github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-02 17:46:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

[hotfix] set return_outputs=False in examples and polish code (#5404)

Browse Source 
* fix: simplify merge_batch

* fix: use return_outputs=False to eliminate extra memory consumption

* feat: add return_outputs warning

* style: remove `return_outputs=False` as it is the default value
This commit is contained in:
Wenhao Chen
2024-03-25 12:31:09 +08:00
committed by GitHub
parent 5fcd7795cd
commit bb0a668fee
24 changed files with 28 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/images/vit/vit_benchmark.py
View File

@@ -120,7 +120,7 @@ def main():
# run pipeline forward backward
batch = iter([batch])
outputs = booster.execute_pipeline(
batch, model, criterion, optimizer, return_loss=True, return_outputs=True
batch, model, criterion, optimizer, return_loss=True
)
else:
outputs = model(**batch)

2
examples/language/bert/finetune.py
View File

@@ -148,7 +148,7 @@ def train_epoch(
for _ in pbar:
if use_pipeline:
outputs = booster.execute_pipeline(
train_dataloader_iter, model, _criterion, optimizer, return_loss=True, return_outputs=True
train_dataloader_iter, model, _criterion, optimizer, return_loss=True
)
# Backward and optimize
if is_pp_last_device:

2
examples/language/gpt/hybridparallelism/finetune.py
View File

@@ -145,7 +145,7 @@ def train_epoch(
for _ in pbar:
if use_pipeline:
outputs = booster.execute_pipeline(
train_dataloader_iter, model, _criterion, optimizer, return_loss=True, return_outputs=True
train_dataloader_iter, model, _criterion, optimizer, return_loss=True
)
# Backward and optimize
if is_pp_last_stage:

2
examples/language/llama2/finetune.py
View File

@@ -271,7 +271,7 @@ def main():
for step in pbar:
if use_pipeline:
outputs = booster.execute_pipeline(
dataloader_iter, model, _criterion, optimizer, return_loss=True, return_outputs=True
dataloader_iter, model, _criterion, optimizer, return_loss=True
)
loss = outputs["loss"]
else:

4
examples/language/llama2/pretrain.py
View File

@@ -185,7 +185,7 @@ def main():
microbatch_size=1,
enable_jit_fused=False,
zero_stage=0,
precision="fp32",
precision=args.mixed_precision,
initial_scale=1,
)
else:
@@ -286,7 +286,7 @@ def main():
for step in pbar:
if use_pipeline:
outputs = booster.execute_pipeline(
dataloader_iter, model, _criterion, optimizer, return_loss=True, return_outputs=True
dataloader_iter, model, _criterion, optimizer, return_loss=True
)
loss = outputs["loss"]
else:

1
examples/language/openmoe/benchmark/benchmark_cai.py
View File

@@ -270,7 +270,6 @@ def main():
lambda x, y: x.loss,
optimizer,
return_loss=True,
return_outputs=True,
)
# Backward and optimize
if is_pp_last_stage:

1
examples/language/openmoe/train.py
View File

@@ -340,7 +340,6 @@ def main():
lambda x, y: x.loss,
optimizer,
return_loss=True,
return_outputs=True,
)
# Backward and optimize
if is_pp_last_stage:

2
examples/language/opt/opt_train_demo.py
View File

@@ -42,7 +42,7 @@ def train_epoch(epoch, model, optimizer, _criterion, lr_scheduler, dataloader, b
for _ in pbar:
if use_pipeline:
outputs = booster.execute_pipeline(
dataloader, model, _criterion, optimizer, return_loss=True, return_outputs=True
dataloader, model, _criterion, optimizer, return_loss=True
)
# Backward and optimize
if is_pp_last_stage: