Enable memory saver for hybrid model (#11974)

This commit is contained in:
Junrong Lin
2025-11-04 14:55:26 +08:00
committed by GitHub
parent f600866a44
commit 173e0f704f
3 changed files with 15 additions and 8 deletions
+2
View File
@@ -42,6 +42,7 @@ class TestMamba(unittest.TestCase):
full_attention_layer_ids=full_attention_layer_ids,
enable_kvcache_transpose=False,
device=device,
enable_memory_saver=False,
mamba_pool=None,
)
assert pool._transfer_full_attention_id(global_interval - 1) == 0
@@ -174,6 +175,7 @@ class TestMamba(unittest.TestCase):
full_attention_layer_ids=full_attention_layer_ids,
enable_kvcache_transpose=False,
device=device,
enable_memory_saver=False,
mamba_pool=req_to_token_pool.mamba_pool,
)