PR #42168 Add afmoe model

Add AFMoE model support

alyosha-swamy committed 219 days ago

Merge remote-tracking branch 'upstream/main' into add_afmoe_model

alyosha-swamy committed 219 days ago

Address review feedback for AFMoE implementation

alyosha-swamy committed 217 days ago

Add flex attention support to AFMoE model

alyosha-swamy committed 213 days ago

Merge branch 'main' into add_afmoe_model

ArthurZucker committed 212 days ago

Fix expert_bias routing in AFMoE

alyosha-swamy committed 212 days ago

Remove test-results directory

alyosha-swamy committed 210 days ago

Merge branch 'main' into add_afmoe_model

ArthurZucker committed 203 days ago

Address PR review feedback for AFMoE model

alyosha-swamy committed 203 days ago

fix(afmoe): ensure RMSNorm output dtype matches input dtype)

alyosha-swamy committed 203 days ago

properly return attn weights