roofAI 🏛️ meets ⚒️ SageMaker + more-8

Next,

@vanwatch ingest
vanwatch update

@batch list status=RUNNING

validate that in the notebook train_label_map.json ends up on the cloud,

using semantic_segmentation_pascalvoc-v2.ipynb,

no trace of train_label_map.json. 🤔 found, commented, added to dataset.py.

on SageMaker,

@seed sagemaker-system
nbs conda create_env
nbs init clear

sagesemseg upload_dataset - \
  suffix=$(today)-v1 \
  --count 2

validate that the assets uploaded by sagesemseg upload_dataset and semantic_segmentation_pascalvoc-v2.ipynb are identical. ✅

full upload,

sagesemseg upload_dataset - suffix=full-v2

while waiting: https://boto3.amazonaws.com/v1/documentation/api/latest/guide/s3-uploading-files.html#the-extraargs-parameter

s3.upload_file(
    'FILE_NAME', 'BUCKET_NAME', 'OBJECT_NAME',
    Callback=ProgressPercentage('FILE_NAME')
)

@ls cloud pascal-voc-v1-full-v2

semantic_segmentation_pascalvoc-v3.ipynb

validate semantic_segmentation_pascalvoc-v3.ipynb: train and deploy ▶️ 📜

continues.