mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-19 17:24:57 +02:00
Xuan Son Nguyen
51 lines
2.0 KiB
Python
51 lines
2.0 KiB
Python
from __future__ import annotations
|
|
|
|
from typing import Callable, Iterable, TYPE_CHECKING
|
|
|
|
if TYPE_CHECKING:
|
|
from torch import Tensor
|
|
|
|
from .base import MOE_HF_MLP, ModelBase, gguf
|
|
|
|
from .llama import LlamaModel
|
|
|
|
|
|
@ModelBase.register("AfmoeForCausalLM")
|
|
class AfmoeModel(LlamaModel):
|
|
model_arch = gguf.MODEL_ARCH.AFMOE
|
|
|
|
def set_gguf_parameters(self):
|
|
super().set_gguf_parameters()
|
|
|
|
# MoE parameters
|
|
if (n_shared_experts := self.hparams.get("num_shared_experts")) is not None:
|
|
self.gguf_writer.add_expert_shared_count(n_shared_experts)
|
|
if (moe_intermediate_size := self.hparams.get("moe_intermediate_size")) is not None:
|
|
self.gguf_writer.add_expert_feed_forward_length(moe_intermediate_size)
|
|
if (n_dense_layers := self.hparams.get("num_dense_layers")) is not None:
|
|
self.gguf_writer.add_leading_dense_block_count(n_dense_layers)
|
|
|
|
# Route normalization and scaling
|
|
if (route_norm := self.hparams.get("route_norm")) is not None:
|
|
self.gguf_writer.add_expert_weights_norm(route_norm)
|
|
if (route_scale := self.hparams.get("route_scale")) is not None:
|
|
self.gguf_writer.add_expert_weights_scale(route_scale)
|
|
|
|
# Sliding window attention
|
|
if (sliding_window := self.hparams.get("sliding_window")) is not None:
|
|
self.gguf_writer.add_sliding_window(sliding_window)
|
|
|
|
@classmethod
|
|
def filter_tensors(cls, item: tuple[str, Callable[[], Tensor]]) -> tuple[str, Callable[[], Tensor]] | None:
|
|
name, gen = item
|
|
|
|
if name.endswith(".expert_bias"):
|
|
name = name.replace(".expert_bias", ".expert_bias.bias")
|
|
|
|
return super().filter_tensors((name, gen))
|
|
|
|
moe_experts = [MOE_HF_MLP._replace(weights=("gate_proj", "up_proj", "down_proj"))]
|
|
|
|
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
|
|
yield from ModelBase.modify_tensors(self, data_torch, name, bid)
|