tensorrt_llm_type

Code from TRT-LLM that export optimized models to the TensorRT-LLM checkpoint.

Classes

LayerNormPositionType	LayerNormPositionType from tensorrt_llm.functional.
LayerNormType	LayerNormType from tensorrt_llm.functional.
MLPType	MLPType from tensorrt_llm.functional.

class LayerNormPositionType

Bases: IntEnum

LayerNormPositionType from tensorrt_llm.functional.

__new__(value)

post_layernorm = 1

pre_layernorm = 0

class LayerNormType

Bases: IntEnum

LayerNormType from tensorrt_llm.functional.

GroupNorm = 2

LayerNorm = 0

RmsNorm = 1

__new__(value)

class MLPType

Bases: IntEnum

MLPType from tensorrt_llm.functional.

FusedGatedMLP = 2

GatedMLP = 1

MLP = 0

__new__(value)