[Docs] add Qwen25vl yaml (#4662)

* Add ci case for min token and max token

* 【CI case】include total_tokens in the last packet of completion interface stream output

* 【CE】add qwen25-vl

* 【CE】add qwen25-vl

---------

Co-authored-by: xujing43 <xujing43@baidu.com>

Author: xjkmfa

benchmarks/yaml/qwen25_7b-vl-32k-bf16.yaml

@@ -0,0 +1,6 @@
+max_model_len: 32768
+max_num_seqs: 128
+gpu_memory_utilization: 0.85
+tensor_parallel_size: 1
+limit_mm_per_prompt: '{"image": 100, "video": 100}'
+enable_mm: True

benchmarks/yaml/request_yaml/qwen25-vl-32k.yaml

@@ -0,0 +1,8 @@
+top_p: 0.8
+temperature: 0.7
+metadata:
+  min_tokens: 1
+max_tokens: 32768
+repetition_penalty: 1.05
+frequency_penalty: 0
+presence_penalty: 0