Accelerate documentation

Working with large models

Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

Working with large models

Dispatching and Offloading Models

accelerate.init_empty_weights

< >

( include_buffers: bool = None )

Parameters

  • include_buffers (bool, optional) — Whether or not to also put all buffers on the meta device while initializing.

A context manager under which models are initialized with all parameters on the meta device, therefore creating an empty model. Useful when just initializing the model would blow the available RAM.

Example:

import torch.nn as nn
from accelerate import init_empty_weights

# Initialize a model with 100 billions parameters in no time and without using any RAM.
with init_empty_weights():
    tst = nn.Sequential(*[nn.Linear(10000, 10000) for _ in range(1000)])

Any model created under this context manager has no weights. As such you can’t do something like model.to(some_device) with it. To load weights inside your empty model, see load_checkpoint_and_dispatch(). Make sure to overwrite the default device_map param for load_checkpoint_and_dispatch(), otherwise dispatch is not called.

accelerate.cpu_offload

< >

( model: Module execution_device: Optional = None offload_buffers: bool = False state_dict: Optional = None preload_module_classes: Optional = None )

Parameters

  • model (torch.nn.Module) — The model to offload.
  • execution_device (torch.device, optional) — The device on which the forward pass of the model will be executed (should be a GPU). Will default to the model first parameter device.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to offload the buffers with the model parameters.
  • state_dict (Dict[str, torch.Tensor], optional) — The state dict of the model that will be kept on CPU.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.

Activates full CPU offload for a model. As a result, all parameters of the model will be offloaded and only one copy of the state dict of the model will be kept. During the forward pass, parameters will be extracted from that state dict and put on the execution device passed as they are needed, then offloaded again.

accelerate.cpu_offload_with_hook

< >

( model: Module execution_device: Union = None prev_module_hook: Optional = None )

Parameters

  • model (torch.nn.Module) — The model to offload.
  • execution_device(str, int or torch.device, optional) — The device on which the model should be executed. Will default to the MPS device if it’s available, then GPU 0 if there is a GPU, and finally to the CPU.
  • prev_module_hook (UserCpuOffloadHook, optional) — The hook sent back by this function for a previous model in the pipeline you are running. If passed, its offload method will be called just before the forward of the model to which this hook is attached.

Offloads a model on the CPU and puts it back to an execution device when executed. The difference with cpu_offload() is that the model stays on the execution device after the forward and is only offloaded again when the offload method of the returned hook is called. Useful for pipelines running a model in a loop.

Example:

model_1, hook_1 = cpu_offload_with_hook(model_1, cuda_device)
model_2, hook_2 = cpu_offload_with_hook(model_2, cuda_device, prev_module_hook=hook_1)
model_3, hook_3 = cpu_offload_with_hook(model_3, cuda_device, prev_module_hook=hook_2)

hid_1 = model_1(input)
for i in range(50):
    # model1 is offloaded on the CPU at the first iteration, model 2 stays on the GPU for this whole loop.
    hid_2 = model_2(hid_1)
# model2 is offloaded to the CPU just before this forward.
hid_3 = model_3(hid_3)

# For model3, you need to manually call the hook offload method.
hook_3.offload()

accelerate.disk_offload

< >

( model: Module offload_dir: Union execution_device: Optional = None offload_buffers: bool = False preload_module_classes: Optional = None )

Parameters

  • model (torch.nn.Module) — The model to offload.
  • offload_dir (str or os.PathLike) — The folder in which to offload the model weights (or where the model weights are already offloaded).
  • execution_device (torch.device, optional) — The device on which the forward pass of the model will be executed (should be a GPU). Will default to the model’s first parameter device.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to offload the buffers with the model parameters.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.

