diff options
| author | Michael Feil <63565275+michaelfeil@users.noreply.github.com> | 2024-12-31 09:41:23 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-12-31 09:41:23 +0100 |
| commit | a594ef669ca5ed82c1f19d2230b4b3dc9cb46f43 (patch) | |
| tree | 8647429f4c0ae7fddbae84a1936819f0c0172514 /candle-flash-attn/src/ffi.rs | |
| parent | 71cd6d55337b1541f602c1afffa6baf6dd75b09c (diff) | |
| download | candle-a594ef669ca5ed82c1f19d2230b4b3dc9cb46f43.tar.gz candle-a594ef669ca5ed82c1f19d2230b4b3dc9cb46f43.tar.bz2 candle-a594ef669ca5ed82c1f19d2230b4b3dc9cb46f43.zip | |
Flash-Attn upgrade / SoftCap Candle-FlashAttn [2/n] (#2689)
* update flash-attn v1
* restore: hdim224
* add 224 flash_fwd_template
* remove whitespace
* softcap is working, including test and api.
* make softcap test case better
---------
Co-authored-by: laurent <laurent.mazare@gmail.com>
Diffstat (limited to 'candle-flash-attn/src/ffi.rs')
| -rw-r--r-- | candle-flash-attn/src/ffi.rs | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/candle-flash-attn/src/ffi.rs b/candle-flash-attn/src/ffi.rs index ca65520b..47e54e2a 100644 --- a/candle-flash-attn/src/ffi.rs +++ b/candle-flash-attn/src/ffi.rs @@ -45,6 +45,8 @@ extern "C" { window_size_left: c_int, window_size_right: c_int, + + softcap: f32, ); } |