`bridge.recipes.nemotronh.nemotron_nano_v2`#

Module Contents#

Functions#

`nemotron_nano_9b_v2_pretrain_config`	Return a pre-training config for Nemotron Nano 9B v2.
`nemotron_nano_12b_v2_pretrain_config`	Return a pre-training config for Nemotron Nano 12B v2.
`nemotron_nano_9b_v2_sft_config`	Return a full SFT config for Nemotron Nano 9B v2.
`nemotron_nano_12b_v2_sft_config`	Return a full SFT config for Nemotron Nano 12B v2.
`nemotron_nano_9b_v2_peft_config`	Return a PEFT config for Nemotron Nano 9B v2.
`nemotron_nano_12b_v2_peft_config`	Return a PEFT config for Nemotron Nano 12B v2.

Data#

__all__

API#

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_9b_v2_pretrain_config() → megatron.bridge.training.config.ConfigContainer#

Return a pre-training config for Nemotron Nano 9B v2.

This recipe is designed for single-node training (1 node). Default parallelism: TP=2, PP=1, SP=True.

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_12b_v2_pretrain_config() → megatron.bridge.training.config.ConfigContainer#

Return a pre-training config for Nemotron Nano 12B v2.

This recipe is designed for single-node training (1 node). Default parallelism: TP=4, PP=1, SP=True.

Note: Uses FP8 precision by default. Communication overlap is disabled by default.

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_9b_v2_sft_config() → megatron.bridge.training.config.ConfigContainer#

Return a full SFT config for Nemotron Nano 9B v2.

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

Returns:: ConfigContainer with all settings pre-configured for Nemotron Nano 9B v2 SFT.

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_12b_v2_sft_config() → megatron.bridge.training.config.ConfigContainer#

Return a full SFT config for Nemotron Nano 12B v2.

Default parallelism: TP=4, PP=1, SP=True

Returns:: ConfigContainer with all settings pre-configured for Nemotron Nano 12B v2 SFT.

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_9b_v2_peft_config( peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora', ) → megatron.bridge.training.config.ConfigContainer#

Return a PEFT config for Nemotron Nano 9B v2.

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

Parameters:: peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.
Returns:: ConfigContainer with all settings pre-configured for Nemotron Nano 9B v2 PEFT.

bridge.recipes.nemotronh.nemotron_nano_v2.nemotron_nano_12b_v2_peft_config( peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora', ) → megatron.bridge.training.config.ConfigContainer#

Return a PEFT config for Nemotron Nano 12B v2.