Activates full disk offload for a model. As a result, all parameters of the model will be offloaded as memory-mapped array in a given folder. During the forward pass, parameters will be accessed from that folder and put on the execution device passed as they are needed, then offloaded again.

accelerate.dispatch_model

< >

( model: Module device_map: Dict main_device: Optional = None state_dict: Optional = None offload_dir: Union = None offload_index: Optional = None offload_buffers: bool = False skip_keys: Union = None preload_module_classes: Optional = None force_hooks: bool = False )

Parameters

  • model (torch.nn.Module) — The model to dispatch.
  • device_map (Dict[str, Union[str, int, torch.device]]) — A dictionary mapping module names in the models state_dict to the device they should go to. Note that "disk" is accepted even if it’s not a proper value for torch.device.
  • main_device (str, int or torch.device, optional) — The main execution device. Will default to the first device in the device_map different from "cpu" or "disk".
  • state_dict (Dict[str, torch.Tensor], optional) — The state dict of the part of the model that will be kept on CPU.
  • offload_dir (str or os.PathLike) — The folder in which to offload the model weights (or where the model weights are already offloaded).
  • offload_index (Dict, optional) — A dictionary from weight name to their information (dtype/ shape or safetensors filename). Will default to the index saved in save_folder.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to offload the buffers with the model parameters.
  • skip_keys (str or List[str], optional) — A list of keys to ignore when moving inputs or outputs between devices.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.
  • force_hooks (bool, optional, defaults to False) — Whether or not to force device hooks to be attached to the model even if all layers are dispatched to a single device.

Dispatches a model according to a given device map. Layers of the model might be spread across GPUs, offloaded on the CPU or even the disk.

accelerate.load_checkpoint_and_dispatch

< >

( model: Module checkpoint: Union device_map: Union = None max_memory: Optional = None no_split_module_classes: Optional = None offload_folder: Union = None offload_buffers: bool = False dtype: Union = None offload_state_dict: Optional = None skip_keys: Union = None preload_module_classes: Optional = None force_hooks: bool = False )

Parameters

  • model (torch.nn.Module) — The model in which we want to load a checkpoint.
  • checkpoint (str or os.PathLike) — The folder checkpoint to load. It can be:

    • a path to a file containing a whole model state dict
    • a path to a .json file containing the index to a sharded checkpoint
    • a path to a folder containing a unique .index.json file and the shards of a checkpoint.
  • device_map (Dict[str, Union[int, str, torch.device]], optional) — A map that specifies where each submodule should go. It doesn’t need to be refined to each parameter/buffer name, once a given module name is inside, every submodule of it will be sent to the same device.

    To have Accelerate compute the most optimized device_map automatically, set device_map="auto". For more information about each option see here. Defaults to None, which means dispatch_model() will not be called.

  • max_memory (Dict, optional) — A dictionary device identifier to maximum memory. Will default to the maximum memory available for each GPU and the available CPU RAM if unset.
  • no_split_module_classes (List[str], optional) — A list of layer class names that should never be split across device (for instance any layer that has a residual connection).
  • offload_folder (str or os.PathLike, optional) — If the device_map contains any value "disk", the folder where we will offload weights.
  • offload_buffers (bool, optional, defaults to False) — In the layers that are offloaded on the CPU or the hard drive, whether or not to offload the buffers as well as the parameters.
  • dtype (str or torch.dtype, optional) — If provided, the weights will be converted to that type when loaded.
  • offload_state_dict (bool, optional) — If True, will temporarily offload the CPU state dict on the hard drive to avoid getting out of CPU RAM if the weight of the CPU state dict + the biggest shard does not fit. Will default to True if the device map picked contains "disk" values.
  • skip_keys (str or List[str], optional) — A list of keys to ignore when moving inputs or outputs between devices.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.
  • force_hooks (bool, optional, defaults to False) — Whether or not to force device hooks to be attached to the model even if all layers are dispatched to a single device.

Loads a (potentially sharded) checkpoint inside a model, potentially sending weights to a given device as they are loaded and adds the various hooks that will make this model run properly (even if split across devices).

