Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
chenpangpang
transformers
Commits
e238e3d5
Unverified
Commit
e238e3d5
authored
Jul 17, 2020
by
Sam Shleifer
Committed by
GitHub
Jul 17, 2020
Browse files
[seq2seq] Don't copy self.source in sortishsampler (#5818)
parent
2e4624b4
Changes
1
Show whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
3 additions
and
10 deletions
+3
-10
examples/seq2seq/utils.py
examples/seq2seq/utils.py
+3
-10
No files found.
examples/seq2seq/utils.py
View file @
e238e3d5
...
@@ -144,16 +144,9 @@ class SummarizationDataset(Dataset):
...
@@ -144,16 +144,9 @@ class SummarizationDataset(Dataset):
batch
=
{
"input_ids"
:
source_ids
,
"attention_mask"
:
source_mask
,
"decoder_input_ids"
:
y
}
batch
=
{
"input_ids"
:
source_ids
,
"attention_mask"
:
source_mask
,
"decoder_input_ids"
:
y
}
return
batch
return
batch
@
property
def
src_lens
(
self
):
# Can delete?
return
lmap
(
len
,
self
.
source
)
@
property
def
tgt_lens
(
self
):
return
lmap
(
len
,
self
.
target
)
def
make_sortish_sampler
(
self
,
batch_size
):
def
make_sortish_sampler
(
self
,
batch_size
):
return
SortishSampler
(
self
.
source
,
batch_size
)
lens
=
[
x
[
"input_ids"
].
ne
(
self
.
pad_token_id
).
sum
()
for
x
in
self
.
source
]
return
SortishSampler
(
lens
,
batch_size
)
class
SortishSampler
(
Sampler
):
class
SortishSampler
(
Sampler
):
...
@@ -163,7 +156,7 @@ class SortishSampler(Sampler):
...
@@ -163,7 +156,7 @@ class SortishSampler(Sampler):
self
.
data
,
self
.
bs
=
data
,
batch_size
self
.
data
,
self
.
bs
=
data
,
batch_size
def
key
(
self
,
i
):
def
key
(
self
,
i
):
return
len
(
self
.
data
[
i
]
)
return
self
.
data
[
i
]
def
__len__
(
self
)
->
int
:
def
__len__
(
self
)
->
int
:
return
len
(
self
.
data
)
return
len
(
self
.
data
)
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment