Nsa trtllm mla sparse fp8 support with Deepseek v3.2 NVFP4 (#18389)
This commit is contained in:
@@ -50,6 +50,7 @@ class TestNSAHiCacheTransfer(unittest.TestCase):
|
||||
layer_num=layer_num,
|
||||
device="cuda",
|
||||
enable_memory_saver=False,
|
||||
kv_cache_dim=576,
|
||||
index_head_dim=128,
|
||||
)
|
||||
pin_memory = io_backend == "kernel"
|
||||
|
||||
Reference in New Issue
Block a user