bridge.recipes.exaone.h100.exaone45

EXAONE 4.5 VL H100 training recipes.

Module Contents

Functions

_set_optimizer_precision
_apply_exaone45_common
exaone45_vl_33b_sft_16gpu_h100_bf16_config	Return a full SFT config for EXAONE 4.5 VL 33B.
exaone45_vl_33b_peft_4gpu_h100_bf16_config	Return a PEFT config for EXAONE 4.5 VL 33B.

Data

_HF_PATH
__all__

API

bridge.recipes.exaone.h100.exaone45._HF_PATH

‘LGAI-EXAONE/EXAONE-4.5-33B’

bridge.recipes.exaone.h100.exaone45._set_optimizer_precision(

cfg: megatron.bridge.training.config.ConfigContainer,

) → None

bridge.recipes.exaone.h100.exaone45._apply_exaone45_common(

cfg: megatron.bridge.training.config.ConfigContainer,

) → None

bridge.recipes.exaone.h100.exaone45.exaone45_vl_33b_sft_16gpu_h100_bf16_config() → megatron.bridge.training.config.ConfigContainer

Return a full SFT config for EXAONE 4.5 VL 33B.

bridge.recipes.exaone.h100.exaone45.exaone45_vl_33b_peft_4gpu_h100_bf16_config(

peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',

) → megatron.bridge.training.config.ConfigContainer

Return a PEFT config for EXAONE 4.5 VL 33B.

bridge.recipes.exaone.h100.exaone45.__all__

[‘exaone45_vl_33b_peft_4gpu_h100_bf16_config’, ‘exaone45_vl_33b_sft_16gpu_h100_bf16_config’]