support HTTP_ADDRESS for url address

experimental/clip_app.py

@@ -10,9 +10,7 @@ from ray import serve
 from clip_retrieval.load_clip import load_clip, get_tokenizer 
 # from clip_retrieval.clip_client import ClipClient, Modality
 
-# @serve.deployment(num_replicas=2, ray_actor_options={"num_cpus": 0.2, "num_gpus": 0.2})
-# @serve.deployment(num_replicas=1, ray_actor_options={"num_cpus": 0.2, "num_gpus": 0.0})
-@serve.deployment(num_replicas=10, ray_actor_options={"num_cpus": .2, "num_gpus": 0.1})
+@serve.deployment(num_replicas=6, ray_actor_options={"num_cpus": .2, "num_gpus": 0.1})
 class CLIPTransform:
     def __init__(self):
         # os.environ["OMP_NUM_THREADS"] = "20"

experimental/clip_app_client.py

@@ -1,19 +1,11 @@
 # File name: graph_client.py
 from concurrent.futures import ThreadPoolExecutor
 import json
+import os
 import requests
 from concurrent.futures import ThreadPoolExecutor, as_completed
 import time
 
-# english_text = (
-#     "It was the best of times, it was the worst of times, it was the age "
-#     "of wisdom, it was the age of foolishness, it was the epoch of belief"
-# )
-# response = requests.post("http://127.0.0.1:8000/", json=english_text)
-# french_text = response.text
-
-# print(french_text)
-
 test_image_url = "https://static.wixstatic.com/media/4d6b49_42b9435ce1104008b1b5f7a3c9bfcd69~mv2.jpg/v1/fill/w_454,h_333,fp_0.50_0.50,q_90/4d6b49_42b9435ce1104008b1b5f7a3c9bfcd69~mv2.jpg"
 english_text = (
     "It was the best of times, it was the worst of times, it was the age "
@@ -23,7 +15,8 @@ english_text = (
 
 def send_text_request(number):
     json = {"text": english_text}
-    response = requests.post("http://127.0.0.1:8000/", json=json)
+    url = os.environ.get("HTTP_ADDRESS", "http://127.0.0.1:8000/")
+    response = requests.post(url, json=json)
     embeddings = response.text
     return number, embeddings
 
@@ -41,10 +34,9 @@ def process_text(numbers, max_workers=10):
 #         result = json.loads(result)
 #         print (f"{n_result} : {len(result[0])}")
 
-
 if __name__ == "__main__":
     # n_calls = 100000
-    n_calls = 1000
+    n_calls = 10000
     numbers = list(range(n_calls))
     start_time = time.monotonic()
     process_text(numbers)