Skip to content

[Misc] Fix ScalarType float4 naming #17690

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
merged 1 commit into from
May 6, 2025
Merged

[Misc] Fix ScalarType float4 naming #17690

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
2 changes: 1 addition & 1 deletion tests/kernels/quantization/test_nvfp4_quant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
SEEDS = [42]
CUDA_DEVICES = ['cuda:0']

FLOAT4_E2M1_MAX = scalar_types.float4_e2m1fn.max()
FLOAT4_E2M1_MAX = scalar_types.float4_e2m1f.max()
FLOAT8_E4M3_MAX = torch.finfo(torch.float8_e4m3fn).max

# E2M1 to float
Expand Down
2 changes: 1 addition & 1 deletion tests/test_scalartype.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@
(0, 15, scalar_types.uint4),
(-8, 7, scalar_types.uint4b8),
(-128, 127, scalar_types.uint8b128),
(-6., 6., scalar_types.float4_e2m1fn),
(-6., 6., scalar_types.float4_e2m1f),
(-28., 28., scalar_types.float6_e3m2f),
(torch.int8, scalar_types.int8),
(torch.uint8, scalar_types.uint8),
Expand Down
2 changes: 1 addition & 1 deletion vllm/scalar_type.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -322,7 +322,7 @@ class scalar_types:
float6_e3m2f = ScalarType.float_(3, 2, True, NanRepr.NONE)

# fp4, https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf
float4_e2m1fn = ScalarType.float_(2, 1, True, NanRepr.NONE)
float4_e2m1f = ScalarType.float_(2, 1, True, NanRepr.NONE)

# "gptq" types
uint2b2 = ScalarType.uint(2, 2)
Expand Down