[Encoder decoder] Add cuda graph support during decoding for encoder-decoder models (#7631)

.buildkite/test-pipeline.yaml

@@ -252,6 +252,13 @@ steps:
   - export VLLM_WORKER_MULTIPROC_METHOD=spawn
   - bash ./run-tests.sh -c configs/models-small.txt -t 1
 
+- label: Encoder Decoder tests # 5min
+  source_file_dependencies:
+  - vllm/
+  - tests/encoder_decoder
+  commands:
+    - pytest -v -s encoder_decoder
+
 - label: OpenAI-Compatible Tool Use # 20 min
   fast_check: false
   mirror_hardwares: [ amd ]

tests/encoder_decoder/test_e2e_correctness.py

+"""E2E tests to verify the correctness of the encoder-decoder framework
+
+Run `pytest tests/encoder_decoder/test_e2e_correctness.py`.
+"""
+from typing import List, Optional, Tuple
+
+import pytest
+from transformers import AutoModelForSeq2SeqLM
+
+from vllm.sequence import SampleLogprobs
+from vllm.utils import is_cpu
+
+from ..conftest import DecoderPromptType
+from ..models.utils import check_logprobs_close
+
+
+def vllm_to_hf_output(
+    vllm_output: Tuple[List[int], str, Optional[SampleLogprobs]],
+    decoder_prompt_type: DecoderPromptType,
+):
+    """Sanitize vllm output to be comparable with hf output."""
+    output_ids, output_str, out_logprobs = vllm_output
+
+    hf_output_str = output_str + "</s>"
+    if decoder_prompt_type == DecoderPromptType.NONE:
+        hf_output_str = "<s>" + hf_output_str
+
+    return output_ids, hf_output_str, out_logprobs
+
+
+@pytest.mark.parametrize("model", ["facebook/bart-large-cnn"])
+@pytest.mark.parametrize("dtype", ["bfloat16"])
+@pytest.mark.parametrize("max_tokens", [128])
+@pytest.mark.parametrize("num_logprobs", [5])
+@pytest.mark.parametrize("decoder_prompt_type", list(DecoderPromptType))
+@pytest.mark.parametrize("enforce_eager", [True, False])
+@pytest.mark.skipif(
+    is_cpu(),
+    reason="CPU backend is not currently supported with encoder/decoder models"
+)
+def test_encoder_decoder_e2e(
+    hf_runner,
+    vllm_runner,
+    example_encoder_decoder_prompts,
+    model: str,
+    dtype: str,
+    max_tokens: int,
+    num_logprobs: int,
+    decoder_prompt_type: DecoderPromptType,
+    enforce_eager: bool,
+) -> None:
+    '''
+    End-to-End (E2E) test for the encoder-decoder framework. 
+    This test evaluates the encoder-decoder functionality using the BART
+    model. We compare the outputs of the Hugging Face and vLLM
+    implementations to ensure that both implementations produce consistent
+    and correct results.
+    '''
+    test_case_prompts = example_encoder_decoder_prompts[decoder_prompt_type]
+
+    # Configuration settings for HF baseline
+    hf_kwargs = {
+        "top_k": None,
+        "num_beams": 1,
+        "repetition_penalty": 1.0,
+        "top_p": 1.0,
+        "length_penalty": 1.0,
+        "early_stopping": False,
+        "no_repeat_ngram_size": None,
+        "min_length": 0
+    }
+
+    with hf_runner(model, dtype=dtype,
+                   auto_cls=AutoModelForSeq2SeqLM) as hf_model:
+        hf_outputs = (hf_model.generate_encoder_decoder_greedy_logprobs_limit(
+            test_case_prompts,
+            max_tokens,
+            num_logprobs,
+            **hf_kwargs,
+        ))
+    with vllm_runner(model, dtype=dtype,
+                     enforce_eager=enforce_eager) as vllm_model:
+        vllm_outputs = vllm_model.generate_encoder_decoder_greedy_logprobs(
+            test_case_prompts, max_tokens, num_logprobs)
+
+    hf_skip_tokens = (1
+                      if decoder_prompt_type == DecoderPromptType.NONE else 0)
+
+    check_logprobs_close(
+        outputs_0_lst=hf_outputs,
+        outputs_1_lst=[
+            vllm_to_hf_output(vllm_output, decoder_prompt_type)
+            for vllm_output in vllm_outputs
+        ],
+        name_0="hf",
+        name_1="vllm",
+        num_outputs_0_skip_tokens=hf_skip_tokens,
+    )

tests/worker/test_encoder_decoder_model_runner.py

+import itertools
 from array import array
 from typing import List
 
@@ -7,13 +8,9 @@ import torch
 from vllm.engine.arg_utils import EngineArgs
 from vllm.sequence import (VLLM_TOKEN_ID_ARRAY_TYPE, SamplingParams,
                            SequenceData, SequenceGroupMetadata)
-from vllm.utils import is_cpu
+from vllm.utils import is_cpu, make_tensor_with_pad
 from vllm.worker.enc_dec_model_runner import EncoderDecoderModelRunner
-
-# CUDA graph scenarios to test
-#
-# Currently CUDA graph is not supported
-ENFORCE_EAGER = [True]
+from vllm.worker.model_runner import _get_graph_batch_size
 
 BATCH_SIZES = [1, 4, 16, 64, 256]
 
@@ -40,8 +37,7 @@ def _create_model_runner(model: str, *args,
                     reason="CPU backend is currently "
                     "unsupported for encoder/ "
                     "decoder models")
-@pytest.mark.parametrize("enforce_eager", ENFORCE_EAGER)
-def test_empty_seq_group(enforce_eager, ):
+def test_empty_seq_group():
     """Verify prepare prompt and decode returns empty output
        for empty seq group list"""
 
