mlx-examples/llms/mlx_lm/models/qwen2_moe.py

import math
from dataclasses import dataclass
from typing import Dict, Optional, Tuple, Union

import mlx.core as mx
import mlx.nn as nn

from .base import BaseModelArgs, KVCache
from .switch_layers import SwitchGLU


@dataclass
class ModelArgs(BaseModelArgs):
    model_type: str
    hidden_size: int
    num_hidden_layers: int
    intermediate_size: int
    num_attention_heads: int
    num_experts_per_tok: int
    num_experts: int
    moe_intermediate_size: int
    shared_expert_intermediate_size: int
    rms_norm_eps: float
    vocab_size: int
    num_key_value_heads: Optional[int] = None
    rope_theta: float = 1000000
    rope_traditional: bool = False
    rope_scaling: Optional[Dict[str, Union[float, str]]] = None
    tie_word_embeddings: bool = False

    def __post_init__(self):
        if self.num_key_value_heads is None:
            self.num_key_value_heads = self.num_attention_heads

        if self.rope_scaling:
            required_keys = {"factor", "type"}
            if not all(key in self.rope_scaling for key in required_keys):
                raise ValueError(f"rope_scaling must contain keys {required_keys}")

            if self.rope_scaling["type"] != "linear":
                raise ValueError("rope_scaling 'type' currently only supports 'linear'")


class Attention(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()

        dim = args.hidden_size
        self.n_heads = n_heads = args.num_attention_heads
        assert args.num_key_value_heads is not None
        self.n_kv_heads = n_kv_heads = args.num_key_value_heads

        head_dim = args.hidden_size // n_heads
        self.scale = head_dim**-0.5

        self.q_proj = nn.Linear(dim, n_heads * head_dim, bias=True)
        self.k_proj = nn.Linear(dim, n_kv_heads * head_dim, bias=True)
        self.v_proj = nn.Linear(dim, n_kv_heads * head_dim, bias=True)
        self.o_proj = nn.Linear(n_heads * head_dim, dim, bias=False)

        self.rope = nn.RoPE(
            head_dim,
            traditional=args.rope_traditional,
            base=args.rope_theta,
        )

    def __call__(
        self,
        x: mx.array,
        mask: Optional[mx.array] = None,
        cache: Optional[KVCache] = None,
    ) -> mx.array:
        B, L, D = x.shape

        queries, keys, values = self.q_proj(x), self.k_proj(x), self.v_proj(x)

        # Prepare the queries, keys and values for the attention computation
        queries = queries.reshape(B, L, self.n_heads, -1).transpose(0, 2, 1, 3)
        keys = keys.reshape(B, L, self.n_kv_heads, -1).transpose(0, 2, 1, 3)
        values = values.reshape(B, L, self.n_kv_heads, -1).transpose(0, 2, 1, 3)

        if cache is not None:
            queries = self.rope(queries, offset=cache.offset)
            keys = self.rope(keys, offset=cache.offset)
            keys, values = cache.update_and_fetch(keys, values)
        else:
            queries = self.rope(queries)
            keys = self.rope(keys)

        output = mx.fast.scaled_dot_product_attention(
            queries, keys, values, scale=self.scale, mask=mask
        )
        output = output.transpose(0, 2, 1, 3).reshape(B, L, -1)
        return self.o_proj(output)


class MLP(nn.Module):
    def __init__(self, dim, hidden_dim):
        super().__init__()
        self.gate_proj = nn.Linear(dim, hidden_dim, bias=False)
        self.down_proj = nn.Linear(hidden_dim, dim, bias=False)
        self.up_proj = nn.Linear(dim, hidden_dim, bias=False)

    def __call__(self, x) -> mx.array:
        return self.down_proj(nn.silu(self.gate_proj(x)) * self.up_proj(x))


class Qwen2MoeSparseMoeBlock(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        dim = args.hidden_size
        intermediate_size = args.moe_intermediate_size
        shared_expert_intermediate_size = args.shared_expert_intermediate_size

        self.num_experts = num_experts = args.num_experts
        self.top_k = args.num_experts_per_tok

