diff --git a/.github/workflows/multi-gpu-e2e.yml b/.github/workflows/multi-gpu-e2e.yml
index 91cbaf957e..ab886c67f1 100644
--- a/.github/workflows/multi-gpu-e2e.yml
+++ b/.github/workflows/multi-gpu-e2e.yml
@@ -1,6 +1,9 @@
 name: docker-multigpu-tests-biweekly
 
 on:
+  pull_request:
+    paths:
+      - 'tests/e2e/multigpu/*.py'
   workflow_dispatch:
   schedule:
     - cron: '0 0 * * 1,4'  # Runs at 00:00 UTC every monday & thursday
diff --git a/requirements.txt b/requirements.txt
index c61216e63b..83116af60f 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,7 +4,7 @@ peft==0.12.0
 transformers==4.44.2
 tokenizers>=0.19.1
 bitsandbytes==0.43.3
-accelerate==0.34.0
+accelerate==0.34.2
 datasets==2.20.0
 deepspeed==0.14.4
 pydantic==2.6.3
diff --git a/tests/e2e/multigpu/test_llama.py b/tests/e2e/multigpu/test_llama.py
index 344c57fb85..61bb8ed327 100644
--- a/tests/e2e/multigpu/test_llama.py
+++ b/tests/e2e/multigpu/test_llama.py
@@ -10,6 +10,7 @@
 import pytest
 import yaml
 from accelerate.test_utils import execute_subprocess_async
+from huggingface_hub import snapshot_download
 
 from axolotl.utils.dict import DictDefault
 
@@ -19,6 +20,12 @@
 os.environ["WANDB_DISABLED"] = "true"
 
 
+@pytest.fixture(scope="session", autouse=True)
+def download_model():
+    # download the model
+    snapshot_download("TinyLlama/TinyLlama_v1.1")
+
+
 class TestMultiGPULlama(unittest.TestCase):
     """
     Test case for Llama models using LoRA