[core][distributed] simplify code to support pipeline parallel (vllm-…

…project#6406)

.buildkite/test-pipeline.yaml

@@ -46,9 +46,7 @@ steps:
   fast_check: true
   commands:
   - pip install https://github.com/flashinfer-ai/flashinfer/releases/download/v0.0.8/flashinfer-0.0.8+cu121torch2.3-cp310-cp310-linux_x86_64.whl
-  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_basic_correctness.py
-  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_basic_correctness.py
-  - VLLM_ATTENTION_BACKEND=FLASHINFER pytest -v -s basic_correctness/test_basic_correctness.py
+  - pytest -v -s basic_correctness/test_basic_correctness.py
   - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_chunked_prefill.py
   - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
   - VLLM_TEST_ENABLE_ARTIFICIAL_PREEMPT=1 pytest -v -s basic_correctness/test_preemption.py

tests/basic_correctness/test_basic_correctness.py

@@ -28,10 +28,8 @@ def test_vllm_gc_ed():
     assert weak_llm() is None
 
 
-@pytest.mark.skipif(is_hip()
-                    and os.getenv("VLLM_ATTENTION_BACKEND") == "FLASHINFER",
-                    reason="Flashinfer does not support ROCm/HIP.")
 @pytest.mark.parametrize("model", MODELS)
+@pytest.mark.parametrize("backend", ["FLASH_ATTN", "XFORMERS", "FLASHINFER"])
 @pytest.mark.parametrize("dtype", ["half"])
 @pytest.mark.parametrize("max_tokens", [5])
 @pytest.mark.parametrize("enforce_eager", [False, True])
@@ -40,10 +38,17 @@ def test_models(
     vllm_runner,
     example_prompts,
     model: str,
+    backend: str,
     dtype: str,
     max_tokens: int,
     enforce_eager: bool,
 ) -> None:
+
+    if backend == "FLASHINFER" and is_hip():
+        pytest.skip("Flashinfer does not support ROCm/HIP.")
+
+    os.environ["VLLM_ATTENTION_BACKEND"] = backend
+
     with hf_runner(model, dtype=dtype) as hf_model:
         hf_outputs = hf_model.generate_greedy(example_prompts, max_tokens)
 

vllm/model_executor/models/gpt2.py

@@ -27,7 +27,6 @@
 from vllm.config import CacheConfig
 from vllm.distributed.parallel_state import (
     get_pp_group, get_tensor_model_parallel_world_size)
-from vllm.distributed.utils import get_pp_indices
 from vllm.model_executor.layers.activation import get_act_fn
 from vllm.model_executor.layers.linear import (ColumnParallelLinear,
                                                QKVParallelLinear,
@@ -42,6 +41,8 @@
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors, SamplerOutput
 
+from .utils import is_pp_missing_parameter, make_layers
+
 
 class GPT2Attention(nn.Module):
 
@@ -183,18 +184,9 @@ def __init__(
         self.embed_dim = config.hidden_size
         self.wte = VocabParallelEmbedding(config.vocab_size, self.embed_dim)
         self.wpe = nn.Embedding(config.max_position_embeddings, self.embed_dim)
-        self.start_layer, self.end_layer = get_pp_indices(
+        self.start_layer, self.end_layer, self.h = make_layers(
             config.num_hidden_layers,
-            get_pp_group().rank_in_group,
-            get_pp_group().world_size)
-        self.h = nn.ModuleList(
-            [nn.Identity() for _ in range(self.start_layer)] + [
-                GPT2Block(config, cache_config, quant_config)
-                for _ in range(self.start_layer, self.end_layer)
-            ] + [
-                nn.Identity()
-                for _ in range(self.end_layer, config.num_hidden_layers)
-            ])
+            lambda: GPT2Block(config, cache_config, quant_config))
         self.ln_f = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
 
     def forward(
@@ -291,19 +283,20 @@ def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
                 continue
             if not name.startswith("transformer."):
                 name = "transformer." + name
-            try:
-                param = params_dict[name]
-                # The HF's GPT-2 implementation uses Conv1D instead of Linear.
-                # Because of this, we need to transpose the weights.
-                # Note(zhuohan): the logic below might break quantized models.
-                for conv1d_weight_name in ["c_attn", "c_proj", "c_fc"]:
-                    if conv1d_weight_name not in name:
-                        continue
-                    if not name.endswith(".weight"):
-                        continue
-                    loaded_weight = loaded_weight.t()
-                weight_loader = getattr(param, "weight_loader",
-                                        default_weight_loader)
-                weight_loader(param, loaded_weight)
-            except KeyError:
+
+            if is_pp_missing_parameter(name, self):
                 continue
+
+            param = params_dict[name]
+            # The HF's GPT-2 implementation uses Conv1D instead of Linear.
+            # Because of this, we need to transpose the weights.
+            # Note(zhuohan): the logic below might break quantized models.
+            for conv1d_weight_name in ["c_attn", "c_proj", "c_fc"]:
+                if conv1d_weight_name not in name:
+                    continue
+                if not name.endswith(".weight"):
+                    continue
+                loaded_weight = loaded_weight.t()
+            weight_loader = getattr(param, "weight_loader",
+                                    default_weight_loader)
+            weight_loader(param, loaded_weight)

