[tests] fix test_nemotron_8b_generation_sdpa (#37665)

add max_new_tokens

tests/models/nemotron/test_modeling_nemotron.py

@@ -195,7 +195,7 @@ class NemotronIntegrationTest(unittest.TestCase):
         tokenizer = AutoTokenizer.from_pretrained(model_id)
         inputs = tokenizer(text, return_tensors="pt").to(torch_device)
 
-        output = model.generate(**inputs, do_sample=False)
+        output = model.generate(**inputs, do_sample=False, max_new_tokens=10)
         output_text = tokenizer.batch_decode(output, skip_special_tokens=True)
         self.assertEqual(EXPECTED_TEXT, output_text)