from megatron.core.fusions.fused_bias_dropout import get_bias_dropout_add
from megatron.core.tensor_parallel.layers import ColumnParallelLinear, RowParallelLinear
from megatron.core.transformer.attention import SelfAttention, SelfAttentionSubmodules
from megatron.core.extensions.transformer_engine import (
TEColumnParallelLinear,
TEDotProductAttention,
TELayerNormColumnParallelLinear,
TENorm,
TERowParallelLinear,
)
from megatron.core.transformer.custom_layers.transformer_engine import TEColumnParallelLinear, TERowParallelLinear
from megatron.core.transformer.dot_product_attention import DotProductAttention
from megatron.core.transformer.enums import AttnMaskType
from megatron.core.transformer.identity_op import IdentityOp
from megatron.core.transformer.spec_utils import ModuleSpec
from megatron.core.models.gpt.gpt_layer_specs import _get_mlp_module_spec
from megatron.training import get_args
from mindspeed_llm.core.transformer.custom_layers.transformer_engine import PTNorm
from megatron.core.transformer import ModuleSpec, TransformerLayer, TransformerLayerSubmodules
args = get_args()
num_experts, moe_grouped_gemm, qk_layernorm = args.num_experts, args.moe_grouped_gemm, args.qk_layernorm
if args.transformer_impl == "transformer_engine":
ColumnLinear = TEColumnParallelLinear
RowLinear = TERowParallelLinear
CoreAttention = TEDotProductAttention
use_te = True
else:
ColumnLinear = ColumnParallelLinear
RowLinear = RowParallelLinear
CoreAttention = DotProductAttention
use_te = False
layer_spec = ModuleSpec(
module=TransformerLayer,
submodules=TransformerLayerSubmodules(
input_layernorm=PTNorm if not use_te else IdentityOp,
self_attention=ModuleSpec(
module=SelfAttention,
params={"attn_mask_type": AttnMaskType.causal},
submodules=SelfAttentionSubmodules(
linear_qkv=ColumnLinear if not use_te else TELayerNormColumnParallelLinear,
core_attention=CoreAttention,
linear_proj=RowLinear,
q_layernorm=PTNorm if qk_layernorm else IdentityOp,
k_layernorm=PTNorm if qk_layernorm else IdentityOp,
),
),
self_attn_bda=get_bias_dropout_add,
pre_mlp_layernorm=IdentityOp if not num_experts and use_te else PTNorm,
mlp=_get_mlp_module_spec(
use_te=use_te, num_experts=num_experts, moe_grouped_gemm=moe_grouped_gemm
),
mlp_bda=get_bias_dropout_add,
sharded_state_dict_keys_map={
'input_layernorm.': 'self_attention.linear_qkv.layer_norm_',
'pre_mlp_layernorm.': 'mlp.linear_fc1.layer_norm_',
},
),
)
