Transformers documentation


Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started


The Trainer is a complete training and evaluation loop for PyTorch models implemented in the Transformers library. You only need to pass it the necessary pieces for training (model, tokenizer, dataset, evaluation function, training hyperparameters, etc.), and the Trainer class takes care of the rest. This makes it easier to start training faster without manually writing your own training loop. But at the same time, Trainer is very customizable and offers a ton of training options so you can tailor it to your exact training needs.

In addition to the Trainer class, Transformers also provides a Seq2SeqTrainer class for sequence-to-sequence tasks like translation or summarization. There is also the SFTTrainer class from the TRL library which wraps the Trainer class and is optimized for training language models like Llama-2 and Mistral with autoregressive techniques. SFTTrainer also supports features like sequence packing, LoRA, quantization, and DeepSpeed for efficiently scaling to any model size.

Feel free to check out the API reference for these other Trainer-type classes to learn more about when to use which one. In general, Trainer is the most versatile option and is appropriate for a broad spectrum of tasks. Seq2SeqTrainer is designed for sequence-to-sequence tasks and SFTTrainer is designed for training language models.

Before you start, make sure Accelerate - a library for enabling and running PyTorch training across distributed environments - is installed.

pip install accelerate

# upgrade
pip install accelerate --upgrade

This guide provides an overview of the Trainer class.

Basic usage

Trainer includes all the code you’ll find in a basic training loop:

  1. perform a training step to calculate the loss
  2. calculate the gradients with the backward method
  3. update the weights based on the gradients
  4. repeat this process until you’ve reached a predetermined number of epochs

The Trainer class abstracts all of this code away so you don’t have to worry about manually writing a training loop every time or if you’re just getting started with PyTorch and training. You only need to provide the essential components required for training, such as a model and a dataset, and the Trainer class handles everything else.

If you want to specify any training options or hyperparameters, you can find them in the TrainingArguments class. For example, let’s define where to save the model in output_dir and push the model to the Hub after training with push_to_hub=True.

from transformers import TrainingArguments

training_args = TrainingArguments(

Pass training_args to the Trainer along with a model, dataset, something to preprocess the dataset with (depending on your data type it could be a tokenizer, feature extractor or image processor), a data collator, and a function to compute the metrics you want to track during training.

Finally, call train() to start training!

from transformers import Trainer

trainer = Trainer(



The Trainer class saves your model checkpoints to the directory specified in the output_dir parameter of TrainingArguments. You’ll find the checkpoints saved in a checkpoint-000 subfolder where the numbers at the end correspond to the training step. Saving checkpoints are useful for resuming training later.

# resume from latest checkpoint

# resume from specific checkpoint saved in output directory

You can save your checkpoints (the optimizer state is not saved by default) to the Hub by setting push_to_hub=True in TrainingArguments to commit and push them. Other options for deciding how your checkpoints are saved are set up in the hub_strategy parameter:

  • hub_strategy="checkpoint" pushes the latest checkpoint to a subfolder named “last-checkpoint” from which you can resume training
  • hug_strategy="all_checkpoints" pushes all checkpoints to the directory defined in output_dir (you’ll see one checkpoint per folder in your model repository)

When you resume training from a checkpoint, the Trainer tries to keep the Python, NumPy, and PyTorch RNG states the same as they were when the checkpoint was saved. But because PyTorch has various non-deterministic default settings, the RNG states aren’t guaranteed to be the same. If you want to enable full determinism, take a look at the Controlling sources of randomness guide to learn what you can enable to make your training fully deterministic. Keep in mind though that by making certain settings deterministic, training may be slower.

Customize the Trainer

While the Trainer class is designed to be accessible and easy-to-use, it also offers a lot of customizability for more adventurous users. Many of the Trainer’s method can be subclassed and overridden to support the functionality you want, without having to rewrite the entire training loop from scratch to accommodate it. These methods include:

For example, if you want to customize the compute_loss() method to use a weighted loss instead.

from torch import nn
from transformers import Trainer

class CustomTrainer(Trainer):
    def compute_loss(self, model, inputs, return_outputs=False):
        labels = inputs.pop("labels")
        # forward pass
        outputs = model(**inputs)
        logits = outputs.get("logits")
        # compute custom loss for 3 labels with different weights
        loss_fct = nn.CrossEntropyLoss(weight=torch.tensor([1.0, 2.0, 3.0], device=model.device))
        loss = loss_fct(logits.view(-1, self.model.config.num_labels), labels.view(-1))
        return (loss, outputs) if return_outputs else loss


Another option for customizing the Trainer is to use callbacks. Callbacks don’t change anything in the training loop. They inspect the training loop state and then execute some action (early stopping, logging results, etc.) depending on the state. In other words, a callback can’t be used to implement something like a custom loss function and you’ll need to subclass and override the compute_loss() method for that.

