Tags: fla-org/flash-linear-attention
Toggle v0.5.1's commit message
[Fix] FLA cache window updates and reset behavior (#910 )
* fix fla cache window updates and reset behavior
* fix legacy cache window update regression
* [Fix] Tidy cache window/reset patch: drop quote churn, fix typos, rename tests
---------
Co-authored-by: Michael-Rusu <89427972+Michael-Rusu@users.noreply.github.com>
Co-authored-by: Yu Zhang <yzhang.cs@outlook.com>
Toggle v0.5.0's commit message
[KDA][TileLang] Add TileLang backend for chunk_kda_bwd_wy_dqkg_fused (#…
…854 )
Toggle v0.4.2's commit message
[Model] Unify cache function (#777 )
Toggle v0.4.1's commit message
[GSA] Fix gate oob bugs (#694 )
Refactor calculations involving 'b_do' and 'b_A' to improve clarity and correctness.
Toggle v0.4.0's commit message
Toggle v0.3.2's commit message
Toggle v0.3.1's commit message
[Mamba] Fix errors in Triton backend (#576 )
Toggle v0.3.0's commit message
[README] Update model list
Toggle v0.2.2's commit message
[ShortConv] Add triton kernels for inference
Toggle v0.2.1's commit message
You can’t perform that action at this time.