[EPLB] Support ernie4.5-moe

vllm/model_executor/models/ernie45_moe.py

@@ -22,7 +22,9 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Inference-only ErineMoE model compatible with HuggingFace weights."""
-from collections.abc import Iterable
+
+import typing
+from collections.abc import Callable, Iterable
 from itertools import islice
 from typing import Any, Optional, Union
 
@@ -32,8 +34,9 @@
 
 from vllm.attention import Attention
 from vllm.compilation.decorators import support_torch_compile
-from vllm.config import CacheConfig, VllmConfig
-from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size
+from vllm.config import CacheConfig, VllmConfig, get_current_vllm_config
+from vllm.distributed import (get_ep_group, get_pp_group,
+                              get_tensor_model_parallel_world_size)
 from vllm.logger import init_logger
 from vllm.model_executor.layers.activation import SiluAndMul
 from vllm.model_executor.layers.fused_moe import FusedMoE
@@ -52,7 +55,7 @@
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import IntermediateTensors
 
-from .interfaces import SupportsLoRA, SupportsPP
+from .interfaces import MixtureOfExperts, SupportsLoRA, SupportsPP
 from .utils import (AutoWeightsLoader, PPMissingLayer, extract_layer_index,
                     is_pp_missing_parameter,
                     make_empty_intermediate_tensors_factory, make_layers,
@@ -104,12 +107,36 @@ def __init__(
         config: PretrainedConfig,
         quant_config: Optional[QuantizationConfig] = None,
         prefix: str = "",
+        enable_eplb: bool = False,
     ):
         super().__init__()
 
         layer_idx = extract_layer_index(prefix)
         self.layer_idx = layer_idx
         self.tp_size = get_tensor_model_parallel_world_size()
+
+        self.moe_num_shared_experts = getattr(config, "moe_num_shared_experts",
+                                              None)
+        self.ep_group = get_ep_group().device_group
+        self.ep_rank = self.ep_group.rank()
+        self.ep_size = self.ep_group.size()
+        self.n_routed_experts: int = config.moe_num_experts
+        self.n_shared_experts: int = self.moe_num_shared_experts
+
+        # Load balancing settings.
+        vllm_config = get_current_vllm_config()
+        parallel_config = vllm_config.parallel_config
+        self.enable_eplb = enable_eplb
+
+        self.n_redundant_experts = parallel_config.num_redundant_experts
+        self.n_logical_experts = self.n_routed_experts
+        self.n_physical_experts = (self.n_logical_experts +
+                                   self.n_redundant_experts)
+        self.n_local_physical_experts = self.n_physical_experts // self.ep_size
+        self.physical_expert_start = (self.ep_rank *
+                                      self.n_local_physical_experts)
+        self.physical_expert_end = (self.physical_expert_start +
+                                    self.n_local_physical_experts)
         self.has_shared_experts = (getattr(config, "moe_num_shared_experts", 0)
                                    > 0)
 
@@ -123,7 +150,6 @@ def __init__(
                                      bias=False,
                                      quant_config=None,
                                      prefix=f"{prefix}.gate")
-
         self.gate.e_score_correction_bias = nn.Parameter(
             torch.empty(config.moe_num_experts))
 
@@ -136,7 +162,9 @@ def __init__(
             renormalize=True,
             quant_config=quant_config,
             prefix=f"{prefix}.experts",
-            e_score_correction_bias=self.gate.e_score_correction_bias)
+            e_score_correction_bias=self.gate.e_score_correction_bias,
+            enable_eplb=self.enable_eplb,
+            num_redundant_experts=self.n_redundant_experts)
 
         if self.has_shared_experts:
             intermediate_size = (config.moe_intermediate_size *
@@ -275,6 +303,7 @@ def __init__(
         cache_config: Optional[CacheConfig] = None,
         quant_config: Optional[QuantizationConfig] = None,
         prefix: str = "",
+        enable_eplb: bool = False,
     ) -> None:
         super().__init__()
         self.hidden_size = config.hidden_size
@@ -313,7 +342,8 @@ def __init__(
                 and layer_idx <= moe_layer_end_index):
             self.mlp = Ernie4_5_MoeMoE(config=config,
                                        quant_config=quant_config,
-                                       prefix=f"{prefix}.mlp")
+                                       prefix=f"{prefix}.mlp",
+                                       enable_eplb=enable_eplb)
         else:
             self.mlp = Ernie4_5_MoeMLP(
                 hidden_size=config.hidden_size,
@@ -370,6 +400,9 @@ def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         self.padding_idx = config.pad_token_id
         self.vocab_size = config.vocab_size
         self.config = config
+        parallel_config = vllm_config.parallel_config
+        enable_eplb = parallel_config.enable_eplb
+        self.num_redundant_experts = parallel_config.num_redundant_experts
 
         if get_pp_group().is_first_rank:
             self.embed_tokens = VocabParallelEmbedding(
@@ -385,7 +418,8 @@ def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
             lambda prefix: Ernie4_5_MoeDecoderLayer(config=config,
                                                     cache_config=cache_config,
                                                     quant_config=quant_config,
-                                                    prefix=prefix),
+                                                    prefix=prefix,
+                                                    enable_eplb=enable_eplb),
             prefix=f"{prefix}.layers",
         )
 
