@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / fc93e56143
Fork 0

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, 18:15 UTC

[Bugfix][TPU] Correct env variable for XLA cache path (#7544)

Woosuk Kwon committed Aug 15, 2024 at 00:02 UTC fc93e5614374688bddc432279244ba7fbf8169c2
1 file changed +1 -1
vllm/envs.py
+1 -1
@@ -334,7 +334,7 @@ environment_variables: Dict[str, Callable[[], Any]] = {
334 "VLLM_XLA_CACHE_PATH":
335 lambda: os.path.expanduser(
336 os.getenv(
337 - "VLLM_ASSETS_CACHE",
337 + "VLLM_XLA_CACHE_PATH",
338 os.path.join(get_default_cache_root(), "vllm", "xla_cache"),
339 )),
340 "VLLM_FUSED_MOE_CHUNK_SIZE":