super-gradients/src/super_gradients/recipes/coco2017_yolox.yaml

# YoloX Detection training on CoCo2017 Dataset:
# YoloX trained in 640x640
# Checkpoints + tensorboards: https://deci-pretrained-models.s3.amazonaws.com/yolox_coco/
# Recipe runs with batch size = 16 X 8 gpus = 128.


# Instructions:
#   0. Make sure that the data is stored in dataset_params.dataset_dir or add "dataset_params.data_dir=<PATH-TO-DATASET>" at the end of the command below (feel free to check ReadMe)
#   1. Move to the project root (where you will find the ReadMe and src folder)
#   2. Run the command you want:
#         yolox_n: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_n
#         yolox_t: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_t
#         yolox_s: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_s
#         yolox_m: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_m
#         yolox_l: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_l
#         yolox_x: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_x
#
# Training times and accuracies (mAP@0.5-0.95 (COCO API, confidence 0.001, IoU threshold 0.6, test on 640x640 images):
#         yolox_n: 1d 16h 33m 9s  on 8 NVIDIA GeForce RTX 3090, mAP: 26.77
#         yolox_t: 20h 43m 37s    on 8 NVIDIA RTX A5000, mAP: 37.18
#         yolox_s: 1d 17h 40m 30s on 8 NVIDIA RTX A5000, mAP: 40.47
#         yolox_m: 1d 22h 23m 43s on 8 NVIDIA GeForce RTX 3090, mAP: 46.40
#         yolox_l: 2d 14h 11m 41s on 8 NVIDIA GeForce RTX 3090, mAP: 49.25

defaults:
  - training_hyperparams: coco2017_yolox_train_params
  - dataset_params: coco_detection_dataset_params
  - arch_params: yolox_s_arch_params
  - checkpoint_params: default_checkpoint_params

train_dataloader: coco2017_train
val_dataloader: coco2017_val

model_checkpoints_location: local

load_checkpoint: False
resume: False
training_hyperparams:
  resume: ${resume}

architecture: yolox_s

multi_gpu: DDP
num_gpus: 8

experiment_suffix: res${dataset_params.train_dataset_params.input_dim}
experiment_name: ${architecture}_coco2017_${experiment_suffix}

ckpt_root_dir: