* update gpu_ids in distributed training * move linear scaling rule after getting correct gpu_ids * Remove support for autoscale_lr |
||
---|---|---|
.. | ||
convert_models | ||
dist_test.sh | ||
dist_train.sh | ||
get_flops.py | ||
publish_model.py | ||
pytorch2onnx.py | ||
slurm_test.sh | ||
slurm_train.sh | ||
test.py | ||
train.py |