mirror of
https://github.com/wassname/multifit.git
synced 2026-10-09 12:20:23 +08:00
33 lines
1.5 KiB
Python
33 lines
1.5 KiB
Python
from fastai.callbacks import *
|
|
from fastai.basic_data import *
|
|
from fastai.datasets import untar_data
|
|
from fastai_contrib.models import get_bilm
|
|
from fastai.text.learner import *
|
|
|
|
|
|
def bilm_learner(data:DataBunch, bptt:int=70, emb_sz:int=400, nh:int=1150, nl:int=3, pad_token:int=1,
|
|
drop_mult:float=1., tie_weights:bool=True, bias:bool=True, qrnn:bool=False, pretrained_model=None,
|
|
pretrained_fnames:OptStrTuple=None, **kwargs) -> 'LanguageLearner':
|
|
"Create a `Learner` with a language model."
|
|
dps = default_dropout['language'] * drop_mult
|
|
vocab_size = len(data.vocab.itos)
|
|
model = get_bilm(vocab_size, emb_sz, nh, nl, pad_token, input_p=dps[0], output_p=dps[1],
|
|
weight_p=dps[2], embed_p=dps[3], hidden_p=dps[4], tie_weights=tie_weights, bias=bias, qrnn=qrnn)
|
|
learn = LanguageLearner(data, model, bptt, split_func=bilm_split, **kwargs)
|
|
if pretrained_model is not None:
|
|
model_path = untar_data(pretrained_model, data=False)
|
|
fnames = [list(model_path.glob(f'*.{ext}'))[0] for ext in ['pth', 'pkl']]
|
|
learn.load_pretrained(*fnames)
|
|
learn.freeze()
|
|
if pretrained_fnames is not None:
|
|
fnames = [learn.path/learn.model_dir/f'{fn}.{ext}' for fn,ext in zip(pretrained_fnames, ['pth', 'pkl'])]
|
|
learn.load_pretrained(*fnames)
|
|
learn.freeze()
|
|
return learn
|
|
|
|
|
|
def bilm_split(model:nn.Module) -> List[nn.Module]:
|
|
"Split a RNN `model` in groups for differential learning rates."
|
|
|
|
return [f+b for f,b in zip(lm_split(model.fwd_lm),lm_split(model.bwd_lm))]
|