vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add BPE models for testing

Browse source
This commit is contained in:
jaime-m-p 2024-05-25 00:19:31 +02:00
parent e013b23102
commit 55e387b2d5
1 changed files with 10 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

12
tests/test-tokenizer-random.py
View file

@ -156,6 +156,7 @@ def generator_custom_text_edge_cases() -> Iterator[str]:
'<s>a', # Phi-3 fail '<s>a', # Phi-3 fail
'<unk><|endoftext|><s>', # Phi-3 fail '<unk><|endoftext|><s>', # Phi-3 fail
'a\na', # TODO: Bert fail 'a\na', # TODO: Bert fail
'a\xa0\xa0\x00b', # jina-v2-es
] ]
@ -328,8 +329,15 @@ if __name__ == "__main__":
# import os # import os
# tokenizers = os.listdir(path_tokenizers) # tokenizers = os.listdir(path_tokenizers)
tokenizers = [ tokenizers = [
"llama-spm", # SPM # "llama-spm", # SPM
"phi-3", # SPM # "phi-3", # SPM
# "bert-bge", # WPM
# "jina-v2-en", # WPM
"gpt-2", # BPE
"llama-bpe", # BPE
"jina-v2-es", # BPE
"jina-v2-de", # BPE
"phi-2", # BPE
] ]
for tokenizer in tokenizers: for tokenizer in tokenizers: