Skip to content

Commit

Permalink
Reverting layer changes
Browse files Browse the repository at this point in the history
  • Loading branch information
sroy745 committed Oct 28, 2024
1 parent d99370c commit 11bda4f
Showing 1 changed file with 0 additions and 2 deletions.
2 changes: 0 additions & 2 deletions vllm/attention/layer.py
Original file line number Diff line number Diff line change
Expand Up @@ -78,11 +78,9 @@ def __init__(
# During model initialization, the default dtype is set as the model
# weight and activation dtype.
dtype = torch.get_default_dtype()
print('dtype ' + str(dtype))
attn_backend = get_attn_backend(head_size, dtype, kv_cache_dtype,
block_size, is_attention_free,
blocksparse_params is not None)
print('attn_backend ' + str(attn_backend))
impl_cls = attn_backend.get_impl_cls()
self.impl = impl_cls(num_heads, head_size, scale, num_kv_heads,
alibi_slopes, sliding_window, kv_cache_dtype,
Expand Down

0 comments on commit 11bda4f

Please sign in to comment.