p60975123
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

add_python_distribute_pretrain_script 

Signed-off-by: GuoMengHao <guomenghao@huawei.com>
This commit is contained in:
GuoMengHao 2020-07-27 11:21:43 +08:00
parent c22792aab1
commit 2309e7369a
11 changed files with 425 additions and 80 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
model_zoo/official/nlp/bert/README.md
View File

@ -21,7 +21,7 @@ This example implements pre-training, fine-tuning and evaluation of [BERT-base](
- Run `run_distribute_pretrain.sh` for distributed pre-training of BERT-base and BERT-NEZHA model.
``` bash
sh scripts/run_distribute_pretrain.sh DEVICE_NUM EPOCH_SIZE DATA_DIR SCHEMA_DIR MINDSPORE_HCCL_CONFIG_PATH
sh scripts/run_distribute_pretrain.sh DATA_DIR MINDSPORE_HCCL_CONFIG_PATH
```
### Fine-Tuning and Evaluation

48
model_zoo/official/nlp/bert/scripts/ascend_distributed_launcher/README.md Normal file
View File

@ -0,0 +1,48 @@
# Run distribute pretrain
## description
The number of D chips can be automatically allocated based on the device_num set in hccl config file, You don not need to specify that.
## how to use
For example, if we want to run the distributed training of Bert model on D chip, we can in `/bert/` dir:
```
python ./scripts/ascend_distributed_launcher/run_distribute_pretrain.py --run_script_dir ./run_pretrain.py --hyper_parameter_config_dir ./scripts/ascend_distributed_launcher/hyper_parameter_config.ini --data_dir /path/dataset/ --hccl_config_dir model_zoo/utils/hccl_tools/hccl_2p_56_x.x.x.x.json
```
output:
```
hccl_config_dir: model_zoo/utils/hccl_tools/hccl_2p_56_x.x.x.x.json
the number of logical core: 192
avg_core_per_rank: 96
rank_size: 2
start training for rank 0, device 5:
rank_id: 0
device_id: 5
core nums: 0-95
epoch_size: 8
data_dir: /data/small_512/
schema_dir:
log file dir: ./LOG5/log.txt
start training for rank 1, device 6:
rank_id: 1
device_id: 6
core nums: 96-191
epoch_size: 8
data_dir: /data/small_512/
schema_dir:
log file dir: ./LOG6/log.txt
```
## Note
1. Note that `hccl_2p_56_x.x.x.x.json` can use [hccl_tools.py](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/utils/hccl_tools) to generate.
2. For hyper parameter, please note that you should customize the scripts `hyper_parameter_config.ini`. Please note that these two hyper parameters are not allowed to be configured here:
device_id
device_num
3. For Other Model, please note that you should customize the option `run_script` and Corresponding `hyper_parameter_config.ini`.

0
model_zoo/official/nlp/bert/scripts/ascend_distributed_launcher/__init__.py Normal file
View File

11
model_zoo/official/nlp/bert/scripts/ascend_distributed_launcher/hyper_parameter_config.ini Normal file
View File

@ -0,0 +1,11 @@
[config]
distribute=true
epoch_size=40
enable_save_ckpt=true
enable_lossscale=true
do_shuffle=true
enable_data_sink=true
data_sink_steps=100
save_checkpoint_path=./checkpoint/
save_checkpoint_steps=10000
save_checkpoint_num=1

142
model_zoo/official/nlp/bert/scripts/ascend_distributed_launcher/run_distribute_pretrain.py Normal file
View File

