wassname/multifit
1
0
Fork 0
mirror of https://github.com/wassname/multifit.git synced 2026-08-31 12:10:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

Use drop_mult properly

Browse Source
This commit is contained in:
Piotr Czapla
2018-11-16 20:02:32 +01:00
parent 23b17da61e
commit 45389be413
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ulmfit/pretrain_lm.py
+2 -2
View File
@@ -104,7 +104,7 @@ def pretrain_lm(dir_path, lang='en', cuda_id=0, qrnn=True, clean=True, max_vocab
if qrnn:
emb_sz, nh, nl = 400, 1550, 3
#dps = np.array([0.0, 0.0, 0.0, 0.0, 0.0])
dps = np.array([0.25, 0.1, 0.2, 0.02, 0.15]) * 0.1
dps = np.array([0.25, 0.1, 0.2, 0.02, 0.15])
drop_mult = 0.1
else:
emb_sz, nh, nl = 400, 1150, 3
@@ -112,7 +112,7 @@ def pretrain_lm(dir_path, lang='en', cuda_id=0, qrnn=True, clean=True, max_vocab
dps = np.array([0.25, 0.1, 0.2, 0.02, 0.15])
drop_mult = 0.1
fastai.text.learner.default_dropout['language'] = dps
fastai.text.learner.default_dropout['language'] = dps * drop_mult
learn = language_model_learner(data_lm, bptt=bptt, emb_sz=emb_sz, nh=nh, nl=nl, pad_token=1,
drop_mult=drop_mult, tie_weights=True, model_dir=model_dir,
bias=True, qrnn=qrnn, clip=0.12)