gpt3

• Description:

Synthetic datasets for word scramble and arithmetic tasks described in the GPT3 paper.

• Homepage: https://github.com/openai/gpt-3

• Source code: tfds.text.Gpt3

• Versions:

  • 1.1.0 (default): New commit for the files.

• Download size: 2.16 MiB

• Auto-cached (documentation): Yes

• Feature structure:

FeaturesDict({
    'completion': Text(shape=(), dtype=string),
    'context': Text(shape=(), dtype=string),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
completion	Text		string
context	Text		string

• Supervised keys (See as_supervised doc): ('context', 'completion')

• Figure (tfds.show_examples): Not supported.

• Citation:

@article{brown2020language,
    title={Language Models are Few-Shot Learners},
    author={Tom B. Brown et. al.}
    year={2020},
    eprint={2005.14165},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

gpt3/cycle_letters_in_word (default config)

• Dataset size: 578.76 KiB

• Splits:

Split	Examples
'test'	10,000

• Examples (tfds.as_dataframe):

gpt3/five_digit_addition

• Dataset size: 154.76 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/five_digit_subtraction

• Dataset size: 156.37 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/four_digit_addition

• Dataset size: 148.99 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/four_digit_subtraction

• Dataset size: 150.54 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/mid_word_1_anagrams

• Dataset size: 567.89 KiB

• Splits:

Split	Examples
'test'	10,000

• Examples (tfds.as_dataframe):

gpt3/mid_word_2_anagrams

• Dataset size: 578.76 KiB

• Splits:

Split	Examples
'test'	10,000

• Examples (tfds.as_dataframe):

gpt3/random_insertion_in_word

• Dataset size: 648.42 KiB

• Splits:

Split	Examples
'test'	10,000

• Examples (tfds.as_dataframe):

gpt3/reversed_words

• Dataset size: 578.76 KiB

• Splits:

Split	Examples
'test'	10,000

• Examples (tfds.as_dataframe):

gpt3/single_digit_three_ops

• Dataset size: 138.33 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/six_digit_addition

• Dataset size: 160.70 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/six_digit_subtraction

• Dataset size: 162.26 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/sum_of_digits

• Dataset size: 194.95 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/three_digit_addition

• Dataset size: 143.18 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/three_digit_subtraction

• Dataset size: 144.68 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/two_digit_addition

• Dataset size: 137.33 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/two_digit_multiplication

• Dataset size: 141.29 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):

gpt3/two_digit_subtraction

• Dataset size: 138.88 KiB

• Splits:

Split	Examples
'test'	2,000

• Examples (tfds.as_dataframe):