emg2tendon — pretrained weights (full 25,253-recording training)
Model checkpoints for the EMG→tendon-control task (16-ch sEMG at 2kHz → 39-ch MyoHand tendon control), trained on the full emg2pose dataset. See the repo README + RESULTS.md for the eval protocol.
| File | Model | val tendon-RMSE | open-loop pose (deg) |
|---|---|---|---|
tds.ckpt |
TDS (time-depth separable conv) | 0.310 | 15.1–16.2 |
sensingdynamics.ckpt |
SensingDynamics | 0.308 | 15.1–16.2 |
neuropose.ckpt |
NeuroPose | 0.308 | 15.1–16.2 |
cldm.ckpt |
Conditional Latent Diffusion (self-contained: both VAEs + UNet) | 0.440 | 16.8–17.6 |
emg_stats.npz |
per-channel EMG mean/std (normalization; required for inference) | — | — |
Load + run: scripts/evaluate_pose.py / scripts/render_model.py (--checkpoint <file> --stats_cache emg_stats.npz).