gsaltintas commited on
Commit
5a62272
·
verified ·
1 Parent(s): 22cb4f0

Uploading tokenizer_robustness_completion_italian_english_keyboard subset

Browse files
README.md CHANGED
@@ -890,6 +890,132 @@ dataset_info:
890
  num_examples: 31
891
  download_size: 35120
892
  dataset_size: 16359
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
893
  configs:
894
  - config_name: tokenizer_robustness_completion_italian_abbreviations
895
  data_files:
@@ -919,6 +1045,10 @@ configs:
919
  data_files:
920
  - split: test
921
  path: tokenizer_robustness_completion_italian_dialects/test-*
 
 
 
 
922
  ---
923
 
924
  # Dataset Card for Tokenization Robustness
 
890
  num_examples: 31
891
  download_size: 35120
892
  dataset_size: 16359
893
+ - config_name: tokenizer_robustness_completion_italian_english_keyboard
894
+ features:
895
+ - name: question
896
+ dtype: string
897
+ - name: choices
898
+ list: string
899
+ - name: answer
900
+ dtype: int64
901
+ - name: answer_label
902
+ dtype: string
903
+ - name: split
904
+ dtype: string
905
+ - name: subcategories
906
+ dtype: string
907
+ - name: category
908
+ dtype: string
909
+ - name: lang
910
+ dtype: string
911
+ - name: second_lang
912
+ dtype: string
913
+ - name: notes
914
+ dtype: string
915
+ - name: id
916
+ dtype: string
917
+ - name: set_id
918
+ dtype: string
919
+ - name: variation_id
920
+ dtype: string
921
+ - name: perturbed_word
922
+ dtype: string
923
+ - name: vanilla_cos_sim_to_canonical
924
+ struct:
925
+ - name: CohereLabs/aya-expanse-8b
926
+ dtype: float64
927
+ - name: Qwen/Qwen3-8B
928
+ dtype: float64
929
+ - name: bigscience/bloom
930
+ dtype: float64
931
+ - name: common-pile/comma-v0.1-1t
932
+ dtype: float64
933
+ - name: facebook/xglm-564M
934
+ dtype: float64
935
+ - name: google-bert/bert-base-multilingual-cased
936
+ dtype: float64
937
+ - name: google/byt5-small
938
+ dtype: float64
939
+ - name: google/gemma-2-2b
940
+ dtype: float64
941
+ - name: gpt2
942
+ dtype: float64
943
+ - name: meta-llama/Llama-3.2-1B
944
+ dtype: float64
945
+ - name: microsoft/Phi-3-mini-4k-instruct
946
+ dtype: float64
947
+ - name: mistralai/tekken
948
+ dtype: float64
949
+ - name: tiktoken/gpt-4o
950
+ dtype: float64
951
+ - name: tokenmonster/englishcode-32000-consistent-v1
952
+ dtype: float64
953
+ - name: trimmed_cos_sim_to_canonical
954
+ struct:
955
+ - name: CohereLabs/aya-expanse-8b
956
+ dtype: float64
957
+ - name: Qwen/Qwen3-8B
958
+ dtype: float64
959
+ - name: bigscience/bloom
960
+ dtype: float64
961
+ - name: common-pile/comma-v0.1-1t
962
+ dtype: float64
963
+ - name: facebook/xglm-564M
964
+ dtype: float64
965
+ - name: google-bert/bert-base-multilingual-cased
966
+ dtype: float64
967
+ - name: google/byt5-small
968
+ dtype: float64
969
+ - name: google/gemma-2-2b
970
+ dtype: float64
971
+ - name: gpt2
972
+ dtype: float64
973
+ - name: meta-llama/Llama-3.2-1B
974
+ dtype: float64
975
+ - name: microsoft/Phi-3-mini-4k-instruct
976
+ dtype: float64
977
+ - name: mistralai/tekken
978
+ dtype: float64
979
+ - name: tiktoken/gpt-4o
980
+ dtype: float64
981
+ - name: tokenmonster/englishcode-32000-consistent-v1
982
+ dtype: float64
983
+ - name: token_counts
984
+ struct:
985
+ - name: CohereLabs/aya-expanse-8b
986
+ dtype: int64
987
+ - name: Qwen/Qwen3-8B
988
+ dtype: int64
989
+ - name: bigscience/bloom
990
+ dtype: int64
991
+ - name: common-pile/comma-v0.1-1t
992
+ dtype: int64
993
+ - name: facebook/xglm-564M
994
+ dtype: int64
995
+ - name: google-bert/bert-base-multilingual-cased
996
+ dtype: int64
997
+ - name: google/byt5-small
998
+ dtype: int64
999
+ - name: google/gemma-2-2b
1000
+ dtype: int64
1001
+ - name: gpt2
1002
+ dtype: int64
1003
+ - name: meta-llama/Llama-3.2-1B
1004
+ dtype: int64
1005
+ - name: microsoft/Phi-3-mini-4k-instruct
1006
+ dtype: int64
1007
+ - name: mistralai/tekken
1008
+ dtype: int64
1009
+ - name: tiktoken/gpt-4o
1010
+ dtype: int64
1011
+ - name: tokenmonster/englishcode-32000-consistent-v1
1012
+ dtype: int64
1013
+ splits:
1014
+ - name: test
1015
+ num_bytes: 36901
1016
+ num_examples: 68
1017
+ download_size: 41589
1018
+ dataset_size: 36901
1019
  configs:
1020
  - config_name: tokenizer_robustness_completion_italian_abbreviations
1021
  data_files:
 
1045
  data_files:
1046
  - split: test
1047
  path: tokenizer_robustness_completion_italian_dialects/test-*
1048
+ - config_name: tokenizer_robustness_completion_italian_english_keyboard
1049
+ data_files:
1050
+ - split: test
1051
+ path: tokenizer_robustness_completion_italian_english_keyboard/test-*
1052
  ---
1053
 
1054
  # Dataset Card for Tokenization Robustness
tokenizer_robustness_completion_italian_english_keyboard/test-00000-of-00001.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cafc061990a6c53fe1630aac33f24bdf1cb1e26a9a544509027804d3f63efeae
3
- size 41603
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50415a551ff5068d6d5ae15e105994e05af1f06f3ead1496c65d1b8fe809595c
3
+ size 41589