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, 17:00 UTC
[CI/Build] Enable InternVL2 PP test only on single node (#8437)
Isotr0py committed
Sep 13, 2024 at 14:35 UTC
9b4a3b235e5bdf0df7901c77a4b01f5358db3638
1 file changed
+5
-4
tests/distributed/test_pipeline_parallel.py
+5
-4
@@ -32,10 +32,11 @@ VLLM_MULTI_NODE = os.getenv("VLLM_MULTI_NODE", "0") == "1"
32
(1, 4, 1, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
33
(2, 2, 1, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
34
(2, 2, 0, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
35
- # TODO: Enable internVL2 in a separate test if needed
36
- # (1, 2, 1, 1, 1, "OpenGVLab/InternVL2-1B", "ray"),
37
- # (1, 2, 1, 1, 1, "OpenGVLab/InternVL2-2B", "ray"),
38
- # (1, 2, 1, 0, 1, "OpenGVLab/InternVL2-4B", "ray"),
35
+ # NOTE: InternVL2 multi-node tests are flaky,
36
+ # use mp backend to skip the multi-node tests
37
+ (1, 2, 1, 1, 1, "OpenGVLab/InternVL2-1B", "mp"),
38
+ (1, 2, 1, 1, 1, "OpenGVLab/InternVL2-2B", "mp"),
39
+ (1, 2, 1, 0, 1, "OpenGVLab/InternVL2-4B", "mp"),
40
],
41
)
42
@fork_new_process_for_each_test