Empty assert hunt (#6056)

* Fixed empty asserts

* black-reformatted stragglers in templates

* More code quality checks

* Update src/transformers/convert_marian_to_pytorch.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/convert_marian_to_pytorch.py

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* removed unused line as per @sshleifer

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

src/transformers/data/datasets/language_modeling.py

@@ -22,7 +22,7 @@ class TextDataset(Dataset):
     def __init__(
         self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int, overwrite_cache=False,
     ):
-        assert os.path.isfile(file_path)
+        assert os.path.isfile(file_path), f"Input file path {file_path} not found"
 
         block_size = block_size - tokenizer.num_special_tokens_to_add(pair=False)
 
@@ -82,7 +82,7 @@ class LineByLineTextDataset(Dataset):
     """
 
     def __init__(self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int):
-        assert os.path.isfile(file_path)
+        assert os.path.isfile(file_path), f"Input file path {file_path} not found"
         # Here, we do not cache the features, operating under the assumption
         # that we will soon use fast multithreaded tokenizers from the
         # `tokenizers` repo everywhere =)

src/transformers/data/metrics/__init__.py

@@ -51,7 +51,9 @@ if _has_sklearn:
         }
 
     def glue_compute_metrics(task_name, preds, labels):
-        assert len(preds) == len(labels)
+        assert len(preds) == len(
+            labels
+        ), f"Predictions and labels have mismatched lengths {len(preds)} and {len(labels)}"
         if task_name == "cola":
             return {"mcc": matthews_corrcoef(labels, preds)}
         elif task_name == "sst-2":
@@ -78,7 +80,9 @@ if _has_sklearn:
             raise KeyError(task_name)
 
     def xnli_compute_metrics(task_name, preds, labels):
-        assert len(preds) == len(labels)
+        assert len(preds) == len(
+            labels
+        ), f"Predictions and labels have mismatched lengths {len(preds)} and {len(labels)}"
         if task_name == "xnli":
             return {"acc": simple_accuracy(preds, labels)}
         else:

src/transformers/data/metrics/squad_metrics.py

@@ -523,7 +523,7 @@ def compute_predictions_logits(
         if not nbest:
             nbest.append(_NbestPrediction(text="empty", start_logit=0.0, end_logit=0.0))
 
-        assert len(nbest) >= 1
+        assert len(nbest) >= 1, "No valid predictions"
 
         total_scores = []
         best_non_null_entry = None
@@ -544,7 +544,7 @@ def compute_predictions_logits(
             output["end_logit"] = entry.end_logit
             nbest_json.append(output)
 
-        assert len(nbest_json) >= 1
+        assert len(nbest_json) >= 1, "No valid predictions"
 
         if not version_2_with_negative:
             all_predictions[example.qas_id] = nbest_json[0]["text"]
@@ -739,8 +739,8 @@ def compute_predictions_log_probs(
             output["end_log_prob"] = entry.end_log_prob
             nbest_json.append(output)
 
-        assert len(nbest_json) >= 1
-        assert best_non_null_entry is not None
+        assert len(nbest_json) >= 1, "No valid predictions"
+        assert best_non_null_entry is not None, "No valid predictions"
 
         score_diff = score_null
         scores_diff_json[example.qas_id] = score_diff

src/transformers/data/processors/utils.py

@@ -194,8 +194,12 @@ class SingleSentenceClassificationProcessor(DataProcessor):
     def add_examples(
         self, texts_or_text_and_labels, labels=None, ids=None, overwrite_labels=False, overwrite_examples=False
     ):
-        assert labels is None or len(texts_or_text_and_labels) == len(labels)
-        assert ids is None or len(texts_or_text_and_labels) == len(ids)
+        assert labels is None or len(texts_or_text_and_labels) == len(
+            labels
+        ), f"Text and labels have mismatched lengths {len(texts_or_text_and_labels)} and {len(labels)}"
+        assert ids is None or len(texts_or_text_and_labels) == len(
+            ids
+        ), f"Text and ids have mismatched lengths {len(texts_or_text_and_labels)} and {len(ids)}"
         if ids is None:
             ids = [None] * len(texts_or_text_and_labels)
         if labels is None:

src/transformers/data/processors/xnli.py

@@ -45,7 +45,9 @@ class XnliProcessor(DataProcessor):
             text_a = line[0]
             text_b = line[1]
             label = "contradiction" if line[2] == "contradictory" else line[2]
-            assert isinstance(text_a, str) and isinstance(text_b, str) and isinstance(label, str)
+            assert isinstance(text_a, str), f"Training input {text_a} is not a string"
+            assert isinstance(text_b, str), f"Training input {text_b} is not a string"
+            assert isinstance(label, str), f"Training label {label} is not a string"
             examples.append(InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
         return examples
 
@@ -63,7 +65,9 @@ class XnliProcessor(DataProcessor):
             text_a = line[6]
             text_b = line[7]
             label = line[1]
-            assert isinstance(text_a, str) and isinstance(text_b, str) and isinstance(label, str)
+            assert isinstance(text_a, str), f"Training input {text_a} is not a string"
+            assert isinstance(text_b, str), f"Training input {text_b} is not a string"
+            assert isinstance(label, str), f"Training label {label} is not a string"
             examples.append(InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
         return examples