exo/inference/mlx/models/sharded_llama.py

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

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

from mlx_lm.models.base import BaseModelArgs, create_additive_causal_mask
from ...shard import Shard


@dataclass
class NormalModelArgs(BaseModelArgs):
    model_type: str
    hidden_size: int
    num_hidden_layers: int
    intermediate_size: int
    num_attention_heads: int
    rms_norm_eps: float
    vocab_size: int
    num_key_value_heads: int = None
    attention_bias: bool = False
    mlp_bias: bool = False
    rope_theta: float = 10000
    rope_traditional: bool = False
    rope_scaling: Optional[Dict[str, Union[float, str]]] = None
    tie_word_embeddings: bool = True

    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'")
@dataclass
class ModelArgs(NormalModelArgs):
    shard: Shard = field(default_factory=lambda: Shard("", 0, 0, 0))

    def __post_init__(self):
        super().__post_init__()  # Ensure parent initializations are respected

        if isinstance(self.shard, Shard):
            return
        if not isinstance(self.shard, dict):
            raise TypeError(f"Expected shard to be a Shard instance or a dict, got {type(self.shard)} instead")

        self.shard = Shard(**self.shard)

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

        dim = args.hidden_size
        self.n_heads = n_heads = args.num_attention_heads
        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
        if hasattr(args, "attention_bias"):
            attention_bias = args.attention_bias
        else:
            attention_bias = False

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

        rope_scale = (
            1 / args.rope_scaling["factor"]
            if args.rope_scaling is not None and args.rope_scaling["type"] == "linear"
            else 1
        )
        self.rope = nn.RoPE(
            head_dim,
            traditional=args.rope_traditional,
            base=args.rope_theta,
            scale=rope_scale,
        )

    def __call__(
        self,
        x: mx.array,
        mask: Optional[mx.array] = None,
        cache: Optional[Tuple[mx.array, mx.array]] = 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, args: ModelArgs):
        super().__init__()

        dim = args.hidden_size
        hidden_dim = args.intermediate_size
        if hasattr(args, "mlp_bias"):
            mlp_bias = args.mlp_bias
        else:
            mlp_bias = False

        self.gate_proj = nn.Linear(dim, hidden_dim, bias=mlp_bias)
        self.down_proj = nn.Linear(hidden_dim, dim, bias=mlp_bias)
        self.up_proj = nn.Linear(dim, hidden_dim, bias=mlp_bias)

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


class TransformerBlock(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        self.num_attention_heads = args.num_attention_heads
        self.hidden_size = args.hidden_size
        self.self_attn = Attention(args)
        self.mlp = MLP(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[Tuple[mx.array, mx.array]] = 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 LlamaModel(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 = [
            TransformerBlock(args=args) for _ in range(args.shard.n_layers)
        ]
        self.norm = nn.RMSNorm(args.hidden_size, eps=args.rms_norm_eps)

    def __call__(
        self,
        inputs: mx.array,
        cache=None,
    ):
        if self.args.shard.is_first_layer():
            h = self.embed_tokens(inputs)
        else:
            h = inputs

        mask = None
        if h.shape[1] > 1:
            mask = create_additive_causal_mask(
                h.shape[1], cache[0].offset if cache is not None else 0
            )
            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, cache=c)

        if self.args.shard.is_last_layer():
            return self.norm(h)
        else:
            return h


class Model(nn.Module):
    def __init__(self, args: ModelArgs):
        super().__init__()
        self.args = args
        self.model_type = args.model_type
        self.model = LlamaModel(args)
        if not args.tie_word_embeddings:
            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)

        if self.args.shard.is_last_layer():
            if self.args.tie_word_embeddings:
                out = self.model.embed_tokens.as_linear(out)
            else:
                out = self.lm_head(out)

        return out


    def sanitize(self, weights):
        # Remove unused precomputed rotary freqs
        return {
            k: v for k, v in weights.items() if "self_attn.rotary_emb.inv_freq" not in k
        }

    @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