Transformers documentation

Efficient Training on Multiple CPUs

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v4.39.3).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

Efficient Training on Multiple CPUs

When training on a single CPU is too slow, we can use multiple CPUs. This guide focuses on PyTorch-based DDP enabling distributed CPU training efficiently on bare metal and Kubernetes.

Intel® oneCCL Bindings for PyTorch

Intel® oneCCL (collective communications library) is a library for efficient distributed deep learning training implementing such collectives like allreduce, allgather, alltoall. For more information on oneCCL, please refer to the oneCCL documentation and oneCCL specification.

Module oneccl_bindings_for_pytorch (torch_ccl before version 1.12) implements PyTorch C10D ProcessGroup API and can be dynamically loaded as external ProcessGroup and only works on Linux platform now

Check more detailed information for oneccl_bind_pt.

Intel® oneCCL Bindings for PyTorch installation

Wheel files are available for the following Python versions:

Extension Version Python 3.6 Python 3.7 Python 3.8 Python 3.9 Python 3.10

Please run pip list | grep torch to get your pytorch_version.

pip install oneccl_bind_pt=={pytorch_version} -f

where {pytorch_version} should be your PyTorch version, for instance 2.1.0. Check more approaches for oneccl_bind_pt installation. Versions of oneCCL and PyTorch must match.

oneccl_bindings_for_pytorch 1.12.0 prebuilt wheel does not work with PyTorch 1.12.1 (it is for PyTorch 1.12.0) PyTorch 1.12.1 should work with oneccl_bindings_for_pytorch 1.12.100

Intel® MPI library

Use this standards-based MPI implementation to deliver flexible, efficient, scalable cluster messaging on Intel® architecture. This component is part of the Intel® oneAPI HPC Toolkit.

oneccl_bindings_for_pytorch is installed along with the MPI tool set. Need to source the environment before using it.

for Intel® oneCCL >= 1.12.0

oneccl_bindings_for_pytorch_path=$(python -c "from oneccl_bindings_for_pytorch import cwd; print(cwd)")
source $oneccl_bindings_for_pytorch_path/env/

for Intel® oneCCL whose version < 1.12.0

torch_ccl_path=$(python -c "import torch; import torch_ccl; import os;  print(os.path.abspath(os.path.dirname(torch_ccl.__file__)))")
source $torch_ccl_path/env/

Intel® Extension for PyTorch installation

Intel Extension for PyTorch (IPEX) provides performance optimizations for CPU training with both Float32 and BFloat16 (refer to the single CPU section to learn more).

The following “Usage in Trainer” takes mpirun in Intel® MPI library as an example.

Usage in Trainer

To enable multi CPU distributed training in the Trainer with the ccl backend, users should add --ddp_backend ccl in the command arguments.

Let’s see an example with the question-answering example

The following command enables training with 2 processes on one Xeon node, with one process running per one socket. The variables OMP_NUM_THREADS/CCL_WORKER_COUNT can be tuned for optimal performance.

 export MASTER_ADDR=
 mpirun -n 2 -genv OMP_NUM_THREADS=23 \
 python3 \
 --model_name_or_path google-bert/bert-large-uncased \
 --dataset_name squad \
 --do_train \
 --do_eval \
 --per_device_train_batch_size 12  \
 --learning_rate 3e-5  \
 --num_train_epochs 2  \
 --max_seq_length 384 \
 --doc_stride 128  \
 --output_dir /tmp/debug_squad/ \
 --no_cuda \
 --ddp_backend ccl \

The following command enables training with a total of four processes on two Xeons (node0 and node1, taking node0 as the main process), ppn (processes per node) is set to 2, with one process running per one socket. The variables OMP_NUM_THREADS/CCL_WORKER_COUNT can be tuned for optimal performance.

In node0, you need to create a configuration file which contains the IP addresses of each node (for example hostfile) and pass that configuration file path as an argument.

 cat hostfile #node0 ip #node1 ip

Now, run the following command in node0 and 4DDP will be enabled in node0 and node1 with BF16 auto mixed precision:

 export #node0 ip
 mpirun -f hostfile -n 4 -ppn 2 \
 -genv OMP_NUM_THREADS=23 \
 python3 \
 --model_name_or_path google-bert/bert-large-uncased \
 --dataset_name squad \
 --do_train \
 --do_eval \
 --per_device_train_batch_size 12  \
 --learning_rate 3e-5  \
 --num_train_epochs 2  \
 --max_seq_length 384 \
 --doc_stride 128  \
 --output_dir /tmp/debug_squad/ \
 --no_cuda \
 --ddp_backend ccl \
 --use_ipex \

Usage with Kubernetes

The same distributed training job from the previous section can be deployed to a Kubernetes cluster using the Kubeflow PyTorchJob training operator.


This example assumes that you have:

  • Access to a Kubernetes cluster with Kubeflow installed
  • kubectl installed and configured to access the Kubernetes cluster
  • A Persistent Volume Claim (PVC) that can be used to store datasets and model files. There are multiple options for setting up the PVC including using an NFS storage class or a cloud storage bucket.
  • A Docker container that includes your model training script and all the dependencies needed to run the script. For distributed CPU training jobs, this typically includes PyTorch, Transformers, Intel Extension for PyTorch, Intel oneCCL Bindings for PyTorch, and OpenSSH to communicate between the containers.

