Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.23402142Inference weights, synthetic motion datasets, and structured numerical evaluation data. Includes 25 final refinement policies (five seeds for each of five regressor/instrumental reward configurations, round 500, reference regularization 0.01), five frozen assessment regressors, 25,600 refined synthetic motions, and 5,120 paired baseline synthetic motions. Version 1.1 adds 300 selected downstream exercise-recognition checkpoints (150 CNN and 150 ST-GCN), the six exact 1,000-motion synthetic input banks, and CSV/JSON evaluation datasets covering all 300 new downstream runs, 200 distinct published reference fits, per-clip validation/test predictions, paired accuracy contrasts, and all 30,720 refinement evaluations with seed/exercise aggregates. The downstream checkpoints cover six synthetic arms, five real-participant budgets, and five classifier seeds; checkpoints were selected using real-validation accuracy. Files are complete and each ZIP opens independently; no numbered archive segments or concatenation are required. Base generator and VAE weights are distributed separately at https://doi.org/10.5281/zenodo.23197021. Original participant recordings, manuscripts, source code, narrative reports, logs, optimizer states, and intermediate training checkpoints are excluded.
Keyword(s):
anonymous review; exercise recognition; model weights; rehabilitation; reinforcement learning.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)