How to continue training with HuggingFace Trainer?

🤗Transformers
1

When training a model with Huggingface Trainer object, e.g. from Neural Plasticity - Bert2Bert on WMT14 | Kaggle

from transformers import Seq2SeqTrainer, Seq2SeqTrainingArguments

import os
os.environ["WANDB_DISABLED"] = "true"

batch_size = 2

# set training arguments - these params are not really tuned, feel free to change
training_args = Seq2SeqTrainingArguments(
    output_dir="./",
    evaluation_strategy="steps",
    per_device_train_batch_size=batch_size,
    per_device_eval_batch_size=batch_size,
    predict_with_generate=True,
    logging_steps=2,  # set to 1000 for full training
    save_steps=16,    # set to 500 for full training
    eval_steps=4,     # set to 8000 for full training
    warmup_steps=1,   # set to 2000 for full training
    max_steps=16,     # delete for full training
    # overwrite_output_dir=True,
    save_total_limit=1,
    #fp16=True, 
)


# instantiate trainer
trainer = Seq2SeqTrainer(
    model=multibert,
    tokenizer=tokenizer,
    args=training_args,
    train_dataset=train_data,
    eval_dataset=val_data,
)

trainer.train()

When it finished training, it outputs:

TrainOutput(global_step=16, training_loss=10.065429925918579, metrics={'train_runtime': 541.4209, 'train_samples_per_second': 0.059, 'train_steps_per_second': 0.03, 'total_flos': 19637939109888.0, 'train_loss': 10.065429925918579, 'epoch': 0.03})

If we want to continue training with more steps, e.g. max_steps=16 (from previous trainer.train() run) and another max_steps=160, do we do something like this?

from transformers import Seq2SeqTrainer, Seq2SeqTrainingArguments

import os
os.environ["WANDB_DISABLED"] = "true"

batch_size = 2

# set training arguments - these params are not really tuned, feel free to change
training_args = Seq2SeqTrainingArguments(
    output_dir="./",
    evaluation_strategy="steps",
    per_device_train_batch_size=batch_size,
    per_device_eval_batch_size=batch_size,
    predict_with_generate=True,
    logging_steps=2,  # set to 1000 for full training
    save_steps=16,    # set to 500 for full training
    eval_steps=4,     # set to 8000 for full training
    warmup_steps=1,   # set to 2000 for full training
    max_steps=16,     # delete for full training
    # overwrite_output_dir=True,
    save_total_limit=1,
    #fp16=True, 
)


# instantiate trainer
trainer = Seq2SeqTrainer(
    model=multibert,
    tokenizer=tokenizer,
    args=training_args,
    train_dataset=train_data,
    eval_dataset=val_data,
)

# First 16 steps.
trainer.train()


# set training arguments - these params are not really tuned, feel free to change
training_args_2 = Seq2SeqTrainingArguments(
    output_dir="./",
    evaluation_strategy="steps",
    per_device_train_batch_size=batch_size,
    per_device_eval_batch_size=batch_size,
    predict_with_generate=True,
    logging_steps=2,  # set to 1000 for full training
    save_steps=16,    # set to 500 for full training
    eval_steps=4,     # set to 8000 for full training
    warmup_steps=1,   # set to 2000 for full training
    max_steps=160,     # delete for full training
    # overwrite_output_dir=True,
    save_total_limit=1,
    #fp16=True, 
)


# instantiate trainer
trainer = Seq2SeqTrainer(
    model=multibert,
    tokenizer=tokenizer,
    args=training_args_2,
    train_dataset=train_data,
    eval_dataset=val_data,
)

# Continue training for 160 steps
trainer.train()

If the above is not the canonical way to continue training a model, how to continue training with HuggingFace Trainer?

1 Like
2

Also asked on python - How to continue training with HuggingFace Trainer? - Stack Overflow

3

There’s a suggestion to use resume_from_checkpoint argument in Trainer but it seems like it’s not available for Seq2SeqTrainer:

trainer = Seq2SeqTrainer(
    model=multibert,
    tokenizer=tokenizer,
    args=training_args,
    train_dataset=train_data,
    eval_dataset=val_data,
    resume_from_checkpoint=True
)

[out]:

TypeError: Seq2SeqTrainer.__init__() got an unexpected keyword argument 'resume_from_checkpoint'
4

The original answer on SO was referencing the correct example in
huggingface docs, but had a bug in describing it. Namely, the parameter resume_from_checkpoint should be used in the train call rather than in the init.

3 Likes