diff --git a/lmms_eval/__init__.py b/lmms_eval/__init__.py
index 88a91ce1..317c0291 100755
--- a/lmms_eval/__init__.py
+++ b/lmms_eval/__init__.py
@@ -1 +1 @@
-from .evaluator import evaluate,simple_evaluate
\ No newline at end of file
+from .evaluator import evaluate, simple_evaluate
diff --git a/lmms_eval/evaluator.py b/lmms_eval/evaluator.py
index fff43fb4..ed0e8115 100755
--- a/lmms_eval/evaluator.py
+++ b/lmms_eval/evaluator.py
@@ -44,8 +44,6 @@
 
 @positional_deprecated
 def simple_evaluate(
-    tuned_model,
-    tuned_model_tokenizer,
     model,
     model_args: Optional[Union[str, dict]] = None,
     tasks: Optional[List[Union[str, dict, object]]] = None,
@@ -75,6 +73,7 @@ def simple_evaluate(
     torch_random_seed: int = 1234,
     fewshot_random_seed: int = 1234,
     cli_args=None,
+    **kwargs,
 ):
     """Instantiate and evaluate a model on a list of tasks.
 
@@ -163,15 +162,16 @@ def simple_evaluate(
         model_args = ""
 
     ModelClass = get_model(model)
-    lm = ModelClass.create_from_arg_string(
-        tuned_model,
-        tuned_model_tokenizer,
-        model_args,
-        {
-            "batch_size": batch_size,
-            "device": device,
-        },
-    )
+    additional_config = {
+        "batch_size": batch_size,
+        "device": device,
+    }
+
+    if "tuned_model" in kwargs and "tuned_model_tokenizer" in kwargs:
+        additional_config["tuned_model"] = kwargs["tuned_model"]
+        additional_config["tuned_model_tokenizer"] = kwargs["tuned_model_tokenizer"]
+
+    lm = ModelClass.create_from_arg_string(model_args, additional_config)
 
     if task_manager is None:
         task_manager = TaskManager(verbosity, model_name=model)
@@ -598,10 +598,9 @@ def evaluate(
     if os.path.exists(f"{cli_args.output_path}/rank{int(os.environ.get('RANK', 0))}_metric_eval_done.txt"):
         os.remove(f"{cli_args.output_path}/rank{int(os.environ.get('RANK', 0))}_metric_eval_done.txt")
 
-
     if not cli_args.output_path.exists():
         cli_args.output_path.mkdir(parents=True, exist_ok=True)
-        
+
     if lm.rank == 0:
         ### Get task ordering for correct sample-wide aggregation
         group_to_task = {}
@@ -714,20 +713,19 @@ def evaluate(
         }
         if log_samples:
             results_dict["samples"] = dict(samples)
-            
+
         with open(f"{cli_args.output_path}/rank{int(os.environ.get('RANK', 0))}_metric_eval_done.txt", "w") as f:
             f.write(f"rank {int(os.environ.get('RANK', 0))} eval done")
         return results_dict
-    
+
     else:
         results_dict = None
 
-
     with open(f"{cli_args.output_path}/rank{int(os.environ.get('RANK', 0))}_metric_eval_done.txt", "w") as f:
         f.write(f"rank {int(os.environ.get('RANK', 0))} eval done")
     while len([file for file in os.listdir(cli_args.output_path) if file.endswith("metric_eval_done.txt")]) < lm._world_size:
         time.sleep(1)
-    
+
     else:
         return None
 
diff --git a/lmms_eval/models/gpt4v.py b/lmms_eval/models/gpt4v.py
index 2eebaf3c..aac62a87 100755
--- a/lmms_eval/models/gpt4v.py
+++ b/lmms_eval/models/gpt4v.py
@@ -109,11 +109,11 @@ def encode_video(self, video_path, for_get_frames_num):
         vr = VideoReader(video_path, ctx=cpu(0))
         total_frame_num = len(vr)
         uniform_sampled_frames = np.linspace(0, total_frame_num - 1, for_get_frames_num, dtype=int)
-        
+
         # Ensure the last frame is included
         if total_frame_num - 1 not in uniform_sampled_frames:
             uniform_sampled_frames = np.append(uniform_sampled_frames, total_frame_num - 1)
-        
+
         frame_idx = uniform_sampled_frames.tolist()
         frames = vr.get_batch(frame_idx).asnumpy()
 
diff --git a/lmms_eval/models/model_utils/load_video.py b/lmms_eval/models/model_utils/load_video.py
index 5bf9b56a..0c4ea23d 100644
--- a/lmms_eval/models/model_utils/load_video.py
+++ b/lmms_eval/models/model_utils/load_video.py
@@ -38,11 +38,11 @@ def read_video_pyav(video_path, num_frm=8):
             total_frames = container.streams.video[0].frames
             sampled_frm = min(total_frames, num_frm)
             indices = np.linspace(0, total_frames - 1, sampled_frm, dtype=int)
-            
+
             # Append the last frame index if not already included
             if total_frames - 1 not in indices:
                 indices = np.append(indices, total_frames - 1)
-                
+
             frames = record_video_length_stream(container, indices)
         except:
             container = av.open(video_path)
@@ -50,11 +50,11 @@ def read_video_pyav(video_path, num_frm=8):
             total_frames = len(frames)
             sampled_frm = min(total_frames, num_frm)
             indices = np.linspace(0, total_frames - 1, sampled_frm, dtype=int)
-            
+
             # Append the last frame index if not already included
             if total_frames - 1 not in indices:
                 indices = np.append(indices, total_frames - 1)
-                
+
             frames = [frames[i] for i in indices]
     else:
         container = av.open(video_path)
@@ -62,10 +62,10 @@ def read_video_pyav(video_path, num_frm=8):
         total_frames = len(frames)
         sampled_frm = min(total_frames, num_frm)
         indices = np.linspace(0, total_frames - 1, sampled_frm, dtype=int)
-        
+
         # Append the last frame index if not already included
         if total_frames - 1 not in indices:
             indices = np.append(indices, total_frames - 1)
-            
+
         frames = [frames[i] for i in indices]
     return np.stack([x.to_ndarray(format="rgb24") for x in frames])
diff --git a/lmms_eval/utils.py b/lmms_eval/utils.py
index e101e2b7..0e940395 100755
--- a/lmms_eval/utils.py
+++ b/lmms_eval/utils.py
@@ -32,6 +32,7 @@
 
 import gc
 from itertools import islice
+
 import numpy as np
 import pytz
 import torch