gsaltintas commited on
Commit
ab8f750
·
verified ·
1 Parent(s): 5b936bd

Uploading tokenizer_robustness_completion_stem_spelled_out subset

Browse files
README.md CHANGED
@@ -1620,6 +1620,130 @@ dataset_info:
1620
  num_examples: 41
1621
  download_size: 40316
1622
  dataset_size: 22025
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1623
  configs:
1624
  - config_name: tokenizer_robustness_completion_stem_canonical
1625
  data_files:
@@ -1673,6 +1797,10 @@ configs:
1673
  data_files:
1674
  - split: test
1675
  path: tokenizer_robustness_completion_stem_space_removal/test-*
 
 
 
 
1676
  ---
1677
 
1678
  # Dataset Card for Tokenization Robustness
 
1620
  num_examples: 41
1621
  download_size: 40316
1622
  dataset_size: 22025
1623
+ - config_name: tokenizer_robustness_completion_stem_spelled_out
1624
+ features:
1625
+ - name: question
1626
+ dtype: string
1627
+ - name: choices
1628
+ list: string
1629
+ - name: answer
1630
+ dtype: int64
1631
+ - name: answer_label
1632
+ dtype: string
1633
+ - name: split
1634
+ dtype: string
1635
+ - name: subcategories
1636
+ dtype: string
1637
+ - name: lang
1638
+ dtype: string
1639
+ - name: second_lang
1640
+ dtype: string
1641
+ - name: notes
1642
+ dtype: string
1643
+ - name: id
1644
+ dtype: string
1645
+ - name: set_id
1646
+ dtype: string
1647
+ - name: variation_id
1648
+ dtype: string
1649
+ - name: question_general_category
1650
+ dtype: string
1651
+ - name: vanilla_cos_sim_to_canonical
1652
+ struct:
1653
+ - name: CohereLabs/aya-expanse-8b
1654
+ dtype: float64
1655
+ - name: Qwen/Qwen3-8B
1656
+ dtype: float64
1657
+ - name: bigscience/bloom
1658
+ dtype: float64
1659
+ - name: common-pile/comma-v0.1-1t
1660
+ dtype: float64
1661
+ - name: facebook/xglm-564M
1662
+ dtype: float64
1663
+ - name: google-bert/bert-base-multilingual-cased
1664
+ dtype: float64
1665
+ - name: google/byt5-small
1666
+ dtype: float64
1667
+ - name: google/gemma-2-2b
1668
+ dtype: float64
1669
+ - name: gpt2
1670
+ dtype: float64
1671
+ - name: meta-llama/Llama-3.2-1B
1672
+ dtype: float64
1673
+ - name: microsoft/Phi-3-mini-4k-instruct
1674
+ dtype: float64
1675
+ - name: mistralai/tekken
1676
+ dtype: float64
1677
+ - name: tiktoken/gpt-4o
1678
+ dtype: float64
1679
+ - name: tokenmonster/englishcode-32000-consistent-v1
1680
+ dtype: float64
1681
+ - name: trimmed_cos_sim_to_canonical
1682
+ struct:
1683
+ - name: CohereLabs/aya-expanse-8b
1684
+ dtype: float64
1685
+ - name: Qwen/Qwen3-8B
1686
+ dtype: float64
1687
+ - name: bigscience/bloom
1688
+ dtype: float64
1689
+ - name: common-pile/comma-v0.1-1t
1690
+ dtype: float64
1691
+ - name: facebook/xglm-564M
1692
+ dtype: float64
1693
+ - name: google-bert/bert-base-multilingual-cased
1694
+ dtype: float64
1695
+ - name: google/byt5-small
1696
+ dtype: float64
1697
+ - name: google/gemma-2-2b
1698
+ dtype: float64
1699
+ - name: gpt2
1700
+ dtype: float64
1701
+ - name: meta-llama/Llama-3.2-1B
1702
+ dtype: float64
1703
+ - name: microsoft/Phi-3-mini-4k-instruct
1704
+ dtype: float64
1705
+ - name: mistralai/tekken
1706
+ dtype: float64
1707
+ - name: tiktoken/gpt-4o
1708
+ dtype: float64
1709
+ - name: tokenmonster/englishcode-32000-consistent-v1
1710
+ dtype: float64
1711
+ - name: token_counts
1712
+ struct:
1713
+ - name: CohereLabs/aya-expanse-8b
1714
+ dtype: int64
1715
+ - name: Qwen/Qwen3-8B
1716
+ dtype: int64
1717
+ - name: bigscience/bloom
1718
+ dtype: int64
1719
+ - name: common-pile/comma-v0.1-1t
1720
+ dtype: int64
1721
+ - name: facebook/xglm-564M
1722
+ dtype: int64
1723
+ - name: google-bert/bert-base-multilingual-cased
1724
+ dtype: int64
1725
+ - name: google/byt5-small
1726
+ dtype: int64
1727
+ - name: google/gemma-2-2b
1728
+ dtype: int64
1729
+ - name: gpt2
1730
+ dtype: int64
1731
+ - name: meta-llama/Llama-3.2-1B
1732
+ dtype: int64
1733
+ - name: microsoft/Phi-3-mini-4k-instruct
1734
+ dtype: int64
1735
+ - name: mistralai/tekken
1736
+ dtype: int64
1737
+ - name: tiktoken/gpt-4o
1738
+ dtype: int64
1739
+ - name: tokenmonster/englishcode-32000-consistent-v1
1740
+ dtype: int64
1741
+ splits:
1742
+ - name: test
1743
+ num_bytes: 13749
1744
+ num_examples: 25
1745
+ download_size: 35840
1746
+ dataset_size: 13749
1747
  configs:
1748
  - config_name: tokenizer_robustness_completion_stem_canonical
1749
  data_files:
 
1797
  data_files:
1798
  - split: test
1799
  path: tokenizer_robustness_completion_stem_space_removal/test-*
1800
+ - config_name: tokenizer_robustness_completion_stem_spelled_out
1801
+ data_files:
1802
+ - split: test
1803
+ path: tokenizer_robustness_completion_stem_spelled_out/test-*
1804
  ---
1805
 
1806
  # Dataset Card for Tokenization Robustness
tokenizer_robustness_completion_stem_spelled_out/test-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31194013b776adc3ad4cd2c27d2939b8f2a39667b55e24dcd27c6898c204863e
3
+ size 35840