squall/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[TPU] Update pallas.py to support trillium (#8871)

Browse Source
This commit is contained in:
Brittany 2024-09-27 01:16:55 -07:00 committed by GitHub
parent 6d792d2f31
commit 8df2dc3c88
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/attention/backends/pallas.py
View File

@ -130,7 +130,7 @@ class PallasAttentionBackendImpl(AttentionImpl):
assert tpu_type is not None assert tpu_type is not None
tpu_type = tpu_type.lower() tpu_type = tpu_type.lower()
if "lite" not in tpu_type: if (("lite" not in tpu_type) and ("v6" not in tpu_type)):
if self.num_kv_heads % 2 == 0: if self.num_kv_heads % 2 == 0:
self.megacore_mode = "kv_head" self.megacore_mode = "kv_head"
else: else: