Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Major] Dataloader: Just-In-Time tabularization #1529

Open
wants to merge 109 commits into
base: main
Choose a base branch
from

Conversation

ourownstory
Copy link
Owner

@ourownstory ourownstory commented Feb 9, 2024

Complete re-write of dataloader: Instead of tabularizing samples ahead of batching, samples are tabularized and features computed during batch creation. Trades Memory use for Compute.

This is an early implementation that works, but that needs to be compute-optimized.

Copy link

codecov bot commented Feb 9, 2024

Codecov Report

Attention: 25 lines in your changes are missing coverage. Please review.

Comparison is base (5a3fabf) 88.51% compared to head (2f38531) 88.24%.

Files Patch % Lines
neuralprophet/time_dataset.py 93.31% 20 Missing ⚠️
neuralprophet/forecaster.py 58.33% 5 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main    #1529      +/-   ##
==========================================
- Coverage   88.51%   88.24%   -0.27%     
==========================================
  Files          41       41              
  Lines        5330     5344      +14     
==========================================
- Hits         4718     4716       -2     
- Misses        612      628      +16     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Copy link

github-actions bot commented Feb 15, 2024

Model Benchmark

Benchmark Metric main current diff
YosemiteTemps MAE_val 0.56412 0.60324 6.94% ⚠️
YosemiteTemps RMSE_val 0.83161 0.88159 6.01% ⚠️
YosemiteTemps Loss_val 0.0004 0.00045 12.29%
YosemiteTemps MAE 0.98449 0.98956 0.52%
YosemiteTemps RMSE 1.75389 1.73958 -0.82%
YosemiteTemps Loss 0.00132 0.0013 -1.5%
YosemiteTemps time 34.5426 833.36 2312.56%
EnergyPriceDaily MAE_val 5.41157 5.65526 4.5% ⚠️
EnergyPriceDaily RMSE_val 6.71538 6.86834 2.28%
EnergyPriceDaily Loss_val 0.02525 0.02645 4.75% ⚠️
EnergyPriceDaily MAE 5.94936 6.2678 5.35% ⚠️
EnergyPriceDaily RMSE 7.9833 8.34813 4.57% ⚠️
EnergyPriceDaily Loss 0.02579 0.02826 9.59%
EnergyPriceDaily time 15.1859 268.35 1667.1%
PeytonManning MAE_val 0.34955 0.35129 0.5%
PeytonManning RMSE_val 0.50049 0.50154 0.21%
PeytonManning Loss_val 0.01771 0.0178 0.53%
PeytonManning MAE 0.34653 0.34697 0.13%
PeytonManning RMSE 0.49312 0.49406 0.19%
PeytonManning Loss 0.01464 0.01468 0.28%
PeytonManning time 11.8971 269.68 2166.76%
AirPassengers MAE_val 30.6285 30.9382 1.01%
AirPassengers RMSE_val 31.5254 32.1915 2.11%
AirPassengers Loss_val 0.01277 0.01332 4.27% ⚠️
AirPassengers MAE 6.16861 6.91875 12.16%
AirPassengers RMSE 7.85225 8.9841 14.41%
AirPassengers Loss 0.00065 0.00078 20.92%
AirPassengers time 7.75698 30.72 296.03%
Model training plots

Model Training

PeytonManning

YosemiteTemps

AirPassengers

EnergyPriceDaily

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.

None yet

2 participants