First version

code/predict.py

@@ -0,0 +1,13 @@
+import time
+from transformers import pipeline
+
+pipeline = pipeline("sentiment-analysis", model="juliensimon/autonlp-imdb-demo-hf-16622767", use_auth_token=True)
+
+data = "Dune is something rarely seen anymore – it’s an epic."
+
+tick=time.time_ns()
+for i in range(0,100):
+    response=pipeline(data)
+tock=time.time_ns()
+
+print("%.2fms" % ((tock-tick)/1000/1000/100))

code/predict_infinity.py

@@ -0,0 +1,15 @@
+import requests as r
+import time
+
+input_data = { "inputs": "Dune is something rarely seen anymore – it’s an epic."}
+
+url = "http://localhost:8080/predict"
+
+sum=0
+for i in range(0,1000):
+    response = r.post(url, json=input_data)
+    resp_time = response.headers["x-compute-time"]
+    sum+=float(resp_time)
+
+sum/=1000
+print(f"{round(float(sum),6)*1000}ms")