Skip to content

Releases: bnosac/tokenizers.bpe

CRAN Releast 0.1.6

13 Jun 21:54

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.6

  • Update parallel-hashmap to commit 48f4c5fb0519e41233f000340039a1a9dd00a1f0 as it fixes the same as what tokenizers.bpe 0.1.5 fixed but now upstream

0.1.5

11 Jun 07:16

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.5

  • phmap_base.h: replace
    using GetIsAlwaysEqual = typename T::is_always_equal;
    with GetIsAlwaysEqual = typename std::allocator_traits::is_always_equal;
    to avoid warning: 'is_always_equal' is deprecated on Debian clang version 21.1.8

CRAN Release 0.1.4

17 May 12:50
831fcab

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.4

  • Update parallel-hashmap to commit 88123934b46b77c3b6d80167382734cbff6eff74 to fix clang 21.1.0 compiler warnings 'pointer' and 'const_pointer' is deprecated

CRAN Release 0.1.3

16 Sep 05:43

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.3

  • Update parallel-hashmap to 1.3.11 which fixes C++17 deprecations about rebind, construct, destroy

CRAN Release 0.1.2

15 Sep 07:45

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.2

  • Drop C++11 specification in Makevars

CRAN Release 0.1.1

06 Jan 10:38

Choose a tag to compare

CHANGES IN tokenizers.bpe VERSION 0.1.1

  • replace move with std::move to fix R CMD check warning on recent versions of clang compilers

CRAN Release 0.1.0

05 Aug 11:26

Choose a tag to compare

  • Build a BPE model

  • Tokenize using the BPE model / Reverse the tokenisation (bpe_decode / bpe_encode)

  • Load a model from disk

  • Based on YouTokenToMe commit 16fcf1d8e208f9bbbc9f522e83e6ff81708dba73
    Replaced third_party/flat_hash_map.h with https://github.com/greg7mdp/parallel-hashmap commit c4abc895b4e51cd5e0a96656bfb925edeae5ecbf