Actions: sustcsonglin/flash-linear-attention
Actions
Showing runs from all workflows
88 workflow runs
88 workflow runs
max_seqlen
when max_position_embeddings
is None
pr
#30:
Pull request #59
opened
by
zhixuan-lin
Mamba2
] Post Merge Fixes - norm_before_gate
and generation with inputs_embeds
pr
#29:
Pull request #57
opened
by
vasqu
__init__.py
in fla/ops/common
for automatic package discovery
pr
#28:
Pull request #56
reopened
by
yzhangcs
__init__.py
in fla/ops/common
for automatic package discovery
pr
#27:
Pull request #56
opened
by
zhixuan-lin
mamba_chunk_scan_combined
triton kernel by simple_gla
triton kernel
pr
#19:
Pull request #49
synchronize
by
yzhangcs
mamba_chunk_scan_combined
triton kernel by simple_gla
triton kernel
pr
#18:
Pull request #49
synchronize
by
yzhangcs
mamba_chunk_scan_combined
triton kernel by simple_gla
triton kernel
pr
#17:
Pull request #49
opened
by
learning-chip