Skip to content

Add Google Colab and Google Cloud training commands for PlaNet model#1

Open
Rayeed221 wants to merge 1 commit into
claude/train-final-model-9RGHVfrom
claude/google-cloud-training-commands-HO2Ex
Open

Add Google Colab and Google Cloud training commands for PlaNet model#1
Rayeed221 wants to merge 1 commit into
claude/train-final-model-9RGHVfrom
claude/google-cloud-training-commands-HO2Ex

Conversation

@Rayeed221
Copy link
Copy Markdown
Owner

  • colab_training.ipynb: step-by-step Jupyter notebook for Google Colab

    • Auto-detects CUDA version and installs a compatible TensorFlow
    • Optional Google Drive mount for checkpoint persistence across sessions
    • Downloads the Zenodo dataset, patches config paths, streams training output
    • TensorBoard integration and resume-from-checkpoint support
    • Model export cell for SavedModel / TFLite output
  • gcloud_train.sh: bash script for two GCP deployment modes

    • MODE=gce: creates a Deep Learning VM (GPU), SSH-installs deps, downloads dataset, streams training, syncs checkpoints to GCS
    • MODE=vertex: uploads dataset to GCS, submits a Vertex AI Custom Training Job using a pre-built TF GPU container
  • planner_learning/config/train_settings_cloud.yaml: cloud-optimised training config (batch_size=16, save_every_n_epochs=5, paths overridden at runtime by the respective launch scripts)

https://claude.ai/code/session_01UCa2UAvFF9LBRLiqjDhTuf

- colab_training.ipynb: step-by-step Jupyter notebook for Google Colab
  - Auto-detects CUDA version and installs a compatible TensorFlow
  - Optional Google Drive mount for checkpoint persistence across sessions
  - Downloads the Zenodo dataset, patches config paths, streams training output
  - TensorBoard integration and resume-from-checkpoint support
  - Model export cell for SavedModel / TFLite output

- gcloud_train.sh: bash script for two GCP deployment modes
  - MODE=gce: creates a Deep Learning VM (GPU), SSH-installs deps,
    downloads dataset, streams training, syncs checkpoints to GCS
  - MODE=vertex: uploads dataset to GCS, submits a Vertex AI Custom
    Training Job using a pre-built TF GPU container

- planner_learning/config/train_settings_cloud.yaml: cloud-optimised
  training config (batch_size=16, save_every_n_epochs=5, paths
  overridden at runtime by the respective launch scripts)

https://claude.ai/code/session_01UCa2UAvFF9LBRLiqjDhTuf
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants