test: do not use staging for cache tests

The HF_ENDPOINT variable is not always taken into account when using the
huggingface_hub client depending on the order of imports.
This modifies the tests to create temporary dorectories under the testing
user account instead.

tests/cache/test_neuronx_cache.py

@@ -18,13 +18,13 @@
 import socket
 import subprocess
 from tempfile import TemporaryDirectory
+from time import time
 
 import PIL
 import pytest
 import torch
 from huggingface_hub import HfApi
 from transformers import AutoTokenizer
-from transformers.testing_utils import ENDPOINT_STAGING
 
 from optimum.neuron import (
     NeuronModelForCausalLM,
@@ -37,27 +37,21 @@
 
 
 @pytest.fixture
-def cache_repos(staging):
+def cache_repos():
     # Setup: create temporary Hub repository and local cache directory
-    token = staging["token"]
-    user = staging["user"]
-    api = HfApi(endpoint=ENDPOINT_STAGING, token=token)
+    api = HfApi()
     hostname = socket.gethostname()
-    cache_repo_id = f"{user}/{hostname}-optimum-neuron-cache"
-    if api.repo_exists(cache_repo_id):
-        api.delete_repo(cache_repo_id)
+    cache_repo_id = f"{hostname}-{time()}-optimum-neuron-cache"
     cache_repo_id = api.create_repo(cache_repo_id, private=True).repo_id
     cache_dir = TemporaryDirectory()
     cache_path = cache_dir.name
     # Modify environment to force neuronx cache to use temporary caches
     previous_env = {}
-    env_vars = ["NEURON_COMPILE_CACHE_URL", "CUSTOM_CACHE_REPO", "HF_ENDPOINT", "HF_TOKEN"]
+    env_vars = ["NEURON_COMPILE_CACHE_URL", "CUSTOM_CACHE_REPO"]
     for var in env_vars:
         previous_env[var] = os.environ.get(var)
     os.environ["NEURON_COMPILE_CACHE_URL"] = cache_path
     os.environ["CUSTOM_CACHE_REPO"] = cache_repo_id
-    os.environ["HF_ENDPOINT"] = ENDPOINT_STAGING
-    os.environ["HF_TOKEN"] = token
     yield (cache_path, cache_repo_id)
     # Teardown
     api.delete_repo(cache_repo_id)
@@ -173,8 +167,7 @@ def check_traced_cache_entry(cache_path):
 
 
 def assert_local_and_hub_cache_sync(cache_path, cache_repo_id):
-    # Since created models are public on the staging endpoint we don't need a token
-    api = HfApi(endpoint=ENDPOINT_STAGING)
+    api = HfApi()
     remote_files = api.list_repo_files(cache_repo_id)
     local_files = get_local_cached_files(cache_path)
     for file in local_files: