search

huggingface / transformers

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
https://huggingface.co/transformers
Apache License 2.0
132.24k stars 26.34k forks source link

[XLA] Cannot restore from checkpoint on TPU #7976

Closed ksjae closed 3 years ago

ksjae commented 3 years ago

Environment info

Who can help

@LysandreJik @sgugger @TevenLeScao

Information

Model I am using (Bert, XLNet ...): GPT2

The problem arises when using:

The tasks I am working on is:

To reproduce

Steps to reproduce the behavior:

  1. Modify examples/language-modeling/run_language_modeling.py to below 
    
import logging
import math
import os
import glob
import datasets
from dataclasses import dataclass, field
from typing import Optional

from datasets import list_datasets, load_dataset

from transformers import ( CONFIG_MAPPING, MODEL_WITH_LM_HEAD_MAPPING, AutoConfig, AutoModelWithLMHead, AutoTokenizer, DataCollatorForLanguageModeling, DataCollatorForPermutationLanguageModeling, HfArgumentParser, LineByLineTextDataset, PreTrainedTokenizer, TextDataset, Trainer, TrainingArguments, set_seed, )

logger = logging.getLogger(name)

MODEL_CONFIG_CLASSES = list(MODEL_WITH_LM_HEAD_MAPPING.keys()) MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)

@dataclass class ModelArguments: """ Arguments pertaining to which model/config/tokenizer we are going to fine-tune, or train from scratch. """

model_name_or_path: Optional[str] = field(
    default=None,
    metadata={
        "help": "The model checkpoint for weights initialization. Leave None if you want to train a model from scratch."
    },
)
model_type: Optional[str] = field(
    default=None,
    metadata={"help": "If training from scratch, pass a model type from the list: " + ", ".join(MODEL_TYPES)},
)
config_name: Optional[str] = field(
    default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
    default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
    default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from s3"}
)

@dataclass class DataTrainingArguments: """ Arguments pertaining to what data we are going to input our model for training and eval. """

train_data_file: Optional[str] = field(
    default=None, metadata={"help": "The input training data file (a text file)."}
)
eval_data_file: Optional[str] = field(
    default=None,
    metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
)
line_by_line: bool = field(
    default=False,
    metadata={"help": "Whether distinct lines of text in the dataset are to be handled as distinct sequences."},
)

mlm: bool = field(
    default=False, metadata={"help": "Train with masked-language modeling loss instead of language modeling."}
)
mlm_probability: float = field(
    default=0.15, metadata={"help": "Ratio of tokens to mask for masked language modeling loss"}
)
plm_probability: float = field(
    default=1 / 6,
    metadata={
        "help": "Ratio of length of a span of masked tokens to surrounding context length for permutation language modeling."
    },
)
max_span_length: int = field(
    default=5, metadata={"help": "Maximum length of a span of masked tokens for permutation language modeling."}
)

block_size: int = field(
    default=-1,
    metadata={
        "help": "Optional input sequence length after tokenization."
        "The training dataset will be truncated in block of this size for training."
        "Default to the model max input length for single sentence inputs (take into account special tokens)."
    },
)
overwrite_cache: bool = field(
    default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
arrow: bool = field(
    default=True,
    metadata={
        "help": "Use Arrow-based HF NLP for optimization."
    },
)

def get_dataset( args: DataTrainingArguments, tokenizer: PreTrainedTokenizer, evaluate: bool = False, cache_dir: Optional[str] = "./cache", ): tokenizer.pad_token = "<|endoftext|>" tokenizer._pad_token = "<|endoftext|>"

tokenizer.pad_token_id = 50256

file_path = args.eval_data_file if evaluate else args.train_data_file
if True:
    dataset = datasets.load_from_disk(file_path)
    dataset.set_format(type='torch', columns=['input_ids'])
    return dataset

if False:
    dataset = load_dataset("text", data_files=[file_path], split='train')
    dataset = dataset.map(lambda ex: tokenizer(ex["text"], add_special_tokens=True,
                                            truncation=True, max_length=args.block_size), batched=True)
    dataset.set_format(type='torch', columns=['input_ids'])
    dataset.save_to_disk(file_path+'.arrow')
    return dataset

