Manoj Bhat Manojbhat09
Manojbhat09
/ flux_caching.py
Last active
September 21, 2025 01:41
Minimal high-performance residual-based caching for DiT based diffusion models that reduces inference time by 30-70% through reuse of outputs on similarity
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # caching.py | |
| import functools | |
| import unittest | |
| import contextlib | |
| import dataclasses | |
| from collections import defaultdict | |
| from typing import DefaultDict, Dict | |
| import torch | |
| from diffusers import DiffusionPipeline, FluxTransformer2DModel |
OlderNewer