Krea2ModelLoader / Krea2LoRA / Krea2Generate / Krea2Unload nodes running Krea-2-Turbo on Apple MLX in-process, returning ComfyUI IMAGE tensors. The krea2_engine/ pipeline is vendored directly in this repo (no external git dependency) — see NOTICE for the full attribution chain back to the original Krea 2 model and the MLX port it builds on. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
16 lines
587 B
Python
16 lines
587 B
Python
"""Krea-2-Turbo — Apple MLX inference engine, vendored into this node package.
|
|
|
|
Pure-MLX, numerically validated bit-faithful to the PyTorch reference. A modified derivative of
|
|
krea/Krea-2-Turbo under the Krea 2 Community License — see ../LICENSE and ../NOTICE.
|
|
"""
|
|
|
|
from .quant_recipes import mixed_4_8, quantize_bulk
|
|
from .sampling import sample, to_pil
|
|
from .text_encoder import Qwen3VLConditioner
|
|
from .transformer import Krea2Config, SingleStreamDiT
|
|
|
|
__all__ = [
|
|
"Krea2Config", "SingleStreamDiT", "Qwen3VLConditioner",
|
|
"sample", "to_pil", "quantize_bulk", "mixed_4_8",
|
|
]
|