@@ -52,7 +48,7 @@ def test_empty_seq_group(enforce_eager, ):
         max_num_batched_tokens=100000,
         max_num_seqs=100000,
         enable_chunked_prefill=False,
-        enforce_eager=enforce_eager,
+        enforce_eager=True,
     )
     seq_group_metadata_list: List[SequenceGroupMetadata] = []
     model_input = model_runner._prepare_model_input_tensors(
@@ -85,11 +81,7 @@ def test_empty_seq_group(enforce_eager, ):
                     "unsupported for encoder/ "
                     "decoder models")
 @pytest.mark.parametrize("batch_size", BATCH_SIZES)
-@pytest.mark.parametrize("enforce_eager", ENFORCE_EAGER)
-def test_prepare_prompt(
-    batch_size,
-    enforce_eager,
-):
+def test_prepare_prompt(batch_size):
     '''
     Test the ability of the encoder/decoder model runner subclass to
     produce prefill-phase model inputs & attention metadata.
@@ -115,7 +107,7 @@ def test_prepare_prompt(
         max_num_batched_tokens=100000,
         max_num_seqs=100000,
         enable_chunked_prefill=False,
-        enforce_eager=enforce_eager,
+        enforce_eager=True,
     )
 
     seq_lens: List[int] = []
@@ -281,11 +273,7 @@ def test_prepare_prompt(
                     "unsupported for encoder/ "
                     "decoder models")
 @pytest.mark.parametrize("batch_size", BATCH_SIZES)
-@pytest.mark.parametrize("enforce_eager", ENFORCE_EAGER)
-def test_prepare_decode(
-    batch_size,
-    enforce_eager,
-):
+def test_prepare_decode(batch_size):
     '''
     Test the ability of the encoder/decoder model runner subclass to
     produce decode-phase model inputs & attention metadata.
@@ -311,7 +299,7 @@ def test_prepare_decode(
         max_num_batched_tokens=100000,
         max_num_seqs=100000,
         enable_chunked_prefill=False,
-        enforce_eager=enforce_eager,
+        enforce_eager=True,
     )
 
     seq_lens: List[int] = []
@@ -428,7 +416,8 @@ def test_prepare_decode(
         expected,
     )
 
-    # Cuda graph should is currently not supported for encoder/decoer.
+    # Model runner's CUDAGraph setting should be propagated to attention
+    # metadata.
     assert attn_metadata.use_cuda_graph is False
 
