nodes/ethanfel-ComfyUI-LoRA-Optimizer
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fix: scoring_speed tooltip — fast is not the default, turbo is

Browse Source 
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Ethanfel
2026-03-07 03:17:36 +01:00
co-authored by Claude Opus 4.6
parent 71d103af30
commit 85ccc52e9f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lora_optimizer.py
+1 -1
View File
@@ -4653,7 +4653,7 @@ class LoRAAutoTuner(LoRAOptimizer):
"tooltip": "Controls how many prefixes Phase 2 scores per candidate. "
"All candidates are scored on the same subset so ranking stays fair.\n\n"
"• full — Score every prefix. Best accuracy, slowest. Use when merging very different LoRAs (e.g. style + character + concept) where block behavior varies a lot.\n"
"• fast — Every 2nd prefix (~50%% faster). Good default for most merges.\n"
"• fast — Every 2nd prefix (~50%% faster). Higher accuracy than turbo, still much faster than full.\n"
"• turbo — Every 3rd prefix (~67%% faster). Works well when your LoRAs have similar conflict across blocks (e.g. multiple characters from the same trainer).\n"
"• turbo+ — Every 4th prefix (~75%% faster). Best for large models (DiT/Flux/WAN) or when iterating quickly. May miss subtle block-level differences on SD/SDXL."
}),