Example:

>>> from accelerate import init_empty_weights, load_checkpoint_and_dispatch
>>> from huggingface_hub import hf_hub_download
>>> from transformers import AutoConfig, AutoModelForCausalLM

>>> # Download the Weights
>>> checkpoint = "EleutherAI/gpt-j-6B"
>>> weights_location = hf_hub_download(checkpoint, "pytorch_model.bin")

>>> # Create a model and initialize it with empty weights
>>> config = AutoConfig.from_pretrained(checkpoint)
>>> with init_empty_weights():
...     model = AutoModelForCausalLM.from_config(config)

>>> # Load the checkpoint and dispatch it to the right devices
>>> model = load_checkpoint_and_dispatch(
...     model, weights_location, device_map="auto", no_split_module_classes=["GPTJBlock"]
... )

accelerate.load_checkpoint_in_model

< >

( model: Module checkpoint: Union device_map: Optional = None offload_folder: Union = None dtype: Union = None offload_state_dict: bool = False offload_buffers: bool = False keep_in_fp32_modules: List = None offload_8bit_bnb: bool = False )

Parameters

  • model (torch.nn.Module) — The model in which we want to load a checkpoint.
  • checkpoint (str or os.PathLike) — The folder checkpoint to load. It can be:
    • a path to a file containing a whole model state dict
    • a path to a .json file containing the index to a sharded checkpoint
    • a path to a folder containing a unique .index.json file and the shards of a checkpoint.
    • a path to a folder containing a unique pytorch_model.bin or a model.safetensors file.
  • device_map (Dict[str, Union[int, str, torch.device]], optional) — A map that specifies where each submodule should go. It doesn’t need to be refined to each parameter/buffer name, once a given module name is inside, every submodule of it will be sent to the same device.
  • offload_folder (str or os.PathLike, optional) — If the device_map contains any value "disk", the folder where we will offload weights.
  • dtype (str or torch.dtype, optional) — If provided, the weights will be converted to that type when loaded.
  • offload_state_dict (bool, optional, defaults to False) — If True, will temporarily offload the CPU state dict on the hard drive to avoid getting out of CPU RAM if the weight of the CPU state dict + the biggest shard does not fit.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to include the buffers in the weights offloaded to disk.
  • keep_in_fp32_modules(List[str], optional) — A list of the modules that we keep in torch.float32 dtype.
  • offload_8bit_bnb (bool, optional) — Whether or not to enable offload of 8-bit modules on cpu/disk.

Loads a (potentially sharded) checkpoint inside a model, potentially sending weights to a given device as they are loaded.

Once loaded across devices, you still need to call dispatch_model() on your model to make it able to run. To group the checkpoint loading and dispatch in one single call, use load_checkpoint_and_dispatch().

accelerate.infer_auto_device_map

< >

( model: Module max_memory: Optional = None no_split_module_classes: Optional = None dtype: Union = None special_dtypes: Optional = None verbose: bool = False clean_result: bool = True offload_buffers: bool = False )

Parameters

  • model (torch.nn.Module) — The model to analyze.
  • max_memory (Dict, optional) — A dictionary device identifier to maximum memory. Will default to the maximum memory available if unset. Example: max_memory={0: "1GB"}.
  • no_split_module_classes (List[str], optional) — A list of layer class names that should never be split across device (for instance any layer that has a residual connection).
  • dtype (str or torch.dtype, optional) — If provided, the weights will be converted to that type when loaded.
  • special_dtypes (Dict[str, Union[str, torch.device]], optional) — If provided, special dtypes to consider for some specific weights (will override dtype used as default for all weights).
  • verbose (bool, optional, defaults to False) — Whether or not to provide debugging statements as the function builds the device_map.
  • clean_result (bool, optional, defaults to True) — Clean the resulting device_map by grouping all submodules that go on the same device together.
  • offload_buffers (bool, optional, defaults to False) — In the layers that are offloaded on the CPU or the hard drive, whether or not to offload the buffers as well as the parameters.

Compute a device map for a given model giving priority to GPUs, then offload on CPU and finally offload to disk, such that:

  • we don’t exceed the memory available of any of the GPU.
  • if offload to the CPU is needed, there is always room left on GPU 0 to put back the layer offloaded on CPU that has the largest size.
  • if offload to the CPU is needed,we don’t exceed the RAM available on the CPU.
  • if offload to the disk is needed, there is always room left on the CPU to put back the layer offloaded on disk that has the largest size.