     # Verify the lengths of input tokens & positions
@@ -484,3 +473,152 @@ def test_prepare_decode(
         dtype=actual.dtype,
     )
     assert torch.equal(actual, expected)
+
+
+@pytest.mark.parametrize("batch_size", list(range(1, 257)))
+def test_prepare_decode_cuda_graph(batch_size):
+    """
+    Tests that for encoder-decoder models with CUDA Graph capture and replay
+    enabled, the tensors used during the decode phase are correctly padded 
+    for varying input batch sizes.
+    """
+    model_runner = _create_model_runner(
+        "facebook/bart-base",
+        seed=0,
+        dtype="float16",
+        max_num_batched_tokens=100000,
+        max_num_seqs=100000,
+        enable_chunked_prefill=False,
+        enforce_eager=False,
+    )
+
+    seq_lens: List[int] = []
+    encoder_seq_lens: List[int] = []
+    seq_group_metadata_list: List[SequenceGroupMetadata] = []
+    block_tables = {0: [1]}
+    cross_block_table = [2]
+    for i in range(batch_size):
+        # make sure all tokens fit into one block
+        seq_len = i % (model_runner.block_size - 1) + 1
+        seq_lens.append(seq_len)
+        seq_data = SequenceData(
+            array(VLLM_TOKEN_ID_ARRAY_TYPE, (range(seq_len))))
+        encoder_seq_len = (i + 1) % (model_runner.block_size - 1) + 1
+        encoder_seq_lens.append(encoder_seq_len)
+        encoder_seq_data = SequenceData(
+            array(VLLM_TOKEN_ID_ARRAY_TYPE, (range(encoder_seq_len))))
+        seq_group_metadata = SequenceGroupMetadata(
+            request_id=f"test_{i}",
+            is_prompt=False,
+            seq_data={0: seq_data},
+            sampling_params=SamplingParams(temperature=0),
+            block_tables=block_tables,
+            encoder_seq_data=encoder_seq_data,
+            cross_block_table=cross_block_table,
+        )
+        assert seq_group_metadata.token_chunk_size == 1
+        seq_group_metadata_list.append(seq_group_metadata)
+
+    model_input = model_runner.prepare_model_input(seq_group_metadata_list)
+    input_tokens = model_input.input_tokens
+    input_positions = model_input.input_positions
+    attn_metadata = model_input.attn_metadata
+    return_seq_lens = model_input.seq_lens
+    slot_mapping = attn_metadata.slot_mapping
+    encoder_input_tokens = model_input.encoder_input_tokens
+    encoder_input_positions = model_input.encoder_input_positions
+    cross_slot_mapping = attn_metadata.cross_slot_mapping
+
+    # With CUDA Graph capture and replay enabled, the decoder and encoder
+    # input sequences will be padded. Create the expected padded tensors
+    # accordingly.
+    graph_batch_size = _get_graph_batch_size(batch_size)
+    cuda_graph_pad_size = graph_batch_size - batch_size
+    padded_seq_lens = seq_lens + list(itertools.repeat(1, cuda_graph_pad_size))
+    padded_encoder_seq_lens = encoder_seq_lens + list(
+        itertools.repeat(1, cuda_graph_pad_size))
+
+    assert return_seq_lens == padded_seq_lens
+    assert len(slot_mapping) == len(input_tokens)
+    assert len(cross_slot_mapping) == len(encoder_input_tokens)
+
+    # Verify attention metadata
+    device = model_runner.device
+    assert attn_metadata.num_prefills == 0
+    assert attn_metadata.num_decode_tokens > 0
+    assert torch.equal(
+        attn_metadata.seq_lens_tensor,
+        torch.tensor(padded_seq_lens, device=device, dtype=torch.int))
+    assert attn_metadata.seq_lens == padded_seq_lens
+    assert attn_metadata.max_prefill_seq_len == 0
+    assert attn_metadata.max_decode_seq_len == max(seq_lens)
+    # - Encoder attention metadata
+    assert attn_metadata.encoder_seq_lens == padded_encoder_seq_lens
+    assert torch.equal(
+        attn_metadata.encoder_seq_lens_tensor,
+        torch.tensor(padded_encoder_seq_lens, device=device, dtype=torch.int))
+    assert attn_metadata.max_encoder_seq_len == max(padded_encoder_seq_lens)
+    assert attn_metadata.num_encoder_tokens == sum(padded_encoder_seq_lens)
+
+    # Verify block tables are correct for prompts
+    # - Decoder self-attention. Pad the block tables as expected.
+    expected = [block_tables[0] for _ in range(batch_size)]
+    expected.extend([[] for _ in range(cuda_graph_pad_size)])
+    expected = make_tensor_with_pad(
+        expected,
+        max_len=64,
+        pad=0,
+        dtype=torch.int32,
+        device=model_runner.device,
+    )
+    assert torch.equal(
+        attn_metadata.block_tables,
+        expected,
+    )
+    # - Encoder/decoder cross-attention. Pad the cross-attention block tables
+    # as expected.
+    expected = [cross_block_table for _ in range(len(seq_group_metadata_list))]
+    expected.extend([[] for _ in range(cuda_graph_pad_size)])
+    expected = make_tensor_with_pad(
+        expected,
+        max_len=64,
+        pad=0,
+        dtype=torch.int32,
+        device=model_runner.device,
+    )
+    assert torch.equal(
+        attn_metadata.cross_block_tables,
+        expected,
+    )
+
+    # Model runner's CUDAGraph setting should be propagated to attention
+    # metadata.
+    assert attn_metadata.use_cuda_graph is True
+
+    # Verify the lengths of input tokens & positions
+    # - Decoder
+    assert len(input_tokens) == len(padded_seq_lens)
+    assert len(input_positions) == len(padded_seq_lens)
+    # -- An indirect check that model_input.input_tokens
+    #    and model_input.input_positions are correct -
+    #    by design of the test, the input tokens are
+    #    equal to the input position values, so if
+    #    the model_input data structure has the correct
+    #    values then these two should be equal
+    assert torch.equal(
+        input_tokens,
+        input_positions,
+    )
+    # - Encoder
+    assert len(encoder_input_tokens) == 0
+    assert len(encoder_input_tokens) == 0
+    # -- An indirect check that model_input.encoder_input_tokens
+    #    and model_input.encoder_input_positions are correct -
+    #    by design of the test, the input tokens are
+    #    equal to the input position values, so if
+    #    the model_input data structure has the correct
+    #    values then these two should be equal
+    assert torch.equal(
+        encoder_input_tokens,
+        encoder_input_positions,
+    )

vllm/attention/backends/abstract.py

@@ -156,18 +156,27 @@ class AttentionState(ABC, Generic[T]):
         ...
 
     @abstractmethod
-    def graph_capture_get_metadata_for_batch(self, batch_size: int) -> T:
+    def graph_capture_get_metadata_for_batch(
+            self,
+            batch_size: int,
+            is_encoder_decoder_model: bool = False) -> T:
         """Get attention metadata for CUDA graph capture of batch_size."""
         ...
 
     @abstractmethod
-    def get_graph_input_buffers(self, attn_metadata: T) -> Dict[str, Any]:
+    def get_graph_input_buffers(
+            self,
+            attn_metadata: T,
+            is_encoder_decoder_model: bool = False) -> Dict[str, Any]:
         """Get attention-specific input buffers for CUDA graph capture."""
         ...
 
     @abstractmethod
-    def prepare_graph_input_buffers(self, input_buffers: Dict[str, Any],
-                                    attn_metadata: T) -> None:
+    def prepare_graph_input_buffers(
+            self,
+            input_buffers: Dict[str, Any],
+            attn_metadata: T,
+            is_encoder_decoder_model: bool = False) -> None:
         """In-place modify input buffers dict for CUDA graph replay."""
         ...
 

vllm/attention/backends/flashinfer.py

@@ -172,7 +172,8 @@ class FlashInferState(AttentionState):
         state._prefill_wrapper = self._get_prefill_wrapper()
         return state
 
-    def graph_capture_get_metadata_for_batch(self, batch_size: int):
+    def graph_capture_get_metadata_for_batch(
+            self, batch_size: int, is_encoder_decoder_model: bool = False):
         assert self._is_graph_capturing
         _indptr_buffer = self._graph_indptr_buffer[:batch_size + 1]
         _last_page_len_buffer = self._graph_last_page_len_buffer[:batch_size]
@@ -232,12 +233,17 @@ class FlashInferState(AttentionState):
         attn_metadata.begin_forward()
         return attn_metadata
 
-    def get_graph_input_buffers(self, attn_metadata):
+    def get_graph_input_buffers(self,
+                                attn_metadata,
+                                is_encoder_decoder_model: bool = False):
         return {
             "slot_mapping": attn_metadata.slot_mapping,
         }
 
