diff --git a/src/llmcompressor/core/session.py b/src/llmcompressor/core/session.py
index 41072feb9..7c489f36f 100644
--- a/src/llmcompressor/core/session.py
+++ b/src/llmcompressor/core/session.py
@@ -1,6 +1,8 @@
 from dataclasses import dataclass
 from typing import Any, Callable, Dict, List, Optional, Union
 
+from loguru import logger
+
 from llmcompressor.core.events import EventType
 from llmcompressor.core.helpers import log_model_info, should_log_model_info
 from llmcompressor.core.lifecycle import CompressionLifecycle
@@ -260,12 +262,16 @@ def reset_stage(self):
         self.lifecycle.initialized_ = False
         self.lifecycle.finalized = False
 
-    def get_serialized_recipe(self) -> str:
+    def get_serialized_recipe(self) -> Optional[str]:
         """
         :return: serialized string of the current compiled recipe
         """
         recipe = self.lifecycle.recipe_container.compiled_recipe
-        return recipe.yaml()
+
+        if recipe is not None and hasattr(recipe, "yaml"):
+            return recipe.yaml()
+
+        logger.warning("Recipe not found in session - it may have been reset")
 
     def _log_model_info(self):
         # Log model level logs if cadence reached
diff --git a/src/llmcompressor/transformers/sparsification/compressed_tensors_utils.py b/src/llmcompressor/transformers/sparsification/compressed_tensors_utils.py
index 001b43b0b..6de89dd8b 100644
--- a/src/llmcompressor/transformers/sparsification/compressed_tensors_utils.py
+++ b/src/llmcompressor/transformers/sparsification/compressed_tensors_utils.py
@@ -191,9 +191,10 @@ def skip(*args, **kwargs):
 
             recipe_path = os.path.join(save_directory, "recipe.yaml")
             session = active_session()
-            recipe_yaml_str = session.get_serialized_recipe()
-            with open(recipe_path, "w") as fp:
-                fp.write(recipe_yaml_str)
+
+            if (recipe_yaml_str := session.get_serialized_recipe()) is not None:
+                with open(recipe_path, "w") as fp:
+                    fp.write(recipe_yaml_str)
 
             # copy python files from cache dir to save_path if any
             copy_python_files_from_model_cache(model, save_directory)
diff --git a/tests/e2e/vLLM/test_vllm.py b/tests/e2e/vLLM/test_vllm.py
index 988793849..316f90ff1 100644
--- a/tests/e2e/vLLM/test_vllm.py
+++ b/tests/e2e/vLLM/test_vllm.py
@@ -1,12 +1,15 @@
+import os
 import shutil
 import unittest
 from typing import Callable
 
 import pytest
 from datasets import load_dataset
+from loguru import logger
 from parameterized import parameterized, parameterized_class
 from transformers import AutoTokenizer
 
+from llmcompressor.core import active_session
 from llmcompressor.modifiers.quantization import QuantizationModifier
 from llmcompressor.transformers import SparseAutoModelForCausalLM, oneshot
 from tests.testing_utils import (
@@ -22,6 +25,7 @@
     vllm_installed = True
 except ImportError:
     vllm_installed = False
+    logger.warning("vllm is not installed. This test will be skipped")
 
 # Defines the file paths to the directories containing the test configs
 # for each of the quantization schemes
@@ -32,6 +36,8 @@
 WNA16_2of4 = "tests/e2e/vLLM/configs/WNA16_2of4"
 CONFIGS = [WNA16, FP8, INT8, ACTORDER, WNA16_2of4]
 
+HF_MODEL_HUB_NAME = "nm-testing"
+
 
 def gen_test_name(testcase_func: Callable, param_num: int, param: dict) -> str:
     return "_".join(
@@ -76,8 +82,8 @@ class TestvLLM(unittest.TestCase):
     save_dir = None
 
     def setUp(self):
-        print("========== RUNNING ==============")
-        print(self.scheme)
+        logger.info("========== RUNNING ==============")
+        logger.debug(self.scheme)
 
         self.device = "cuda:0"
         self.oneshot_kwargs = {}
@@ -88,6 +94,7 @@ def setUp(self):
             "The president of the US is",
             "My name is",
         ]
+        self.session = active_session()
 
     def test_vllm(self):
         import torch
@@ -124,16 +131,20 @@ def test_vllm(self):
             )
 
         # Apply quantization.
-        print("ONESHOT KWARGS", self.oneshot_kwargs)
+        logger.debug("ONESHOT KWARGS", self.oneshot_kwargs)
         oneshot(
             **self.oneshot_kwargs,
-            clear_sparse_session=True,
             oneshot_device=self.device,
         )
+
         self.oneshot_kwargs["model"].save_pretrained(self.save_dir)
         tokenizer.save_pretrained(self.save_dir)
+
+        # Reset after session info is extracted on save -- recipe
+        self.session.reset()
+
         # Run vLLM with saved model
-        print("================= RUNNING vLLM =========================")
+        logger.info("================= RUNNING vLLM =========================")
         sampling_params = SamplingParams(temperature=0.80, top_p=0.95)
         if "W4A16_2of4" in self.scheme:
             # required by the kernel
@@ -141,16 +152,19 @@ def test_vllm(self):
         else:
             llm = LLM(model=self.save_dir)
         outputs = llm.generate(self.prompts, sampling_params)
-        print("================= vLLM GENERATION ======================")
+
+        logger.info("================= vLLM GENERATION ======================")
         for output in outputs:
             assert output
             prompt = output.prompt
             generated_text = output.outputs[0].text
-            print("PROMPT", prompt)
-            print("GENERATED TEXT", generated_text)
-        print("================= UPLOADING TO HUB ======================")
-        self.oneshot_kwargs["model"].push_to_hub(f"nm-testing/{self.save_dir}-e2e")
-        tokenizer.push_to_hub(f"nm-testing/{self.save_dir}-e2e")
+            logger.debug("PROMPT", prompt)
+            logger.debug("GENERATED TEXT", generated_text)
+
+        logger.info("================= UPLOADING TO HUB ======================")
+        hf_upload_path = os.path.join(HF_MODEL_HUB_NAME, f"{self.save_dir}-e2e")
+        self.oneshot_kwargs["model"].push_to_hub(hf_upload_path)
+        tokenizer.push_to_hub(hf_upload_path)
 
     def tearDown(self):
         if self.save_dir is not None: