Merge pull request #5546 from chengchengpei/cpei/refactor

1, log exceptions in details; 2, check processor is None before calling it

Former-commit-id: f8a2cc61054cec215d658159dc2d6fbacf6624e9

src/llamafactory/model/loader.py

@@ -82,6 +82,8 @@ def load_tokenizer(model_args: "ModelArguments") -> "TokenizerModule":
             padding_side="right",
             **init_kwargs,
         )
+    except Exception as e:
+        raise OSError("Failed to load tokenizer") from e
 
     if model_args.new_special_tokens is not None:
         num_added_tokens = tokenizer.add_special_tokens(
@@ -97,12 +99,13 @@ def load_tokenizer(model_args: "ModelArguments") -> "TokenizerModule":
     try:
         processor = AutoProcessor.from_pretrained(model_args.model_name_or_path, **init_kwargs)
         patch_processor(processor, config, tokenizer, model_args)
-    except Exception:
+    except Exception as e:
+        logger.warning("Failed to load processor. Error: {}".format(e))
         processor = None
 
     # Avoid load tokenizer, see:
     # https://github.com/huggingface/transformers/blob/v4.40.0/src/transformers/models/auto/processing_auto.py#L324
-    if "Processor" not in processor.__class__.__name__:
+    if processor and "Processor" not in processor.__class__.__name__:
         processor = None
 
     return {"tokenizer": tokenizer, "processor": processor}

src/llamafactory/train/tuner.py

@@ -139,5 +139,5 @@ def export_model(args: Optional[Dict[str, Any]] = None) -> None:
                     model_args.export_hub_model_id, token=model_args.hf_hub_token
                 )
 
-    except Exception:
-        logger.warning("Cannot save tokenizer, please copy the files manually.")
+    except Exception as e:
+        logger.warning("Cannot save tokenizer, please copy the files manually. Error: {}".format(e))