Skip to main content

torchft

Easy Per Step Fault Tolerance for PyTorch

| Documentation | Poster | Design Doc |

PyPI - Stable Version PyPI - Nightly Version


This repository implements techniques for doing a per-step fault tolerance so you can keep training if errors occur without interrupting the entire training job.

This is based on the large scale training techniques presented at PyTorch Conference 2024.

Overview

torchft is designed to provide the primitives required to implement fault tolerance in any application/train script as well as the primitives needed to implement custom fault tolerance strategies.

Out of the box, torchft provides the following algorithms:

  • Fault Tolerant DDP
  • Fault Tolerant HSDP: fault tolerance across the replicated dimension with any mix of FSDP/TP/etc across the other dimensions.
  • LocalSGD
  • DiLoCo

To implement these, torchft provides some key reusable components:

  1. Coordination primitives that can determine which workers are healthy via heartbeating on a per-step basis
  2. Fault tolerant ProcessGroup implementations that report errors sanely and be reinitialized gracefully.
  3. Checkpoint transports that can be used to do live recovery from a healthy peer when doing scale up operations.

The following component diagram shows the high level components and how they relate to each other:

Component Diagram

See torchft's documentation for more details.

Examples

torchtitan (Fault Tolerant HSDP)

torchtitan provides an out of the box fault tolerant HSDP training loop built on top of torchft that can be used to train models such as Llama 3 70B.

It also serves as a good example of how you can integrate torchft into your own training script for use with HSDP.

See torchtitan's documentation for end to end usage.

Fault Tolerant DDP

We have a minimal DDP train loop that highlights all of the key components in torchft.

See train_ddp.py for more info.

DiLoCo

LocalSGD and DiLoCo are currently experimental.

See the diloco_train_loop/local_sgd_train_loop tests for an example on how to integrate these algorithms into your training loop.

Design

torchft is designed to allow for fault tolerance when using training with replicated weights such as in DDP or HSDP (FSDP with DDP).

See the design doc for the most detailed explanation.

Lighthouse

torchft implements a lighthouse server that coordinates across the different replica groups and then a per replica group manager and fault tolerance library that can be used in a standard PyTorch training loop.

This allows for membership changes at the training step granularity which can greatly improve efficiency by avoiding stopping the world training on errors.

Lighthouse Diagram

Fault Tolerant HSDP Algorithm

torchft provides an implementation of a fault tolerant HSDP/DDP algorithm. The following diagram shows the high level operations that need to happen in the train loop to ensure everything stays consistent during a healing operation.

HSDP Diagram

See the design doc linked above for more details.

Installing from PyPI

Install the latest stable release:

pip install torchft

Nightly builds are available at https://pypi.org/project/torchft-nightly/

To install a nightly build:

pip install torchft-nightly

If you want all development dependencies you can install:

pip install torchft-nightly[dev]

Installing from Source

Prerequisites

Before proceeding, ensure you have the following installed:

  • Rust (with necessary dependencies)
  • protobuf-compiler and the corresponding development package for Protobuf.
  • PyTorch 2.7 RC+ or Nightly

Note that the Rust versions available in many conda environments may be outdated. To install the latest version of Rust, we recommend downloading it directly from the official website as shown in the below command:

curl --proto '=https' --tlsv1.2 https://sh.rustup.rs -sSf | sh

To install the required packages on a Debian-based system (such as Ubuntu) using apt, run:

sudo apt install protobuf-compiler libprotobuf-dev

or for a Red Hat-based system, run:

sudo dnf install protobuf-compiler protobuf-devel

Installation

pip install .

This uses pyo3+maturin to build the package, you'll need maturin installed.

If the installation command fails to invoke cargo update due to an inability to fetch the manifest, it may be caused by the proxy, proxySSLCert, and proxySSLKey settings in your .gitconfig file affecting the cargo command. To resolve this issue, try temporarily removing these fields from your .gitconfig before running the installation command.

To install in editable mode w/ the Rust extensions and development dependencies, you can use the normal pip install command:

pip install -e '.[dev]'

Usage

Lighthouse

The lighthouse is used for fault tolerance across replicated workers (DDP/FSDP) when using synchronous training.

You can start a lighthouse server by running:

RUST_BACKTRACE=1 torchft_lighthouse --min_replicas 1 --quorum_tick_ms 100 --join_timeout_ms 10000

Example Training Loop (DDP)

See train_ddp.py for the full example.

