Are sliding window attention (SWA) and/or AutoFit enabled by default or not?

[alex-ie]

I usually run kcpp linux64 and leave all on defaults, changing only context size and now after TurboQuant tend to set KV at Q4. In the launcher I see AutoFit and SWA checkboxes are empty by default in v1.111.2 (running kcpp without arguments from command line in bash). In model load log (terminal output) I see for e.g. Gemma 4 E4B
creating SWA KV cache, size =

So I don't understand, is SWA enabled when there is such line in log or not? What does above line mean?
I've asked Gemma 4 E4B Q4_K_M about lines from the log - it tells it means rebuilding KV cache for SWA (it told it's Structured / Sparse Weight Averaging).

For AutoFit https://github.com/LostRuins/koboldcpp/wiki

Using --autofit ... also enabled if you set --gpulayers to -1 and do not set any incompatible flags (Autofit is not compatible with manual tensor overrides, tensor splits or --moecpu).

Is 'Force AutoFit' checkbox in the launcher same as --autofit? If yes, why is it not ON by default when I start the launcher? The launcher shows 'GPU Layers' box = -1 by default.

TIA

[LostRuins]

SWA is disabled by default, to enable it you use the flag --useswa or the checkbox in the launcher