[1/n][CI] Load models in CI from S3 instead of HF (#13205)

Signed-off-by: <>
Co-authored-by: EC2 Default User <ec2-user@ip-172-31-20-117.us-west-2.compute.internal>

tests/entrypoints/llm/test_encode.py

@@ -6,9 +6,10 @@ from typing import List
 import pytest
 
 from vllm import LLM, PoolingParams, PoolingRequestOutput
+from vllm.config import LoadFormat
 from vllm.distributed import cleanup_dist_env_and_memory
 
-MODEL_NAME = "intfloat/e5-mistral-7b-instruct"
+MODEL_NAME = "s3://vllm-ci-model-weights/e5-mistral-7b-instruct"
 
 PROMPTS = [
     "Hello, my name is",
@@ -32,6 +33,7 @@ def llm():
     # pytest caches the fixture so we use weakref.proxy to
     # enable garbage collection
     llm = LLM(model=MODEL_NAME,
+              load_format=LoadFormat.RUNAI_STREAMER,
               max_num_batched_tokens=32768,
               tensor_parallel_size=1,
               gpu_memory_utilization=0.75,