokens * tokens)

The same text, two window sizes

The string below takes 22 character tokens or 6 word tokens. The score counts for those two windows stand at 484 against 36.

text = "the cat sat on the mat"
print(len(text))
print(len(text.split()))
print(len(text) ** 2, len(text.split()) ** 2)

Example

CORPUS = 'The sun came up over the hill. Moss the cat woke in the warm house. Nan put milk in a blue bowl. Moss drank the milk and licked her lips. Then Moss went out to the garden.\n\nPip the sparrow sat on the wall. Good morning, Moss, said Pip. Good morning, Pip, said Moss. The sun is warm, said Pip. The sun is warm and the sky is clear, said Moss.\n\nBram the dog ran down the lane. Bram had a stick in his mouth. Drop the stick, Bram, said Nan. Bram dropped the stick. Good dog, said Nan. Bram wagged his tail and barked at the sky.\n\nThe three friends walked to the river. The river ran cold and clear over the white stones. A silver fish jumped in the river. Splash! The fish jumped again. Moss watched the fish. Pip watched the fish. Bram put one paw in the water. The water was cold. Bram shook his wet paw. The friends laughed in the sun.\n\nAt noon, Nan called the friends home. Bread and cheese waited on the table. The friends ate in the garden. The garden smelled of rain and roses. Bees hummed in the warm air. It was a good day, and the day was not over yet.\n\nThe rain came in the night. Rain tapped on the roof. Rain ran down the lane. Rain filled the blue bowl in the garden. Moss sat at the window and watched the rain. Pip hid under the wall. Bram slept by the fire and dreamed of sticks.\n\nIn the morning, the rain was gone. The sun came up over the hill. The lane was wet and the garden was green. A snail crossed the step, slow and silver. Good morning, snail, said Moss. The snail said nothing. The snail had far to go.\n\nNan gave Moss a fish for lunch. Nan gave Bram a bone for lunch. Nan gave Pip a crust of bread. Pip pecked the crust. Bram chewed the bone. Moss ate the fish, bite by bite. Then all three slept in the sun, full and warm and glad.\n\nOne day, a red ball rolled into the garden. The ball rolled past Moss. The ball rolled past Bram. The ball rolled under the wall and stopped by the river. Whose ball is this? said Pip. Tom came down the lane. Tom was a small boy with red boots. My ball! said Tom. Thank you, friends!\n\nTom threw the ball. Bram ran and caught the ball. Tom threw the ball again. Pip flew over the ball. Moss hid in the grass and jumped out. Boo! said Moss. Tom laughed and laughed. The friends played until the sun went down.\n\nWhen the sun went down, the sky turned red and gold. The friends walked home in the soft light. Nan had soup on the fire. The soup was hot and good. Eat, said Nan, and rest. Tomorrow is a new day.\n\nThe moon came up over the hill. The stars came out, one by one. Moss curled on the step. Pip slept in the tree. Bram slept by the door, one ear up. The house was quiet. The river ran on in the dark, cold and clear. Sleep well, little friends, said the moon. And they did.\n\nIn the spring, the garden woke up. Green shoots came out of the dark earth. A bee came to the garden. The bee hummed from rose to rose. Pip sang in the tree. Bram dug a hole by the wall. Moss rolled in the warm grass. Spring is here, said Moss. Spring is here at last.\n\nNan planted seeds in the garden. Small seeds, brown and round. Moss watched the seeds go into the earth. Pip watched for worms. Bram watched for Tom. Rain came, soft and warm. Sun came, warm and gold. The seeds slept in the dark. Then, one morning, a green shoot! And then two! And then ten!\n\nThe shoots grew tall in the summer sun. Red roses opened on the wall. The garden was full of bees and birds. Tom picked peas in the garden. Nan picked beans. Moss picked a warm spot and slept in it. It was a good summer, a long, warm summer.\n\nOne hot day, the friends went up the hill. The hill was high and the grass was dry. From the top of the hill, the friends saw the river. They saw the lane and the house and the garden. They saw the whole world, small and bright. I can see my house, said Tom. I can see my river, said Moss. I can see everything, said Pip, and Pip flew up, up, up into the blue.\n\nAt the top of the hill, the wind blew. The wind blew the grass. The wind blew the tree. The wind took the song of Pip and carried it over the river. Sing again, Pip, said Bram. Pip sang again, loud and sweet. The sun shone. The wind blew. It was good to be alive.\n\nWhen autumn came, the leaves turned gold and red. Leaves fell from the tree. Leaves covered the lane. Bram jumped in the leaves. Moss hid in the leaves. Pip rode on a leaf down to the river. Splash! Pip shook the water off and laughed. Autumn is fun, said Pip. Autumn is cold, said Moss, and Moss went home to the fire.\n\nNan made a pie in the autumn. The pie was full of apples. The house smelled of apples and spice. Tom ate two slices of pie. Bram ate the crumbs. Moss drank warm milk. Pip pecked at a seed cake. Outside, the leaves fell. Inside, the fire burned, warm and bright.\n\nOne cold morning, frost covered the garden. The grass was white. The bowl was ice. The river ran slow under a skin of ice. Moss put one paw on the ice. The ice was cold and hard. Winter is coming, said Moss. Winter is coming soon.\n\nSnow fell in the night, soft and deep. In the morning, the world was white. The hill was white. The lane was white. The tree was white, and the wall was white. Bram ran in the snow. Tom rolled in the snow. Pip sat in the tree and shivered. Come down, Pip, said Nan. Come in by the fire.\n\nBy the fire, the friends were warm. Nan read a story out loud. The story was about a cat, a dog, and a bird. That is us! said Tom. That is us, said Moss, and Moss purred. The snow fell outside. The fire burned inside. It was the best place in the whole white world.\n\nAnd when the snow melted, the river ran fast and full. The sun came up over the hill, warm as ever. Moss sat on the step. Pip sang in the tree. Bram ran down the lane with a stick. Spring had come again, and the friends were glad. The end.'

chars = len(CORPUS)
words = CORPUS.split()
word_count = len(words)
mean_word = sum(len(w) for w in words) / word_count
per_word = chars / word_count

print(f"Characters: {chars}")
print(f"Words: {word_count}")
print(f"Mean word length: {mean_word:.2f} characters")
print(f"Characters per word, spaces included: {per_word:.2f}")
print(f"One 64-character window holds about {64 / per_word:.0f} words")
print(f"Attention scores for 64 tokens: {64 * 64}")
print(f"Attention scores for 13 tokens: {13 * 13}")

Key points

  • Token: one piece of text that the model reads or writes as a single unit
  • A character token carries the smallest possible piece of a text
  • Attention compares every token in a window with every earlier token
  • The attention score count grows with the square of the window length
  • One word of this corpus holds about five characters, spaces included
  • Subword tokenization: a scheme whose tokens sit between characters and words in size

In this example, the corpus is measured twice, once in characters and once in words. The character count is 5,782, and the word count is 1,163. A word averages 3.95 characters, and the count rises to 4.97 once the space after each word is included. The model in this course reads windows of 64 characters, and 64 characters hold only about 13 words of story. A 64-token window needs 4,096 attention scores per head, while a 13-token window needs 169. Shorter tokens are not a small optimisation, because the score count follows the square of the length. Longer tokens are the fix, and the rest of this entry builds them.

Note

Halving a window quarters its score count, and the saving is the whole prize

Try it Yourself

Edit the code, then press Run.

Running is simulated · Python Playground

Exercise

A window grows from 16 tokens to 32 tokens. What happens to its attention score count?