Fixes integration test (deepjavalibrary#779)

tests/integration/llm/client.py

@@ -10,7 +10,6 @@
 import numpy as np
 from datetime import datetime
 from io import BytesIO
-import hashlib
 
 logging.basicConfig(level=logging.INFO)
 parser = argparse.ArgumentParser(description="Build the LLM configs")
@@ -53,11 +52,10 @@
 args = parser.parse_args()
 
 
-def compute_model_name_hash(model_name):
-    # This mirrors the Utils.hash implementation from DJL Core
-    m = hashlib.sha256()
-    m.update(model_name)
-    return m.hexdigest()[:40]
+def get_model_name():
+    endpoint = f"http://127.0.0.1:8080/models"
+    res = requests.get(endpoint).json()
+    return res["models"][0]["modelName"]
 
 
 ds_raw_model_spec = {
@@ -118,21 +116,18 @@ def compute_model_name_hash(model_name):
         "batch_size": [1, 4],
         "seq_length": [16, 32],
         "worker": 1,
-        "model_name": compute_model_name_hash(b"nomic-ai/gpt4all-j"),
     },
     "no-code/databricks/dolly-v2-7b": {
         "max_memory_per_gpu": [10.0, 12.0],
         "batch_size": [1, 4],
         "seq_length": [16, 32],
         "worker": 2,
-        "model_name": compute_model_name_hash(b"databricks/dolly-v2-7b"),
     },
     "no-code/google/flan-t5-xl": {
         "max_memory_per_gpu": [7.0, 7.0],
         "batch_size": [1, 4],
         "seq_length": [16, 32],
         "worker": 2,
-        "model_name": compute_model_name_hash(b"google/flan-t5-xl")
     }
 }
 
@@ -266,7 +261,8 @@ def compute_model_name_hash(model_name):
 }
 
 
-def check_worker_number(desired, model_name="test"):
+def check_worker_number(desired):
+    model_name = get_model_name()
     endpoint = f"http://127.0.0.1:8080/models/{model_name}"
     res = requests.get(endpoint).json()
     if desired == len(res[0]["models"][0]["workerGroups"]):
@@ -278,9 +274,9 @@ def check_worker_number(desired, model_name="test"):
             f"Worker number does not meet requirements! {res}")
 
 
-def send_json(data, model_name="test"):
+def send_json(data):
     headers = {'content-type': 'application/json'}
-    endpoint = f"http://127.0.0.1:8080/predictions/{model_name}"
+    endpoint = f"http://127.0.0.1:8080/invocations"
     resp = requests.post(endpoint, headers=headers, json=data)
 
     if resp.status_code >= 300:
@@ -289,12 +285,12 @@ def send_json(data, model_name="test"):
     return resp
 
 
-def send_image_json(img_url, data, model_name="test"):
+def send_image_json(img_url, data):
     multipart_form_data = {
         'data': BytesIO(requests.get(img_url, stream=True).content),
         'json': (None, json.dumps(data), 'application/json')
     }
-    endpoint = f"http://127.0.0.1:8080/predictions/{model_name}"
+    endpoint = f"http://127.0.0.1:8080/invocations"
     resp = requests.post(endpoint, files=multipart_form_data)
 
     if resp.status_code >= 300:
@@ -459,8 +455,7 @@ def test_handler(model, model_spec):
         )
     spec = model_spec[args.model]
     if "worker" in spec:
-        check_worker_number(spec["worker"],
-                            model_name=spec.get("model_name", "test"))
+        check_worker_number(spec["worker"])
     for i, batch_size in enumerate(spec["batch_size"]):
         for seq_length in spec["seq_length"]:
             if "t5" in model:
@@ -470,7 +465,7 @@ def test_handler(model, model_spec):
             params = {"max_new_tokens": seq_length}
             req["parameters"] = params
             logging.info(f"req {req}")
-            res = send_json(req, model_name=spec.get("model_name", "test"))
+            res = send_json(req)
             if spec.get("stream_output", False):
                 logging.info(f"res: {res.content}")
                 result = res.content.decode().split("\n")[:-1]

tests/integration/llm/sagemaker-endpoint-tests.py

@@ -129,9 +129,7 @@
     }
 }
 
-ENGINE_TO_METRIC_CONFIG_ENGINE = {
-    "Python" : "Accelerate"
-}
+ENGINE_TO_METRIC_CONFIG_ENGINE = {"Python": "Accelerate"}
 
 
 def get_sagemaker_session(default_bucket=DEFAULT_BUCKET,