fix typos in the tests directory (#36717)

tests/generation/test_configuration_utils.py

@@ -193,7 +193,7 @@ class GenerationConfigTest(unittest.TestCase):
             generation_config_bad_temperature.update(temperature=None)
         self.assertEqual(len(captured_warnings), 0)
 
-        # Impossible sets of contraints/parameters will raise an exception
+        # Impossible sets of constraints/parameters will raise an exception
         with self.assertRaises(ValueError):
             GenerationConfig(do_sample=False, num_beams=1, num_return_sequences=2)
         with self.assertRaises(ValueError):

tests/generation/test_logits_process.py

@@ -751,7 +751,7 @@ class LogitsProcessorTest(unittest.TestCase):
         scores = self._get_uniform_logits(batch_size, vocab_size)
         processed_scores = logits_processor(input_ids, scores)
         self.assertTrue(torch.isneginf(processed_scores[:, bos_token_id + 1 :]).all())
-        # score for bos_token_id shold be zero
+        # score for bos_token_id should be zero
         self.assertListEqual(processed_scores[:, bos_token_id].tolist(), 4 * [0])
 
         # processor should not change logits in-place
@@ -972,7 +972,7 @@ class LogitsProcessorTest(unittest.TestCase):
 
         watermark = WatermarkLogitsProcessor(vocab_size=vocab_size, device=input_ids.device)
 
-        # use fixed id for last token, needed for reprodicibility and tests
+        # use fixed id for last token, needed for reproducibility and tests
         input_ids[:, -1] = 10
         scores_wo_bias = scores[:, -1].clone()
         out = watermark(input_ids=input_ids, scores=scores)

tests/generation/test_stopping_criteria.py

@@ -256,7 +256,7 @@ class StoppingCriteriaTestCase(unittest.TestCase):
             ]
         )
 
-        # trigger stopping when at leat one criteria is satisfied, one value per batch
+        # trigger stopping when at least one criteria is satisfied, one value per batch
         self.assertTrue(criteria(inputs["input_ids"], scores))
 
         # return False when neither is satisfied
@@ -283,7 +283,7 @@ class StoppingCriteriaTestCase(unittest.TestCase):
             ]
         )
 
-        # trigger stopping when at leat one criteria is satisfied
+        # trigger stopping when at least one criteria is satisfied
         self.assertListEqual(criteria(inputs["input_ids"], scores).tolist(), [True, False, False])
 
         # False when neither is satisfied

tests/generation/test_utils.py

@@ -173,7 +173,7 @@ class GenerationTesterMixin:
     def _check_similar_generate_outputs(self, output_1, output_2, atol=1e-5, rtol=1e-5):
         """
         Checks whether a pair of generate outputs are similar. Two `generate` call outputs are considered similar in
-        the following siturations:
+        the following situations:
         1. The sequences are the same
         2. The sequences are different, but the scores up to (and including) the first mismatch are nearly identical
         """
@@ -1617,7 +1617,7 @@ class GenerationTesterMixin:
             embed_dim = getattr(text_config, "d_model", text_config.hidden_size)
             per_head_embed_dim = embed_dim // num_attention_heads
 
-            # some models have diffent num-head for query vs key/value so we need to assign correct value
+            # some models have different num-head for query vs key/value so we need to assign correct value
             # BUT only after `per_head_embed_dim` is set
             num_attention_heads = (
                 text_config.num_key_value_heads
@@ -2316,7 +2316,7 @@ class GenerationTesterMixin:
     def _test_attention_implementation(self, attn_implementation):
         """
         Compares the output of generate with the eager attention implementation against other implementations.
-        NOTE: despite the test logic being the same, different implementations actually need diferent decorators, hence
+        NOTE: despite the test logic being the same, different implementations actually need different decorators, hence
         this separate function.
         """
         max_new_tokens = 30
@@ -4619,7 +4619,7 @@ class GenerationIntegrationTests(unittest.TestCase):
         self.assertTrue(diff < 1e-4)
 
     def test_generate_input_ids_as_kwarg(self):
-        """Test that `input_ids` work equaly as a positional and keyword argument in decoder-only models"""
+        """Test that `input_ids` work equally as a positional and keyword argument in decoder-only models"""
         article = "I need input_ids to generate"
         tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-gpt2")
         model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2", max_length=15)
@@ -4636,7 +4636,7 @@ class GenerationIntegrationTests(unittest.TestCase):
         self.assertEqual(output_sequences.shape, (1, 15))
 
     def test_generate_input_ids_as_encoder_kwarg(self):
-        """Test that `input_ids` work equaly as a positional and keyword argument in encoder-decoder models"""
+        """Test that `input_ids` work equally as a positional and keyword argument in encoder-decoder models"""
         article = "Justin Timberlake and Jessica Biel, welcome to parenthood."
         tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-bart")
         model = AutoModelForSeq2SeqLM.from_pretrained("hf-internal-testing/tiny-random-bart")