        self.gate = nn.Linear(dim, num_experts, bias=False)
        self.switch_mlp = SwitchGLU(dim, intermediate_size, num_experts)

        self.shared_expert = MLP(dim, shared_expert_intermediate_size)
        self.shared_expert_gate = nn.Linear(dim, 1, bias=False)

    def __call__(
        self,
        x: mx.array,
    ):
        gates = self.gate(x)
        gates = mx.softmax(gates, axis=-1, precise=True)

        k = self.top_k
        inds = mx.stop_gradient(mx.argpartition(-gates, kth=k - 1, axis=-1)[..., :k])
        scores = mx.take_along_axis(gates, inds, axis=-1)

        y = self.switch_mlp(x, inds)
        y = (y * scores[..., None]).sum(axis=-2)

        shared_expert_output = self.shared_expert(x)
        shared_expert_output = (
            mx.sigmoid(self.shared_expert_gate(x)) * shared_expert_output
        )

        return y + shared_expert_output


class Qwen2MoeDecoderLayer(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        self.hidden_size = args.hidden_size
        self.self_attn = Attention(args)
        self.mlp = Qwen2MoeSparseMoeBlock(args)

        self.input_layernorm = nn.RMSNorm(args.hidden_size, eps=args.rms_norm_eps)
        self.post_attention_layernorm = nn.RMSNorm(
            args.hidden_size, eps=args.rms_norm_eps
        )
        self.args = args

    def __call__(
        self,
        x: mx.array,
        mask: Optional[mx.array] = None,
        cache: Optional[KVCache] = None,
    ) -> mx.array:
        r = self.self_attn(self.input_layernorm(x), mask, cache)
        h = x + r
        r = self.mlp(self.post_attention_layernorm(h))
        out = h + r
        return out


class Qwen2MoeModel(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        self.args = args
        self.vocab_size = args.vocab_size
        self.num_hidden_layers = args.num_hidden_layers
        assert self.vocab_size > 0
        self.embed_tokens = nn.Embedding(args.vocab_size, args.hidden_size)
        self.layers = [
            Qwen2MoeDecoderLayer(args=args) for _ in range(args.num_hidden_layers)
        ]
        self.norm = nn.RMSNorm(args.hidden_size, eps=args.rms_norm_eps)

    def __call__(
        self,
        inputs: mx.array,
        cache=None,
    ):
        h = self.embed_tokens(inputs)

        mask = None
        if h.shape[1] > 1:
            mask = nn.MultiHeadAttention.create_additive_causal_mask(h.shape[1])
            mask = mask.astype(h.dtype)

        if cache is None:
            cache = [None] * len(self.layers)

        for layer, c in zip(self.layers, cache):
            h = layer(h, mask, c)

        return self.norm(h)


class Model(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        self.args = args
        self.model_type = args.model_type
        self.model = Qwen2MoeModel(args)
        self.lm_head = nn.Linear(args.hidden_size, args.vocab_size, bias=False)

    def __call__(
        self,
        inputs: mx.array,
        cache=None,
    ):
        out = self.model(inputs, cache)
        return self.lm_head(out)

    def sanitize(self, weights):
        if "model.layers.0.mlp.experts.0.up_proj.weight" not in weights:
            return weights
        for l in range(self.args.num_hidden_layers):
            prefix = f"model.layers.{l}"
            for n in ["up_proj", "down_proj", "gate_proj"]:
                for k in ["weight", "scales", "biases"]:
                    if f"{prefix}.mlp.experts.0.{n}.{k}" in weights:
                        to_join = [
                            weights.pop(f"{prefix}.mlp.experts.{e}.{n}.{k}")
                            for e in range(self.args.num_experts)
                        ]
                        weights[f"{prefix}.mlp.switch_mlp.{n}.{k}"] = mx.stack(to_join)
        return weights

    @property
    def layers(self):
        return self.model.layers

    @property
    def head_dim(self):
        return self.args.hidden_size // self.args.num_attention_heads

    @property
    def n_kv_heads(self):
        return self.args.num_key_value_heads