For example, if you want to add an early stopping callback to the training loop after 10 steps.

from transformers import TrainerCallback

class EarlyStoppingCallback(TrainerCallback):
    def __init__(self, num_steps=10):
        self.num_steps = num_steps
    def on_step_end(self, args, state, control, **kwargs):
        if state.global_step >= self.num_steps:
            return {"should_training_stop": True}
            return {}

Then pass it to the Trainer’s callback parameter.

from transformers import Trainer

trainer = Trainer(


Check out the logging API reference for more information about the different logging levels.

The Trainer is set to logging.INFO by default which reports errors, warnings, and other basic information. A Trainer replica - in distributed environments - is set to logging.WARNING which only reports errors and warnings. You can change the logging level with the log_level and log_level_replica parameters in TrainingArguments.

To configure the log level setting for each node, use the log_on_each_node parameter to determine whether to use the log level on each node or only on the main node.

Trainer sets the log level separately for each node in the Trainer.__init__() method, so you may want to consider setting this sooner if you’re using other Transformers functionalities before creating the Trainer object.

For example, to set your main code and modules to use the same log level according to each node:

logger = logging.getLogger(__name__)

    format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
    datefmt="%m/%d/%Y %H:%M:%S",

log_level = training_args.get_process_log_level()

trainer = Trainer(...)

Use different combinations of log_level and log_level_replica to configure what gets logged on each of the nodes.

single node
multi-node ... --log_level warning --log_level_replica error


NEFTune is a technique that can improve performance by adding noise to the embedding vectors during training. To enable it in Trainer, set the neftune_noise_alpha parameter in TrainingArguments to control how much noise is added.

from transformers import TrainingArguments, Trainer

training_args = TrainingArguments(..., neftune_noise_alpha=0.1)
trainer = Trainer(..., args=training_args)

NEFTune is disabled after training to restore the original embedding layer to avoid any unexpected behavior.

Accelerate and Trainer

The Trainer class is powered by Accelerate, a library for easily training PyTorch models in distributed environments with support for integrations such as FullyShardedDataParallel (FSDP) and DeepSpeed.

Learn more about FSDP sharding strategies, CPU offloading, and more with the Trainer in the Fully Sharded Data Parallel guide.

To use Accelerate with Trainer, run the accelerate.config command to set up training for your training environment. This command creates a config_file.yaml that’ll be used when you launch your training script. For example, some example configurations you can setup are:

DeepSpeed with Accelerate plugin
compute_environment: LOCAL_MACHINE                                                                                             
distributed_type: MULTI_GPU                                                                                                    
downcast_bf16: 'no'
gpu_ids: all
machine_rank: 0 #change rank as per the node
main_process_port: 9898
main_training_function: main
mixed_precision: fp16
num_machines: 2
num_processes: 8
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false

The accelerate_launch command is the recommended way to launch your training script on a distributed system with Accelerate and Trainer with the parameters specified in config_file.yaml. This file is saved to the Accelerate cache folder and automatically loaded when you run accelerate_launch.

For example, to run the training script with the FSDP configuration:

accelerate launch \
    ./examples/pytorch/text-classification/ \
    --model_name_or_path google-bert/bert-base-cased \
    --task_name $TASK_NAME \
    --do_train \
    --do_eval \
    --max_seq_length 128 \
    --per_device_train_batch_size 16 \
    --learning_rate 5e-5 \
    --num_train_epochs 3 \
    --output_dir /tmp/$TASK_NAME/ \

You could also specify the parameters from the config_file.yaml file directly in the command line:

accelerate launch --num_processes=2 \
    --use_fsdp \
    --mixed_precision=bf16 \
    --fsdp_auto_wrap_policy=TRANSFORMER_BASED_WRAP  \
    --fsdp_transformer_layer_cls_to_wrap="BertLayer" \
    --fsdp_sharding_strategy=1 \
    --fsdp_state_dict_type=FULL_STATE_DICT \
    --model_name_or_path google-bert/bert-base-cased \
    --task_name $TASK_NAME \
    --do_train \
    --do_eval \
    --max_seq_length 128 \
    --per_device_train_batch_size 16 \
    --learning_rate 5e-5 \
    --num_train_epochs 3 \
    --output_dir /tmp/$TASK_NAME/ \

Check out the Launching your Accelerate scripts tutorial to learn more about accelerate_launch and custom configurations.