`bridge.recipes.gemma.gemma2`#

Module Contents#

Classes#

`Gemma2CommonKwargs`	Typed options accepted by Gemma2 recipe helper functions.
`Gemma2FinetuneKwargs`	Typed options accepted by Gemma2 finetuning recipe helper functions.

Functions#

`gemma2_2b_pretrain_config`	Return a pre-training config for Gemma2 2B.
`gemma2_9b_pretrain_config`	Return a pre-training config for Gemma2 9B.
`gemma2_27b_pretrain_config`	Return a pre-training config for Gemma2 27B.
`_gemma2_common`	Create a pre-training configuration for Gemma2 models.
`gemma2_2b_finetune_config`	Return a finetuning config for Gemma2 2B.
`gemma2_9b_finetune_config`	Return a finetuning config for Gemma2 9B.
`gemma2_27b_finetune_config`	Return a finetuning config for Gemma2 27B.
`_gemma2_finetune_common`	Common finetuning configuration for all Gemma2 models.

API#

class bridge.recipes.gemma.gemma2.Gemma2CommonKwargs#

Bases: typing_extensions.TypedDict

Typed options accepted by Gemma2 recipe helper functions.

Initialization

Initialize self. See help(type(self)) for accurate signature.

hf_path: str#: None

dir: Optional[str]#: None

name: str#: None

data_paths: Optional[List[str]]#: None

data_args_path: Optional[str]#: None

train_data_path: Optional[List[str]]#: None

valid_data_path: Optional[List[str]]#: None

test_data_path: Optional[str]#: None

per_split_data_args_path: Optional[str]#: None

mock: bool#: None

tensor_model_parallel_size: int#: None

pipeline_model_parallel_size: int#: None

pipeline_dtype: Optional[torch.dtype]#: None

virtual_pipeline_model_parallel_size: Optional[int]#: None

context_parallel_size: int#: None

sequence_parallel: bool#: None

use_megatron_fsdp: bool#: None

train_iters: int#: None

global_batch_size: int#: None

micro_batch_size: int#: None

seq_length: int#: None

lr: float#: None

min_lr: float#: None

lr_warmup_iters: int#: None

lr_decay_iters: Optional[int]#: None

eval_interval: int#: None

save_interval: int#: None

use_null_tokenizer: bool#: None

precision_config: Optional[Union[megatron.bridge.training.mixed_precision.MixedPrecisionConfig, str]]#: None

comm_overlap_config: Optional[megatron.bridge.training.comm_overlap.CommOverlapConfig]#: None

class bridge.recipes.gemma.gemma2.Gemma2FinetuneKwargs#

Bases: typing_extensions.TypedDict

Typed options accepted by Gemma2 finetuning recipe helper functions.

Initialization

Initialize self. See help(type(self)) for accurate signature.

hf_path: str#: None

dir: Optional[str]#: None

name: str#: None

pretrained_checkpoint: Optional[str]#: None

peft: Union[str, megatron.bridge.peft.base.PEFT, None]#: None

packed_sequence: bool#: None

train_iters: int#: None

global_batch_size: Optional[int]#: None

micro_batch_size: int#: None

seq_length: Optional[int]#: None

eval_interval: int#: None

save_interval: int#: None

finetune_lr: Optional[float]#: None

min_lr: float#: None

lr_warmup_iters: int#: None

lr_decay_iters: Optional[int]#: None

wandb_project: Optional[str]#: None

wandb_entity: Optional[str]#: None

wandb_exp_name: Optional[str]#: None

precision_config: Optional[Union[megatron.bridge.training.mixed_precision.MixedPrecisionConfig, str]]#: None

bridge.recipes.gemma.gemma2.gemma2_2b_pretrain_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2CommonKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a pre-training config for Gemma2 2B.

Default parallelism: TP=2, PP=1

bridge.recipes.gemma.gemma2.gemma2_9b_pretrain_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2CommonKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a pre-training config for Gemma2 9B.

Default parallelism: TP=8, PP=1

bridge.recipes.gemma.gemma2.gemma2_27b_pretrain_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2CommonKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a pre-training config for Gemma2 27B.

