Add ulysses to supported attention comment in WAN configs

Author: csgoogle

src/maxdiffusion/configs/base_wan_14b.yml

@@ -60,7 +60,7 @@ jit_initializers: True
 # Set true to load weights from pytorch
 from_pt: True
 split_head_dim: True
-attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring
+attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring, ulysses
 flash_min_seq_length: 0
 
 # If mask_padding_tokens is True, we pass in segment ids to splash attention to avoid attending to padding tokens.

src/maxdiffusion/configs/base_wan_1_3b.yml

@@ -60,7 +60,7 @@ jit_initializers: True
 # Set true to load weights from pytorch
 from_pt: True
 split_head_dim: True
-attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring
+attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring, ulysses
 flash_min_seq_length: 0
 
 # If mask_padding_tokens is True, we pass in segment ids to splash attention to avoid attending to padding tokens.

src/maxdiffusion/configs/base_wan_27b.yml

@@ -60,7 +60,7 @@ jit_initializers: True
 # Set true to load weights from pytorch
 from_pt: True
 split_head_dim: True
-attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring
+attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring, ulysses
 flash_min_seq_length: 4096
 # If mask_padding_tokens is True, we pass in segment ids to splash attention to avoid attending to padding tokens.
 # Else we do not pass in segment ids and on vpu bound hardware like trillium this is faster.

src/maxdiffusion/configs/base_wan_i2v_14b.yml

@@ -60,7 +60,7 @@ jit_initializers: True
 # Set true to load weights from pytorch
 from_pt: True
 split_head_dim: True
-attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring
+attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring, ulysses
 flash_min_seq_length: 4096
 dropout: 0.0
 

src/maxdiffusion/configs/base_wan_i2v_27b.yml

@@ -60,7 +60,7 @@ jit_initializers: True
 # Set true to load weights from pytorch
 from_pt: True
 split_head_dim: True
-attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring
+attention: 'flash' # Supported attention: dot_product, flash, cudnn_flash_te, ring, ulysses
 flash_min_seq_length: 4096
 dropout: 0.0