未验证 提交 d2110579 编写于 作者: K Kaipeng Deng 提交者: GitHub

Merge pull request #2053 from heavengate/tsm_gc

gc flag for TSM.
...@@ -34,6 +34,9 @@ TSM的训练数据采用由DeepMind公布的Kinetics-400动作识别数据集。 ...@@ -34,6 +34,9 @@ TSM的训练数据采用由DeepMind公布的Kinetics-400动作识别数据集。
数据准备完毕后,可以通过如下两种方式启动训练: 数据准备完毕后,可以通过如下两种方式启动训练:
export FLAGS_fast_eager_deletion_mode=1
export FLAGS_eager_delete_tensor_gb=0.0
export FLAGS_fraction_of_gpu_memory_to_use=0.98
python train.py --model_name=TSM python train.py --model_name=TSM
--config=./configs/tsm.txt --config=./configs/tsm.txt
--save_dir=checkpoints --save_dir=checkpoints
......
export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
# activate eager gc to reduce memory use
export FLAGS_fast_eager_deletion_mode=1
export FLAGS_eager_delete_tensor_gb=0.0
export FLAGS_fraction_of_gpu_memory_to_use=0.98
python train.py --model_name="TSM" --config=./configs/tsm.txt --epoch_num=65 \ python train.py --model_name="TSM" --config=./configs/tsm.txt --epoch_num=65 \
--valid_interval=1 --log_interval=10 --valid_interval=1 --log_interval=10
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册