diff options
author | Zack Angelo <zackangelo@gmail.com> | 2024-10-23 11:07:09 -0700 |
---|---|---|
committer | GitHub <noreply@github.com> | 2024-10-23 20:07:09 +0200 |
commit | a2e9d41b2062be5b45c84d24fe2bf4527ec27cee (patch) | |
tree | f587b1c8dc547d2213076adc653505df0f116711 /candle-book | |
parent | 7c09215ef443256523d2de2579db56d1b59fd683 (diff) | |
download | candle-a2e9d41b2062be5b45c84d24fe2bf4527ec27cee.tar.gz candle-a2e9d41b2062be5b45c84d24fe2bf4527ec27cee.tar.bz2 candle-a2e9d41b2062be5b45c84d24fe2bf4527ec27cee.zip |
use softmax_last_dim (metal and cuda kernel) in llama attention layer (#2572)
Diffstat (limited to 'candle-book')
0 files changed, 0 insertions, 0 deletions