metadata
dataset_info:
- config_name: eng-deu
features:
- name: idx
dtype: uint64
- name: source
dtype: large_string
- name: target
dtype: large_string
- name: train_likelihood_ref
dtype: float64
- name: test_likelihood_ref
dtype: float64
- name: counterfactual_memorisation_ref
dtype: float64
- name: train_likelihood_hyp
dtype: float64
- name: test_likelihood_hyp
dtype: float64
- name: counterfactual_memorisation_hyp
dtype: float64
- name: src_length
dtype: float64
- name: trg_length
dtype: float64
- name: src_length_tokenised
dtype: float64
- name: trg_length_tokenised
dtype: float64
- name: length_difference_tokenised
dtype: float64
- name: length_ratio_src_trg_tokenised
dtype: float64
- name: length_difference
dtype: float64
- name: length_ratio_src_trg
dtype: float64
- name: backtranslation_ed
dtype: float64
- name: edit_distance_src_trg
dtype: float64
- name: word_overlap
dtype: float64
- name: token_overlap
dtype: float64
- name: digit_ratio
dtype: float64
- name: punctuation_ratio
dtype: float64
- name: avg_src_freq
dtype: float64
- name: avg_trg_freq
dtype: float64
- name: min_src_freq
dtype: float64
- name: min_trg_freq
dtype: float64
- name: avg_src_freq_tokenised
dtype: float64
- name: avg_trg_freq_tokenised
dtype: float64
- name: min_src_freq_tokenised
dtype: float64
- name: min_trg_freq_tokenised
dtype: float64
- name: trg_repeated
dtype: float64
- name: unaligned_fwd
dtype: float64
- name: unaligned_rev
dtype: float64
- name: reordering
dtype: float64
- name: segmentation_src
dtype: float64
- name: segmentation_trg
dtype: float64
splits:
- name: train
num_bytes: 557267936
num_examples: 1000000
download_size: 312382809
dataset_size: 557267936
- config_name: eng-fra
features:
- name: idx
dtype: uint64
- name: source
dtype: large_string
- name: target
dtype: large_string
- name: train_likelihood_ref
dtype: float64
- name: test_likelihood_ref
dtype: float64
- name: counterfactual_memorisation_ref
dtype: float64
- name: train_likelihood_hyp
dtype: float64
- name: test_likelihood_hyp
dtype: float64
- name: counterfactual_memorisation_hyp
dtype: float64
- name: src_length
dtype: float64
- name: trg_length
dtype: float64
- name: src_length_tokenised
dtype: float64
- name: trg_length_tokenised
dtype: float64
- name: length_difference_tokenised
dtype: float64
- name: length_ratio_src_trg_tokenised
dtype: float64
- name: length_difference
dtype: float64
- name: length_ratio_src_trg
dtype: float64
- name: backtranslation_ed
dtype: float64
- name: edit_distance_src_trg
dtype: float64
- name: word_overlap
dtype: float64
- name: token_overlap
dtype: float64
- name: digit_ratio
dtype: float64
- name: punctuation_ratio
dtype: float64
- name: avg_src_freq
dtype: float64
- name: avg_trg_freq
dtype: float64
- name: min_src_freq
dtype: float64
- name: min_trg_freq
dtype: float64
- name: avg_src_freq_tokenised
dtype: float64
- name: avg_trg_freq_tokenised
dtype: float64
- name: min_src_freq_tokenised
dtype: float64
- name: min_trg_freq_tokenised
dtype: float64
- name: trg_repeated
dtype: float64
- name: unaligned_fwd
dtype: float64
- name: unaligned_rev
dtype: float64
- name: reordering
dtype: float64
- name: segmentation_src
dtype: float64
- name: segmentation_trg
dtype: float64
splits:
- name: train
num_bytes: 559656387
num_examples: 1000000
download_size: 314210551
dataset_size: 559656387
- config_name: eng-ita
features:
- name: idx
dtype: uint64
- name: source
dtype: large_string
- name: target
dtype: large_string
- name: train_likelihood_ref
dtype: float64
- name: test_likelihood_ref
dtype: float64
- name: counterfactual_memorisation_ref
dtype: float64
- name: train_likelihood_hyp
dtype: float64
- name: test_likelihood_hyp
dtype: float64
- name: counterfactual_memorisation_hyp
dtype: float64
- name: src_length
dtype: float64
- name: trg_length
dtype: float64
- name: src_length_tokenised
dtype: float64
- name: trg_length_tokenised
dtype: float64
- name: length_difference_tokenised
dtype: float64
- name: length_ratio_src_trg_tokenised
dtype: float64
- name: length_difference
dtype: float64
- name: length_ratio_src_trg
dtype: float64
- name: backtranslation_ed
dtype: float64
- name: edit_distance_src_trg
dtype: float64
- name: word_overlap
dtype: float64
- name: token_overlap
dtype: float64
- name: digit_ratio
dtype: float64
- name: punctuation_ratio
dtype: float64