-    def prepare_graph_input_buffers(self, input_buffers, attn_metadata):
+    def prepare_graph_input_buffers(self,
+                                    input_buffers,
+                                    attn_metadata,
+                                    is_encoder_decoder_model: bool = False):
         return
 
     def begin_forward(self, model_input):

vllm/attention/backends/utils.py

@@ -304,7 +304,8 @@ class CommonAttentionState(AttentionState):
         assert self._is_graph_capturing
         return self.__class__(self.runner)
 
-    def graph_capture_get_metadata_for_batch(self, batch_size: int):
+    def graph_capture_get_metadata_for_batch(
+            self, batch_size: int, is_encoder_decoder_model: bool = False):
         assert self._is_graph_capturing
         attn_metadata = self.runner.attn_backend.make_metadata(
             num_prefills=0,
@@ -322,21 +323,121 @@ class CommonAttentionState(AttentionState):
             block_tables=self._graph_block_tables[:batch_size],
             use_cuda_graph=True,
         )
+        if is_encoder_decoder_model:
+            # The encoder decoder model works only with XFormers backend.
+            # Assert the same.
+            assert self.runner.attn_backend.get_name() == "xformers", \
+            f"Expected attn_backend name to be 'xformers', but "\
+            f" got '{self.runner.attn_backend.get_name()}'"
+            self._update_captured_metadata_for_enc_dec_model(
+                batch_size=batch_size, attn_metadata=attn_metadata)
+
         return attn_metadata
 
-    def get_graph_input_buffers(self, attn_metadata) -> Dict[str, Any]:
-        return {
+    def get_graph_input_buffers(
+            self,
+            attn_metadata,
+            is_encoder_decoder_model: bool = False) -> Dict[str, Any]:
+        input_buffers = {
             "slot_mapping": attn_metadata.slot_mapping,
             "seq_lens_tensor": attn_metadata.decode_metadata.seq_lens_tensor,
             "block_tables": attn_metadata.decode_metadata.block_tables,
         }
-
-    def prepare_graph_input_buffers(self, input_buffers,
-                                    attn_metadata) -> None:
+        if is_encoder_decoder_model:
+            # The encoder decoder model works only with XFormers backend.
+            # Assert the same.
+            assert self.runner.attn_backend.get_name() == "xformers", \
+            f"Expected attn_backend name to be 'xformers', but "\
+            f" got '{self.runner.attn_backend.get_name()}'"
+            self._add_additonal_input_buffers_for_enc_dec_model(
+                attn_metadata=attn_metadata, input_buffers=input_buffers)
+        return input_buffers
+
+    def prepare_graph_input_buffers(
+            self,
+            input_buffers,
+            attn_metadata,
+            is_encoder_decoder_model: bool = False) -> None:
         input_buffers["seq_lens_tensor"].copy_(
             attn_metadata.decode_metadata.seq_lens_tensor, non_blocking=True)
         input_buffers["block_tables"].copy_(
             attn_metadata.decode_metadata.block_tables, non_blocking=True)
+        if is_encoder_decoder_model:
+            # The encoder decoder model works only with XFormers backend.
+            # Assert the same.
+            assert self.runner.attn_backend.get_name() == "xformers", \
+            f"Expected attn_backend name to be 'xformers', but "\
+            f" got '{self.runner.attn_backend.get_name()}'"
+            self._prepare_input_buffers_for_enc_dec_model(
+                attn_metadata, input_buffers)
 
     def begin_forward(self, model_input) -> None:
         return
+
+    def _update_captured_metadata_for_enc_dec_model(self, batch_size: int,
+                                                    attn_metadata):
+        """
+        Updates the attention metadata parameters for CUDA graph capture in an
+        encoder-decoder model.
+
+        This method modifies attention-related tensors and metadata required
+        for CUDA graph capture in encoder-decoder models. Specifically, it
+        updates the cross-attention and encoder sequence tensors in the 
+        AttentionMetadata object.
+        """
+        # During decode phase the cross_slot_mapping will be empty. Hence set
+        # an empty tensor for CUDA Graph capture.
+        attn_metadata.cross_slot_mapping = torch.tensor(
+            [], dtype=torch.int).cuda()
+        attn_metadata.cross_block_tables = torch.full(
+            (batch_size, self.runner.get_max_block_per_batch()),
+            1,
+            dtype=torch.int).cuda()
+        attn_metadata.encoder_seq_lens = torch.full((batch_size, ),
+                                                    1,
+                                                    dtype=torch.int).cuda()
+        attn_metadata.encoder_seq_lens_tensor = torch.full(
+            (batch_size, ), 1, dtype=torch.int).cuda()
+        attn_metadata.max_encoder_seq_len = self.runner.max_seq_len_to_capture
+
+    def _add_additonal_input_buffers_for_enc_dec_model(
+            self, attn_metadata, input_buffers: Dict[str, Any]):
+        """
+        Saves additional input buffers specific to the encoder-decoder model
+        from the attention metadata.
+
+        This method extracts and stores encoder-decoder related input buffers
+        from the `attn_metadata` into the `input_buffers` dictionary. The
+        buffers include encoder sequence lengths, cross-slot mappings, and
+        cross-block tables, which are essential for the encoder-decoder model
+        during CUDA graph replay.
+        """
+        input_buffers["encoder_seq_lens_tensor"] = (
+            attn_metadata.decode_metadata.encoder_seq_lens_tensor)
+        input_buffers["cross_slot_mapping"] = (
+            attn_metadata.decode_metadata.cross_slot_mapping)
+        input_buffers["cross_block_tables"] = (
+            attn_metadata.decode_metadata.cross_block_tables)
+
+    def _prepare_input_buffers_for_enc_dec_model(self, attn_metadata,
+                                                 input_buffers: Dict[str,
+                                                                     Any]):
+        """
+        Populates input buffers with data from the encoder-decoder model's
+        attention metadata.
+
+        This method fills the input buffers with encoder-decoder specific
+        tensors. It copies data from the `attn_metadata` and keyword arguments
+        (`kwargs`) into corresponding buffers in the `input_buffers` dictionary.
+        The copied data includes attention-related metadata as well as input 
+        IDs and positional information for the encoder.
+        """
+        input_buffers["encoder_seq_lens_tensor"].copy_(
+            attn_metadata.decode_metadata.encoder_seq_lens_tensor,
+            non_blocking=True)
+        input_buffers["cross_slot_mapping"].copy_(
+            attn_metadata.decode_metadata.cross_slot_mapping,
+            non_blocking=True)
+        input_buffers["cross_block_tables"].copy_(
+            attn_metadata.decode_metadata.cross_block_tables,
+            non_blocking=True)