All computation is done analyzing sizes and dtypes of the model parameters. As a result, the model can be on the meta device (as it would if initialized within the init_empty_weights context manager).

Model Hooks

Hook Classes

class accelerate.hooks.ModelHook

< >

( )

A hook that contains callbacks to be executed just before and after the forward method of a model. The difference with PyTorch existing hooks is that they get passed along the kwargs.

Class attribute:

  • no_grad (bool, optional, defaults to False) — Whether or not to execute the actual forward pass under the torch.no_grad() context manager.

detach_hook

< >

( module )

Parameters

  • module (torch.nn.Module) — The module detached from this hook.

To be executed when the hook is detached from a module.

init_hook

< >

( module )

Parameters

  • module (torch.nn.Module) — The module attached to this hook.

To be executed when the hook is attached to the module.

post_forward

< >

( module output ) Any

Parameters

  • module (torch.nn.Module) — The module whose forward pass been executed just before this event.
  • output (Any) — The output of the module.

Returns

Any

The processed output.

To be executed just after the forward method of the model.

pre_forward

< >

( module *args **kwargs ) Tuple[Tuple[Any], Dict[Str, Any]]

Parameters

  • module (torch.nn.Module) — The module whose forward pass will be executed just after this event.
  • args (Tuple[Any]) — The positional arguments passed to the module.
  • kwargs (Dict[Str, Any]) — The keyword arguments passed to the module.

Returns

Tuple[Tuple[Any], Dict[Str, Any]]

A tuple with the treated args and kwargs.

To be executed just before the forward method of the model.

class accelerate.hooks.AlignDevicesHook

< >

( execution_device: Union = None offload: bool = False io_same_device: bool = False weights_map: Optional = None offload_buffers: bool = False place_submodules: bool = False skip_keys: Union = None tied_params_map: Optional = None )

Parameters

  • execution_device (torch.device, optional) — The device on which inputs and model weights should be placed before the forward pass.
  • offload (bool, optional, defaults to False) — Whether or not the weights should be offloaded after the forward pass.
  • io_same_device (bool, optional, defaults to False) — Whether or not the output should be placed on the same device as the input was.
  • weights_map (Mapping[str, torch.Tensor], optional) — When the model weights are offloaded, a (potentially lazy) map from param names to the tensor values.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to include the associated module’s buffers when offloading.
  • place_submodules (bool, optional, defaults to False) — Whether to place the submodules on execution_device during the init_hook event.

A generic ModelHook that ensures inputs and model weights are on the same device for the forward pass of the associated module, potentially offloading the weights after the forward pass.

class accelerate.hooks.SequentialHook

< >

( *hooks )

A hook that can contain several hooks and iterates through them at each event.

Adding Hooks

accelerate.hooks.add_hook_to_module

< >

( module: Module hook: ModelHook append: bool = False ) torch.nn.Module

Parameters

  • module (torch.nn.Module) — The module to attach a hook to.
  • hook (ModelHook) — The hook to attach.
  • append (bool, optional, defaults to False) — Whether the hook should be chained with an existing one (if module already contains a hook) or not.

Returns

torch.nn.Module

The same module, with the hook attached (the module is modified in place, so the result can be discarded).

Adds a hook to a given module. This will rewrite the forward method of the module to include the hook, to remove this behavior and restore the original forward method, use remove_hook_from_module.

If the module already contains a hook, this will replace it with the new hook passed by default. To chain two hooks together, pass append=True, so it chains the current and new hook into an instance of the SequentialHook class.

accelerate.hooks.attach_execution_device_hook

< >

( module: Module execution_device: Union skip_keys: Union = None preload_module_classes: Optional = None tied_params_map: Optional = None )

Parameters

  • module (torch.nn.Module) — The module where we want to attach the hooks.
  • execution_device (int, str or torch.device) — The device on which inputs and model weights should be placed before the forward pass.
  • skip_keys (str or List[str], optional) — A list of keys to ignore when moving inputs or outputs between devices.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.
  • tied_params_map (Optional[Dict[int, Dict[torch.device, torch.Tensor]]], optional, defaults to None) — A map of data pointers to dictionaries of devices to already dispatched tied weights. For a given execution device, this parameter is useful to reuse the first available pointer of a shared weight for all others, instead of duplicating memory.

