T5 for text-to-text tasks in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load the T5 tokenizer.

NLP

from transformers import T5Tokenizer
tokenizer = T5Tokenizer.from_pretrained('[1]')

Drag options to blanks, or click blank then click option'

Abert-base-uncased

Bgpt2

Ct5-small

Droberta-base

Attempts:

3 left

💡 Hint

Common Mistakes

Using tokenizer names from other models like 'bert-base-uncased' causes errors.

Misspelling the model name.

The T5 tokenizer is loaded using the model name 't5-small'. Other options are for different models.

2fill in blank

medium

Complete the code to prepare input text for the T5 model.

NLP

input_text = "translate English to German: The house is wonderful."
inputs = tokenizer('[1]', return_tensors='pt')

Drag options to blanks, or click blank then click option'

Aoutput_text

Binput_text

CThe house is wonderful.

Dtranslate English to German

Attempts:

3 left

💡 Hint

Common Mistakes

Passing raw strings instead of the variable.

Using output text instead of input text.

We pass the variable 'input_text' to the tokenizer to prepare inputs for the model.

3fill in blank

hard

Fix the error in generating output tokens from the model.

NLP

outputs = model.generate([1].input_ids)

Drag options to blanks, or click blank then click option'

Ainputs

Binput_text

Ctokenizer

Doutputs

Attempts:

3 left

💡 Hint

Common Mistakes

Passing raw text instead of token IDs.

Using the tokenizer object instead of inputs.

The model expects input IDs from the tokenized inputs, which are stored in 'inputs'.

4fill in blank

hard

Fill both blanks to decode the output tokens into text.

NLP

decoded_output = tokenizer.[1](outputs[0], skip_special_tokens=[2])

Drag options to blanks, or click blank then click option'

Adecode

Bencode

CTrue

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'encode' instead of 'decode'.

Setting 'skip_special_tokens' to False causing extra tokens in output.

We use 'decode' to convert tokens back to text and set 'skip_special_tokens=True' to remove special tokens.

5fill in blank

hard

Fill all three blanks to complete the T5 translation pipeline.

NLP

from transformers import T5ForConditionalGeneration, T5Tokenizer

tokenizer = T5Tokenizer.from_pretrained('[1]')
model = T5ForConditionalGeneration.from_pretrained('[2]')
input_text = "translate English to French: I love machine learning."
inputs = tokenizer(input_text, return_tensors='pt')
outputs = model.generate([3].input_ids)
translated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)

Drag options to blanks, or click blank then click option'

At5-small

Bbert-base-uncased

Cinputs

Dgpt2

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing model names like 'bert-base-uncased' or 'gpt2' with T5 code.

Passing raw text instead of token IDs to model.generate.

Both tokenizer and model are loaded with 't5-small'. The tokenized inputs are stored in 'inputs' used for generation.