nemo_automodel.components.models.qwen3_moe.model

Module Contents

Classes

Name	Description
`Block`	-
`Qwen3MoeForCausalLM`	-
`Qwen3MoeModel`	-

Data

ModelClass

API

class nemo_automodel.components.models.qwen3_moe.model.Block(
    layer_idx: int,
    config: transformers.models.qwen3_moe.configuration_qwen3_moe.Qwen3MoeConfig,
    moe_config: nemo_automodel.components.moe.config.MoEConfig,
    backend: nemo_automodel.components.models.common.BackendConfig
)

Bases: Module

input_layernorm

mlp

= MoE(moe_config, backend)

post_attention_layernorm

self_attn

= Qwen3MoeAttention(config, backend)

nemo_automodel.components.models.qwen3_moe.model.Block._mlp(
    x: torch.Tensor,
    padding_mask: torch.Tensor | None
) -> torch.Tensor

nemo_automodel.components.models.qwen3_moe.model.Block.forward(
    x: torch.Tensor,
    freqs_cis: torch.Tensor,
    attention_mask: torch.Tensor | None = None,
    padding_mask: torch.Tensor | None = None,
    attn_kwargs: typing.Any = {}
) -> torch.Tensor

nemo_automodel.components.models.qwen3_moe.model.Block.init_weights(
    buffer_device: torch.device
)

class nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM(
    config: transformers.models.qwen3_moe.configuration_qwen3_moe.Qwen3MoeConfig,
    moe_config: nemo_automodel.components.moe.config.MoEConfig | None = None,
    backend: nemo_automodel.components.models.common.BackendConfig | None = None,
    kwargs = {}
)

Bases: HFCheckpointingMixin, Module, MoEFSDPSyncMixin

_pp_keep_self_forward

bool = True

backend

= backend or BackendConfig()

lm_head

model

state_dict_adapter

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.forward(
    input_ids: torch.Tensor,
    position_ids: torch.Tensor | None = None,
    attention_mask: torch.Tensor | None = None,
    padding_mask: torch.Tensor | None = None,
    logits_to_keep: typing.Union[int, torch.Tensor] = 0,
    output_hidden_states: typing.Optional[bool] = None,
    attn_kwargs: typing.Any = {}
) -> transformers.modeling_outputs.CausalLMOutputWithPast

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.from_config(
    config: transformers.models.qwen3_moe.configuration_qwen3_moe.Qwen3MoeConfig,
    moe_config: nemo_automodel.components.moe.config.MoEConfig | None = None,
    backend: nemo_automodel.components.models.common.BackendConfig | None = None,
    kwargs = {}
)

classmethod

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.from_pretrained(
    pretrained_model_name_or_path: str,
    model_args = (),
    kwargs = {}
)

classmethod

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.get_input_embeddings()

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.get_output_embeddings()

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.initialize_weights(
    buffer_device: torch.device | None = None,
    dtype: torch.dtype = torch.bfloat16
) -> None

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.set_input_embeddings(
    value
)

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeForCausalLM.set_output_embeddings(
    new_embeddings
)

class nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeModel(
    config: transformers.models.qwen3_moe.configuration_qwen3_moe.Qwen3MoeConfig,
    backend: nemo_automodel.components.models.common.BackendConfig,
    moe_config: nemo_automodel.components.moe.config.MoEConfig | None = None,
    moe_overrides: dict | None = None
)

Bases: Module

embed_tokens

head_dim

layers

= torch.nn.ModuleDict()

max_seq_len

= config.max_position_embeddings

moe_config

= moe_config or MoEConfig(**moe_defaults)

norm

rotary_emb

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeModel.forward(
    input_ids: torch.Tensor,
    position_ids: torch.Tensor | None = None,
    attention_mask: torch.Tensor | None = None,
    padding_mask: torch.Tensor | None = None,
    attn_kwargs: typing.Any = {}
) -> torch.Tensor

nemo_automodel.components.models.qwen3_moe.model.Qwen3MoeModel.init_weights(
    buffer_device: torch.device | None = None
) -> None

nemo_automodel.components.models.qwen3_moe.model.ModelClass = Qwen3MoeForCausalLM