test(hfjobs): various fix

.hfjobs/Dockerfile

@@ -5,10 +5,10 @@ ARG ENDPOINT_URL="http://host.docker.internal:8000"
 ARG ENDPOINT_NUM_SAMPLES=50
 ARG ENDPOINT_TEST_SEED=31415926
 
-ENV ENDPOINT_API_KEY=$ENDPOINT_API_KEY
 ENV ENDPOINT_URL=$ENDPOINT_URL
 ENV ENDPOINT_NUM_SAMPLES=$ENDPOINT_NUM_SAMPLES
 ENV ENDPOINT_TEST_SEED=$ENDPOINT_TEST_SEED
+ENV HF_HOME=/tmp/huggingface
 
 RUN --mount=type=bind,source=.hfjobs/requirements.txt,target=/opt/endpoint/tests/requirements.txt \
     python -m pip install -r /opt/endpoint/tests/requirements.txt

Dockerfile

@@ -2,10 +2,10 @@ ARG SDK_VERSION=latest
 FROM huggingface/hfendpoints-sdk:${SDK_VERSION} AS sdk
 
 FROM vllm/vllm-openai:v0.8.4
-RUN --mount=type=bind,from=sdk,source=/usr/local/hfendpoints/dist,target=/usr/local/hfendpoints/dist \
+RUN --mount=type=bind,from=sdk,source=/opt/hfendpoints/dist,target=/usr/local/endpoints/dist \
     --mount=type=bind,source=requirements.txt,target=/tmp/requirements.txt \
     python3 -m pip install -r /tmp/requirements.txt && \
-    python3 -m pip install /usr/local/hfendpoints/dist/*.whl
+    python3 -m pip install /usr/local/endpoints/dist/*.whl
 
 COPY handler.py /usr/local/endpoint/
 

tests/test_openai.py

@@ -7,7 +7,6 @@ import soundfile as sf
 import pytest
 
 # Global variables defining endpoint targets
-ENDPOINT_API_KEY = environ["ENDPOINT_API_KEY"]
 ENDPOINT_URL = environ["ENDPOINT_URL"]
 ENDPOINT_NUM_SAMPLES = int(environ["ENDPOINT_NUM_SAMPLES"])
 ENDPOINT_TEST_SEED = int(environ["ENDPOINT_TEST_SEED"])
@@ -20,7 +19,7 @@ from datasets import load_dataset, Dataset
 from openai import OpenAI
 
 # Global client to make requests
-client = OpenAI(api_key=ENDPOINT_API_KEY, base_url=ENDPOINT_URL)
+client = OpenAI(base_url=ENDPOINT_URL)
 
 
 @pytest.fixture