vllm/model_executor/models/llama.py

@@ -29,8 +29,7 @@
 
 from vllm.attention import Attention, AttentionMetadata
 from vllm.config import CacheConfig, LoRAConfig
-from vllm.distributed import (get_pp_group, get_pp_indices,
-                              get_tensor_model_parallel_rank,
+from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank,
                               get_tensor_model_parallel_world_size)
 from vllm.model_executor.layers.activation import SiluAndMul
 from vllm.model_executor.layers.layernorm import RMSNorm
@@ -51,6 +50,7 @@
 from vllm.utils import is_hip, print_warning_once
 
 from .interfaces import SupportsLoRA
+from .utils import is_pp_missing_parameter, make_layers
 
 
 class LlamaMLP(nn.Module):
@@ -262,20 +262,11 @@ def __init__(
             config.hidden_size,
             org_num_embeddings=config.vocab_size,
         )
-        self.start_layer, self.end_layer = get_pp_indices(
+        self.start_layer, self.end_layer, self.layers = make_layers(
             config.num_hidden_layers,
-            get_pp_group().rank_in_group,
-            get_pp_group().world_size)
-        self.layers = nn.ModuleList(
-            [nn.Identity() for _ in range(self.start_layer)] + [
-                LlamaDecoderLayer(config=config,
-                                  cache_config=cache_config,
-                                  quant_config=quant_config)
-                for _ in range(self.start_layer, self.end_layer)
-            ] + [
-                nn.Identity()
-                for _ in range(self.end_layer, config.num_hidden_layers)
-            ])
+            lambda: LlamaDecoderLayer(config=config,
+                                      cache_config=cache_config,
+                                      quant_config=quant_config))
         self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
 
     def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
@@ -455,12 +446,14 @@ def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
                 # Skip loading extra bias for GPTQ models.
                 if name.endswith(".bias") and name not in params_dict:
                     continue
-                try:
-                    param = params_dict[name]
-                    weight_loader = param.weight_loader
-                    weight_loader(param, loaded_weight, shard_id)
-                except KeyError:
-                    pass
+
+                if is_pp_missing_parameter(name, self):
+                    continue
+
+                param = params_dict[name]
+                weight_loader = param.weight_loader
+                weight_loader(param, loaded_weight, shard_id)
+
                 break
             else:
                 # Skip loading extra bias for GPTQ models.
@@ -479,13 +472,14 @@ def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
                         continue
                     else:
                         name = remapped_kv_scale_name
-                try:
-                    param = params_dict[name]
-                    weight_loader = getattr(param, "weight_loader",
-                                            default_weight_loader)
-                    weight_loader(param, loaded_weight)
-                except KeyError:
-                    pass
+
+                if is_pp_missing_parameter(name, self):
+                    continue
+
+                param = params_dict[name]
+                weight_loader = getattr(param, "weight_loader",
+                                        default_weight_loader)
+                weight_loader(param, loaded_weight)
 
     # If this function is called, it should always initialize KV cache scale
     # factors (or else raise an exception). Thus, handled exceptions should

vllm/model_executor/models/utils.py

@@ -1,3 +1,5 @@
+from typing import Callable, Dict, List, Tuple
+
 import torch
 
 from vllm.multimodal import BatchedTensors
@@ -39,3 +41,57 @@ def merge_vision_embeddings(input_ids: torch.Tensor,
         inputs_embeds[mask] = torch.cat(vision_embeddings)
 
     return inputs_embeds
+
+
+class PPMissingLayer(torch.nn.Identity):
+    """
+    A placeholder layer for missing layers in a pipeline parallel model.
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__()
+
+
+def make_layers(
+    num_hidden_layers: int, layer_fn: Callable[[], torch.nn.Module]
+) -> Tuple[int, int, torch.nn.ModuleList]:
+    """Make a list of layers with the given layer function, taking
+    pipeline parallelism into account.
+    """
+    from vllm.distributed.parallel_state import get_pp_group
+    from vllm.distributed.utils import get_pp_indices
+    start_layer, end_layer = get_pp_indices(num_hidden_layers,
+                                            get_pp_group().rank_in_group,
+                                            get_pp_group().world_size)
+    modules = torch.nn.ModuleList(
+        [PPMissingLayer() for _ in range(start_layer)] +
+        [layer_fn() for _ in range(start_layer, end_layer)] +
+        [PPMissingLayer() for _ in range(end_layer, num_hidden_layers)])
+    return start_layer, end_layer, modules
+
+
+# NOTE: don't use lru_cache here because it can prevent garbage collection
+_model_to_pp_missing_layer_names: Dict[int, List[str]] = {}
+
+
+def get_pp_missing_layer_names(model: torch.nn.Module) -> List[str]:
+    """Get the names of the missing layers in a pipeline parallel model."""
+    model_id = id(model)
+    if model_id in _model_to_pp_missing_layer_names:
+        return _model_to_pp_missing_layer_names[model_id]
+
+    missing_layer_names = []
+    for name, module in model.named_modules():
+        if isinstance(module, PPMissingLayer):
+            missing_layer_names.append(name)
+    _model_to_pp_missing_layer_names[model_id] = missing_layer_names
+
+    return missing_layer_names
+
+
+def is_pp_missing_parameter(name: str, model: torch.nn.Module) -> bool:
+    """Check if a parameter is missing in a pipeline parallel model."""
+    for missing_layer_name in get_pp_missing_layer_names(model):
+        if name.startswith(missing_layer_name):
+            return True
+    return False