vllm/config.py

@@ -16,9 +16,8 @@ from vllm.tracing import is_otel_available, otel_import_error_traceback
 from vllm.transformers_utils.config import (ConfigFormat, get_config,
                                             get_hf_image_processor_config,
                                             get_hf_text_config)
-from vllm.utils import (STR_NOT_IMPL_ENC_DEC_CUDAGRAPH, GiB_bytes,
-                        cuda_device_count_stateless, get_cpu_memory, is_cpu,
-                        is_hip, is_neuron, is_openvino, is_xpu,
+from vllm.utils import (GiB_bytes, cuda_device_count_stateless, get_cpu_memory,
+                        is_cpu, is_hip, is_neuron, is_openvino, is_xpu,
                         print_warning_once)
 
 if TYPE_CHECKING:
@@ -96,15 +95,15 @@ class ModelConfig:
         enforce_eager: Whether to enforce eager execution. If True, we will
             disable CUDA graph and always execute the model in eager mode.
             If False, we will use CUDA graph and eager execution in hybrid.
-            If None, the user did not specify, so default to False -
-            except for encoder/decoder models, which currently require
-            eager mode.
+            If None, the user did not specify, so default to False.
         max_context_len_to_capture: Maximum context len covered by CUDA graphs.
             When a sequence has context length larger than this, we fall back
             to eager mode (DEPRECATED. Use max_seq_len_to_capture instead).
         max_seq_len_to_capture: Maximum sequence len covered by CUDA graphs.
             When a sequence has context length larger than this, we fall back
-            to eager mode
+            to eager mode. Additionally for encoder-decoder models, if the
+            sequence length of the encoder input is larger than this, we fall
+            back to the eager mode.
         disable_sliding_window: Whether to disable sliding window. If True,
             we will disable the sliding window functionality of the model.
             If the model does not support sliding window, this argument is
@@ -186,32 +185,8 @@ class ModelConfig:
         self.dtype = _get_and_verify_dtype(self.hf_text_config, dtype)
         self.use_async_output_proc = use_async_output_proc
 
-        # Choose a default enforce_eager value if the user did not specify
-        # a value (enforce_eager is None)
-        if getattr(self.hf_config, 'is_encoder_decoder', False):
-            if self.enforce_eager is None:
-                # *Only for encoder/decoder models* and
-                # *only if enforce_eager is unset*, override
-                # to enforce_eager=True
-                #
-                # Add a logger message since it is *somewhat* non-intuitive that
-                # enforce_eager is True when the user has not specified its
-                # value.
-                logger.info("Forcing enforce_eager == True because "
-                            "enforce_eager setting was unspecified and "
-                            "CUDAGraph is not supported with encoder/ "
-                            "decoder models.")
-                self.enforce_eager = True
-
-            if not self.enforce_eager:
-                # Eager mode explicitly disabled by user for an encoder/
-                # decoder model; however CUDAGRAPH + encoder/decoder is
-                # not currently supported
-                raise ValueError(STR_NOT_IMPL_ENC_DEC_CUDAGRAPH)
-        elif self.enforce_eager is None:
-            # *Only for decoder-only models*, enforce_eager
-            # defaults to False if unset. This is intuitive
-            # so no logging message needed.
+        # Set enforce_eager to False if the value is unset.
+        if self.enforce_eager is None:
             self.enforce_eager = False
 
         if (not self.disable_sliding_window

vllm/engine/arg_utils.py

@@ -472,7 +472,10 @@ class EngineArgs:
                             default=EngineArgs.max_seq_len_to_capture,
                             help='Maximum sequence length covered by CUDA '
                             'graphs. When a sequence has context length '
-                            'larger than this, we fall back to eager mode.')
+                            'larger than this, we fall back to eager mode. '
+                            'Additionally for encoder-decoder models, if the '
+                            'sequence length of the encoder input is larger '
+                            'than this, we fall back to the eager mode.')
         parser.add_argument('--disable-custom-all-reduce',
                             action='store_true',
                             default=EngineArgs.disable_custom_all_reduce,

vllm/entrypoints/llm.py

@@ -88,7 +88,9 @@ class LLM:
             to eager mode (DEPRECATED. Use `max_seq_len_to_capture` instead).
         max_seq_len_to_capture: Maximum sequence len covered by CUDA graphs.
             When a sequence has context length larger than this, we fall back
-            to eager mode.
+            to eager mode. Additionally for encoder-decoder models, if the
+            sequence length of the encoder input is larger than this, we fall
+            back to the eager mode.
         disable_custom_all_reduce: See ParallelConfig
         **kwargs: Arguments for :class:`~vllm.EngineArgs`. (See
             :ref:`engine_args`)
@@ -137,9 +139,7 @@ class LLM:
         LLM constructor.
 
