Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
OpenDAS
vllm_cscc
Commits
1c160841
Unverified
Commit
1c160841
authored
Oct 22, 2025
by
Wentao Ye
Committed by
GitHub
Oct 22, 2025
Browse files
[Bug] Fix DeepSeek-V2.5-1210-FP8 issue (#27267)
Signed-off-by:
yewentao256
<
zhyanwentao@126.com
>
parent
4ca13a86
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
11 additions
and
5 deletions
+11
-5
vllm/model_executor/layers/fused_moe/cutlass_moe.py
vllm/model_executor/layers/fused_moe/cutlass_moe.py
+11
-5
No files found.
vllm/model_executor/layers/fused_moe/cutlass_moe.py
View file @
1c160841
...
@@ -511,13 +511,19 @@ def cutlass_moe_fp8(
...
@@ -511,13 +511,19 @@ def cutlass_moe_fp8(
assert
quant_config
is
not
None
assert
quant_config
is
not
None
if
quant_config
.
a1_scale
is
not
None
:
if
quant_config
.
a1_scale
is
not
None
:
assert
quant_config
.
per_act_token_quant
==
quant_config
.
a1_scale
.
numel
()
!=
1
assert
quant_config
.
per_act_token_quant
==
(
quant_config
.
a1_scale
.
numel
()
!=
1
)
if
quant_config
.
a2_scale
is
not
None
:
if
quant_config
.
a2_scale
is
not
None
:
assert
quant_config
.
per_act_token_quant
==
quant_config
.
a2_scale
.
numel
()
!=
1
assert
quant_config
.
per_act_token_quant
==
(
quant_config
.
a2_scale
.
numel
()
!=
1
)
assert
quant_config
.
w1_scale
is
None
or
(
if
quant_config
.
w1_scale
is
not
None
:
quant_config
.
per_out_ch_quant
==
(
quant_config
.
w1_scale
.
size
(
1
)
==
w1_q
.
size
(
1
))
if
quant_config
.
per_out_ch_quant
:
)
assert
quant_config
.
w1_scale
.
dim
()
>
1
and
quant_config
.
w1_scale
.
size
(
1
)
==
w1_q
.
size
(
1
)
else
:
assert
(
quant_config
.
w1_scale
.
dim
()
==
1
or
quant_config
.
w1_scale
.
size
(
1
)
==
1
)
num_experts
=
global_num_experts
if
global_num_experts
!=
-
1
else
w1_q
.
size
(
0
)
num_experts
=
global_num_experts
if
global_num_experts
!=
-
1
else
w1_q
.
size
(
0
)
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment