afmoe.py

# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
"""Inference-only AfMoE model compatible with HuggingFace weights."""

import typing
from collections.abc import Callable, Iterable
from itertools import islice

import torch
from torch import nn

from vllm.compilation.decorators import support_torch_compile
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.attention import Attention
from vllm.model_executor.layers.fused_moe.shared_fused_moe import SharedFusedMoE
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.model_executor.layers.linear import (
    ColumnParallelLinear,
    QKVParallelLinear,
    RowParallelLinear,
)
from vllm.model_executor.layers.logits_processor import LogitsProcessor
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.rotary_embedding import get_rope
from vllm.model_executor.layers.vocab_parallel_embedding import (
    ParallelLMHead,
    VocabParallelEmbedding,
)
from vllm.model_executor.model_loader.weight_utils import (
    default_weight_loader,
    maybe_remap_kv_scale_name,
)
from vllm.model_executor.models.interfaces import (
    EagleModelMixin,
    SupportsEagle3,
    SupportsLoRA,
    SupportsPP,
)
from vllm.model_executor.models.llama import LlamaMLP as AfmoeMLP
from vllm.model_executor.models.utils import (
    AutoWeightsLoader,
    PPMissingLayer,
    WeightsMapper,
    extract_layer_index,
    is_pp_missing_parameter,
    make_empty_intermediate_tensors_factory,
    make_layers,
    maybe_prefix,
)
from vllm.sequence import IntermediateTensors
from vllm.v1.attention.backend import AttentionType

logger = init_logger(__name__)


class AfmoeMoE(nn.Module):
    def __init__(
        self,
        config,  # AfmoeConfig
        quant_config: QuantizationConfig | None = None,
        prefix: str = "",
        enable_eplb: bool = False,
    ):
        super().__init__()
        self.tp_size = get_tensor_model_parallel_world_size()
        self.route_scale = config.route_scale
        self.score_func = config.score_func
        self.route_norm = config.route_norm

        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.num_experts
        self.n_shared_experts: int = config.num_shared_experts

        if config.hidden_act != "silu":
            raise ValueError(
                f"Unsupported activation: {config.hidden_act}. "
                "Only silu is supported for now."
            )

        # Router gate
        self.gate = nn.Linear(
            config.hidden_size,
            config.num_experts,
            bias=False,
            dtype=torch.float32,
        )
        self.expert_bias = nn.Parameter(
            torch.empty(config.num_experts, dtype=torch.float32)
        )

        # Load balancing settings
        vllm_config = get_current_vllm_config()
        eplb_config = vllm_config.parallel_config.eplb_config
        self.enable_eplb = enable_eplb

        self.n_redundant_experts = eplb_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.shared_experts = None
        # Shared experts
        if config.num_shared_experts > 0:
            intermediate_size = config.moe_intermediate_size * config.num_shared_experts
            self.shared_experts = AfmoeMLP(
                hidden_size=config.hidden_size,
                intermediate_size=intermediate_size,
                hidden_act=config.hidden_act,
                quant_config=quant_config,
                reduce_results=False,
                prefix=f"{prefix}.shared_experts",
            )

        # Routed experts using SharedFusedMoE
        self.experts = SharedFusedMoE(
            shared_experts=self.shared_experts,
            num_experts=config.num_experts,
            top_k=config.num_experts_per_tok,
            hidden_size=config.hidden_size,
            intermediate_size=config.moe_intermediate_size,
            reduce_results=False,
            renormalize=self.route_norm if self.score_func == "sigmoid" else False,
            quant_config=quant_config,
            use_grouped_topk=True,
            num_expert_group=config.n_group,
            topk_group=config.topk_group,
            prefix=f"{prefix}.experts",
            scoring_func=self.score_func,
            routed_scaling_factor=self.route_scale,
            e_score_correction_bias=self.expert_bias,
            enable_eplb=self.enable_eplb,
            num_redundant_experts=self.n_redundant_experts,
            router_logits_dtype=torch.float32,
        )

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        num_tokens, hidden_dim = hidden_states.shape
        hidden_states = hidden_states.view(-1, hidden_dim)

        router_logits = self.gate(hidden_states.to(dtype=torch.float32))

        fused_moe_out = self.experts(
            hidden_states=hidden_states, router_logits=router_logits
        )

        if self.shared_experts is not None:
            shared_output, final_hidden_states = fused_moe_out
            final_hidden_states = final_hidden_states + shared_output
        else:
            final_hidden_states = fused_moe_out
        if self.tp_size > 1:
            final_hidden_states = self.experts.maybe_all_reduce_tensor_model_parallel(
                final_hidden_states
            )

        return final_hidden_states.view(num_tokens, hidden_dim)


class AfmoeAttention(nn.Module):
    def __init__(
        self,
        config,  # AfmoeConfig
        layer_idx: int,
        hidden_size: int,
        num_heads: int,
        num_kv_heads: int,
        max_position_embeddings: int = 131072,
        head_dim: int | None = None,
        rms_norm_eps: float = 1e-05,
        cache_config: CacheConfig | None = None,
        quant_config: QuantizationConfig | None = None,
        prefix: str = "",
        attn_type: str = AttentionType.DECODER,
    ) -> None:
        super().__init__()
        self.layer_idx = layer_idx
        self.hidden_size = hidden_size
        tp_size = get_tensor_model_parallel_world_size()
        self.total_num_heads = num_heads
        assert self.total_num_heads % tp_size == 0
        self.num_heads = self.total_num_heads // tp_size
        self.total_num_kv_heads = num_kv_heads
        if self.total_num_kv_heads >= tp_size:
            # Number of KV heads is greater than TP size, so we partition
            # the KV heads across multiple tensor parallel GPUs.
            assert self.total_num_kv_heads % tp_size == 0
        else:
            # Number of KV heads is less than TP size, so we replicate
            # the KV heads across multiple tensor parallel GPUs.
            assert tp_size % self.total_num_kv_heads == 0
        self.num_kv_heads = max(1, self.total_num_kv_heads // tp_size)
        self.head_dim = head_dim or (hidden_size // self.total_num_heads)
        self.q_size = self.num_heads * self.head_dim
        self.kv_size = self.num_kv_heads * self.head_dim
        self.scaling = self.head_dim**-0.5
        self.max_position_embeddings = max_position_embeddings

        # Check if this is a local attention layer
        self.is_local_attention = config.layer_types[layer_idx] == "sliding_attention"
        self.sliding_window = config.sliding_window if self.is_local_attention else None

        self.qkv_proj = QKVParallelLinear(
            self.hidden_size,
            self.head_dim,
            self.total_num_heads,
            self.total_num_kv_heads,
            bias=False,
            quant_config=quant_config,
            prefix=f"{prefix}.qkv_proj",
        )

        self.o_proj = RowParallelLinear(
            self.total_num_heads * self.head_dim,
            self.hidden_size,
            bias=False,
            quant_config=quant_config,
            prefix=f"{prefix}.o_proj",
        )

        # Gating projection
        self.gate_proj = ColumnParallelLinear(
            hidden_size,
            self.total_num_heads * self.head_dim,
            bias=False,
            quant_config=quant_config,
            prefix=f"{prefix}.gate_proj",
        )

        # Q/K normalization
        self.q_norm = RMSNorm(self.head_dim, eps=config.rms_norm_eps)
        self.k_norm = RMSNorm(self.head_dim, eps=config.rms_norm_eps)

        # Only create rotary embeddings for local attention
        if self.is_local_attention:
            self.rotary_emb = get_rope(
                self.head_dim,
                max_position=max_position_embeddings,
                rope_parameters=config.rope_parameters,
                is_neox_style=True,
            )
        else:
            self.rotary_emb = None

        self.attn = Attention(
            self.num_heads,
            self.head_dim,
            self.scaling,
            num_kv_heads=self.num_kv_heads,
            cache_config=cache_config,
            quant_config=quant_config,
            per_layer_sliding_window=self.sliding_window,
            prefix=f"{prefix}.attn",
            attn_type=attn_type,
        )

