Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
OpenDAS
apex
Commits
45cd1001
Unverified
Commit
45cd1001
authored
Jan 21, 2022
by
eqy
Committed by
GitHub
Jan 21, 2022
Browse files
Fix missing `dtype` in `recv_forward` (#1276)
CC @crcrpar @ptrblck
parent
c4e85f7b
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
1 addition
and
1 deletion
+1
-1
apex/transformer/pipeline_parallel/schedules/fwd_bwd_pipelining_without_interleaving.py
...llel/schedules/fwd_bwd_pipelining_without_interleaving.py
+1
-1
No files found.
apex/transformer/pipeline_parallel/schedules/fwd_bwd_pipelining_without_interleaving.py
View file @
45cd1001
...
@@ -227,7 +227,7 @@ def forward_backward_pipelining_without_interleaving(
...
@@ -227,7 +227,7 @@ def forward_backward_pipelining_without_interleaving(
for
i
in
range
(
num_warmup_microbatches
):
for
i
in
range
(
num_warmup_microbatches
):
_logger
.
debug
(
f
"warmup iter:
{
i
}
/
{
num_warmup_microbatches
}
"
)
_logger
.
debug
(
f
"warmup iter:
{
i
}
/
{
num_warmup_microbatches
}
"
)
_logger
.
debug
(
"receive fwd"
)
_logger
.
debug
(
"receive fwd"
)
input_tensor
=
recv_forward
(
tensor_shapes
=
recv_tensor_shapes
)
input_tensor
=
recv_forward
(
tensor_shapes
=
recv_tensor_shapes
,
dtype
=
dtype
)
cur_microbatch
=
get_kth_microbatch
(
batch
,
i
)
cur_microbatch
=
get_kth_microbatch
(
batch
,
i
)
output_tensor
=
forward_step
(
forward_step_func
,
cur_microbatch
,
model
,
input_tensor
,
losses_reduced
)
output_tensor
=
forward_step
(
forward_step_func
,
cur_microbatch
,
model
,
input_tensor
,
losses_reduced
)
_logger
.
debug
(
"send fwd"
)
_logger
.
debug
(
"send fwd"
)
...
...
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