[Model] Add support for MPT (#334)
This commit is contained in:
@@ -395,6 +395,9 @@ void single_query_cached_kv_attention_launcher(
|
||||
case 96:
|
||||
LAUNCH_ATTENTION_KERNEL(T, 96, BLOCK_SIZE, NUM_THREADS);
|
||||
break;
|
||||
case 112:
|
||||
LAUNCH_ATTENTION_KERNEL(T, 112, BLOCK_SIZE, NUM_THREADS);
|
||||
break;
|
||||
case 128:
|
||||
LAUNCH_ATTENTION_KERNEL(T, 128, BLOCK_SIZE, NUM_THREADS);
|
||||
break;
|
||||
|
||||
Reference in New Issue
Block a user