MCPcopy Create free account
hub / github.com/AI-Hypercomputer/maxtext / get_tokenizer

Method get_tokenizer

src/MaxText/maxengine.py:1489–1501  ·  view source on GitHub ↗

Return a protobuf of tokenizer info, callable from Py or C++.

(self)

Source from the content-addressed store, hash-verified

1487 }
1488
1489 def get_tokenizer(self) -> TokenizerParameters:
1490 """Return a protobuf of tokenizer info, callable from Py or C++."""
1491 try:
1492 tokenizer_type_val = TokenizerType.DESCRIPTOR.values_by_name[self.config.tokenizer_type].number
1493 return TokenizerParameters(
1494 path=self.config.tokenizer_path,
1495 tokenizer_type=tokenizer_type_val,
1496 access_token=self.config.hf_access_token,
1497 use_chat_template=self.config.use_chat_template,
1498 extra_ids=0,
1499 )
1500 except KeyError as _:
1501 raise KeyError(f"Unsupported tokenizer type: {self.config.tokenizer_type}") from None
1502
1503 def build_tokenizer(self, metadata: TokenizerParameters) -> tokenizer_api.Tokenizer:
1504 """Return a tokenizer"""

Callers 13

prepare_settingFunction · 0.95
mainFunction · 0.95
run_benchmarksFunction · 0.95
mainFunction · 0.95
mainFunction · 0.95
mainFunction · 0.95
__init__Method · 0.80
_init_tokenizerMethod · 0.80
preprocessing_pipelineFunction · 0.80
setUpClassMethod · 0.80
setUpClassMethod · 0.80
setUpClassMethod · 0.80

Calls 1

TokenizerParametersClass · 0.85

Tested by 3

setUpClassMethod · 0.64
setUpClassMethod · 0.64
setUpClassMethod · 0.64