]> Gitweb @ Texas Instruments - Open Source Git Repositories - git.TI.com/gitweb - jacinto-ai/pytorch-jacinto-ai-devkit.git/blobdiff - docs/Calibration.md
epoch_size - meaning has changed to number of images instead of number of iterations...
[jacinto-ai/pytorch-jacinto-ai-devkit.git] / docs / Calibration.md
index 2c97bb95680856e5fba7a820b181cf8b746eca5c..385c0c7bde8515ef9d6ade7da081a5ac3c2009f1 100644 (file)
@@ -75,18 +75,18 @@ Few examples of calibration are provided below. These commands are also listed i
 
 - Calibration of ImageNet Classification MobileNetV2 model
 ```
-python ./scripts/train_classification_main.py --phase calibration --dataset_name image_folder_classification --model_name mobilenetv2_tv_x1 --data_path ./data/datasets/image_folder_classification --pretrained https://download.pytorch.org/models/mobilenet_v2-b0353104.pth --batch_size 64 --quantize True --epochs 1 --epoch_size 100
+python ./scripts/train_classification_main.py --phase calibration --dataset_name image_folder_classification --model_name mobilenetv2_tv_x1 --data_path ./data/datasets/image_folder_classification --pretrained https://download.pytorch.org/models/mobilenet_v2-b0353104.pth --batch_size 64 --quantize True --epochs 1 --epoch_size 0.1
 ```
 
 - Calibration of ImageNet Classification ResNet50 model
 ```
-python ./scripts/train_classification_main.py --phase calibration --dataset_name image_folder_classification --model_name mobilenetv2_tv_x1 --data_path ./data/datasets/image_folder_classification --pretrained https://download.pytorch.org/models/resnet50-19c8e357.pth --batch_size 64 --quantize True --epochs 1 --epoch_size 100
+python ./scripts/train_classification_main.py --phase calibration --dataset_name image_folder_classification --model_name mobilenetv2_tv_x1 --data_path ./data/datasets/image_folder_classification --pretrained https://download.pytorch.org/models/resnet50-19c8e357.pth --batch_size 64 --quantize True --epochs 1 --epoch_size 0.1
 ```
 
 - Calibration of Cityscapes Semantic Segmentation model
 ```
 python ./scripts/train_segmentation_main.py --phase calibration --dataset_name cityscapes_segmentation --model_name deeplabv3lite_mobilenetv2_tv --data_path ./data/datasets/cityscapes/data --img_resize 384 768 --output_size 1024 2048 --gpus 0 1 
 --pretrained ./data/modelzoo/pytorch/semantic_segmentation/cityscapes/jacinto_ai/deeplabv3lite_mobilenetv2_tv_resize768x384_best.pth 
---batch_size 12 --quantize True --epochs 1 --epoch_size 100
+--batch_size 12 --quantize True --epochs 1
 ```