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 15, 14:45 UTC
[CI][Bugfix] Pin triton version for CPU (#16384)
Signed-off-by: Roger Wang <ywang@roblox.com>
Roger Wang committed
Apr 9, 2025 at 21:35 UTC
0d4d06fe2fe248dbd5c4a360e8c4ffa3b386279c
1 file changed
+3
requirements/cpu.txt
+3
@@ -15,3 +15,6 @@ torchaudio==2.6.0; platform_machine == "ppc64le"
15
torchvision; platform_machine != "ppc64le" and platform_machine != "s390x"
16
torchvision==0.21.0; platform_machine == "ppc64le"
17
datasets # for benchmark scripts
18
+
19
+# cpu cannot use triton 3.3.0
20
+triton==3.2.0