[GuideLLM Refactor] Propagate valid failures from HuggingFace datasets loading (#413)

## Summary

Propagate errors not related to not found dataset errors to surface to
the user that the dataset configuration is incorrect and how to
address/fix it

## Details

<!--
Provide a detailed list of all changes introduced in this pull request.
-->
- [ ]

## Test Plan

<!--
List the steps needed to test this PR.
-->
-

## Related Issues

<!--
Link any relevant issues that this PR addresses.
-->
- Resolves #

---

- [ ] "I certify that all code in this PR is my own, except as noted
below."

## Use of AI

- [ ] Includes AI-assisted code completion
- [ ] Includes code generated by an AI application
- [ ] Includes AI-generated tests (NOTE: AI written tests should have a
docstring that includes `## WRITTEN BY AI ##`)

Author: markurtz

src/guidellm/data/deserializers/deserializer.py

@@ -77,7 +77,7 @@ def deserialize(
         if dataset is None:
             raise DataNotSupportedError(
                 f"No suitable deserializer found for data {data} "
-                f"with kwargs {data_kwargs} and type_ {type_}."
+                f"with kwargs {data_kwargs} and deserializer type {type_}."
             )
 
         if resolve_split:

src/guidellm/data/deserializers/huggingface.py

@@ -12,6 +12,11 @@
     load_dataset,
     load_from_disk,
 )
+from datasets.exceptions import (
+    DataFilesNotFoundError,
+    DatasetNotFoundError,
+    FileNotFoundDatasetsError,
+)
 from transformers import PreTrainedTokenizerBase
 
 from guidellm.data.deserializers.deserializer import (
@@ -35,38 +40,45 @@ def __call__(
         _ = (processor_factory, random_seed)
 
         if isinstance(
-            data, (Dataset, IterableDataset, DatasetDict, IterableDatasetDict)
+            data, Dataset | IterableDataset | DatasetDict | IterableDatasetDict
         ):
             return data
 
         load_error = None
 
         if (
-            isinstance(data, (str, Path))
+            isinstance(data, str | Path)
             and (path := Path(data)).exists()
             and ((path.is_file() and path.suffix == ".py") or path.is_dir())
         ):
             # Handle python script or nested python script in a directory
             try:
                 return load_dataset(str(data), **data_kwargs)
-            except Exception as err:  # noqa: BLE001
-                load_error = err
-
-        if (
-            isinstance(data, (str, Path))
-            and (path := Path(data)).exists()
-            and path.is_dir()
-        ):
-            # Handle local dataset directory
-            try:
-                return load_from_disk(str(data), **data_kwargs)
-            except Exception as err:  # noqa: BLE001
+            except (
+                FileNotFoundDatasetsError,
+                DatasetNotFoundError,
+                DataFilesNotFoundError,
+            ) as err:
                 load_error = err
+            except Exception:  # noqa: BLE001
+                # Try loading as a local dataset directory next
+                try:
+                    return load_from_disk(str(data), **data_kwargs)
+                except (
+                    FileNotFoundDatasetsError,
+                    DatasetNotFoundError,
+                    DataFilesNotFoundError,
+                ) as err2:
+                    load_error = err2
 
         try:
             # Handle dataset identifier from the Hugging Face Hub
             return load_dataset(str(data), **data_kwargs)
-        except Exception as err:  # noqa: BLE001
+        except (
+            FileNotFoundDatasetsError,
+            DatasetNotFoundError,
+            DataFilesNotFoundError,
+        ) as err:
             load_error = err
 
         not_supported = DataNotSupportedError(