From 24768bfbe3996425943133c731ad935d498f2212 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Kamil=20Murat=20Y=C4=B1lmaz?= <kamilmuratyilmaz@gmail.com>
Date: Tue, 27 Feb 2024 21:07:19 +0300
Subject: [PATCH] Kamilmuratyilmaz/fix import error libcuda.so.1 (#11442)
---
.gitignore | 1 +
.../llama_index/llms/vllm/base.py | 11 ++++++-----
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/.gitignore b/.gitignore
index c40424b7fa..aa14bce105 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,6 +2,7 @@
dist/
migration_scripts/
venv/
+.idea
.venv/
.ipynb_checkpoints
.__pycache__
diff --git a/llama-index-integrations/llms/llama-index-llms-vllm/llama_index/llms/vllm/base.py b/llama-index-integrations/llms/llama-index-llms-vllm/llama_index/llms/vllm/base.py
index e08543fb62..5b34ca956f 100644
--- a/llama-index-integrations/llms/llama-index-llms-vllm/llama_index/llms/vllm/base.py
+++ b/llama-index-integrations/llms/llama-index-llms-vllm/llama_index/llms/vllm/base.py
@@ -214,13 +214,14 @@ class Vllm(LLM):
def __del__(self) -> None:
import torch
- from vllm.model_executor.parallel_utils.parallel_state import (
- destroy_model_parallel,
- )
- destroy_model_parallel()
- del self._client
if torch.cuda.is_available():
+ from vllm.model_executor.parallel_utils.parallel_state import (
+ destroy_model_parallel,
+ )
+
+ destroy_model_parallel()
+ del self._client
torch.cuda.synchronize()
def _get_all_kwargs(self, **kwargs: Any) -> Dict[str, Any]:
--
GitLab