thu-nics MixDQ issues - Githubissues

thu-nics / MixDQ

[ECCV24] MixDQ: Memory-Efficient Few-Step Text-to-Image Diffusion Models with Metric-Decoupled Mixed Precision Quantization

https://a-suozhang.xyz/mixdq.github.io/

25 stars 3 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

How reproduce the FID

#13 suppwilly opened 1 week ago
1
libpeng error for mixed-precision process

#12 Harahan opened 4 weeks ago
1
is MixDQ a PTQ or QAT method？

#11 LiMa-cas opened 1 month ago
6
how many gpu hours it costs during your The Mixed Precision Search Process?

#10 greasebig opened 1 month ago
1
got bad results: quant with 1024 image using sdxl-tubo, using 1.3.1 Normal infer, follow your steps

#9 greasebig opened 1 month ago
17
mat1 and mat2 must have the same dtype, but got Float and Half

#8 greasebig closed 1 month ago
2
FileNotFoundError: [Errno 2] No such file or directory: './mixed_precision_scripts/optimize/tensor_ratio/sdxl_turbo/weight_ratio_config.yaml'

#7 Flying-Cloud closed 1 month ago
1
can your code use w4a8?

#6 greasebig closed 2 months ago
5
AttributeError: 'Linear' object has no attribute 'act_quant'

#5 majiajiong closed 2 months ago
1
Can it be used in Controlnet and IP adapter? Why am I using an error message? Is there a special method for this

#4 libai-lab opened 2 months ago
1
does onnx can support the model an inference with onnxruntime?

#3 dragen1860 opened 2 months ago
2
Missing qdiff folder

#2 colorjam closed 2 months ago
0
Can it be used in sd1.5 and can it be combined with other acceleration methods such as ByteDance/Hyper SD

#1 libai-lab opened 2 months ago
6