diff options
| author | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2024-07-27 07:55:01 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-07-27 07:55:01 +0200 |
| commit | 154e0d75fccf1784fe9ff6fd76a630b66563da3d (patch) | |
| tree | 81ce6dbb5b1900c1aa78a879f0593c694cab9d27 /src/unicode-data.h | |
| parent | 0684c3e9c70d49323b4fc517128cbe222cab7f96 (diff) | |
Merge mainline llama.cpp (#3)
* Merging mainline - WIP
* Merging mainline - WIP
AVX2 and CUDA appear to work.
CUDA performance seems slightly (~1-2%) lower as it is so often
the case with llama.cpp/ggml after some "improvements" have been made.
* Merging mainline - fix Metal
* Remove check
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'src/unicode-data.h')
| -rw-r--r-- | src/unicode-data.h | 20 |
1 files changed, 20 insertions, 0 deletions
diff --git a/src/unicode-data.h b/src/unicode-data.h new file mode 100644 index 00000000..e27fe177 --- /dev/null +++ b/src/unicode-data.h @@ -0,0 +1,20 @@ +#pragma once + +#include <cstdint> +#include <vector> +#include <unordered_map> +#include <unordered_set> + +struct range_nfd { + uint32_t first; + uint32_t last; + uint32_t nfd; +}; + +static const uint32_t MAX_CODEPOINTS = 0x110000; + +extern const std::vector<std::pair<uint32_t, uint16_t>> unicode_ranges_flags; +extern const std::unordered_set<uint32_t> unicode_set_whitespace; +extern const std::unordered_map<uint32_t, uint32_t> unicode_map_lowercase; +extern const std::unordered_map<uint32_t, uint32_t> unicode_map_uppercase; +extern const std::vector<range_nfd> unicode_ranges_nfd; |