Nodes/ComfyUI-LTX-Attention-Toolkit/LTX Attn — Head Freeze
ComfyUI Node

LTX Attn — Head Freeze

A ComfyUI node in g_raw/LTX/Profiler with 7 inputs and 1 output.

By g-raw·Created 2 months ago·Updated 29 days ago· 2
LTX Attn — Head Freeze
  • model
  • steered_model
targets24:8
freeze_from_step3
freeze_step_source3
attn_typesa
blend_weight1.00
store_handle
Categoryg_raw/LTX/Profiler

Inputs (7)

NameTypeDefaultDescription
modelMODEL
targetsSTRING24:8One or more (block, head) pairs to freeze in the same run. Paste Head Candidates' candidates_csv directly (one 'block,head' per line), or type manually as 'block:head | block:head | ...' (or 'block:h1,h2,h3' for several heads of one block in a single entry). Use 'block:all' to freeze every captured head of that block. Leave blank to disable the freeze entirely -- this is the reliable way to turn it off; ComfyUI's node bypass/mute skips this node's cleanup, so the diffusion_model (shared across runs) can be left patched from a previous run.
freeze_from_stepINT30–255
freeze_step_sourceINT30–255
attn_typeCOMBOsa1 options: sa
blend_weightFLOAT1.000–1Shared across all targets — no per-head override yet.
store_handleSTRING

Outputs (1)

NameTypeDescription
steered_modelMODEL