MCPcopy Create free account
hub / github.com/NVIDIA/TensorRT-LLM / to_mapping

Method to_mapping

tensorrt_llm/llmapi/llm_args.py:577–590  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

575 return self.world_size > 1
576
577 def to_mapping(self) -> Mapping:
578 return Mapping(world_size=self.world_size,
579 rank=mpi_rank(),
580 gpus_per_node=self.gpus_per_node,
581 tp_size=self.tp_size,
582 pp_size=self.pp_size,
583 pp_partition=self.pp_partition,
584 cp_size=self.cp_size,
585 cp_config=self.cp_config,
586 enable_attention_dp=self.enable_attention_dp,
587 enable_lm_head_tp_in_adp=self.enable_lm_head_tp_in_adp,
588 moe_cluster_size=self.moe_cluster_size,
589 moe_tp_size=self.moe_tp_size,
590 moe_ep_size=self.moe_ep_size)
591
592
593class CalibConfig(StrictBaseModel):

Callers 4

__init__Method · 0.80
get_pretrained_configMethod · 0.80
create_py_executorFunction · 0.80
_create_py_executorMethod · 0.80

Calls 2

MappingClass · 0.85
mpi_rankFunction · 0.50

Tested by

no test coverage detected