ADE20K batchsize

[RainHxj]

For ADE20K dataset, the batchsize is 32 (4*8gpus) ?

[raoyongming]

@RainHxj Yes, we use 8 gpus in our experiments. The global batch size is 32.

[RainHxj]

Thanks for your reply. The text encoder is frozen in the training phase, which is implemented by lr=0.0. Compared with no grad, does this method affect the training speed?

[raoyongming]

Although the text encoder is fixed during training, we still need to compute its gradients to update the learnable context (the input of the text encoder). Therefore, the torch.no_grad() mode cannot be used to accelerate the training.

[RainHxj]

Thanks. Does the red box denote using a fixed prompt, such as "a photo of"?

[raoyongming]

We use learnable context in this case (Eq. 3 in our paper). Using the hard-craft prompt (a photo of class name) will lead to a slightly worse result.

[RainHxj]

Thanks for your reply.