@ -0,0 +1,142 @@
# Copyright 2020 Huawei Technologies Co., Ltd
#
# 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.
# ============================================================================
"""distribute pretrain script"""
import os
import json
import configparser
import multiprocessing
from argparse import ArgumentParser
def parse_args():
"""
parse args .
Args:
Returns:
args.
Examples:
>>> parse_args()
"""
parser = ArgumentParser(description="mindspore distributed training")
parser.add_argument("--run_script_dir", type=str, default="",
help="Run script path, it is better to use absolute path")
parser.add_argument("--hyper_parameter_config_dir", type=str, default="",
help="Hyper Parameter config path, it is better to use absolute path")
parser.add_argument("--data_dir", type=str, default="",
help="Data path, it is better to use absolute path")
parser.add_argument("--hccl_config_dir", type=str, default="",
help="Hccl config path, it is better to use absolute path")
args = parser.parse_args()
return args
def distribute_pretrain():
"""
distribute pretrain scripts. The number of D chips can be automatically allocated
based on the device_num set in hccl config file, You don not need to specify that.
"""
print("start", __file__)
args = parse_args()
run_script = args.run_script_dir
data_dir = args.data_dir
cf = configparser.ConfigParser()
cf.read(args.hyper_parameter_config_dir)
cfg = dict(cf.items("config"))
print("hccl_config_dir:", args.hccl_config_dir)
os.environ['MINDSPORE_HCCL_CONFIG_PATH'] = args.hccl_config_dir
os.environ['RANK_TABLE_FILE'] = args.hccl_config_dir
cores = multiprocessing.cpu_count()
print("the number of logical core:", cores)
# get device_ips
device_ips = {}
with open('/etc/hccn.conf', 'r') as fin:
for hccn_item in fin.readlines():
if hccn_item.strip().startswith('address_'):
device_id, device_ip = hccn_item.split('=')
device_id = device_id.split('_')[1]
device_ips[device_id] = device_ip.strip()
with open(args.hccl_config_dir, "r", encoding="utf-8") as fin:
hccl_config = json.loads(fin.read())
rank_size = 0
for server in hccl_config["server_list"]:
rank_size += len(server["device"])
if server["device"][0]["device_ip"] in device_ips.values():
this_server = server
os.environ['RANK_SIZE'] = str(rank_size)
print("total rank size:", rank_size)
print("this server rank size:", len(this_server["device"]))
avg_core_per_rank = int(int(cores) / len(this_server["device"]))
core_gap = avg_core_per_rank - 1
print("avg_core_per_rank:", avg_core_per_rank)
count = 0
for instance in this_server["device"]:
device_id = instance["device_id"]
rank_id = instance["rank_id"]
print("\nstart training for rank " + str(rank_id) + ", device " + str(device_id) + ":")
print("rank_id:", rank_id)
print("device_id:", device_id)
start = count * int(avg_core_per_rank)
count += 1
end = start + core_gap
cmdopt = str(start) + "-" + str(end)
os.environ["DEVICE_ID"] = device_id
os.environ["RANK_ID"] = rank_id
os.environ["DEPLOY_MODE"] = "0"
os.environ["GE_USE_STATIC_MEMORY"] = "1"
os.system("rm -rf LOG" + str(device_id))
os.system("mkdir ./LOG" + str(device_id))
os.system("cp *.py ./LOG" + str(device_id))
os.system("mkdir -p ./LOG" + str(device_id) + "/ms_log")
os.system("env > ./LOG" + str(device_id) + "/env.log")
cur_dir = os.getcwd()
os.environ["GLOG_log_dir"] = cur_dir + "/LOG" + str(device_id) + "/ms_log"
os.environ["GLOG_logtostderr"] = "0"
print("core_nums:", cmdopt)
print("epoch_size:", str(cfg['epoch_size']))
print("data_dir:", data_dir)
print("log_file_dir: ./LOG" + str(device_id) + "/log.txt")
cmd = 'taskset -c ' + cmdopt + ' python ' + run_script + " "
opt = " ".join(["--" + key + "=" + str(cfg[key]) for key in cfg.keys()])
if ('device_id' in opt) or ('device_num' in opt) or ('data_dir' in opt):
raise ValueError("hyper_parameter_config.ini can not setting 'device_id',"
" 'device_num' or 'data_dir'! ")
cmd += opt
cmd += " --data_dir=" + data_dir
cmd += ' --device_id=' + str(device_id) + ' --device_num=' \
+ str(rank_size) + ' >./LOG' + str(device_id) + '/log.txt 2>&1 &'
os.system(cmd)
if __name__ == "__main__":
distribute_pretrain()