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
    ) -> torch.Tensor:
        qkv, _ = self.qkv_proj(hidden_states)
        gate, _ = self.gate_proj(hidden_states)
        q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)

        # Apply Q/K normalization
        q = self.q_norm(q.reshape(-1, self.num_heads, self.head_dim)).reshape(q.shape)
        k = self.k_norm(k.reshape(-1, self.num_kv_heads, self.head_dim)).reshape(
            k.shape
        )

        # Apply rotary embeddings only for local attention
        if self.is_local_attention and self.rotary_emb is not None:
            q, k = self.rotary_emb(positions, q, k)

        attn_output = self.attn(q, k, v)

        # Apply gating
        attn_output = attn_output * torch.sigmoid(gate)
        output, _ = self.o_proj(attn_output)
        return output


class AfmoeDecoderLayer(nn.Module):
    def __init__(
        self,
        config,  # AfmoeConfig
        cache_config: CacheConfig | None = None,
        quant_config: QuantizationConfig | None = None,
        prefix: str = "",
        enable_eplb: bool = False,
    ) -> None:
        super().__init__()
        self.hidden_size = config.hidden_size
        max_position_embeddings = getattr(config, "max_position_embeddings", 131072)

        # DecoderLayers are created with `make_layers` which passes the prefix
        # with the layer's index.
        self.layer_idx = extract_layer_index(prefix)

        self.self_attn = AfmoeAttention(
            config=config,
            layer_idx=self.layer_idx,
            hidden_size=self.hidden_size,
            num_heads=config.num_attention_heads,
            num_kv_heads=config.num_key_value_heads,
            max_position_embeddings=max_position_embeddings,
            head_dim=config.head_dim,
            rms_norm_eps=config.rms_norm_eps,
            cache_config=cache_config,
            quant_config=quant_config,
            prefix=f"{prefix}.self_attn",
        )

        # MoE or dense FFN
        self.moe_enabled = self.layer_idx >= config.num_dense_layers
        if self.moe_enabled:
            self.mlp = AfmoeMoE(
                config=config,
                quant_config=quant_config,
                prefix=f"{prefix}.mlp",
                enable_eplb=enable_eplb,
            )
        else:
            self.mlp = AfmoeMLP(
                hidden_size=config.hidden_size,
                intermediate_size=config.intermediate_size,
                hidden_act=config.hidden_act,
                quant_config=quant_config,
                prefix=f"{prefix}.mlp",
            )

        self.input_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.post_attention_layernorm = RMSNorm(
            config.hidden_size, eps=config.rms_norm_eps
        )
        self.pre_mlp_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.post_mlp_layernorm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
        residual: torch.Tensor | None,
    ) -> tuple[torch.Tensor, torch.Tensor]:
        if residual is None:
            residual = hidden_states
            hidden_states = self.input_layernorm(hidden_states)
        else:
            hidden_states, residual = self.input_layernorm(hidden_states, residual)

        hidden_states = self.self_attn(
            positions=positions,
            hidden_states=hidden_states,
        )
        hidden_states = self.post_attention_layernorm(hidden_states)  # attn norm b

        # Fully Connected
        hidden_states, residual = self.pre_mlp_layernorm(  # ffn norm a
            hidden_states, residual
        )
        hidden_states = self.mlp(hidden_states)
        hidden_states = self.post_mlp_layernorm(hidden_states)  # ffn norm b

        return hidden_states, residual


@support_torch_compile(
    dynamic_arg_dims={
        "input_ids": 0,
        "positions": -1,
        "intermediate_tensors": 0,
        "inputs_embeds": 0,
    }
)
class AfmoeModel(nn.Module, EagleModelMixin):
    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()

