Skip to content

Commit eaba4fc

Browse files
Update tests/v1/worker/test_gpu_model_runner.py
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
1 parent b38a35b commit eaba4fc

File tree

1 file changed

+6
-1
lines changed

1 file changed

+6
-1
lines changed

tests/v1/worker/test_gpu_model_runner.py

Lines changed: 6 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -825,7 +825,12 @@ def test_hybrid_attention_mamba_tensor_shapes(monkeypatch):
825825
vllm_ctx[layer].kv_cache[0][1][
826826
blocks1, :] = ssm_blocks_constant.detach().clone()
827827

828-
# verify attention contents is unchanged
828+
# verify attention and mamba contents are correct
829829
for layer in [layer_0, layer_1]:
830830
assert torch.equal(vllm_ctx[layer].kv_cache[0][blocks0, :],
831831
attn_blocks_constant)
832+
for layer in [layer_2, layer_3, layer_4, layer_5]:
833+
assert torch.equal(vllm_ctx[layer].kv_cache[0][0][blocks1, :],
834+
conv_blocks_constant)
835+
assert torch.equal(vllm_ctx[layer].kv_cache[0][1][blocks1, :],
836+
ssm_blocks_constant)

0 commit comments

Comments
 (0)