61
model_zoo/official/nlp/bert/scripts/run_distribute_pretrain.sh
View File

@ -16,57 +16,16 @@
echo "=============================================================================================================="
echo "Please run the scipt as: "
echo "bash run_distribute_pretrain.sh DEVICE_NUM EPOCH_SIZE DATA_DIR SCHEMA_DIR MINDSPORE_HCCL_CONFIG_PATH"
echo "for example: bash run_distribute_pretrain.sh 8 40 /path/zh-wiki/ /path/Schema.json /path/hccl.json"
echo "bash run_distribute_pretrain.sh DATA_DIR MINDSPORE_HCCL_CONFIG_PATH"
echo "for example: bash run_distribute_pretrain.sh /path/dataset /path/hccl.json"
echo "It is better to use absolute path."
echo "For hyper parameter, please note that you should customize the scripts:
'{CUR_DIR}/scripts/ascend_distributed_launcher/hyper_parameter_config.ini' "
echo "=============================================================================================================="
CUR_DIR=`pwd`
EPOCH_SIZE=$2
DATA_DIR=$3
SCHEMA_DIR=$4
PROJECT_DIR=$(cd "$(dirname "$0")" || exit; pwd)
export RANK_TABLE_FILE=$5
export RANK_SIZE=$1
cores=`cat /proc/cpuinfo|grep "processor" |wc -l`
echo "the number of logical core" $cores
avg_core_per_rank=`expr $cores \/ $RANK_SIZE`
core_gap=`expr $avg_core_per_rank \- 1`
echo "avg_core_per_rank" $avg_core_per_rank
echo "core_gap" $core_gap
for((i=0;i<RANK_SIZE;i++))
do
start=`expr $i \* $avg_core_per_rank`
export DEVICE_ID=$i
export RANK_ID=$i
export DEPLOY_MODE=0
export GE_USE_STATIC_MEMORY=1
end=`expr $start \+ $core_gap`
cmdopt=$start"-"$end
rm -rf LOG$i
mkdir ./LOG$i
cp *.py ./LOG$i
cd ./LOG$i || exit
echo "start training for rank $i, device $DEVICE_ID"
mkdir -p ms_log
CUR_DIR=`pwd`
export GLOG_log_dir=${CUR_DIR}/ms_log
export GLOG_logtostderr=0
env > env.log
taskset -c $cmdopt python ${PROJECT_DIR}/../run_pretrain.py \
--distribute="true" \
--epoch_size=$EPOCH_SIZE \
--device_id=$DEVICE_ID \
--device_num=$RANK_SIZE \
--enable_save_ckpt="true" \
--enable_lossscale="true" \
--do_shuffle="true" \
--enable_data_sink="true" \
--data_sink_steps=100 \
--load_checkpoint_path="" \
--save_checkpoint_steps=10000 \
--save_checkpoint_num=1 \
--data_dir=$DATA_DIR \
--schema_dir=$SCHEMA_DIR > log.txt 2>&1 &
cd ../
done
python ${CUR_DIR}/scripts/ascend_distributed_launcher/run_distribute_pretrain.py \
--run_script_dir=${CUR_DIR}/run_pretrain.py \
--hyper_parameter_config_dir=${CUR_DIR}/scripts/ascend_distributed_launcher/hyper_parameter_config.ini \
--data_dir=$1 \
--hccl_config_dir=$2

