Add to overload for GGMLTensor, so calling to on the model moves the quantized data.#7949

Merged

psychedelicious merged 12 commits intoinvoke-ai:mainfrom

Vargol:gruff_to

May 19, 2025

Commits on May 19, 2025

Add to overload for GGMLTensor, so calling to on the model moves the quantized data as well

Vargol
authored and
psychedelicious
committed
raise exected exception when attempting to change dtype

Vargol
authored and
psychedelicious
committed
fix missing bracket

Vargol
authored and
psychedelicious
committed
fix picky ruff issue

Vargol
authored and
psychedelicious
committed
revert to overload due to failing tests, use Torch futures instead

Vargol
authored and
psychedelicious
committed
fix offload device

Vargol
authored and
psychedelicious
committed
add check for state_dict, required to load TI's

Vargol
authored and
psychedelicious
committed
Add to overload for GGMLTensor, so calling to on the model moves the quantized data as well

Vargol
authored and
psychedelicious
committed
raise exected exception when attempting to change dtype

Vargol
authored and
psychedelicious
committed
fix missing bracket

Vargol
authored and
psychedelicious
committed
fix picky ruff issue

Vargol
authored and
psychedelicious
committed
fix import ordering, remove code I reverted that the resync added back

Vargol
authored and
psychedelicious
committed