Ethanfel 65dc549494 feat: add reference audio comparison metrics to LoRA trainer eval ...

New _reference_metrics() computes LSD, MCD, and per-band correlation
between eval samples and the original source audio at each checkpoint.
Loads reference audio once before the training loop and logs metrics
alongside existing spectral metrics.

Also fix batch_size in lora_optimized_dataset.json (4 -> 16).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-04-10 15:04:07 +02:00

..

alpha_scale_sweep.json

feat: add alpha_scale_sweep to fix LoRA noise contamination

2026-04-08 17:55:05 +02:00

bigvgan_disc_fm_retest.json

chore: sanitize tooltips/comments + add experiment configs

2026-04-10 13:44:37 +02:00

bigvgan_overnight.json

feat: add BigVGAN vocoder sweep scheduler node

2026-04-10 02:39:56 +02:00

eval_r128_candidates.json

feat: add eval_r128_candidates.json

2026-04-08 17:28:28 +02:00

lora_optimized_dataset.json

feat: add reference audio comparison metrics to LoRA trainer eval

2026-04-10 15:04:07 +02:00

pissa_sweep.json

feat: add PiSSA/rsLoRA support to scheduler and PiSSA sweep experiment

2026-04-09 22:07:27 +02:00

r64_overnight.json

feat: r64_overnight sweep — focused rank-64 ablation at 8000 steps

2026-04-08 01:32:23 +02:00

r128_sweet_spot.json

feat: add cosine LR decay schedule to trainer and scheduler

2026-04-08 13:25:01 +02:00

ti_sweep_1.json

fix(ti-trainer): clamp token norm to CLIP manifold to prevent buzz artifacts

2026-04-08 23:54:23 +02:00

tier1_sweep.json

fix: point data_dir to features/ subdir where .npz and audio live

2026-04-08 00:45:32 +02:00

tier1_thorough.json

feat: sweep resume + 5 additional experiments (LR, target, extended)

2026-04-08 00:59:16 +02:00

vocoder_finetune.json

fix: use soundfile for WAV/FLAC/OGG to bypass torchcodec/FFmpeg dependency

2026-04-09 15:16:22 +02:00