bridge.recipes.moonlight.moonlight_16b

Module Contents

Functions

_get_moonlight_pipeline_layout	Get pipeline layout for Moonlight-16B based on PP and VP size.
moonlight_16b_pretrain_config	Return a pre-training config for Moonlight-16B.
moonlight_16b_sft_config	Return a full SFT config for Moonlight-16B.
moonlight_16b_peft_config	Return a PEFT config for Moonlight-16B.

Data

__all__

API

bridge.recipes.moonlight.moonlight_16b._get_moonlight_pipeline_layout(pp_size: int, vp_size: int)

Get pipeline layout for Moonlight-16B based on PP and VP size.

bridge.recipes.moonlight.moonlight_16b.moonlight_16b_pretrain_config() → megatron.bridge.training.config.ConfigContainer

Return a pre-training config for Moonlight-16B.

Recommended parallelism: TP=2, PP=1, EP=8 Uses precision-aware optimizer with bf16 gradients/moments.

bridge.recipes.moonlight.moonlight_16b.moonlight_16b_sft_config() → megatron.bridge.training.config.ConfigContainer

Return a full SFT config for Moonlight-16B.

Default parallelism: TP=2, PP=1, EP=8, SP=True

Returns:

ConfigContainer with all settings pre-configured for Moonlight-16B SFT.

bridge.recipes.moonlight.moonlight_16b.moonlight_16b_peft_config(

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

) → megatron.bridge.training.config.ConfigContainer

Return a PEFT config for Moonlight-16B.

Default parallelism: TP=1, PP=1, EP=2, SP=False

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

Returns:

ConfigContainer with all settings pre-configured for Moonlight-16B PEFT.

bridge.recipes.moonlight.moonlight_16b.__all__

[‘moonlight_16b_pretrain_config’, ‘moonlight_16b_sft_config’, ‘moonlight_16b_peft_config’]