Skip to content

Fix rope_theta arg for diffusers_attention #4656

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mrwyattii merged 2 commits into from
Nov 8, 2023
Merged

Fix rope_theta arg for diffusers_attention #4656

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d09b5cd
Fix rope_theta arg for diffusers_attention
lekurile Nov 8, 2023
1434340
Merge branch 'master' into lekurile/fix_rope_theta
mrwyattii Nov 8, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions deepspeed/ops/transformer/inference/diffusers_attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ class DeepSpeedDiffusersAttentionFunction(Function):
@staticmethod
def forward(ctx, input, context, input_mask, config, attn_qkvw, attn_qw, attn_kw, attn_vw, attn_qkvb,
num_attention_heads_per_partition, norm_factor, hidden_size_per_partition, attn_ow, attn_ob,
do_out_bias, score_context_func, linear_func, triton_flash_attn_kernel):
do_out_bias, score_context_func, linear_func, triton_flash_attn_kernel, rope_theta):

def _transpose_for_context(x):
x = x.permute(0, 2, 1, 3)
Expand All @@ -59,7 +59,7 @@ def selfAttention_fp(input, context, input_mask):
scale = (1 / norm_factor) * (1 / norm_factor)
if do_flash_attn and context is None:
qkv_out = linear_func(input, attn_qkvw, attn_qkvb if attn_qkvb is not None else attn_qkvw, attn_qkvb
is not None, do_flash_attn, config.heads, False)
is not None, do_flash_attn, config.heads, False, rope_theta)

context_layer = triton_flash_attn_kernel(qkv_out[0], qkv_out[1], qkv_out[2], scale,
input.shape[-2] % 128 == 0)
Expand All @@ -81,7 +81,7 @@ def selfAttention_fp(input, context, input_mask):
attention_scores = (torch.matmul(query, key.transpose(-1, -2)) * scale).softmax(dim=-1)
context_layer = _transpose_for_context(torch.matmul(attention_scores, value))

output = linear_func(context_layer, attn_ow, attn_ob, do_out_bias, False, config.heads, False)
output = linear_func(context_layer, attn_ow, attn_ob, do_out_bias, False, config.heads, False, rope_theta)
return output

output = selfAttention_fp(input, context, input_mask)
Expand Down Expand Up @@ -191,6 +191,6 @@ def forward(self, input, context=None, input_mask=None):
self.num_attention_heads_per_partition, self.norm_factor,
self.hidden_size_per_partition, self.attn_ow, self.attn_ob,
self.do_out_bias, self.score_context_func, self.linear_func,
self.triton_flash_attn_kernel)
self.triton_flash_attn_kernel, self.config.rope_theta)

return output