Commits · 609ba690e6b1cdac6b0c6a19b10fd89c620e0733 · Elphel / imagej_elphel_dnn

26 Jun, 2026 5 commits

CLAUDE: default L2 = runs/mexhat_gaps_boost40 (current production L2, was l2_v1) · 609ba690
Andrey Filippov authored Jun 26, 2026
```
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
```
609ba690

CLAUDE: add run_infer_local.sh — local GPU infer_server (no DGX/docker/transfer) · 49975486

Andrey Filippov authored Jun 26, 2026

Run the PyTorch L1+L2 server on the workstation 5060 Ti; Java pipeline points at 127.0.0.1:5577.
Verified: server loads L1(weighted9_pm_s)+L2(l2_v1) on CUDA, warm-up + pyramid build OK.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

49975486

CLAUDE: broaden scope — general PyTorch DNN-companions repo (not CUAS-only) · 9262d943

Andrey Filippov authored Jun 26, 2026

Per Andrey: future DNN companions to imagej-elphel (all PyTorch, not Java) live here too. README + per-file
header tagline generalized accordingly.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

9262d943

CLAUDE: add export_torchscript.py — TorchScript export for native LibTorch inference · 76dbfda2

Andrey Filippov authored Jun 26, 2026

Validated in PoC: L1 (weighted9_pm_s) -> TorchScript -> C++/CUDA on Blackwell matches PyTorch (7.6e-4).
Writes raw-f32 reference vectors for the native probe.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

76dbfda2

CLAUDE: initial import — DNN training/eval/export (migrated from imagej-elphel-internal/c5p_dnn) · 782ef529

Andrey Filippov authored Jun 26, 2026

L1 RawFCN + L2 ConvGRU(torus), synthetic data gen, training/eval, infer_server,
and export_torchscript.py (self-contained TorchScript for native LibTorch inference).
GPLv3 (Elphel norm); headers on all .py/.sh; LICENSE = GPLv3. runs/ checkpoints untracked.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

782ef529