yangchao

yeap: In the training phase, I set batch_size=48, and each device will assign 12 batches about 4GB in my TITAN V (12G) In inference phase, the 'generate_z' operator need repeat...

In CelebA-HQ I got -6.5.

Where can I change this batch-size parameter?

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa8 in position 269495: invalid start byte

Any idea here? or redump and upload your mil_data.zip and tell me your pickle's version.

I also have the same problem, as follow : data/ucf101/jpegs_256/v_YoYo_g10_c05/frame000000.jpg: error '' data/ucf101/jpegs_256/v_FieldHockeyPenalty_g12_c04/frame000000.jpg: error '' data/ucf101/jpegs_256/v_HandstandPushups_g16_c06/frame000000.jpg: error '' data/ucf101/jpegs_256/v_PlayingDaf_g12_c04/frame000000.jpg: error '' data/ucf101/jpegs_256/v_UnevenBars_g16_c04/frame000000.jpg: error '' data/ucf101/jpegs_256/v_HandstandPushups_g20_c04/frame000000.jpg: error '' data/ucf101/jpegs_256/v_TrampolineJumping_g11_c01/frame000000.jpg: error ''...

The work will be very strong if rewrite the twostreamfusion framework using keras. I just want to see the result of the twostreamfusion, and I just want to know how...

I can download from the link

Here are some values of variables in sft.py. When going once gradient optimizing,  batch_size: 1 strategy: ddp accimulation_steps: 8 lr: 2e-5 And if **DO NOT** pass gradient updating (through...

too