mirror of
https://github.com/leejet/stable-diffusion.cpp.git
synced 2026-11-12 11:56:36 +00:00
fix: use tokenizer-specific pre-tokenization rules (#1975)
This commit is contained in:
1 parent
07a85c74cb
commit
59c23bce0d
9 files changed
+92
-1021
No files matched your search
@@ -87,7 +87,8 @@ Qwen2Tokenizer::Qwen2Tokenizer(const std::string& merges_utf8_str)
|
||||
}
|
||||
|
||||
Qwen2Tokenizer::Qwen2Tokenizer(const std::string& merges_utf8_str,
|
||||
const std::vector<std::string>& special_tokens_override) {
|
||||
const std::vector<std::string>& special_tokens_override)
|
||||
: BPETokenizer(R"((?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+)") {
|
||||
UNK_TOKEN = "<|endoftext|>";
|
||||
EOS_TOKEN = "<|endoftext|>";
|
||||
PAD_TOKEN = "<|endoftext|>";
|
||||
|
||||
Reference in new issue
Block a user