def transformer(model='base', **kwargs):
"""
Load Malaya transformer encoder-decoder model to generate a paraphrase given a string.
Parameters
----------
model : str, optional (default='base')
Model architecture supported. Allowed values:
* ``'malaya-small'`` - Malaya Transformer SMALL parameters.
* ``'malaya-base'`` - Malaya Transformer BASE parameters.
Returns
-------
result: malaya.model.tf.PARAPHRASE class
"""
model = model.lower()
if model not in _transformer_availability:
raise ValueError(
'model not supported, please check supported models from malaya.paraphrase.available_transformer()'
)
from malaya.model.tf import PARAPHRASE
return transformer_load.load_lm(
path=PATH_PARAPHRASE['transformer'],
s3_path=S3_PATH_PARAPHRASE['transformer'],
model=model,
model_class=PARAPHRASE,
**kwargs,
)
def transformer(model: str = 't2t', quantized: bool = False, **kwargs):
"""
Load Malaya transformer encoder-decoder model to generate a paraphrase given a string.
Parameters
----------
model : str, optional (default='t2t')
Model architecture supported. Allowed values:
* ``'t2t'`` - Malaya Transformer BASE parameters.
* ``'small-t2t'`` - Malaya Transformer SMALL parameters.
* ``'t5'`` - T5 BASE parameters.
* ``'small-t5'`` - T5 SMALL parameters.
quantized : bool, optional (default=False)
if True, will load 8-bit quantized model.
Quantized model not necessary faster, totally depends on the machine.
Returns
-------
result: model
List of model classes:
* if `t2t` in model, will return `malaya.model.tf.Paraphrase`.
* if `t5` in model, will return `malaya.model.t5.Paraphrase`.
"""
model = model.lower()
if model not in _transformer_availability:
raise ValueError(
'model not supported, please check supported models from `malaya.paraphrase.available_transformer()`.'
)
if 't2t' in model:
return transformer_load.load_lm(
module='paraphrase',
model=model,
model_class=TF_Paraphrase,
quantized=quantized,
**kwargs,
)
if 't5' in model:
return t5_load.load(
module='paraphrase',
model=model,
model_class=T5_Paraphrase,
quantized=quantized,
**kwargs,
)
def transformer(model: str = 'base', **kwargs):
model = model.lower()
if model not in _transformer_availability:
raise ValueError(
'model not supported, please check supported models from malaya.summarization.abstractive.available_transformer()'
)
from malaya.model.tf import SUMMARIZATION
return transformer_load.load_lm(
path = PATH_SUMMARIZE['transformer'],
s3_path = S3_PATH_SUMMARIZE['transformer'],
model = model,
model_class = SUMMARIZATION,
**kwargs,
)
def transformer(model: str = 'base', quantized: bool = False, **kwargs):
"""
Load Malaya transformer encoder-decoder model to generate a paraphrase given a string.
Parameters
----------
model : str, optional (default='base')
Model architecture supported. Allowed values:
* ``'small'`` - Malaya Transformer SMALL parameters.
* ``'base'`` - Malaya Transformer BASE parameters.
quantized : bool, optional (default=False)
if True, will load 8-bit quantized model.
Quantized model not necessary faster, totally depends on the machine.
Returns
-------
result: malaya.model.tf.PARAPHRASE class
"""
model = model.lower()
if model not in _transformer_availability:
raise ValueError(
'model not supported, please check supported models from `malaya.paraphrase.available_transformer()`.'
)
from malaya.model.tf import PARAPHRASE
return transformer_load.load_lm(
path=PATH_PARAPHRASE['transformer'],
s3_path=S3_PATH_PARAPHRASE['transformer'],
model=model,
model_class=PARAPHRASE,
quantized=quantized,
**kwargs,
)
def transformer(model: str = 't2t', quantized: bool = False, **kwargs):
"""
Load Malaya transformer encoder-decoder model to generate a summary given a string.
Parameters
----------
model : str, optional (default='base')
Model architecture supported. Allowed values:
* ``'t2t'`` - Malaya Transformer BASE parameters.
* ``'small-t2t'`` - Malaya Transformer SMALL parameters.
* ``'t5'`` - T5 BASE parameters.
* ``'small-t5'`` - T5 SMALL parameters.
* ``'bigbird'`` - BigBird + Pegasus BASE parameters.
* ``'small-bigbird'`` - BigBird + Pegasus SMALL parameters.
quantized : bool, optional (default=False)
if True, will load 8-bit quantized model.
Quantized model not necessary faster, totally depends on the machine.
Returns
-------
result: model
List of model classes:
* if `t2t` in model, will return `malaya.model.tf.Summarization`.
* if `t5` in model, will return `malaya.model.t5.Summarization`.
* if `bigbird` in model, will return `malaya.model.bigbird.Summarization`.
"""
model = model.lower()
if model not in _transformer_availability:
raise ValueError(
'model not supported, please check supported models from `malaya.summarization.abstractive.available_transformer()`.'
)
if 't2t' in model:
return transformer_load.load_lm(
module='abstractive-summarization',
model=model,
model_class=TF_Summarization,
quantized=quantized,
**kwargs,
)
if 't5' in model:
return t5_load.load(
module='abstractive-summarization',
model=model,
model_class=T5_Summarization,
quantized=quantized,
**kwargs,
)
if 'bigbird' in model:
return bigbird_load.load_lm(
module='abstractive-summarization',
model=model,
model_class=BigBird_Summarization,
maxlen=_transformer_availability[model]['Suggested length'],
quantized=quantized,
**kwargs,
)