Ethanfel
  • Joined on 2026-01-04
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 00:38:32 +02:00
95923cdf42 feat: add activation steering pipeline (extractor, loader, sampler injection)
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 00:07:59 +02:00
28ee3db337 feat(sampler): add ti_strength blend for TI injection
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:54:25 +02:00
b89167cfae fix(ti-trainer): clamp token norm to CLIP manifold to prevent buzz artifacts
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:42:24 +02:00
f9d092158a fix(ti): lower default lr/batch, add lr_batch sweep group
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:39:33 +02:00
92535deab2 fix(ti-scheduler): save comparison image after each completed experiment
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:33:29 +02:00
0b24207ca5 feat(ti-trainer): generate baseline.wav once before training starts
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:31:54 +02:00
e1a2f0ed7d feat: add inject_mode (suffix/prefix) to TI pipeline
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:20:47 +02:00
f96265da23 feat(ti-trainer): add loss curve IMAGE output
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:14:34 +02:00
c0d95ce356 feat: add ti_sweep_1 experiment file
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:13:06 +02:00
e37bfe1b1c feat: add SelVA TI Scheduler for sweep-based textual inversion experiments
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:10:21 +02:00
bb07bc8169 fix(ti-trainer): guard spectral metrics, drop unused imports
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:08:18 +02:00
e36cdd7947 fix(ti-trainer): fix gradient flow and spectral metric shapes
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 23:01:49 +02:00
e56ece9c1c feat: add SelVA Textual Inversion Trainer and Loader nodes
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 20:28:18 +02:00
eed7eefeac feat: add SelVA HF Smoother and Spectral Matcher preprocessing nodes
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 20:05:47 +02:00
107bb05f17 fix(vae-roundtrip): pass bigvgan path to encoder-only FeaturesUtils
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 19:50:05 +02:00
10e6095e31 fix(vae-roundtrip): use model feature_utils for decode, add normalize/unnormalize, normalize output
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 19:22:11 +02:00
528d33be39 fix: trim/pad latent to seq_cfg.latent_seq_len before decoding
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 19:15:22 +02:00
8195c3114a feat: add SelVA VAE Roundtrip node
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 17:55:07 +02:00
c8e6b91f67 feat: add alpha_scale_sweep to fix LoRA noise contamination
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-08 17:42:58 +02:00
fdce9cbbf1 feat: evaluate adapters on all dataset clips, not just clip_001