Ludan Ruan
Ludan Ruan
The results in the paper `FVD=117.20, KVD=5.78, FAD=10.72 `are produced by **DDPM sampling** (1000 steps) with a video size of 64x64. The default sampling method in multimodal_sample_sr.sh is **DPM solver**,...
Sorry， it should be "guided-diffusion_64_256_upsampler.pt". README and scripts have been updated.
Training multimodal-generation model requires no initialization, it has been updated now.
The batchsize aims at one GPU. For example, set "--GPU 0,1,2,3,4,5,6,7 mpiexec -n 8 python..." , the total batchsize equals 4\*8=32. Our training requires 4 Nodes, that is 32\*GPUs, you...
In my experiments, updating to 50000 steps will have meaningful results. I recomand you to set `--save_interval 10000 ` to save the storage. Set` --sample_fn ddpm` to test the intermediate...
In my experiments, 50,000 iter brings meaningful results. ---- Replied Message ---- | From | Ahmet Selim ***@***.***> | | Date | 06/19/2023 03:47 | | To | ***@***.***> |...
In my experiments setting(32x32GV100), 30,000 iter takes one day. In other words, I get meaningful results?within 2 days. ---- Replied Message ---- | From | Ahmet Selim ***@***.***> | |...
Thanks for interesting in our paper, We will release the code and pretrained models once the paper is accepted.