Modify unit test

Author: jeejeelee

tests/lora/test_minicpmv.py

@@ -56,7 +56,7 @@ def do_sample(llm: vllm.LLM, lora_path: str, lora_id: int) -> List[str]:
 def test_minicpmv_lora(minicpmv_lora_files):
     llm = vllm.LLM(
         MODEL_PATH,
-        max_model_len=1024,
+        max_num_seqs=2,
         enable_lora=True,
         max_loras=4,
         max_lora_rank=64,

tests/lora/test_minicpmv_tp.py

@@ -13,8 +13,7 @@
 PROMPT_TEMPLATE = (
     "<|begin_of_text|><|start_header_id|>user<|end_header_id|>\n\n"
     "(<image>./</image>)\nWhat is in the image?<|eot_id|>"
-    "<|start_header_id|>assistant<|end_header_id|>\n\n"
-)
+    "<|start_header_id|>assistant<|end_header_id|>\n\n")
 
 IMAGE_ASSETS = [
     ImageAsset("stop_sign"),
@@ -35,20 +34,18 @@ def do_sample(llm: vllm.LLM, lora_path: str, lora_id: int) -> List[str]:
         stop_token_ids=[128001, 128009],  # eos_id, eot_id
     )
 
-    inputs = [
-        {
-            "prompt": PROMPT_TEMPLATE,
-            "multi_modal_data": {"image": asset.pil_image},
-        }
-        for asset in IMAGE_ASSETS
-    ]
+    inputs = [{
+        "prompt": PROMPT_TEMPLATE,
+        "multi_modal_data": {
+            "image": asset.pil_image
+        },
+    } for asset in IMAGE_ASSETS]
 
     outputs = llm.generate(
         inputs,
         sampling_params,
         lora_request=LoRARequest(str(lora_id), lora_id, lora_path)
-        if lora_id
-        else None,
+        if lora_id else None,
     )
     # Print the outputs.
     generated_texts: List[str] = []
@@ -59,13 +56,14 @@ def do_sample(llm: vllm.LLM, lora_path: str, lora_id: int) -> List[str]:
         print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
     return generated_texts
 
+
 @multi_gpu_test(num_gpus=2)
 @pytest.mark.parametrize("fully_sharded", [True, False])
 def test_minicpmv_tp2(minicpmv_lora_files, fully_sharded):
     llm = vllm.LLM(
         MODEL_PATH,
         enable_lora=True,
-        max_num_seqs=16,
+        max_num_seqs=2,
         max_loras=4,
         max_lora_rank=64,
         tensor_parallel_size=2,
@@ -85,7 +83,7 @@ def test_minicpmv_tp4(minicpmv_lora_files, fully_sharded):
     llm = vllm.LLM(
         MODEL_PATH,
         enable_lora=True,
-        max_num_seqs=16,
+        max_num_seqs=2,
         max_loras=4,
         max_lora_rank=64,
         tensor_parallel_size=4,