Skip to content

[None][feat] add specdec to nemotron nas #8985

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[None][feat] add specdec to nemotron nas #8985

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 15 additions & 8 deletions tensorrt_llm/_torch/models/modeling_nemotron_nas.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,8 +17,9 @@
from ..modules.gated_mlp import GatedMLP
from ..modules.linear import Linear, TensorParallelMode
from ..modules.rms_norm import RMSNorm
from .modeling_utils import (DecoderModel, DecoderModelForCausalLM,
register_auto_model)
from ..speculative import SpecMetadata
from .modeling_speculative import SpecDecOneEngineForCausalLM
from .modeling_utils import DecoderModel, register_auto_model


def _ffn_mult_to_intermediate_size(ffn_mult: float, n_embd: int) -> int:
Expand Down Expand Up @@ -117,6 +118,7 @@ def __init__(self, model_config: ModelConfig[PretrainedConfig],
block_config: Dict[str, Any], layer_idx: int):
super().__init__()
config = model_config.pretrained_config
self.layer_idx = layer_idx
self.block_config = block_config
if not self.block_config.attention.no_op:
self.input_layernorm = RMSNorm(hidden_size=config.hidden_size,
Expand Down Expand Up @@ -150,6 +152,7 @@ def forward(
hidden_states: torch.Tensor,
attn_metadata: AttentionMetadata,
residual: Optional[torch.Tensor] = None,
spec_metadata: Optional[SpecMetadata] = None,
**kwargs,
) -> torch.Tensor:
if not self.block_config.attention.no_op:
Expand Down Expand Up @@ -178,6 +181,11 @@ def forward(
hidden_states, residual)
hidden_states = self.mlp(hidden_states, **kwargs)

# Capture hidden states for speculative decoding
if spec_metadata is not None:
spec_metadata.maybe_capture_hidden_states(self.layer_idx,
hidden_states, residual)

return hidden_states, residual


Expand Down Expand Up @@ -238,6 +246,7 @@ def forward(
input_ids: Optional[torch.IntTensor] = None,
position_ids: Optional[torch.IntTensor] = None,
inputs_embeds: Optional[torch.FloatTensor] = None,
spec_metadata: Optional[SpecMetadata] = None,
lora_params: Optional[dict] = None,
**kwargs,
) -> torch.Tensor:
Expand All @@ -259,6 +268,7 @@ def forward(
hidden_states=hidden_states,
attn_metadata=attn_metadata,
residual=residual,
spec_metadata=spec_metadata,
lora_params=lora_params,
)

Expand All @@ -267,11 +277,8 @@ def forward(


@register_auto_model("DeciLMForCausalLM")
class NemotronNASForCausalLM(DecoderModelForCausalLM[NemotronNASModel,
PretrainedConfig]):
class NemotronNASForCausalLM(SpecDecOneEngineForCausalLM[NemotronNASModel,
PretrainedConfig]):

def __init__(self, model_config: ModelConfig[PretrainedConfig]):
super().__init__(NemotronNASModel(model_config),
config=model_config,
hidden_size=model_config.pretrained_config.hidden_size,
vocab_size=model_config.pretrained_config.vocab_size)
super().__init__(NemotronNASModel(model_config), model_config)