EfficientDM
EfficientDM copied to clipboard
ThisisBillhe
[ICLR 2024 Spotlight] This is the official PyTorch implementation of "EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models"
Thank you for your paper! I am very interested in your work and would like to conduct some ablation studies based on your project. However, I couldn't find the implementation...
Thank you for sharing the code! I generated and tested 50,000 samples with the LDM-4 checkpoint that you provided, but I got IS 58.2. I think I did something wrong,...
Hi, I executed the W2A8 ImageNet finetuing using the script by directly use **n_bits_w = 2, n_bits_a = 8**. But it produces unexpected results in the W2A8 setting. Could you...
Hi, sorry to bother you. In the released code of `smaple_lora_int_model.py`, is seems only one image is sampled as a toy example. How to reproduce the results of 500K samples...
Hi, sorry to bother you. In the class `QuantModel_intnlora`, I notice you have used the count number to decide to employ `QuantModule`([this line](https://github.com/ThisisBillhe/EfficientDM/blob/293e212b8c62c356a6da44d6550cd296c87b9f63/quant_scripts/quant_model.py#L94)) or `QuantModule_intnlora` ([this line](https://github.com/ThisisBillhe/EfficientDM/blob/293e212b8c62c356a6da44d6550cd296c87b9f63/quant_scripts/quant_model.py#L96C43-L96C63)). The obtain of...
Hi, Thanks very much for sharing this awesome repo. In the paper, you have given the results on CIFAR-10 with DDIM and ImageNet with LDM-1.4. I only find the code...
Hello, I would like to use your code to quantify sd-v15. Can you provide me with your training log ， I am trying to train models that not with TALSQ，but...
There is dimension error. How can I resolve this? I think it's because context_dim in configs(yaml file) of cin256-v2 model is 512. What is the shape of **cali_y[:32]** in your...
Hi, I see in your code that you have set lr to avg_delta / 2500. How is the 2500 determined?
Hi. After setting both n_bits_w and n_bits_a to 8, when I fine-tune, it fails to generate images properly. Here is the sampled image at epoch 0. Any idea how to...
Metadata
Owner
Metadata
[ICLR 2024 Spotlight] This is the official PyTorch implementation of "EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models"