Fix test for multi-call generate()

tests/test_full_model.py

@@ -78,8 +78,8 @@ def test_full_model_exact_match(tokenizer, model, ref_model, use_peft, pass_empt
 
 
 @pytest.mark.forked
-@pytest.mark.parametrize("multiple_steps", [False, True])
-def test_greedy_generation(tokenizer, model, ref_model, multiple_steps, max_new_tokens=4):
+@pytest.mark.parametrize("multiple_calls", [False, True])
+def test_greedy_generation(tokenizer, model, ref_model, multiple_calls, max_new_tokens=4):
     inputs_single = tokenizer("A cat sat on a mat", return_tensors="pt")["input_ids"]
 
     if tokenizer.pad_token_id is None:
@@ -89,15 +89,15 @@ def test_greedy_generation(tokenizer, model, ref_model, multiple_steps, max_new_
     ]
 
     for inputs in [inputs_single, inputs_batch]:
-        if not multiple_steps:
+        if not multiple_calls:
             outputs = model.generate(inputs, max_new_tokens=max_new_tokens, do_sample=False)
         else:
             with model.inference_session(max_length=inputs.shape[1] + max_new_tokens) as sess:
                 outputs = [
                     # Sessions provided both explicitly and implicitly should work
                     model.generate(inputs, max_new_tokens=1, do_sample=False, session=sess),
-                    model.generate(inputs, max_new_tokens=max_new_tokens - 2, do_sample=False),
-                    model.generate(inputs, max_new_tokens=1, do_sample=False),
+                    model.generate(None, max_new_tokens=max_new_tokens - 2, do_sample=False),
+                    model.generate(None, max_new_tokens=1, do_sample=False),
                 ]
                 outputs = torch.cat(outputs, dim=1)