init v0.23.0

Signed-off-by: Sun Ruoxi <sunruoxi@4paradigm.com>

tests/e2e/models/configs/Qwen3-VL-8B-Instruct-W8A8.yaml

@@ -0,0 +1,22 @@
+model_name: "vllm-ascend/Qwen3-VL-8B-Instruct-W8A8"
+model_type: "vllm-vlm"
+hardware: "Atlas A2 Series"
+
+serve:
+  tensor_parallel_size: 1
+  dtype: auto
+  max_model_len: 8192
+  gpu_memory_utilization: 0.8
+  trust_remote_code: false
+  quantization: ascend
+
+tasks:
+- name: "mmmu_val"
+  metrics:
+  - name: "acc,none"
+    value: 0.52
+
+num_fewshot: 0
+apply_chat_template: true
+fewshot_as_multiturn: false
+batch_size: 32