A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 00:30 UTC
[mis][ci/test] fix flaky test in test_sharded_state_loader.py (#5361)
[mis][ci/test] fix flaky test in tests/test_sharded_state_loader.py (#5361)
youkaichao committed
Jun 8, 2024 at 20:50 UTC
5d7e3d0176e0dbcf144c64b7d14d996c55e36c50
1 file changed
+2
-1
tests/test_sharded_state_loader.py
+2
-1
@@ -39,7 +39,8 @@ def test_filter_subtensors():
39
filtered_state_dict = ShardedStateLoader._filter_subtensors(state_dict)
40
assert tuple(filtered_state_dict.keys()) == ("a", "b", "c")
41
for key, tensor in filtered_state_dict.items():
42
- assert tensor.equal(state_dict[key])
42
+ # NOTE: don't use `euqal` here, as the tensor might contain NaNs
43
+ assert tensor is state_dict[key]
44
45
46
@pytest.fixture(scope="module")