Default parallelism: TP=8, PP=2

bridge.recipes.gemma.gemma2._gemma2_common( hf_path: str, dir: Optional[str] = None, name: str = 'default', data_paths: Optional[List[str]] = None, data_args_path: Optional[str] = None, train_data_path: Optional[List[str]] = None, valid_data_path: Optional[List[str]] = None, test_data_path: Optional[List[str]] = None, per_split_data_args_path: Optional[str] = None, mock: bool = False, tensor_model_parallel_size: int = 1, pipeline_model_parallel_size: int = 1, pipeline_dtype: Optional[torch.dtype] = None, virtual_pipeline_model_parallel_size: Optional[int] = None, context_parallel_size: int = 1, sequence_parallel: bool = False, use_megatron_fsdp: bool = False, train_iters: int = 300000, global_batch_size: int = 32, micro_batch_size: int = 2, seq_length: int = 4096, lr: float = 0.0003, min_lr: float = 3e-05, lr_warmup_iters: int = 500, lr_decay_iters: Optional[int] = None, eval_interval: int = 500, save_interval: int = 500, use_null_tokenizer: bool = False, precision_config: Optional[Union[megatron.bridge.training.mixed_precision.MixedPrecisionConfig, str]] = 'bf16_mixed', comm_overlap_config: Optional[megatron.bridge.training.comm_overlap.CommOverlapConfig] = None, ) → megatron.bridge.training.config.ConfigContainer#: Create a pre-training configuration for Gemma2 models.

bridge.recipes.gemma.gemma2.gemma2_2b_finetune_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2FinetuneKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a finetuning config for Gemma2 2B.

Default configuration: 1 node, 8 GPUs

LoRA/DoRA: TP=1, PP=1, LR=1e-4
Full SFT: TP=1, PP=1, LR=5e-6

bridge.recipes.gemma.gemma2.gemma2_9b_finetune_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2FinetuneKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a finetuning config for Gemma2 9B.

Default configuration: 1 node, 8 GPUs

LoRA/DoRA: TP=1, PP=1, LR=1e-4
Full SFT: TP=4, PP=1, LR=5e-6

bridge.recipes.gemma.gemma2.gemma2_27b_finetune_config(

**user_kwargs: typing_extensions.Unpack[bridge.recipes.gemma.gemma2.Gemma2FinetuneKwargs],

) → megatron.bridge.training.config.ConfigContainer#

Return a finetuning config for Gemma2 27B.

Default configuration: 2 nodes (SFT) or 1 node (LoRA), 8 GPUs per node

LoRA/DoRA: TP=4, PP=1, LR=1e-4
Full SFT: TP=8, PP=2, LR=5e-6

bridge.recipes.gemma.gemma2._gemma2_finetune_common( hf_path: str, dir: Optional[str] = None, name: str = 'default', tensor_model_parallel_size: int = 1, pipeline_model_parallel_size: int = 1, pipeline_dtype: Optional[torch.dtype] = None, virtual_pipeline_model_parallel_size: Optional[int] = None, context_parallel_size: int = 1, sequence_parallel: bool = False, pretrained_checkpoint: Optional[str] = None, peft: Union[str, megatron.bridge.peft.base.PEFT, None] = 'lora', packed_sequence: bool = False, train_iters: int = 100, global_batch_size: Optional[int] = None, micro_batch_size: int = 1, seq_length: Optional[int] = None, eval_interval: int = 50, save_interval: int = 100, finetune_lr: Optional[float] = None, min_lr: float = 0.0, lr_warmup_iters: int = 10, lr_decay_iters: Optional[int] = None, wandb_project: Optional[str] = None, wandb_entity: Optional[str] = None, wandb_exp_name: Optional[str] = None, precision_config: Optional[Union[megatron.bridge.training.mixed_precision.MixedPrecisionConfig, str]] = None, ) → megatron.bridge.training.config.ConfigContainer#: Common finetuning configuration for all Gemma2 models.

bridge.recipes.gemma.gemma2#

Module Contents#

Classes#

Functions#

API#

`bridge.recipes.gemma.gemma2`#