masterpiece, 1girl, schoolgirl uniform, drawn by a***, t***
Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 2506480929, Size: 512x512, Model hash: 925997e9, ENSD: 31337

This allows processing multiple HN sequentially, or in parallel.

How?

.hns file extension can be used for Hypernetwork Structures. This structure cannot be trained.

.hns file uses standard python syntaxes. Only str, list, set, dict, and tuple is allowed.

Example file

[{('a-1', 0.45):0.5, ('ta', 0.45):0.5}, 'p']

Definition

Tuples ( ) are only used to specify Hypernetwork Strength.

('HYPER', 0.1) Apply HYPER hypernetwork with 0.1 strength

('HYPER') Equals to Singletons, 'HYPER', {'HYPER'}, ['HYPER'], or ('HYPER, 0.1)

Set are used when parallel processing with equal weighted sum is desired.

{"HN A", "HN B", "HN C"} Process HN A(x)+ HN B(x)+ HN C(x) then do average

Dictionaries (or, Maps) are used when parallel processing with Custom weights are desired.
NOTE: Weights sum should be 1, or it will produce weird results.

{"HN A": 0.1, "HN B" : 0.2, "HN C": 0.7} Process HN A(x)*0.1 + HN B(x)*0.2 + HN C(x)* 0.7

String-covered dictionaries / etc
To do some complex processing, dictionaries can be covered by ' or ".

{'{"First A" : 0.5, "First B" : 0.5}' : 0.2, "Next" : 0.8} Process First A and First B as Parallel. Then get the result, do the weighted sum with Next as 0.2, 0.8

As you expect, you can see those files in selection menu, and do processing as if its normal Hypernetwork File.

Fixed compatibility with 3.7 and tested.

Needs check for memory managements