Flash-Attn: fix generation when no attention mask or no pading (#32241)

* fix

* fix prev test (half of failures)

* [run-slow] llama, gemma2

* [run-slow] llama, gemma2

src/transformers/modeling_flash_attention_utils.py

@@ -264,9 +264,11 @@ def _flash_attention_forward(
         )
         attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length)
 
-    # if position_ids is provided and check not all examples (row) contain only 1 sequence,
+    # if position_ids is provided and check not all examples (row) contain only 1 sequence, and is in pre-fill/training stage
     # then use `flash_attn_varlen_func` to prevent cross-example attention and also allow padding free approach
-    elif position_ids is not None and not (position_ids[:, -1] == position_ids.size(1) - 1).all():
+    elif (
+        position_ids is not None and not (position_ids[:, -1] == position_ids.size(1) - 1).all() and query_length != 1
+    ):
         batch_size = query_states.size(0)
         query_states, key_states, value_states, indices_q, cu_seq_lens, max_seq_lens = prepare_fa2_from_position_ids(
             query_states, key_states, value_states, position_ids

tests/test_modeling_common.py

@@ -4270,6 +4270,18 @@ class ModelTesterMixin:
                     use_cache=True,
                 )
 
+                # Generate with one batch only to test generation when attention mask will be None
+                # when real inputs are used, because there is no padding. See issue #32237 for more
+                dummy_input = dummy_input[:1, ...]
+                dummy_attention_mask = torch.ones_like(dummy_attention_mask[:1, ...])
+                _ = model.generate(
+                    dummy_input,
+                    attention_mask=dummy_attention_mask,
+                    max_new_tokens=max_new_tokens,
+                    do_sample=False,
+                    use_cache=True,
+                )
+
     @require_flash_attn
     @require_torch_gpu
     @require_bitsandbytes
@@ -4342,6 +4354,8 @@ class ModelTesterMixin:
                 self.skipTest(f"{model_class.__name__} does not support Flash Attention 2")
 
             config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
+            if 0 not in inputs_dict.get("attention_mask", []) or "attention_mask" not in inputs_dict:
+                self.skipTest("Model dummy inputs should contain padding in their attention mask")
 
             dummy_input = inputs_dict[model_class.main_input_name]
             if dummy_input.dtype in [torch.float32, torch.bfloat16]:
@@ -4356,7 +4370,6 @@ class ModelTesterMixin:
             with tempfile.TemporaryDirectory() as tmpdirname:
                 model.save_pretrained(tmpdirname)
 
-                assert 0 in inputs_dict["attention_mask"], "assert padding in testing inputs"
                 # ensure left padding, to adapt for some models
                 if 0 in inputs_dict["attention_mask"][:, -1]:
                     inputs_dict["attention_mask"] = inputs_dict["attention_mask"].flip(1)