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 16, 10:45 UTC
Fix failing spec decode test (#9054)
sroy745 committed
Oct 3, 2024 at 16:07 UTC
91add85ec409a3628d01a1e4d4b3230e0fd3aa3f
1 file changed
+1
tests/spec_decode/e2e/test_compatibility.py
+1
@@ -100,6 +100,7 @@ def test_spec_decode_xfail_spec_max_model_len(test_llm_generator):
100
"model": "JackFram/llama-68m",
101
"speculative_model": "JackFram/llama-68m",
102
"num_speculative_tokens": 5,
103
+ "use_v2_block_manager": False,
104
}])
105
@pytest.mark.parametrize("per_test_common_llm_kwargs", [{}])
106
@pytest.mark.parametrize("test_llm_kwargs", [{}])