fix: Correct logic for checkpoint_format handling in create_input_processor

- Updated the condition to check for "HF" explicitly, removing the None case handling.

Signed-off-by: Robin Kobus <[email protected]>

Author: Funatiq

tensorrt_llm/inputs/registry.py

@@ -419,7 +419,7 @@ def create_input_processor(
         model_path_or_dir: Path or repo id used to locate pretrained config/tokenizer.
         tokenizer: Tokenizer instance.
         checkpoint_format: Checkpoint format identifier. "HF" uses Hugging Face-style
-            config loading; any other value skips HF config loading. None is treated as "HF".
+            config loading; any other value skips HF config loading.
 
     Returns:
         An InputProcessor implementation (model-specific if registered; otherwise DefaultInputProcessor).
@@ -429,7 +429,7 @@ def create_input_processor(
 
     model_config = None
 
-    if checkpoint_format is None or checkpoint_format == "HF":
+    if checkpoint_format == "HF":
         try:
             config = ModelConfig.from_pretrained(model_path_or_dir,
                                                  trust_remote_code=True)