Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add CE for Distributed Hybrid Parallel #7782

Merged
merged 8 commits into from
Jan 9, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=CE_gpt-345m_seqlen1024_pretrain
dp_degree=2
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-PP2-DP2-mbs8-acc2
device_num=N1C8
max_iter=50000
sharding=False
sharding_degree=1
virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=False

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-PP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=50000
sharding=stage2
sharding_degree=1

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=False

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=CE_gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-SP2-PP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=50000
sharding=stage1
sharding_degree=2

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=True

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=CE_gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=1
bs_item=32
fp_item=bf16
run_mode=MP2-SP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=50000
sharding=stage1
sharding_degree=2

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=False
sequence_parallel=True

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=gpt-345m_seqlen1024_pretrain
dp_degree=2
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-PP2-DP2-mbs8-acc2
device_num=N1C8
max_iter=100
sharding=False
sharding_degree=1
virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=False

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-PP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=100
sharding=stage1
sharding_degree=2

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=False

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=2
bs_item=32
fp_item=bf16
run_mode=MP2-SP2-PP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=100
sharding=stage1
sharding_degree=2

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=True
sequence_parallel=True

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

model_item=gpt-345m_seqlen1024_pretrain
dp_degree=1
mp_degree=2
pp_degree=1
bs_item=32
fp_item=bf16
run_mode=MP2-SP2-SD2-Stage1-mbs8-acc2
device_num=N1C8
max_iter=100
sharding=stage1
sharding_degree=2

virtual_pp_degree=1
use_recompute=True
eval_freq=25
use_pipeline_parallel=False
sequence_parallel=True

model=gpt
micro_bs=8

bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/prepare.sh
# run
bash ./test_tipc/dygraph/hybrid_parallelism/ce_gpt/benchmark_common/run_benchmark.sh ${model_item} ${fp_item} ${dp_degree} ${mp_degree} ${pp_degree} ${micro_bs} ${bs_item} ${run_mode} ${device_num} \
${max_iter} ${sharding} ${sharding_degree} ${virtual_pp_degree} ${use_recompute} ${eval_freq} ${use_pipeline_parallel} ${sequence_parallel} 2>&1;
Original file line number Diff line number Diff line change
Expand Up @@ -151,11 +151,16 @@ function _train(){
run_pretrain.py ${train_cmd}"
workerlog_id=0
;;
DP8-mbs2-acc2|SD8-stage1-mbs2-acc2|SD8-stage2-mbs2-acc2|SD8-stage3-mbs2-acc2|MP2-SD4-stage1-mbs4-acc2|MP2-SP2-PP2-DP2-mbs8-acc2|MP8-mbs16-acc2) echo "run run_mode: ${run_mode}"
DP8-mbs2-acc2|SD8-stage1-mbs2-acc2|SD8-stage2-mbs2-acc2|SD8-stage3-mbs2-acc2|MP2-SD4-stage1-mbs4-acc2|MP2-SP2-PP2-DP2-mbs8-acc2|MP8-mbs16-acc2|MP2-PP2-DP2-mbs8-acc2|MP2-PP2-SD2-Stage1-mbs8-acc2|MP2-SP2-PP2-SD2-Stage1-mbs8-acc2) echo "run run_mode: ${run_mode}"
train_cmd="python -m paddle.distributed.launch --log_dir=./mylog --devices=0,1,2,3,4,5,6,7 ${PADDLE_RANK_OPTION}\
run_pretrain.py ${train_cmd}"
workerlog_id=0
;;
MP2-SP2-SD2-Stage1-mbs8-acc2) echo "run run_mode: ${run_mode}"
train_cmd="python -m paddle.distributed.launch --log_dir=./mylog --devices=0,1,2,3 ${PADDLE_RANK_OPTION}\
run_pretrain.py ${train_cmd}"
workerlog_id=0
;;
*) echo "choose run_mode "; exit 1;
esac
cd ../llm/gpt-3
Expand Down