Update README.md

pull/66/head
Ross Wightman 5 years ago committed by GitHub
parent a4497afa21
commit 0554b7942c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -264,7 +264,7 @@ This params are for dual Titan RTX cards with NVIDIA Apex installed:
### SE-ResNeXt-26-D and SE-ResNeXt-26-T ### SE-ResNeXt-26-D and SE-ResNeXt-26-T
These hparams (or similar) work well for a wide range of ResNet architecture, generally a good idea to increase the epoch # as the model size increases... ie approx 180-200 for ResNe(X)t50, and 220+ for larger. Increase batch size and LR proportionally for better GPUs or with AMP enabled. These params were for 2 1080Ti cards: These hparams (or similar) work well for a wide range of ResNet architecture, generally a good idea to increase the epoch # as the model size increases... ie approx 180-200 for ResNe(X)t50, and 220+ for larger. Increase batch size and LR proportionally for better GPUs or with AMP enabled. These params were for 2 1080Ti cards:
i
`./distributed_train.sh 2 /imagenet/ --model seresnext26t_32x4d --lr 0.1 --warmup-epochs 5 --epochs 160 --weight-decay 1e-4 --sched cosine --reprob 0.4 --remode pixel -b 112` `./distributed_train.sh 2 /imagenet/ --model seresnext26t_32x4d --lr 0.1 --warmup-epochs 5 --epochs 160 --weight-decay 1e-4 --sched cosine --reprob 0.4 --remode pixel -b 112`
### EfficientNet-B3 with RandAugment - 81.5 top-1, 95.7 top-5 ### EfficientNet-B3 with RandAugment - 81.5 top-1, 95.7 top-5
@ -279,13 +279,14 @@ The training of this model started with the same command line as EfficientNet-B2
All development and testing has been done in Conda Python 3 environments on Linux x86-64 systems, specifically Python 3.6.x and 3.7.x. Little to no care has been taken to be Python 2.x friendly and I don't plan to support it. If you run into any challenges running on Windows, or other OS, I'm definitely open to looking into those issues so long as it's in a reproducible (read Conda) environment. All development and testing has been done in Conda Python 3 environments on Linux x86-64 systems, specifically Python 3.6.x and 3.7.x. Little to no care has been taken to be Python 2.x friendly and I don't plan to support it. If you run into any challenges running on Windows, or other OS, I'm definitely open to looking into those issues so long as it's in a reproducible (read Conda) environment.
PyTorch versions 1.0 and 1.1 have been tested with this code. PyTorch versions 1.2 and 1.3.1 have been tested with this code.
I've tried to keep the dependencies minimal, the setup is as per the PyTorch default install instructions for Conda: I've tried to keep the dependencies minimal, the setup is as per the PyTorch default install instructions for Conda:
``` ```
conda create -n torch-env conda create -n torch-env
conda activate torch-env conda activate torch-env
conda install -c pytorch pytorch torchvision cudatoolkit=10.0 conda install -c pytorch pytorch torchvision cudatoolkit=10
conda install pyyaml
``` ```
### Pip ### Pip
@ -331,5 +332,4 @@ To run inference from a checkpoint:
## TODO ## TODO
A number of additions planned in the future for various projects, incl A number of additions planned in the future for various projects, incl
* Do a model performance (speed + accuracy) benchmarking across all models (make runable as script) * Do a model performance (speed + accuracy) benchmarking across all models (make runable as script)
* Add usage examples to comments, good hyper params for training * Complete feature map extraction across all model types and build obj detection/segmentation models and scripts (or integrate backbones with mmdetection, detectron2)
* Comments, cleanup and the usual things that get pushed back

Loading…
Cancel
Save