Add Mellum (Mellum 2) model support (ml-explore#1339)

Author: jedisct1

mlx_lm/models/mellum.py

@@ -0,0 +1,264 @@
+# Copyright © 2026 Apple Inc.
+
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Optional
+
+import mlx.core as mx
+import mlx.nn as nn
+
+from .activations import swiglu
+from .base import BaseModelArgs, create_attention_mask, scaled_dot_product_attention
+from .cache import KVCache, RotatingKVCache
+from .rope_utils import initialize_rope
+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: int
+    num_experts_per_tok: int
+    moe_intermediate_size: int
+    rms_norm_eps: float
+    vocab_size: int
+    num_key_value_heads: int
+    head_dim: int
+    tie_word_embeddings: bool
+    max_position_embeddings: int
+    norm_topk_prob: bool
+    sliding_window: int
+    layer_types: List[str]
+    rope_parameters: Dict[str, Any] = field(default_factory=dict)
+
+
+def _rope_for(layer_type: str, args: ModelArgs):
+    params = args.rope_parameters[layer_type]
+    base = params["rope_theta"]
+    rope_type = params.get("rope_type", "default")
+    if rope_type in ("default", "linear"):
+        return initialize_rope(args.head_dim, base=base, traditional=False)
+    scaling_config = dict(params)
+    scaling_config["type"] = rope_type
+    return initialize_rope(
+        args.head_dim,
+        base=base,
+        traditional=False,
+        scaling_config=scaling_config,
+        max_position_embeddings=args.max_position_embeddings,
+    )
+
+
+class Attention(nn.Module):
+    def __init__(self, args: ModelArgs, layer_idx: int):
+        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.head_dim
+        self.scale = head_dim**-0.5
+
+        self.q_proj = nn.Linear(dim, n_heads * head_dim, bias=False)
+        self.k_proj = nn.Linear(dim, n_kv_heads * head_dim, bias=False)
+        self.v_proj = nn.Linear(dim, n_kv_heads * head_dim, bias=False)
+        self.o_proj = nn.Linear(n_heads * head_dim, dim, bias=False)
+
+        self.q_norm = nn.RMSNorm(head_dim, eps=args.rms_norm_eps)
+        self.k_norm = nn.RMSNorm(head_dim, eps=args.rms_norm_eps)
+
+        self.rope = _rope_for(args.layer_types[layer_idx], args)
+
+    def __call__(
+        self,
+        x: mx.array,
+        mask: Optional[mx.array] = None,
+        cache: Optional[Any] = None,
+    ) -> mx.array:
+        B, L, D = x.shape
+
+        queries, keys, values = self.q_proj(x), self.k_proj(x), self.v_proj(x)
+
+        queries = self.q_norm(queries.reshape(B, L, self.n_heads, -1)).transpose(
+            0, 2, 1, 3
+        )
+        keys = self.k_norm(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 = scaled_dot_product_attention(
+            queries, keys, values, cache=cache, scale=self.scale, mask=mask
+        )
+        output = output.transpose(0, 2, 1, 3).reshape(B, L, -1)
+        return self.o_proj(output)
+
+
+class MellumSparseMoeBlock(nn.Module):
+    def __init__(self, args: ModelArgs):
+        super().__init__()
+        dim = args.hidden_size
+        self.num_experts = args.num_experts
+        self.top_k = args.num_experts_per_tok
+        self.norm_topk_prob = args.norm_topk_prob
+
+        self.gate = nn.Linear(dim, self.num_experts, bias=False)
+        self.switch_mlp = SwitchGLU(dim, args.moe_intermediate_size, self.num_experts)
+
+    def __call__(self, x: mx.array) -> mx.array:
+        gates = self.gate(x)
+        gates = mx.softmax(gates, axis=-1, precise=True)
+
+        k = self.top_k
+        inds = mx.argpartition(gates, kth=-k, axis=-1)[..., -k:]
+        scores = mx.take_along_axis(gates, inds, axis=-1)
+        if self.norm_topk_prob:
+            scores /= mx.sum(scores, axis=-1, keepdims=True)
+
+        y = self.switch_mlp(x, inds)
+        y = (y * scores[..., None]).sum(axis=-2)
+        return y
+
+
+class MellumDecoderLayer(nn.Module):
+    def __init__(self, args: ModelArgs, layer_idx: int):
+        super().__init__()
+        self.self_attn = Attention(args, layer_idx)
+        self.mlp = MellumSparseMoeBlock(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
+        )
+
+    def __call__(
+        self,
+        x: mx.array,
+        mask: Optional[mx.array] = None,
+        cache: Optional[Any] = None,
+    ) -> mx.array:
+        r = self.self_attn(self.input_layernorm(x), mask, cache)
+        h = x + r
+        r = self.mlp(self.post_attention_layernorm(h))
+        return h + r
+
+
+class MellumModel(nn.Module):
+    def __init__(self, args: ModelArgs):
+        super().__init__()
+        self.args = args
+        self.vocab_size = args.vocab_size
+        assert self.vocab_size > 0
+        self.embed_tokens = nn.Embedding(args.vocab_size, args.hidden_size)
+        self.layers = [
+            MellumDecoderLayer(args=args, layer_idx=i)
+            for i in range(args.num_hidden_layers)
+        ]
+        self.norm = nn.RMSNorm(args.hidden_size, eps=args.rms_norm_eps)
+
+        self._first_full = next(
+            i for i, t in enumerate(args.layer_types) if t == "full_attention"
+        )
+        self._first_sliding = next(
+            (i for i, t in enumerate(args.layer_types) if t == "sliding_attention"),
+            None,
+        )
+
+    def __call__(
+        self,
+        inputs: mx.array,
+        cache=None,
+        input_embeddings: Optional[mx.array] = None,
+    ) -> mx.array:
+        if input_embeddings is not None:
+            h = input_embeddings
+        else:
+            h = self.embed_tokens(inputs)
+
+        if cache is None:
+            cache = [None] * len(self.layers)
+
+        full_mask = create_attention_mask(h, cache[self._first_full])
+        if self._first_sliding is not None:
+            sliding_mask = create_attention_mask(
+                h, cache[self._first_sliding], window_size=self.args.sliding_window
+            )
+        else:
+            sliding_mask = None
+
+        for layer, c, t in zip(self.layers, cache, self.args.layer_types):
+            mask = full_mask if t == "full_attention" else sliding_mask
+            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 = MellumModel(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,
+        input_embeddings: Optional[mx.array] = None,
+    ) -> mx.array:
+        out = self.model(inputs, cache, input_embeddings)
+        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):
+        if self.args.tie_word_embeddings:
+            weights.pop("lm_head.weight", None)
+        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"]:
+                if f"{prefix}.mlp.experts.0.{n}.weight" in weights:
+                    to_join = [
+                        weights.pop(f"{prefix}.mlp.experts.{e}.{n}.weight")
+                        for e in range(self.args.num_experts)
+                    ]
+                    weights[f"{prefix}.mlp.switch_mlp.{n}.weight"] = mx.stack(to_join)
+        return weights
+
+    @property
+    def quant_predicate(self):
+        def predicate(path, _):
+            if path.endswith("mlp.gate"):
+                return {"group_size": 64, "bits": 8}
+            return True
+
+        return predicate
+
+    @property
+    def layers(self):
+        return self.model.layers
+
+    def make_cache(self):
+        caches = []
+        for t in self.args.layer_types:
+            if t == "full_attention":
+                caches.append(KVCache())
+            else:
+                caches.append(RotatingKVCache(max_size=self.args.sliding_window))
+        return caches