theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

support qwen-image-distill-dmd2-lora

Browse Source
This commit is contained in:
Artiprocher
2025-09-16 19:43:07 +08:00
parent d93e8738cd
commit 1a7f482fbd
3 changed files with 33 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
diffsynth/pipelines/qwen_image.py
View File

@@ -371,6 +371,7 @@ class QwenImagePipeline(BasePipeline):
rand_device: str = "cpu",
# Steps
num_inference_steps: int = 30,
exponential_shift_mu: float = None,
# Blockwise ControlNet
blockwise_controlnet_inputs: list[ControlNetInput] = None,
# EliGen
@@ -393,7 +394,7 @@ class QwenImagePipeline(BasePipeline):
progress_bar_cmd = tqdm,
):
# Scheduler
self.scheduler.set_timesteps(num_inference_steps, denoising_strength=denoising_strength, dynamic_shift_len=(height // 16) * (width // 16))
self.scheduler.set_timesteps(num_inference_steps, denoising_strength=denoising_strength, dynamic_shift_len=(height // 16) * (width // 16), exponential_shift_mu=exponential_shift_mu)
# Parameters
inputs_posi = {

9
diffsynth/schedulers/flow_match.py
View File

@@ -31,7 +31,7 @@ class FlowMatchScheduler():
self.set_timesteps(num_inference_steps)
def set_timesteps(self, num_inference_steps=100, denoising_strength=1.0, training=False, shift=None, dynamic_shift_len=None):
def set_timesteps(self, num_inference_steps=100, denoising_strength=1.0, training=False, shift=None, dynamic_shift_len=None, exponential_shift_mu=None):
if shift is not None:
self.shift = shift
sigma_start = self.sigma_min + (self.sigma_max - self.sigma_min) * denoising_strength
@@ -42,7 +42,12 @@ class FlowMatchScheduler():
if self.inverse_timesteps:
self.sigmas = torch.flip(self.sigmas, dims=[0])
if self.exponential_shift:
mu = self.calculate_shift(dynamic_shift_len) if dynamic_shift_len is not None else self.exponential_shift_mu
if exponential_shift_mu is not None:
mu = exponential_shift_mu
elif dynamic_shift_len is not None:
mu = self.calculate_shift(dynamic_shift_len)
else:
mu = self.exponential_shift_mu
self.sigmas = math.exp(mu) / (math.exp(mu) + (1 / self.sigmas - 1))
else:
self.sigmas = self.shift * self.sigmas / (1 + (self.shift - 1) * self.sigmas)