[0.11.0][Fix] Fix Qwen2-Audio-7B-Instruct accuracy test (#4018)

### What this PR does / why we need it?

Fix Qwen2-Audio-7B-Instruct accuracy test

Backport:https://github.com/vllm-project/vllm-ascend/pull/4017

Signed-off-by: hfadzxy <starmoon_zhang@163.com>

tests/e2e/models/configs/Qwen2-Audio-7B-Instruct.yaml

@@ -8,3 +8,4 @@ tasks:
   - name: "exact_match,flexible-extract"
     value: 0.45
 num_fewshot: 5
+gpu_memory_utilization: 0.8