add dataset split and config to model-index in TrainingSummary.from_trainer (#18064)

* added metadata to training summary

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/modelcard.py

@@ -384,6 +384,7 @@ class TrainingSummary:
     dataset: Optional[Union[str, List[str]]] = None
     dataset_tags: Optional[Union[str, List[str]]] = None
     dataset_args: Optional[Union[str, List[str]]] = None
+    dataset_metadata: Optional[Dict[str, Any]] = None
     eval_results: Optional[Dict[str, float]] = None
     eval_lines: Optional[List[str]] = None
     hyperparameters: Optional[Dict[str, Any]] = None
@@ -412,10 +413,12 @@ class TrainingSummary:
         dataset_names = _listify(self.dataset)
         dataset_tags = _listify(self.dataset_tags)
         dataset_args = _listify(self.dataset_args)
+        dataset_metadata = _listify(self.dataset_metadata)
         if len(dataset_args) < len(dataset_tags):
             dataset_args = dataset_args + [None] * (len(dataset_tags) - len(dataset_args))
         dataset_mapping = {tag: name for tag, name in zip(dataset_tags, dataset_names)}
         dataset_arg_mapping = {tag: arg for tag, arg in zip(dataset_tags, dataset_args)}
+        dataset_metadata_mapping = {tag: metadata for tag, metadata in zip(dataset_tags, dataset_metadata)}
 
         task_mapping = {
             task: TASK_TAG_TO_NAME_MAPPING[task] for task in _listify(self.tasks) if task in TASK_TAG_TO_NAME_MAPPING
@@ -438,7 +441,12 @@ class TrainingSummary:
                 result["task"] = {"name": task_mapping[task_tag], "type": task_tag}
 
             if ds_tag is not None:
-                result["dataset"] = {"name": dataset_mapping[ds_tag], "type": ds_tag}
+                metadata = dataset_metadata_mapping.get(ds_tag, {})
+                result["dataset"] = {
+                    "name": dataset_mapping[ds_tag],
+                    "type": ds_tag,
+                    **metadata,
+                }
                 if dataset_arg_mapping[ds_tag] is not None:
                     result["dataset"]["args"] = dataset_arg_mapping[ds_tag]
 
@@ -565,6 +573,7 @@ class TrainingSummary:
         finetuned_from=None,
         tasks=None,
         dataset_tags=None,
+        dataset_metadata=None,
         dataset=None,
         dataset_args=None,
     ):
@@ -574,6 +583,8 @@ class TrainingSummary:
             default_tag = one_dataset.builder_name
             # Those are not real datasets from the Hub so we exclude them.
             if default_tag not in ["csv", "json", "pandas", "parquet", "text"]:
+                if dataset_metadata is None:
+                    dataset_metadata = [{"config": one_dataset.config_name, "split": str(one_dataset.split)}]
                 if dataset_tags is None:
                     dataset_tags = [default_tag]
                 if dataset_args is None:
@@ -618,9 +629,10 @@ class TrainingSummary:
             model_name=model_name,
             finetuned_from=finetuned_from,
             tasks=tasks,
-            dataset_tags=dataset_tags,
             dataset=dataset,
+            dataset_tags=dataset_tags,
             dataset_args=dataset_args,
+            dataset_metadata=dataset_metadata,
             eval_results=eval_results,
             eval_lines=eval_lines,
             hyperparameters=hyperparameters,
@@ -751,7 +763,7 @@ def parse_log_history(log_history):
         while idx >= 0 and "eval_loss" not in log_history[idx]:
             idx -= 1
 
-        if idx > 0:
+        if idx >= 0:
             return None, None, log_history[idx]
         else:
             return None, None, None