48
model_zoo/utils/ascend_distributed_launcher/README.md Normal file
View File

@ -0,0 +1,48 @@
# Run distribute pretrain
## description
The number of D chips can be automatically allocated based on the device_num set in hccl config file, You don not need to specify that.
## how to use
For example, if we want to run the distributed training of Bert model on D chip, we can in `/bert/` dir:
```
python model_zoo/utils/ascend_distributed_launcher/run_distribute_pretrain.py --run_script_dir ./run_pretrain.py --hyper_parameter_config_dir model_zoo/utils/ascend_distributed_launcher/hyper_parameter_config.ini --data_dir /path/dataset/ --hccl_config_dir model_zoo/utils/hccl_tools/hccl_2p_56_x.x.x.x.json
```
output:
```
hccl_config_dir: model_zoo/utils/hccl_tools/hccl_2p_56_x.x.x.x.json
the number of logical core: 192
avg_core_per_rank: 96
rank_size: 2
start training for rank 0, device 5:
rank_id: 0
device_id: 5
core nums: 0-95
epoch_size: 8
data_dir: /data/small_512/
schema_dir:
log file dir: ./LOG5/log.txt
start training for rank 1, device 6:
rank_id: 1
device_id: 6
core nums: 96-191
epoch_size: 8
data_dir: /data/small_512/
schema_dir:
log file dir: ./LOG6/log.txt
```
## Note
1. Note that `hccl_2p_56_x.x.x.x.json` can use [hccl_tools.py](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/utils/hccl_tools) to generate.
2. For hyper parameter, please note that you should customize the scripts `hyper_parameter_config.ini`. Please note that these two hyper parameters are not allowed to be configured here:
device_id
device_num
3. For Other Model, please note that you should customize the option `run_script` and Corresponding `hyper_parameter_config.ini`.

0
model_zoo/utils/ascend_distributed_launcher/__init__.py Normal file
View File

11
model_zoo/utils/ascend_distributed_launcher/hyper_parameter_config.ini Normal file
View File

@ -0,0 +1,11 @@
[config]
distribute=true
epoch_size=40
enable_save_ckpt=true
enable_lossscale=true
do_shuffle=true
enable_data_sink=true
data_sink_steps=100
save_checkpoint_path=./checkpoint/
save_checkpoint_steps=10000
save_checkpoint_num=1

142
model_zoo/utils/ascend_distributed_launcher/run_distribute_pretrain.py Normal file
View File

@ -0,0 +1,142 @@
# Copyright 2020 Huawei Technologies Co., Ltd
#
# 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.
# ============================================================================
"""distribute pretrain script"""
import os
import json
import configparser
import multiprocessing
from argparse import ArgumentParser
def parse_args():
"""
parse args .
Args:
Returns:
args.
Examples:
>>> parse_args()
"""
parser = ArgumentParser(description="mindspore distributed training")
parser.add_argument("--run_script_dir", type=str, default="",
help="Run script path, it is better to use absolute path")
parser.add_argument("--hyper_parameter_config_dir", type=str, default="",
help="Hyper Parameter config path, it is better to use absolute path")
parser.add_argument("--data_dir", type=str, default="",
help="Data path, it is better to use absolute path")
parser.add_argument("--hccl_config_dir", type=str, default="",
help="Hccl config path, it is better to use absolute path")
args = parser.parse_args()
return args
def distribute_pretrain():
"""
distribute pretrain scripts. The number of D chips can be automatically allocated
based on the device_num set in hccl config file, You don not need to specify that.
"""
print("start", __file__)
args = parse_args()
run_script = args.run_script_dir
data_dir = args.data_dir
cf = configparser.ConfigParser()
cf.read(args.hyper_parameter_config_dir)
cfg = dict(cf.items("config"))
print("hccl_config_dir:", args.hccl_config_dir)
os.environ['MINDSPORE_HCCL_CONFIG_PATH'] = args.hccl_config_dir
os.environ['RANK_TABLE_FILE'] = args.hccl_config_dir
cores = multiprocessing.cpu_count()
print("the number of logical core:", cores)
# get device_ips
device_ips = {}
with open('/etc/hccn.conf', 'r') as fin:
for hccn_item in fin.readlines():
if hccn_item.strip().startswith('address_'):
device_id, device_ip = hccn_item.split('=')
device_id = device_id.split('_')[1]
device_ips[device_id] = device_ip.strip()
with open(args.hccl_config_dir, "r", encoding="utf-8") as fin:
hccl_config = json.loads(fin.read())
rank_size = 0
for server in hccl_config["server_list"]:
rank_size += len(server["device"])
if server["device"][0]["device_ip"] in device_ips.values():
this_server = server
os.environ['RANK_SIZE'] = str(rank_size)
print("total rank size:", rank_size)
print("this server rank size:", len(this_server["device"]))
avg_core_per_rank = int(int(cores) / len(this_server["device"]))
core_gap = avg_core_per_rank - 1
print("avg_core_per_rank:", avg_core_per_rank)
count = 0
for instance in this_server["device"]:
device_id = instance["device_id"]
rank_id = instance["rank_id"]
print("\nstart training for rank " + str(rank_id) + ", device " + str(device_id) + ":")
print("rank_id:", rank_id)
print("device_id:", device_id)
start = count * int(avg_core_per_rank)
count += 1
end = start + core_gap
cmdopt = str(start) + "-" + str(end)
os.environ["DEVICE_ID"] = device_id
os.environ["RANK_ID"] = rank_id
os.environ["DEPLOY_MODE"] = "0"
os.environ["GE_USE_STATIC_MEMORY"] = "1"
os.system("rm -rf LOG" + str(device_id))
os.system("mkdir ./LOG" + str(device_id))
os.system("cp *.py ./LOG" + str(device_id))
os.system("mkdir -p ./LOG" + str(device_id) + "/ms_log")
os.system("env > ./LOG" + str(device_id) + "/env.log")
cur_dir = os.getcwd()
os.environ["GLOG_log_dir"] = cur_dir + "/LOG" + str(device_id) + "/ms_log"
os.environ["GLOG_logtostderr"] = "0"
print("core_nums:", cmdopt)
print("epoch_size:", str(cfg['epoch_size']))
print("data_dir:", data_dir)
print("log_file_dir: ./LOG" + str(device_id) + "/log.txt")
cmd = 'taskset -c ' + cmdopt + ' python ' + run_script + " "
opt = " ".join(["--" + key + "=" + str(cfg[key]) for key in cfg.keys()])
if ('device_id' in opt) or ('device_num' in opt) or ('data_dir' in opt):
raise ValueError("hyper_parameter_config.ini can not setting 'device_id',"
" 'device_num' or 'data_dir'! ")
cmd += opt
cmd += " --data_dir=" + data_dir
cmd += ' --device_id=' + str(device_id) + ' --device_num=' \
+ str(rank_size) + ' >./LOG' + str(device_id) + '/log.txt 2>&1 &'
os.system(cmd)
if __name__ == "__main__":
distribute_pretrain()

