nasa-nccs-hpda / astrotime / 13401484631

Builds Branch Commit Type Ran Committer Via Coverage
13401484631 main Reduce max_series_length to 6000 in all workflow scripts. This change standardizes the max_series_length parameter across all scripts, lowering it from 7000 to 6000. This adjustment ensures consistency and may help optimize memory usage or comput... push ThomasMaxwell github
80.0
13401444317 main Comment out debug print statement in wavelet encoder Removed an unnecessary debug print statement to clean up the code. This ensures no extraneous output is generated during normal execution. push ThomasMaxwell github
80.0
13401425013 main Increase max_series_length and adjust batch sizes Updated `max_series_length` to 7000 across various scripts to enhance series processing capability. Adjusted batch sizes to 50 in multiple workflows for improved processing consistency. Removed de... push ThomasMaxwell github
80.0
13401222112 main Refactor workflow scripts to include sparsity-based downsampling. Added `RandomDownsample` filter with a configurable sparsity parameter to relevant enc push ThomasMaxwell github
80.0
13401151836 main Add debug logging for WWZ batch encoding shapes This change introduces a print statement to log the shapes of inputs and outputs during WWZ batch encoding. It aids in debugging and ensures correctness by providing visibility into the data transfo... push ThomasMaxwell github
80.0
13401116824 main Refactor wavelet encoder logging and return variable names Improve clarity of debug print statements by including reshaped input dimensions and consistent variable naming. Update return push ThomasMaxwell github
80.0
13401025092 main Refactor encoders to support dynamic `nfeatures` parameter. Updated the `WaveletEncoder` to accept `nfeatures` as a parameter instead of using a fixed value. Adjusted initialization and configuration in relevant scripts to reflect these changes. ... push ThomasMaxwell github
80.0
13400862356 main Adjust batch size from 50 to 25 in wavelet-embedding script. The batch size change optimizes memory usage and training stability. This update helps improve model performance on large datasets without running into memory constraints. No other para... push ThomasMaxwell github
80.0
13400761508 main Add `encode_batch` method to `wavelet.py` encoder Introduces a new `encode_batch` function for processing batched inputs in the `wavelet.py` encoder, leveraging filtering, normalization, and stacking. Additionally, cleanup in `baseline.py` remove... push ThomasMaxwell github
80.0
13400613159 main Update sinusoid loader to ncSinusoidLoader with new parameters Replaced `SinusoidLoader` with `ncSinusoidLoader` to support working with `.nc` files. Added parameters for dataset root, file pattern, file size, and batch size configuration. This e... push ThomasMaxwell github
80.0
13400566506 main "Add wavelet-based model and refactor max series length logic Introduce a new wavelet-based CNN model workflow for sinusoid data processing. Refactor encoders to support configurable max series length, improving flexibility and consistency. Updat... push ThomasMaxwell github
80.0
13400304088 main Adjust batch size in baseline-nc workflow to 25. Reduced the batch size from 100 to 25 to better align with processing requirements or system constraints. This change ensures better handling of data within the specified workflow and may improve t... push ThomasMaxwell github
80.0
13400286184 main Remove debug print statements from baseline encoder. The debug print statements were commented out to clean up output during runtime. These statements were likely used for development purposes and are no longer necessary in the production code. push ThomasMaxwell github
80.0
13400243338 main Comment out debug print statements in data processing. Removed debug print statements from `get_batch` and `encode_batch` methods to clean up the logs during execution. This helps ensure the code outputs only necessary information in production e... push ThomasMaxwell github
80.0
13400181784 main Update training shuffle option and remove unused DataPreprocessor class Set `shuffle=False` in training arguments to ensure consistent batch order during training. Removed the unused `DataPreprocessor` class from `astrotime/loaders/base.py` to cl... push ThomasMaxwell github
80.0