Invoke with:

TORCHFT_LIGHTHOUSE=http://localhost:29510 torchrun --master_port 29501 --nnodes 1 --nproc_per_node 1 train_ddp.py

train.py:

from torchft import Manager, DistributedDataParallel, Optimizer, ProcessGroupGloo

manager = Manager(
    pg=ProcessGroupGloo(),
    load_state_dict=...,
    state_dict=...,
)

m = nn.Linear(2, 3)
m = DistributedDataParallel(manager, m)
optimizer = Optimizer(manager, optim.AdamW(m.parameters()))

for i in range(1000):
    batch = torch.rand(2, 2, device=device)

    optimizer.zero_grad()

    out = m(batch)
    loss = out.sum()

    loss.backward()

    optimizer.step()

Running DDP

After starting the lighthouse server by running:

RUST_BACKTRACE=1 torchft_lighthouse --min_replicas 1 --quorum_tick_ms 100 --join_timeout_ms 10000

A test DDP script can be launched with torchX with:

torchx run

Or Diloco with:

USE_STREAMING=True torchx run ./torchft/torchx.py:hsdp --script='train_diloco.py'

See .torchxconfig, torchx.py and the torchX documentation to understand how DDP is being ran.

torchx.py could also launch HSDP jobs when workers_per_replica is set > 1, if the training script supports it. For an example HSDP training implementation with torchFT enabled, see torchtitan.

Alternatively, to test on a node with two GPUs, you can launch two replica groups running train_ddp.py by:

On shell 1 (one replica groups starts initial training):

export REPLICA_GROUP_ID=0
export NUM_REPLICA_GROUPS=2

CUDA_VISIBLE_DEVICES=0 TORCHFT_LIGHTHOUSE=http://localhost:29510 torchrun --master_port=29600 --nnodes=1 --nproc_per_node=1 -- train_ddp.py

On shell 2 (a second replica group joins):

export REPLICA_GROUP_ID=1
export NUM_REPLICA_GROUPS=2

CUDA_VISIBLE_DEVICES=1 TORCHFT_LIGHTHOUSE=http://localhost:29510 torchrun --master_port=29601 --nnodes=1 --nproc_per_node=1 -- train_ddp.py

By observing the outputs from both shells, you should observe process group reconfiguration and live checkpoint recovery.

Example Parameter Server

torchft has a fault tolerant parameter server implementation built on it's reconfigurable ProcessGroups. This does not require/use a Lighthouse server.

See parameter_server_test.py for an example.

Contributing

We welcome PRs! See the CONTRIBUTING file.

License

torchft is BSD 3-Clause licensed. See LICENSE for more details.

Metadata

Release files for torchft-nightly 2026.11.8

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distributions (wheels)

Table of built distributions (wheels) for torchft-nightly 2026.11.8
File Actions
torchft_nightly-2026.11.8-cp315-cp315-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.15 CPython 3.15 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp315-cp315-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
torchft_nightly-2026.11.8-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.14 CPython 3.14 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
torchft_nightly-2026.11.8-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.13 CPython 3.13 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
torchft_nightly-2026.11.8-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.12 CPython 3.12 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
torchft_nightly-2026.11.8-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.11 CPython 3.11 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
torchft_nightly-2026.11.8-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.10 CPython 3.10 Linux glibc 2.17+ x86-64 Details for torchft_nightly-2026.11.8-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Total release size: 15.0 MB

Release files / torchft_nightly-2026.11.8-cp315-cp315-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp315-cp315-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.15 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
1ea50339b2367342b1006d6c29b34baec9908b517640f6eca8e61109e756b79b
BLAKE2b-256 checksum
How to use checksums
b36e006b25a857cc0f8fc5548beace54dc07aff3c45683ebe59da784663ee293
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Release files / torchft_nightly-2026.11.8-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.14 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
6633d97f06a148d6a72639f22d894ae27679b78f554c093c3d0eea101ce4e8ee
BLAKE2b-256 checksum
How to use checksums
17861d0b42ac8fc6a5c30e9b213448bc9afe4100fc2663f3412d750ca03cdf08
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Release files / torchft_nightly-2026.11.8-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.13 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
b8057454e2167b74a6d4388a4f332b0f248bfa1c02d914b4d019b1ccc1fa2798
BLAKE2b-256 checksum
How to use checksums
052c425f760f9789260bef51ae7556e1770e6b9ae85f5770e4899d77940dc8ee
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Release files / torchft_nightly-2026.11.8-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.12 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
f92a59a804171431ee04745d1f755c6bf5291ca3514222eaab139e065058a513
BLAKE2b-256 checksum
How to use checksums
3c566f200ebb7a0e5f7077c018012106cfc9a3e1fe09b8b9fd158b9fdbd4d3b2
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Release files / torchft_nightly-2026.11.8-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.11 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
5efc40a5e8f3b75df1c38d65b388f1f3ba5aea4321a51e9af8ef05380886e6c1
BLAKE2b-256 checksum
How to use checksums
6ce03c5538d89e207270b07c35329a0bc153abc269cd625f9b954485f11fbc8f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Release files / torchft_nightly-2026.11.8-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL torchft_nightly-2026.11.8-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 2.5 MB
Tags CPython 3.10 Linux glibc 2.17+ x86-64
SHA-256 checksum
How to use checksums
3267ca99ed9a069ec208358d8f5a6f9a8b6f1ea372dedc98572f8d0e4c80aa85
BLAKE2b-256 checksum
How to use checksums
aaab12d0786d911a31a6004d098ecf6185ba7276d925af1c8387462579b6ba83
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.10.22

Security for torchft-nightly 2026.11.8

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

2026.11.8 This release

6 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page