Recursively attaches AlignDevicesHook to all submodules of a given model to make sure they have the right execution device

accelerate.hooks.attach_align_device_hook

< >

( module: Module execution_device: Optional = None offload: bool = False weights_map: Optional = None offload_buffers: bool = False module_name: str = '' skip_keys: Union = None preload_module_classes: Optional = None tied_params_map: Optional = None )

Parameters

  • module (torch.nn.Module) — The module where we want to attach the hooks.
  • execution_device (torch.device, optional) — The device on which inputs and model weights should be placed before the forward pass.
  • offload (bool, optional, defaults to False) — Whether or not the weights should be offloaded after the forward pass.
  • weights_map (Mapping[str, torch.Tensor], optional) — When the model weights are offloaded, a (potentially lazy) map from param names to the tensor values.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to include the associated module’s buffers when offloading.
  • module_name (str, optional, defaults to "") — The name of the module.
  • skip_keys (str or List[str], optional) — A list of keys to ignore when moving inputs or outputs between devices.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.
  • tied_params_map (Optional[Dict[int, Dict[torch.device, torch.Tensor]]], optional, defaults to None) — A map of data pointers to dictionaries of devices to already dispatched tied weights. For a given execution device, this parameter is useful to reuse the first available pointer of a shared weight for all others, instead of duplicating memory.

Recursively attaches AlignDevicesHook to all submodules of a given model that have direct parameters and/or buffers.

accelerate.hooks.attach_align_device_hook_on_blocks

< >

( module: Module execution_device: Union = None offload: Union = False weights_map: Mapping = None offload_buffers: bool = False module_name: str = '' skip_keys: Union = None preload_module_classes: Optional = None tied_params_map: Optional = None )

Parameters

  • module (torch.nn.Module) — The module where we want to attach the hooks.
  • execution_device (torch.device or Dict[str, torch.device], optional) — The device on which inputs and model weights should be placed before the forward pass. It can be one device for the whole module, or a dictionary mapping module name to device.
  • offload (bool, optional, defaults to False) — Whether or not the weights should be offloaded after the forward pass. It can be one boolean for the whole module, or a dictionary mapping module name to boolean.
  • weights_map (Mapping[str, torch.Tensor], optional) — When the model weights are offloaded, a (potentially lazy) map from param names to the tensor values.
  • offload_buffers (bool, optional, defaults to False) — Whether or not to include the associated module’s buffers when offloading.
  • module_name (str, optional, defaults to "") — The name of the module.
  • skip_keys (str or List[str], optional) — A list of keys to ignore when moving inputs or outputs between devices.
  • preload_module_classes (List[str], optional) — A list of classes whose instances should load all their weights (even in the submodules) at the beginning of the forward. This should only be used for classes that have submodules which are registered but not called directly during the forward, for instance if a dense linear layer is registered, but at forward, dense.weight and dense.bias are used in some operations instead of calling dense directly.
  • tied_params_map (Optional[Dict[int, Dict[torch.device, torch.Tensor]]], optional, defaults to None) — A map of data pointers to dictionaries of devices to already dispatched tied weights. For a given execution device, this parameter is useful to reuse the first available pointer of a shared weight for all others, instead of duplicating memory.

Attaches AlignDevicesHook to all blocks of a given model as needed.

Removing Hooks

accelerate.hooks.remove_hook_from_module

< >

( module: Module recurse = False ) torch.nn.Module

Parameters

  • module (torch.nn.Module) — The module to attach a hook to.
  • recurse (bool, optional) — Whether to remove the hooks recursively

Returns

torch.nn.Module

The same module, with the hook detached (the module is modified in place, so the result can be discarded).

Removes any hook attached to a module via add_hook_to_module.

accelerate.hooks.remove_hook_from_submodules

< >

( module: Module )

Parameters

  • module (torch.nn.Module) — The module on which to remove all hooks.

Recursively removes all hooks attached on the submodules of a given model.