         Note: if enforce_eager is unset (enforce_eager is None)
-        it defaults to False for decoder-only models and True
-        for encoder/decoder models, since encoder/decoder models
-        do not currently support CUDAGraph.
+        it defaults to False.
         '''
 
         if "disable_log_stats" not in kwargs:

vllm/model_executor/models/bart.py

@@ -848,11 +848,13 @@ class BartForConditionalGeneration(nn.Module):
         self,
         input_ids: torch.Tensor,
         positions: torch.Tensor,
-        encoder_input_ids: torch.Tensor,
-        encoder_positions: torch.Tensor,
         kv_caches: List[torch.Tensor],
         attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors] = None,
+        *,
+        encoder_input_ids: torch.Tensor,
+        encoder_positions: torch.Tensor,
+        **kwargs,
     ) -> torch.Tensor:
         r"""
         Args:

vllm/utils.py

@@ -71,10 +71,6 @@ STR_NOT_IMPL_ENC_DEC_SPEC_DEC = ("Speculative decoding is not "
                                  "currently supported with encoder/"
                                  "decoder models.")
 
-STR_NOT_IMPL_ENC_DEC_CUDAGRAPH = ("CUDAGraph is not "
-                                  "currently supported with encoder/"
-                                  "decoder models.")
-
 STR_NOT_IMPL_ENC_DEC_BACKEND = ("XFormers is the only backend "
                                 "currently supported with encoder/"
                                 "decoder models.")
@@ -98,7 +94,6 @@ STR_NOT_IMPL_ENC_DEC_ERR_STRS = {
     "STR_NOT_IMPL_ENC_DEC_PP": STR_NOT_IMPL_ENC_DEC_PP,
     "STR_NOT_IMPL_ENC_DEC_MM": STR_NOT_IMPL_ENC_DEC_MM,
     "STR_NOT_IMPL_ENC_DEC_SPEC_DEC": STR_NOT_IMPL_ENC_DEC_SPEC_DEC,
-    "STR_NOT_IMPL_ENC_DEC_CUDA_GRAPH": STR_NOT_IMPL_ENC_DEC_CUDAGRAPH,
     "STR_NOT_IMPL_ENC_DEC_BACKEND": STR_NOT_IMPL_ENC_DEC_BACKEND,
     "STR_NOT_IMPL_ENC_DEC_PROMPT_ADAPTER": STR_NOT_IMPL_ENC_DEC_PROMPT_ADAPTER,
     "STR_NOT_IMPL_ENC_DEC_CPU": STR_NOT_IMPL_ENC_DEC_CPU

vllm/worker/enc_dec_model_runner.py

 import dataclasses
+import itertools
 from typing import Any, Dict, List, Optional, Tuple, Type, cast
 
 import torch
@@ -24,7 +25,8 @@ from vllm.sequence import (IntermediateTensors, PoolerOutput,
 from vllm.utils import STR_NOT_IMPL_ENC_DEC_BACKEND, make_tensor_with_pad
 from vllm.worker.model_runner import (GPUModelRunnerBase,
                                       ModelInputForGPUBuilder,
-                                      ModelInputForGPUWithSamplingMetadata)
+                                      ModelInputForGPUWithSamplingMetadata,
+                                      _get_graph_batch_size)
 from vllm.worker.model_runner_base import (
     _add_attn_metadata_broadcastable_dict,
     _add_sampling_metadata_broadcastable_dict)
@@ -178,7 +180,15 @@ class EncoderDecoderModelRunner(GPUModelRunnerBase[EncoderDecoderModelInput]):
             raise ValueError("num_steps > 1 is not supported in "
                              "EncoderDecoderModelRunner")
 
-        model_executable = self.model
+        if (model_input.attn_metadata is not None
+                and model_input.attn_metadata.prefill_metadata is None
+                and model_input.attn_metadata.decode_metadata.use_cuda_graph):
+            assert model_input.input_tokens is not None
+            graph_batch_size = model_input.input_tokens.shape[0]
+            model_executable = self.graph_runners[
+                model_input.virtual_engine][graph_batch_size]
+        else:
+            model_executable = self.model
 
         seqlen_agnostic_kwargs = {
             "finished_requests_ids": model_input.finished_requests_ids,
@@ -200,6 +210,9 @@ class EncoderDecoderModelRunner(GPUModelRunnerBase[EncoderDecoderModelInput]):
         if not self.is_driver_worker:
             return []
 
+        if model_input.async_callback is not None:
+            model_input.async_callback()
+
         # Sample the next token.
         output: SamplerOutput = self.model.sample(
             logits=logits,
@@ -231,14 +244,12 @@ class EncoderDecoderModelRunner(GPUModelRunnerBase[EncoderDecoderModelInput]):
         """
         model_input = self._prepare_model_input_tensors(
             seq_group_metadata_list, finished_requests_ids)
-
         (
             attn_metadata,
             encoder_input_tokens_tensor,
             encoder_input_positions_tensor,
         ) = (self._prepare_encoder_model_input_tensors(seq_group_metadata_list,
                                                        model_input))
-
         # Inject attn_metadata encoder/cross-attention fields &
         # encoder input tokens/positions into model_input.
         # Frozen dataclass fields cannot be modified, so use
@@ -437,11 +448,29 @@ class EncoderDecoderModelRunner(GPUModelRunnerBase[EncoderDecoderModelInput]):
                 cross_block_tables.append([] if (
                     cross_block_table is None) else cross_block_table)
 