        config = vllm_config.model_config.hf_config
        cache_config = vllm_config.cache_config
        quant_config = vllm_config.quant_config
        enable_eplb = vllm_config.parallel_config.enable_eplb
        self.config = config

        self.vocab_size = config.vocab_size
        self.mup_enabled = config.mup_enabled

        if get_pp_group().is_first_rank:
            self.embed_tokens = VocabParallelEmbedding(
                config.vocab_size, config.hidden_size, prefix=f"{prefix}.embed_tokens"
            )
        else:
            self.embed_tokens = PPMissingLayer()

        self.start_layer, self.end_layer, self.layers = make_layers(
            config.num_hidden_layers,
            lambda prefix: AfmoeDecoderLayer(
                config=config,
                cache_config=cache_config,
                quant_config=quant_config,
                prefix=prefix,
                enable_eplb=enable_eplb,
            ),
            prefix=f"{prefix}.layers",
        )

        if get_pp_group().is_last_rank:
            self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        else:
            self.norm = PPMissingLayer()

        self.make_empty_intermediate_tensors = make_empty_intermediate_tensors_factory(
            ["hidden_states", "residual"], config.hidden_size
        )

    def embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.embed_tokens(input_ids)

    def forward(
        self,
        input_ids: torch.Tensor | None,
        positions: torch.Tensor,
        intermediate_tensors: IntermediateTensors | None = None,
        inputs_embeds: torch.Tensor | None = None,
    ) -> torch.Tensor | IntermediateTensors | tuple[torch.Tensor, list[torch.Tensor]]:
        if get_pp_group().is_first_rank:
            if inputs_embeds is not None:
                hidden_states = inputs_embeds
            else:
                hidden_states = self.embed_input_ids(input_ids)

            # Apply muP input scaling if enabled
            if self.mup_enabled:
                hidden_states = hidden_states * (self.config.hidden_size**0.5)

            residual = None
        else:
            assert intermediate_tensors is not None
            hidden_states = intermediate_tensors["hidden_states"]
            residual = intermediate_tensors["residual"]

        aux_hidden_states = self._maybe_add_hidden_state([], 0, hidden_states, residual)
        for idx, layer in enumerate(
            islice(self.layers, self.start_layer, self.end_layer)
        ):
            hidden_states, residual = layer(positions, hidden_states, residual)
            self._maybe_add_hidden_state(
                aux_hidden_states, idx + 1, hidden_states, residual
            )

        if not get_pp_group().is_last_rank:
            return IntermediateTensors(
                {"hidden_states": hidden_states, "residual": residual}
            )

        hidden_states, _ = self.norm(hidden_states, residual)

        if len(aux_hidden_states) > 0:
            return hidden_states, aux_hidden_states

        return hidden_states

    def make_empty_intermediate_tensors(
        self, batch_size: int, dtype: torch.dtype, device: torch.device
    ) -> IntermediateTensors:
        return IntermediateTensors(
            {
                "hidden_states": torch.zeros(
                    (batch_size, self.config.hidden_size), dtype=dtype, device=device
                ),
                "residual": torch.zeros(
                    (batch_size, self.config.hidden_size), dtype=dtype, device=device
                ),
            }
        )

    def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
        # Params for weights, fp8 weight scales, fp8 activation scales
        # (param_name, weight_name, expert_id, shard_id)
        return SharedFusedMoE.make_expert_params_mapping(
            self,
            ckpt_gate_proj_name="gate_proj",
            ckpt_down_proj_name="down_proj",
            ckpt_up_proj_name="up_proj",
            num_experts=self.config.num_experts,
        )

    def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
        stacked_params_mapping = [
            # (param_name, shard_name, shard_id)
            ("qkv_proj", "q_proj", "q"),
            ("qkv_proj", "k_proj", "k"),
            ("qkv_proj", "v_proj", "v"),
            ("gate_up_proj", "gate_proj", 0),
            ("gate_up_proj", "up_proj", 1),
        ]

        params_dict = dict(self.named_parameters())
        loaded_params: set[str] = set()
        expert_params_mapping = self.get_expert_mapping()