The snippet below is an example of a Dockerfile that uses a base image that supports distributed CPU training and then extracts a Transformers release to the /workspace directory, so that the example scripts are included in the image:

FROM intel/ai-workflows:torch-2.0.1-huggingface-multinode-py3.9

WORKDIR /workspace

# Download and extract the transformers code
RUN mkdir transformers && \
    curl -sSL --retry 5${HF_TRANSFORMERS_VER}.tar.gz | tar -C transformers --strip-components=1 -xzf -

The image needs to be built and copied to the cluster’s nodes or pushed to a container registry prior to deploying the PyTorchJob to the cluster.

PyTorchJob Specification File

The Kubeflow PyTorchJob is used to run the distributed training job on the cluster. The yaml file for the PyTorchJob defines parameters such as:

  • The name of the PyTorchJob
  • The number of replicas (workers)
  • The python script and it’s parameters that will be used to run the training job
  • The types of resources (node selector, memory, and CPU) needed for each worker
  • The image/tag for the Docker container to use
  • Environment variables
  • A volume mount for the PVC

The volume mount defines a path where the PVC will be mounted in the container for each worker pod. This location can be used for the dataset, checkpoint files, and the saved model after training completes.

The snippet below is an example of a yaml file for a PyTorchJob with 4 workers running the question-answering example.

apiVersion: ""
kind: PyTorchJob
  name: transformers-pytorchjob
  namespace: kubeflow
    rdzvBackend: c10d
    minReplicas: 1
    maxReplicas: 4
    maxRestarts: 10
      replicas: 4  # The number of worker pods
      restartPolicy: OnFailure
            - name: pytorch
              image: <image name>:<tag>  # Specify the docker image to use for the worker pods
              imagePullPolicy: IfNotPresent
                - torchrun
                - /workspace/transformers/examples/pytorch/question-answering/
                - --model_name_or_path
                - "google-bert/bert-large-uncased"
                - --dataset_name
                - "squad"
                - --do_train
                - --do_eval
                - --per_device_train_batch_size
                - "12"
                - --learning_rate
                - "3e-5"
                - --num_train_epochs
                - "2"
                - --max_seq_length
                - "384"
                - --doc_stride
                - "128"
                - --output_dir
                - "/tmp/pvc-mount/output"
                - --no_cuda
                - --ddp_backend
                - "ccl"
                - --use_ipex
                - --bf16  # Specify --bf16 if your hardware supports bfloat16
              - name: LD_PRELOAD
                value: "/usr/lib/x86_64-linux-gnu/"
              - name: TRANSFORMERS_CACHE
                value: "/tmp/pvc-mount/transformers_cache"
              - name: HF_DATASETS_CACHE
                value: "/tmp/pvc-mount/hf_datasets_cache"
              - name: LOGLEVEL
                value: "INFO"
              - name: CCL_WORKER_COUNT
                value: "1"
              - name: OMP_NUM_THREADS  # Can be tuned for optimal performance
-                value: "56"
                  cpu: 200  # Update the CPU and memory limit values based on your nodes
                  memory: 128Gi
                  cpu: 200  # Update the CPU and memory request values based on your nodes
                  memory: 128Gi
              - name: pvc-volume
                mountPath: /tmp/pvc-mount
              - mountPath: /dev/shm
                name: dshm
          restartPolicy: Never
          nodeSelector:  #  Optionally use the node selector to specify what types of nodes to use for the workers
            node-type: spr
          - name: pvc-volume
              claimName: transformers-pvc
          - name: dshm
              medium: Memory

To run this example, update the yaml based on your training script and the nodes in your cluster.

The CPU resource limits/requests in the yaml are defined in cpu units where 1 CPU unit is equivalent to 1 physical CPU core or 1 virtual core (depending on whether the node is a physical host or a VM). The amount of CPU and memory limits/requests defined in the yaml should be less than the amount of available CPU/memory capacity on a single machine. It is usually a good idea to not use the entire machine’s capacity in order to leave some resources for the kubelet and OS. In order to get “guaranteed” quality of service for the worker pods, set the same CPU and memory amounts for both the resource limits and requests.


After the PyTorchJob spec has been updated with values appropriate for your cluster and training job, it can be deployed to the cluster using:

kubectl create -f pytorchjob.yaml

The kubectl get pods -n kubeflow command can then be used to list the pods in the kubeflow namespace. You should see the worker pods for the PyTorchJob that was just deployed. At first, they will probably have a status of “Pending” as the containers get pulled and created, then the status should change to “Running”.

NAME                                                     READY   STATUS                  RESTARTS          AGE
transformers-pytorchjob-worker-0                         1/1     Running                 0                 7m37s
transformers-pytorchjob-worker-1                         1/1     Running                 0                 7m37s
transformers-pytorchjob-worker-2                         1/1     Running                 0                 7m37s
transformers-pytorchjob-worker-3                         1/1     Running                 0                 7m37s

The logs for worker can be viewed using kubectl logs -n kubeflow <pod name>. Add -f to stream the logs, for example:

kubectl logs -n kubeflow transformers-pytorchjob-worker-0 -f

After the training job completes, the trained model can be copied from the PVC or storage location. When you are done with the job, the PyTorchJob resource can be deleted from the cluster using kubectl delete -f pytorchjob.yaml.


This guide covered running distributed PyTorch training jobs using multiple CPUs on bare metal and on a Kubernetes cluster. Both cases utilize Intel Extension for PyTorch and Intel oneCCL Bindings for PyTorch for optimal training performance, and can be used as a template to run your own workload on multiple nodes.