-            # Convert cross-attention block tables to encoder input tensor
+            if (model_input.attn_metadata is not None
+                    and model_input.attn_metadata.use_cuda_graph):
+                # We will be using CUDA graph replay for this decode.
+                max_len_of_block_table = self.get_max_block_per_batch()
+                batch_size = len(encoder_seq_lens)
+                graph_batch_size = _get_graph_batch_size(batch_size)
+                assert graph_batch_size >= batch_size
+                cuda_graph_pad_size = graph_batch_size - batch_size
+                # extend the cross_block_tables and encoder_seq_lens to match
+                # the graph_batch_size.
+                cross_block_tables.extend([[]
+                                           for _ in range(cuda_graph_pad_size)
+                                           ])
+                encoder_seq_lens.extend(
+                    itertools.repeat(1, cuda_graph_pad_size))
+
+            else:
+                max_len_of_block_table = max(
+                    len(block_table) for block_table in cross_block_tables)
+
             cross_block_tables = make_tensor_with_pad(
                 cross_block_tables,
-                max_len=max(
-                    len(block_table) for block_table in cross_block_tables),
+                max_len=max_len_of_block_table,
                 pad=0,
                 dtype=torch.int32,
                 device=self.device,

vllm/worker/model_runner.py

@@ -243,6 +243,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             prefix_cache_hit: bool = False,
             reinit: bool = False,
             reinit_use_defaults: bool = False,
+            encoder_seq_len: int = 0,
         ):
             if reinit:
                 assert len(self.seq_ids) == len(seq_ids)  # type: ignore
@@ -256,6 +257,7 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             self.block_tables = block_tables
             self.computed_block_nums = computed_block_nums
             self.n_seqs = n_seqs
+            self.encoder_seq_len = encoder_seq_len
 
             if reinit:
                 if len(self.seq_ids) == 1 and reinit_use_defaults:
@@ -702,6 +704,11 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             assert n_seqs == 1
             self.decode_only = False
 
+        encoder_seq_len = 0
+
+        if self.runner.model_config.is_encoder_decoder_model:
+            encoder_seq_len = seq_group_metadata.encoder_seq_data.get_len()
+
         inter_data = self.init_cached_inter_data(
             request_id=seq_group_metadata.request_id,
             seq_ids=seq_ids,
@@ -709,7 +716,8 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
             block_tables=seq_group_metadata.block_tables,
             computed_block_nums=seq_group_metadata.computed_block_nums,
             reinit=True,
-            reinit_use_defaults=True)
+            reinit_use_defaults=True,
+            encoder_seq_len=encoder_seq_len)
 
         self.inter_data_list.append(inter_data)
 
@@ -719,11 +727,15 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
         for per_seq_group_fn in self.per_seq_group_compute_fns:
             per_seq_group_fn(inter_data, seq_group_metadata)
 
-    def _use_captured_graph(self, batch_size: int,
-                            max_decode_seq_len: int) -> bool:
+    def _use_captured_graph(self,
+                            batch_size: int,
+                            max_decode_seq_len: int,
+                            max_encoder_seq_len: int = 0) -> bool:
         return (self.decode_only and not self.runner.model_config.enforce_eager
-                and batch_size <= self.runner.max_batchsize_to_capture
-                and max_decode_seq_len <= self.runner.max_seq_len_to_capture)
+                and batch_size <= _BATCH_SIZES_TO_CAPTURE[-1]
+                and max_decode_seq_len <= self.runner.max_seq_len_to_capture
+                and max_encoder_seq_len <= self.runner.max_seq_len_to_capture
+                and batch_size <= self.runner.max_batchsize_to_capture)
 
     def build(self) -> ModelInputForGPU:
         """Finalize the builder intermediate data and
@@ -763,15 +775,18 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
                     input_positions.extend(cur_input_positions)
 
         seq_lens = []
+        query_lens = []
         max_decode_seq_len = 0
+        max_encoder_seq_len = 0
         for inter_data in self.inter_data_list:
             seq_lens.extend(inter_data.seq_lens)
+            query_lens.extend(inter_data.query_lens)
             if not inter_data.is_prompt:
                 max_decode_seq_len = max(max_decode_seq_len,
                                          max(inter_data.seq_lens))
-        query_lens = []
-        for inter_data in self.inter_data_list:
-            query_lens.extend(inter_data.query_lens)
+                if self.runner.model_config.is_encoder_decoder_model:
+                    max_encoder_seq_len = max(max_encoder_seq_len,
+                                              inter_data.encoder_seq_len)
 
         # Mapping from request IDs to sequence IDs. Used for Jamba models
         # that manages the cache by itself.
@@ -781,8 +796,10 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
         }
 
         batch_size = len(input_tokens)
-        use_captured_graph = self._use_captured_graph(batch_size,
-                                                      max_decode_seq_len)
+        use_captured_graph = self._use_captured_graph(
+            batch_size,
+            max_decode_seq_len,
+            max_encoder_seq_len=max_encoder_seq_len)
 
         # If cuda graph can be used, pad tensors accordingly.
         # See `capture_model` API for more details.
@@ -1364,7 +1381,9 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
                 for batch_size in reversed(batch_size_capture_list):
                     attn_metadata = (
                         self.attn_state.graph_capture_get_metadata_for_batch(
-                            batch_size))
+                            batch_size,
+                            is_encoder_decoder_model=self.model_config.
+                            is_encoder_decoder_model))
 
                     if self.lora_config:
                         lora_mapping = LoRAMapping(
@@ -1380,10 +1399,10 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
                         )
                         self.set_active_prompt_adapters(
                             set(), prompt_adapter_mapping)
-
                     graph_runner = CUDAGraphRunner(
                         self.model, self.attn_backend.get_name(),
-                        self.attn_state.graph_clone(batch_size))
+                        self.attn_state.graph_clone(batch_size),
+                        self.model_config.is_encoder_decoder_model)
 
                     capture_inputs = {
                         "input_ids":
@@ -1420,6 +1439,12 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
                             self.model.get_seqlen_agnostic_capture_inputs(
                                 batch_size)
                         })
+                    if self.model_config.is_encoder_decoder_model:
+                        # add the additional inputs to capture for
+                        # encoder-decoder models.
+                        self._update_inputs_to_capture_for_enc_dec_model(
+                            capture_inputs)
+
                     graph_runner.capture(**capture_inputs)
                     self.graph_memory_pool = graph_runner.graph.pool()
                     self.graph_runners[virtual_engine][batch_size] = (
@@ -1430,6 +1455,24 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
         # This usually takes < 10 seconds.
         logger.info("Graph capturing finished in %.0f secs.", elapsed_time)
 
