SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

[CodeLlamaTokenizer] Nit, update __init__ to make sure the AddedTokens are not normalized because they are special (#27359)

Browse Source 
* make sure tokens are properly initialized for codellama slow

* add m ore pretrained models

* style

* test more tokenizers checkpoints
This commit is contained in:
Arthur
2023-11-09 10:15:10 +01:00
committed by GitHub
parent 7ecd229ba4
commit 085ea7e56c
2 changed files with 5 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/transformers/models/code_llama/tokenization_code_llama.py
View File

@@ -149,9 +149,9 @@ class CodeLlamaTokenizer(PreTrainedTokenizer):
): ):
requires_backends(self, "protobuf") requires_backends(self, "protobuf")
self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
bos_token = AddedToken(bos_token, lstrip=False, rstrip=False) if isinstance(bos_token, str) else bos_token bos_token = AddedToken(bos_token, normalized=False, special=True) if isinstance(bos_token, str) else bos_token
eos_token = AddedToken(eos_token, lstrip=False, rstrip=False) if isinstance(eos_token, str) else eos_token eos_token = AddedToken(eos_token, normalized=False, special=True) if isinstance(eos_token, str) else eos_token
unk_token = AddedToken(unk_token, lstrip=False, rstrip=False) if isinstance(unk_token, str) else unk_token unk_token = AddedToken(unk_token, normalized=False, special=True) if isinstance(unk_token, str) else unk_token
self.use_default_system_prompt = use_default_system_prompt self.use_default_system_prompt = use_default_system_prompt
# mark tokens special to skip them # mark tokens special to skip them

2
tests/models/code_llama/test_tokenization_code_llama.py
View File

@@ -150,6 +150,8 @@ class CodeLlamaTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
self.tokenizers_list = [ self.tokenizers_list = [
(self.rust_tokenizer_class, "hf-internal-testing/llama-code-tokenizer", {}), (self.rust_tokenizer_class, "hf-internal-testing/llama-code-tokenizer", {}),
(self.tokenizer_class, "hf-internal-testing/llama-code-tokenizer", {}), (self.tokenizer_class, "hf-internal-testing/llama-code-tokenizer", {}),
(self.tokenizer_class, "codellama/CodeLlama-34b-Instruct-hf", {}),
(self.rust_tokenizer_class, "codellama/CodeLlama-34b-Instruct-hf", {}),
] ]
for tokenizer, pretrained_name, kwargs in self.tokenizers_list: for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"): with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):