@@ -441,7 +475,9 @@ def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
             ckpt_gate_proj_name="gate_proj",
             ckpt_down_proj_name="down_proj",
             ckpt_up_proj_name="up_proj",
-            num_experts=self.config.moe_num_experts)
+            num_experts=self.config.moe_num_experts,
+            num_redundant_experts=self.num_redundant_experts,
+        )
 
     def load_weights(self, weights: Iterable[tuple[str,
                                                    torch.Tensor]]) -> set[str]:
@@ -490,31 +526,51 @@ def load_weights(self, weights: Iterable[tuple[str,
                 weight_loader(param, loaded_weight, shard_id)
                 break
             else:
+                is_expert_weight = False
                 for mapping in expert_params_mapping:
                     param_name, weight_name, expert_id, shard_id = mapping
 
                     if weight_name not in name:
                         continue
 
-                    name = name.replace(weight_name, param_name)
+                    # Anyway, this is an expert weight and should not be
+                    # attempted to load as other weights later
+                    is_expert_weight = True
+
+                    # Do not modify `name` since the loop may continue here
+                    # Instead, create a new variable
+                    name_mapped = name.replace(weight_name, param_name)
                     # Skip layers on other devices.
-                    if is_pp_missing_parameter(name, self):
+                    if is_pp_missing_parameter(name_mapped, self):
                         continue
 
                     # Skip loading extra bias for GPTQ models.
-                    if ((name.endswith(".bias") or name.endswith("_bias"))
-                            and name not in params_dict):
+                    if ((name_mapped.endswith(".bias")
+                         or name_mapped.endswith("_bias"))
+                            and name_mapped not in params_dict):
                         continue
-                    param = params_dict[name]
-
-                    weight_loader = param.weight_loader
-                    weight_loader(param,
-                                  loaded_weight,
-                                  name,
-                                  shard_id=shard_id,
-                                  expert_id=expert_id)
-                    break
+                    param = params_dict[name_mapped]
+                    # We should ask the weight loader to return success or not
+                    # here since otherwise we may skip experts with other
+                    # available replicas.
+                    weight_loader = typing.cast(Callable[..., bool],
+                                                param.weight_loader)
+                    success = weight_loader(param,
+                                            loaded_weight,
+                                            name_mapped,
+                                            shard_id=shard_id,
+                                            expert_id=expert_id,
+                                            return_success=True)
+                    if success:
+                        name = name_mapped
+                        break
                 else:
+                    if is_expert_weight:
+                        # We've checked that this is an expert weight
+                        # However it's not mapped locally to this rank
+                        # So we simply skip it
+                        continue
+
                     # Skip loading extra bias for GPTQ models.
                     if ((name.endswith(".bias") or name.endswith("_bias"))
                             and name not in params_dict):
@@ -535,7 +591,8 @@ def load_weights(self, weights: Iterable[tuple[str,
         return loaded_params
 
 
-class Ernie4_5_MoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
+class Ernie4_5_MoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA,
+                              MixtureOfExperts):
     packed_modules_mapping = {
         "qkv_proj": [
             "q_proj",
@@ -572,6 +629,73 @@ def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         self.make_empty_intermediate_tensors = (
             self.model.make_empty_intermediate_tensors)
 
+        self.expert_weights = []
+
+        # Set MoE hyperparameters
+        moe_layers_indices = [
+            i for i in range(config.num_hidden_layers)
+            if (i >= config.moe_layer_start_index
+                and i <= config.moe_layer_end_index and (i + 1) %
+                config.moe_layer_interval == 0)
+        ]
+        self.num_moe_layers = len(moe_layers_indices)
+        self.num_expert_groups = 1
+
+        self.moe_layers: list[FusedMoE] = []
+        for layer in self.model.layers:
+            if isinstance(layer, PPMissingLayer):
+                continue
+
+            assert isinstance(layer, Ernie4_5_MoeDecoderLayer)
+            if isinstance(layer.mlp, Ernie4_5_MoeMoE):
+                example_moe = layer.mlp
+                self.moe_layers.append(layer.mlp.experts)
+
+        if example_moe is None:
+            raise RuntimeError(
+                "No Ernie4_5_MoeMoE layer found in model.layers.")
+
+        self.num_logical_experts = example_moe.n_logical_experts
+        self.num_physical_experts = example_moe.n_physical_experts
+        self.num_local_physical_experts = example_moe.n_local_physical_experts
+        self.num_routed_experts = example_moe.n_routed_experts
+        self.num_shared_experts = example_moe.n_shared_experts
+        self.num_redundant_experts = example_moe.n_redundant_experts
+
+    def set_eplb_state(
+        self,
+        expert_load_view: torch.Tensor,
+        logical_to_physical_map: torch.Tensor,
+        logical_replica_count: torch.Tensor,
+    ) -> None:
+        for layer_idx, layer in enumerate(self.moe_layers):
+            # Register the expert weights.
+            self.expert_weights.append(layer.get_expert_weights())
+            layer.set_eplb_state(
+                moe_layer_idx=layer_idx,
+                expert_load_view=expert_load_view,
+                logical_to_physical_map=logical_to_physical_map,
+                logical_replica_count=logical_replica_count,
+            )
+
+    def update_physical_experts_metadata(
+        self,
+        num_physical_experts: int,
+        num_local_physical_experts: int,
+    ) -> None:
+        assert self.num_local_physical_experts == num_local_physical_experts
+        self.num_physical_experts = num_physical_experts
+        self.num_local_physical_experts = num_local_physical_experts
+        self.num_redundant_experts = (num_physical_experts -
+                                      self.num_logical_experts)
+        for layer in self.model.layers:
+            if isinstance(layer.mlp, Ernie4_5_MoeMoE):
+                moe = layer.mlp
+                moe.n_local_physical_experts = num_local_physical_experts
+                moe.n_physical_experts = num_physical_experts
+                moe.n_redundant_experts = self.num_redundant_experts
+                moe.experts.update_expert_map()
+
     def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
         return self.model.get_input_embeddings(input_ids)