+    def _update_inputs_to_capture_for_enc_dec_model(self,
+                                                    capture_inputs: Dict[str,
+                                                                         Any]):
+        """
+        Updates the set of input tensors needed for CUDA graph capture in an
+        encoder-decoder model.
+
+        This method modifies the provided `capture_inputs` dictionary by
+        adding tensors specific to encoder-decoder specific models that
+        need to be captured for CUDA Graph replay.
+        """
+        # During the decode phase encoder_input_ids and encoder_positions are
+        # unset. Do the same thing for graph capture.
+        capture_inputs["encoder_input_ids"] = torch.tensor(
+            [], dtype=torch.long).cuda()
+        capture_inputs["encoder_positions"] = torch.tensor(
+            [], dtype=torch.long).cuda()
+
     @property
     def vocab_size(self) -> int:
         return self.model_config.get_vocab_size()
@@ -1629,7 +1672,7 @@ class ModelRunner(GPUModelRunnerBase[ModelInputForGPUWithSamplingMetadata]):
 class CUDAGraphRunner:
 
     def __init__(self, model: nn.Module, backend_name: str,
-                 attn_state: AttentionState):
+                 attn_state: AttentionState, is_encoder_decoder_model: bool):
         self.model = model
         self.backend_name = backend_name
         self.attn_state = attn_state
@@ -1638,6 +1681,7 @@ class CUDAGraphRunner:
         self.output_buffers: Dict[str, torch.Tensor] = {}
 
         self._graph: Optional[torch.cuda.CUDAGraph] = None
+        self._is_encoder_decoder_model = is_encoder_decoder_model
 
     @property
     def graph(self):
@@ -1671,8 +1715,9 @@ class CUDAGraphRunner:
                 intermediate_tensors=intermediate_inputs,
                 **kwargs,
             )
+        # Wait for the warm up operations to finish before proceeding with
+        # Graph Capture.
         torch.cuda.synchronize()
-
         # Capture the graph.
         self._graph = torch.cuda.CUDAGraph()
         with torch.cuda.graph(self._graph, pool=memory_pool, stream=stream):
@@ -1704,10 +1749,14 @@ class CUDAGraphRunner:
 
         # Save the input and output buffers.
         self.input_buffers = {
-            "input_ids": input_ids,
-            "positions": positions,
-            "kv_caches": kv_caches,
-            **self.attn_state.get_graph_input_buffers(attn_metadata),
+            "input_ids":
+            input_ids,
+            "positions":
+            positions,
+            "kv_caches":
+            kv_caches,
+            **self.attn_state.get_graph_input_buffers(
+                attn_metadata, self._is_encoder_decoder_model),
             **kwargs,
         }
         if intermediate_inputs is not None:
@@ -1737,8 +1786,8 @@ class CUDAGraphRunner:
         self.input_buffers["positions"].copy_(positions, non_blocking=True)
         self.input_buffers["slot_mapping"].copy_(attn_metadata.slot_mapping,
                                                  non_blocking=True)
-        self.attn_state.prepare_graph_input_buffers(self.input_buffers,
-                                                    attn_metadata)
+        self.attn_state.prepare_graph_input_buffers(
+            self.input_buffers, attn_metadata, self._is_encoder_decoder_model)
         if "seqlen_agnostic_capture_inputs" in self.input_buffers:
             self.model.copy_inputs_before_cuda_graphs(self.input_buffers,
                                                       **kwargs)
@@ -1752,6 +1801,12 @@ class CUDAGraphRunner:
                 if key != "model_execute_time" and key != "model_forward_time":
                     self.input_buffers[key].copy_(intermediate_tensors[key],
                                                   non_blocking=True)
+        if self._is_encoder_decoder_model:
+            self.input_buffers["encoder_input_ids"].copy_(
+                kwargs['encoder_input_ids'], non_blocking=True)
+            self.input_buffers["encoder_positions"].copy_(
+                kwargs['encoder_positions'], non_blocking=True)
+
         # Run the graph.
         self.graph.replay()
         # Return the output tensor.

vllm/worker/utils.py

@@ -47,10 +47,6 @@ def assert_enc_dec_mr_supported_scenario(
         raise NotImplementedError(
             STR_NOT_IMPL_ENC_DEC_ERR_STRS['STR_NOT_IMPL_ENC_DEC_SPEC_DEC'])
 
-    if not enc_dec_mr.model_config.enforce_eager:
-        raise NotImplementedError(
-            STR_NOT_IMPL_ENC_DEC_ERR_STRS['STR_NOT_IMPL_ENC_DEC_CUDA_GRAPH'])
-
     if enc_dec_mr.prompt_adapter_config is not None:
         raise NotImplementedError(STR_NOT_IMPL_ENC_DEC_ERR_STRS[
             'STR_NOT_IMPL_ENC_DEC_PROMPT_ADAPTER'])