40
model_zoo/utils/hccl_tools/hccl_tools.py
View File

@ -17,7 +17,6 @@ import os
import sys
import json
import socket
import platform
from argparse import ArgumentParser
from typing import Dict, Any
@ -114,40 +113,25 @@ def main():
device_id = device_id.split('_')[1]
device_ips[device_id] = device_ip.strip()
arch = platform.processor()
hccn_table = {'board_id': {'aarch64': '0x002f', 'x86_64': '0x0000'}[arch],
'chip_info': '910',
'deploy_mode': 'lab',
'group_count': '1',
'group_list': []}
instance_list = []
hccn_table = {'version': '1.0',
'server_count': '1',
'server_list': []}
device_list = []
rank_id = 0
for instance_id in device_num_list:
instance = {'devices': []}
device_id = visible_devices[instance_id]
device_ip = device_ips[device_id]
instance['devices'].append({
'device_id': device_id,
'device_ip': device_ip,
})
device = {'device_id': device_id,
'device_ip': device_ip,
'rank_id': str(rank_id)}
print('rank_id:{}, device_id:{}, device_ip:{}'.format(rank_id, device_id, device_ip))
instance['rank_id'] = str(rank_id)
rank_id += 1
instance['server_id'] = server_id
instance_list.append(instance)
hccn_table['group_list'].append({
'device_num': str(len(device_num_list)),
'server_num': '1',
'group_name': '',
'instance_count': str(len(device_num_list)),
'instance_list': instance_list,
device_list.append(device)
hccn_table['server_list'].append({
'server_id': server_id,
'device': device_list,
'host_nic_ip': 'reserve'
})
hccn_table['para_plane_nic_location'] = 'device'
hccn_table['para_plane_nic_name'] = []
for instance_id in device_num_list:
eth_id = visible_devices[instance_id]
hccn_table['para_plane_nic_name'].append('eth{}'.format(eth_id))
hccn_table['para_plane_nic_num'] = str(len(device_num_list))
hccn_table['status'] = 'completed'
# save hccn_table to file