- name: avg_src_freq
dtype: float64
- name: avg_trg_freq
dtype: float64
- name: min_src_freq
dtype: float64
- name: min_trg_freq
dtype: float64
- name: avg_src_freq_tokenised
dtype: float64
- name: avg_trg_freq_tokenised
dtype: float64
- name: min_src_freq_tokenised
dtype: float64
- name: min_trg_freq_tokenised
dtype: float64
- name: trg_repeated
dtype: float64
- name: unaligned_fwd
dtype: float64
- name: unaligned_rev
dtype: float64
- name: reordering
dtype: float64
- name: segmentation_src
dtype: float64
- name: segmentation_trg
dtype: float64
splits:
- name: train
num_bytes: 551390517
num_examples: 1000000
download_size: 309540898
dataset_size: 551390517
- config_name: eng-nld
features:
- name: idx
dtype: uint64
- name: source
dtype: large_string
- name: target
dtype: large_string
- name: train_likelihood_ref
dtype: float64
- name: test_likelihood_ref
dtype: float64
- name: counterfactual_memorisation_ref
dtype: float64
- name: train_likelihood_hyp
dtype: float64
- name: test_likelihood_hyp
dtype: float64
- name: counterfactual_memorisation_hyp
dtype: float64
- name: src_length
dtype: float64
- name: trg_length
dtype: float64
- name: src_length_tokenised
dtype: float64
- name: trg_length_tokenised
dtype: float64
- name: length_difference_tokenised
dtype: float64
- name: length_ratio_src_trg_tokenised
dtype: float64
- name: length_difference
dtype: float64
- name: length_ratio_src_trg
dtype: float64
- name: backtranslation_ed
dtype: float64
- name: edit_distance_src_trg
dtype: float64
- name: word_overlap
dtype: float64
- name: token_overlap
dtype: float64
- name: digit_ratio
dtype: float64
- name: punctuation_ratio
dtype: float64
- name: avg_src_freq
dtype: float64
- name: avg_trg_freq
dtype: float64
- name: min_src_freq
dtype: float64
- name: min_trg_freq
dtype: float64
- name: avg_src_freq_tokenised
dtype: float64
- name: avg_trg_freq_tokenised
dtype: float64
- name: min_src_freq_tokenised
dtype: float64
- name: min_trg_freq_tokenised
dtype: float64
- name: trg_repeated
dtype: float64
- name: unaligned_fwd
dtype: float64
- name: unaligned_rev
dtype: float64
- name: reordering
dtype: float64
- name: segmentation_src
dtype: float64
- name: segmentation_trg
dtype: float64
splits:
- name: train
num_bytes: 551862036
num_examples: 1000000
download_size: 307377496
dataset_size: 551862036
- config_name: eng-spa
features:
- name: idx
dtype: uint64
- name: source
dtype: large_string
- name: target
dtype: large_string
- name: train_likelihood_ref
dtype: float64
- name: test_likelihood_ref
dtype: float64
- name: counterfactual_memorisation_ref
dtype: float64
- name: train_likelihood_hyp
dtype: float64
- name: test_likelihood_hyp
dtype: float64
- name: counterfactual_memorisation_hyp
dtype: float64
- name: src_length
dtype: float64
- name: trg_length
dtype: float64
- name: src_length_tokenised
dtype: float64
- name: trg_length_tokenised
dtype: float64
- name: length_difference_tokenised
dtype: float64
- name: length_ratio_src_trg_tokenised
dtype: float64
- name: length_difference
dtype: float64
- name: length_ratio_src_trg
dtype: float64
- name: backtranslation_ed
dtype: float64
- name: edit_distance_src_trg
dtype: float64
- name: word_overlap
dtype: float64
- name: token_overlap
dtype: float64
- name: digit_ratio
dtype: float64
- name: punctuation_ratio
dtype: float64
- name: avg_src_freq
dtype: float64
- name: avg_trg_freq
dtype: float64
- name: min_src_freq
dtype: float64
- name: min_trg_freq
dtype: float64
- name: avg_src_freq_tokenised
dtype: float64
- name: avg_trg_freq_tokenised
dtype: float64
- name: min_src_freq_tokenised
dtype: float64
- name: min_trg_freq_tokenised
dtype: float64
- name: trg_repeated
dtype: float64
- name: unaligned_fwd
dtype: float64
- name: unaligned_rev
dtype: float64
- name: reordering
dtype: float64
- name: segmentation_src
dtype: float64
- name: segmentation_trg
dtype: float64
splits:
- name: train
num_bytes: 553589612
num_examples: 1000000
download_size: 310717691
dataset_size: 553589612
configs:
- config_name: eng-deu
data_files:
- split: train
path: eng-deu/train-*
- config_name: eng-fra
data_files:
- split: train
path: eng-fra/train-*
- config_name: eng-ita
data_files:
- split: train
path: eng-ita/train-*
- config_name: eng-nld
data_files:
- split: train
path: eng-nld/train-*
- config_name: eng-spa
data_files:
- split: train
path: eng-spa/train-*