Ethanfel
0 Followers · 0 Following
  • Joined on 2026-01-04
Repositories
30
 Projects Packages Public Activity Starred Repositories
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 19:36:06 +02:00
1d1ae61409 fix: move only VAE+vocoder to GPU during eval to prevent device mismatch
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:57:12 +02:00
8fa2699551 fix: correct DITTO reference latent space mismatch
Ethanfel pushed to master at Ethanfel/Comfyui-Nodes-Stats 2026-04-09 18:47:18 +02:00
4ccc772bb9 chore: bump version to 1.1.0
Ethanfel pushed to master at Ethanfel/Comfyui-Nodes-Stats 2026-04-09 18:43:25 +02:00
777e4890ae docs: update README for model tracking feature
9477777124 fix: scope input-scanning loop inside per-node try/except in extract_models_from_prompt
831ed302e9 fix: update model_type on upsert conflict and scope close button to dialog
c7003df0bb fix: scope badge querySelector to dialog, update dialog title
bf5c8683a7 feat: add Models tab with per-type usage stats
Compare 16 commits »
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:40:22 +02:00
14fabf01f9 fix: reduce opt_lr step to 0.001 to allow finer lr control in DITTO
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:30:07 +02:00
445da1e69b fix: replace std clamp with anchor regularization to prevent OOD noise
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:23:41 +02:00
fa6c4fa834 fix: clamp x0 std after each optimizer step to prevent OOD noise
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:18:42 +02:00
286681edff fix: cast mel to model dtype before VAE encode in DITTO reference loading
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:15:29 +02:00
056a7b973d fix: enable VAE encoder in model loader — required for DITTO reference encoding
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:12:33 +02:00
633fe36fbb fix: compute DITTO style loss in latent space to eliminate VAE decoder noise
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:10:12 +02:00
8862089fd0 fix: remove 32-clip cap on DITTO reference loading — use all available clips
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 18:03:33 +02:00
608e7df04b feat: add gram_weight param to DITTO, reduce default style_weight to 0.1
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:57:02 +02:00
101b1bdb41 fix: _do_optimize returns dict not tuple — prevent double-wrapping AUDIO output
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:48:42 +02:00
732df151b0 fix: cast ref_mean/ref_gram to model dtype before loss computation
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:44:37 +02:00
817b75df49 fix: bypass @torch.inference_mode() on decode to preserve gradient chain
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:40:02 +02:00
1f02d73a3e fix: remove checkpoint wrapper on decode — direct call preserves grad chain
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:35:16 +02:00
fb255edaf0 fix: strip inference-mode tensor flags in DITTO before conditions computation
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 17:18:39 +02:00
8ccc2438e4 fix: remove FlashSR (audiosr incompatible with Python 3.12), add training loss CSV
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 16:39:05 +02:00
8371466e44 fix: guarantee length preservation in _ActivationWithGAFilter
Ethanfel pushed to feature/lora-timestep-sampling at Ethanfel/ComfyUI-SelVA 2026-04-09 16:32:03 +02:00
ba0499b77c fix: FlashSR device handling and remove unused tmp_out