support chunked prefill and fix minor bug

2026-04-14 02:47:35 +08:00
parent 9e8507ef41
commit 8d63a98c03
8 changed files with 65 additions and 53 deletions
@@ -54,8 +54,6 @@ def get_rope(
    rotary_dim: int,
    max_position: int,
    base: float,
-    rope_scaling: dict | None = None,
 ):
-    assert rope_scaling is None
    rotary_emb = RotaryEmbedding(head_size, rotary_dim, max_position, base)
    return rotary_emb