diff --git a/cv/classification/resnest50/pytorch/train_resnest50_amp_dist.sh b/cv/classification/resnest50/pytorch/train_resnest50_amp_dist.sh index af76da2a5e6c0f6a844d1d34773df719b5e56157..3331e037e5ee93655a0746ff4fb3c9f40252aeef 100755 --- a/cv/classification/resnest50/pytorch/train_resnest50_amp_dist.sh +++ b/cv/classification/resnest50/pytorch/train_resnest50_amp_dist.sh @@ -26,5 +26,5 @@ fi cd ${ROOT_DIR} python3 $PYTHONARG ${ROOT_DIR}/run_train.py \ --model resnest50 --dali --dali-cpu --data-path $DATA_PATH \ - --opt fused_sgd --batch-size 512 --lr 0.0125 \ + --opt fused_sgd --batch-size 256 --lr 0.0125 \ --amp --nhwc "$@" diff --git a/cv/classification/resnet50/pytorch/README.md b/cv/classification/resnet50/pytorch/README.md index c388ce51cec2e50c43c1404bb6d5c5ede3cbe435..85bd0f571601058f1e1bf07ebddf4d7e59d90fd5 100644 --- a/cv/classification/resnet50/pytorch/README.md +++ b/cv/classification/resnet50/pytorch/README.md @@ -21,17 +21,17 @@ bash amp_1card.sh --data-path /path/to/imagenet ``` ### Multiple GPUs on one machine ```bash -bash fp32_4card.sh --data-path /path/to/imagenet -bash fp32_8card.sh --data-path /path/to/imagenet +bash fp32_4cards.sh --data-path /path/to/imagenet +bash fp32_8cards.sh --data-path /path/to/imagenet ``` ### Multiple GPUs on one machine (AMP) ```bash -bash amp_4card.sh --data-path /path/to/imagenet -bash amp_8card.sh --data-path /path/to/imagenet +bash amp_4cards.sh --data-path /path/to/imagenet +bash amp_8cards.sh --data-path /path/to/imagenet ``` ### Multiple GPUs on two machines ```bash -bash fp32_16card.sh --data-path /path/to/imagenet +bash fp32_16cards.sh --data-path /path/to/imagenet ``` ## Results on BI-V100 diff --git a/cv/detection/autoassign/pytorch/README.md b/cv/detection/autoassign/pytorch/README.md index 3037ca1ca821829814d5f9b488b5bea2a564d8d2..02a110b624dbedca9f8c49fb28e12f9fb12806bf 100755 --- a/cv/detection/autoassign/pytorch/README.md +++ b/cv/detection/autoassign/pytorch/README.md @@ -17,6 +17,7 @@ $ MMCV_WITH_OPS=1 python3 setup.py build && cp build/lib.linux*/mmcv/_ext.cpytho ```bash $ cd /path/to/modelzoo/cv/detection/autoassign/pytorch $ mkdir -p data && cd data +$ apt install dos2unix # Download from homepage of coco: https://cocodataset.org/ ``` @@ -30,7 +31,7 @@ $ python3 train.py [training args] # config file can be found in ### Multiple GPUs on one machine ```bash -$ bash train_dist.sh [training args] # config file can be found in the configs directory +$ bash dist_train.sh [training args] # config file can be found in the configs directory ``` ## Reference