Hugging Face
Models
Datasets
Spaces
Buckets
new
Docs
Enterprise
Pricing
Log In
Sign Up
Motif-Technologies
/
activation
like
30
Follow
Motif Technologies
155
Kernels
License:
apache-2.0
Model card
Files
Files and versions
xet
Community
3
Use this model
main
activation
/
tests
63.9 kB
Ctrl+K
Ctrl+K
11 contributors
History:
24 commits
Jangwoong Kim
Merge pull request #22 from MotifTechnologies/jangwoong/mla-rope-fa4-port
5adea7d
unverified
about 15 hours ago
__init__.py
Safe
0 Bytes
feat(poly-norm): Add PolyNorm
11 months ago
allclose_default.py
Safe
512 Bytes
Add fusion (#3)
7 months ago
pytest.ini
71 Bytes
fix: unify all backward kernels to input-based math + fix test import
3 days ago
test_fused_add_rms_norm.py
Safe
3.6 kB
refactor(activation): change fused_add_rms_norm and fused_add_rms_norm_backward to out-place operations
6 months ago
test_fused_add_rms_norm_sequence_parallel.py
Safe
5.85 kB
fix: fix fused add rms norm sharding strategy
5 months ago
test_fused_mul_grouped_poly_norm.py
Safe
32.4 kB
style: fix yapf/isort formatting for CI --all-files check
3 days ago
test_fused_mul_poly_norm.py
Safe
4.13 kB
Add fusion (#3)
7 months ago
test_mla_rope_grad.py
5.96 kB
test: numerical parity for MLA RoPE fused kernels vs PyTorch reference
about 22 hours ago
test_poly_norm.py
Safe
2.55 kB
Add fusion (#3)
7 months ago
test_rms_norm.py
Safe
1.94 kB
feat: make rms_norm as out-place
7 months ago
test_rms_norm_sequence_parallel.py
Safe
4.37 kB
fix: fix rms norm sharding strategy
5 months ago
utils.py
Safe
2.56 kB
Add fusion (#3)
7 months ago