        for name, loaded_weight in weights:
            for param_name, weight_name, shard_id in stacked_params_mapping:
                # Skip non-stacked layers and experts (experts handled below).
                if (weight_name not in name) or ("self_attn.gate_proj" in name):
                    continue
                # We have mlp.experts[0].gate_proj in the checkpoint.
                # Since we handle the experts below in expert_params_mapping,
                # we need to skip here BEFORE we update the name, otherwise
                # name will be updated to mlp.experts[0].gate_up_proj, which
                # will then be updated below in expert_params_mapping
                # for mlp.experts[0].gate_gate_up_proj, which breaks load.
                if ("mlp.experts." in name) and name not in params_dict:
                    continue

                name = name.replace(weight_name, param_name)
                # Skip loading extra bias for GPTQ models.
                if name.endswith(".bias") and name not in params_dict:
                    continue
                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:
                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

                    # 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)

                    if is_pp_missing_parameter(name_mapped, self):
                        continue

                    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") and name not in params_dict:
                        continue

                    # Remapping the name of FP8 kv-scale.
                    name = maybe_remap_kv_scale_name(name, params_dict)
                    if name is None:
                        continue

                    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)
            loaded_params.add(name)

        return loaded_params


class AfmoeForCausalLM(nn.Module, SupportsPP, SupportsEagle3, SupportsLoRA):
    packed_modules_mapping = {
        "qkv_proj": [
            "q_proj",
            "k_proj",
            "v_proj",
        ],
        "gate_up_proj": [
            "gate_proj",
            "up_proj",
        ],
    }

    hf_to_vllm_mapper = WeightsMapper(
        orig_to_new_suffix={
            ".router.gate.weight": ".gate.weight",
        },
    )

    fall_back_to_pt_during_load = False

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()
        config = vllm_config.model_config.hf_config
        quant_config = vllm_config.quant_config
        self.config = config
        self.quant_config = quant_config
        self.model = AfmoeModel(
            vllm_config=vllm_config, prefix=maybe_prefix(prefix, "model")
        )
        if get_pp_group().is_last_rank:
            self.lm_head = ParallelLMHead(
                config.vocab_size, config.hidden_size, quant_config=quant_config
            )
        else:
            self.lm_head = PPMissingLayer()
        self.logits_processor = LogitsProcessor(config.vocab_size)
        self.make_empty_intermediate_tensors = (
            self.model.make_empty_intermediate_tensors
        )
        self.expert_weights = []

        # Set MoE hyperparameters
        self.num_moe_layers = config.num_hidden_layers - config.num_dense_layers
        self.num_expert_groups = config.n_group

        self.moe_layers: list[SharedFusedMoE] = []
        example_moe = None
        for layer in self.model.layers:
            if isinstance(layer, PPMissingLayer):
                continue

            assert isinstance(layer, AfmoeDecoderLayer)
            if layer.moe_enabled:
                example_moe = layer.mlp
                self.moe_layers.append(layer.mlp.experts)

        if example_moe is None and self.num_moe_layers > 0:
            raise RuntimeError("No AfmoeMoE layer found in model.layers.")

        if example_moe is not None:
            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 embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.model.embed_input_ids(input_ids)

    def forward(
        self,
        input_ids: torch.Tensor | None,
        positions: torch.Tensor,
        intermediate_tensors: IntermediateTensors | None = None,
        inputs_embeds: torch.Tensor | None = None,
    ) -> torch.Tensor | IntermediateTensors | tuple[torch.Tensor, list[torch.Tensor]]:
        hidden_states = self.model(
            input_ids, positions, intermediate_tensors, inputs_embeds
        )
        return hidden_states

    def compute_logits(self, hidden_states: torch.Tensor) -> torch.Tensor | None:
        logits = self.logits_processor(self.lm_head, hidden_states)
        return logits

    def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
        loader = AutoWeightsLoader(self)
        return loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)

    def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
        return self.model.get_expert_mapping()