Accelerate documentation

Fully Sharded Data Parallel

Join the Hugging Face community

to get started

# Fully Sharded Data Parallel

To accelerate training huge models on larger batch sizes, we can use a fully sharded data parallel model. This type of data parallel paradigm enables fitting more data and larger models by sharding the optimizer states, gradients and parameters. To read more about it and the benefits, check out the Fully Sharded Data Parallel blog. We have integrated the latest PyTorch’s Fully Sharded Data Parallel (FSDP) training feature. All you need to do is enable it through the config.

## How it works out the box

accelerate config

and answer the questions asked. This will generate a config file that will be used automatically to properly set the default options when doing

accelerate launch my_script.py --args_to_my_script

For instance, here is how you would run the NLP example (from the root of the repo) with FSDP enabled:

compute_environment: LOCAL_MACHINE
deepspeed_config: {}
distributed_type: FSDP
fsdp_config:
min_num_params: 2000
sharding_strategy: 1
machine_rank: 0
main_process_ip: null
main_process_port: null
main_training_function: main
mixed_precision: 'no'
num_machines: 1
num_processes: 2
use_cpu: false
accelerate launch examples/nlp_example.py

Currently, Accelerate supports following config through the CLI:

Sharding Strategy: [1] FULL_SHARD, [2] SHARD_GRAD_OP
Min Num Params: FSDP\'s minimum number of parameters for Default Auto Wrapping.
Offload Params: Decides Whether to offload parameters and gradients to CPU.

## Few caveats to be aware of

• PyTorch FSDP auto wraps sub-modules, flattens the parameters and shards the parameters in place. Due to this, any optimizer created before model wrapping gets broken and occupies more memory. Hence, it is highly recommended and efficient to prepare model before creating optimizer. Accelerate will automatically wrap the model and create an optimizer for you in case of single model with a warning message.

FSDP Warning: When using FSDP, it is efficient and recommended to call prepare for the model before creating the optimizer

However, below is the recommended way to prepare model and optimizer while using FSDP:

model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", return_dict=True)
+ model = accelerator.prepare(model)

-    )

+        )

• In case of a single model, if you have created optimizer with multiple parameter groups and called prepare with them together, then the parameter groups will be lost and the following warning is displayed:

FSDP Warning: When using FSDP, several parameter groups will be conflated into a single one due to nested module wrapping and parameter flattening.

This is because parameter groups created before wrapping will have no meaning post wrapping due parameter flattening of nested FSDP modules into 1D arrays (which can consume many layers). For instance, below are the named parameters of FSDP model on GPU 0 (When using 2 GPUs. Around 55M (110M/2) params in 1D arrays as this will have the 1st shard of the parameters). Here, if one has applied no weight decay for [bias, LayerNorm.weight] named parameters of unwrapped BERT model, it can’t be applied to the below FSDP wrapped model as there are no named parameters with either of those strings and the parameters of those layers are concatenated with parameters of various other layers.

{
'_fsdp_wrapped_module.flat_param': torch.Size([494209]),
'_fsdp_wrapped_module._fpw_module.bert.embeddings.word_embeddings._fsdp_wrapped_module.flat_param': torch.Size([11720448]),
'_fsdp_wrapped_module._fpw_module.bert.encoder._fsdp_wrapped_module.flat_param': torch.Size([42527232])
}
• In case of multiple models, it is necessary to prepare the models before creating optimizers else it will throw an error.
• Mixed precision is currently not supported with FSDP.

For more control, users can leverage the FullyShardedDataParallelPlugin wherein they can specify auto_wrap_policy, backward_prefetch and ignored_modules. After creating an instance of this class, users can pass it to the Accelerator class instantiation. For more information on these options, please refer to the PyTorch FullyShardedDataParallel code.

### class accelerate.FullyShardedDataParallelPlugin

< >

( sharding_strategy: typing.Any = None backward_prefetch: typing.Any = None auto_wrap_policy: typing.Any = None cpu_offload: typing.Optional[typing.Callable] = None min_num_params: int = None ignored_modules: typing.Optional[typing.Iterable[torch.nn.modules.module.Module]] = None )

This plugin is used to enable fully sharded data parallelism.