if args.line_by_line:
    return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)
else:
    return TextDataset(
        tokenizer=tokenizer,
        file_path=file_path,
        block_size=args.block_size,
        overwrite_cache=args.overwrite_cache,
        cache_dir=cache_dir,
    )
"""
dataset = load_dataset("text", data_files=file_path, split="train")
dataset = dataset.map(lambda ex: tokenizer(ex["text"], add_special_tokens=True,
                                        truncation=True, max_length=args.block_size), batched=True)
dataset.set_format(type='torch', columns=['input_ids'])
return dataset
"""

def main():

See all possible arguments in src/transformers/training_args.py

# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.

parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
model_args, data_args, training_args = parser.parse_args_into_dataclasses()

if data_args.eval_data_file is None and training_args.do_eval:
    raise ValueError(
        "Cannot do evaluation without an evaluation data file. Either supply a file to --eval_data_file "
        "or remove the --do_eval argument."
    )

if (
    os.path.exists(training_args.output_dir)
    and os.listdir(training_args.output_dir)
    and training_args.do_train
    and not training_args.overwrite_output_dir
):
    raise ValueError(
        f"Output directory ({training_args.output_dir}) already exists and is not empty. Use --overwrite_output_dir to overcome."
    )

# Setup logging
logging.basicConfig(
    format="%(asctime)s - %(levelname)s - %(name)s -   %(message)s",
    datefmt="%m/%d/%Y %H:%M:%S",
    level=logging.INFO if training_args.local_rank in [-1, 0] else logging.WARN,
)
logger.warning(
    "Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
    training_args.local_rank,
    training_args.device,
    training_args.n_gpu,
    bool(training_args.local_rank != -1),
    training_args.fp16,
)
logger.info("Training/evaluation parameters %s", training_args)

# Set seed
set_seed(training_args.seed)

# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.

if model_args.config_name:
    config = AutoConfig.from_pretrained(model_args.config_name, cache_dir=model_args.cache_dir)
elif model_args.model_name_or_path:
    config = AutoConfig.from_pretrained(model_args.model_name_or_path, cache_dir=model_args.cache_dir)
else:
    config = CONFIG_MAPPING[model_args.model_type]()
    logger.warning("You are instantiating a new config instance from scratch.")

if model_args.tokenizer_name:
    tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name, cache_dir=model_args.cache_dir)
elif model_args.model_name_or_path:
    tokenizer = AutoTokenizer.from_pretrained(model_args.model_name_or_path, cache_dir=model_args.cache_dir)
else:
    raise ValueError(
        "You are instantiating a new tokenizer from scratch. This is not supported, but you can do it from another script, save it,"
        "and load it from here, using --tokenizer_name"
    )

tokenizer.pad_token = "<|endoftext|>"
tokenizer._pad_token = "<|endoftext|>"

if model_args.model_name_or_path:
    model = AutoModelWithLMHead.from_pretrained(
        model_args.model_name_or_path,
        from_tf=bool(".ckpt" in model_args.model_name_or_path),
        config=config,
        cache_dir=model_args.cache_dir,
    )
else:
    logger.info("Training new model from scratch")
    model = AutoModelWithLMHead.from_config(config)

model.resize_token_embeddings(len(tokenizer))

if config.model_type in ["bert", "roberta", "distilbert", "camembert"] and not data_args.mlm:
    raise ValueError(
        "BERT and RoBERTa-like models do not have LM heads but masked LM heads. They must be run using the"
        "--mlm flag (masked language modeling)."
    )

if data_args.block_size <= 0:
    data_args.block_size = tokenizer.max_len
    # Our input block size will be the max possible for the model
else:
    data_args.block_size = min(data_args.block_size, tokenizer.max_len)

# Get datasets

train_dataset = (
    get_dataset(data_args, tokenizer=tokenizer, cache_dir=model_args.cache_dir) if training_args.do_train else None
)
eval_dataset = (
    get_dataset(data_args, tokenizer=tokenizer, evaluate=True, cache_dir=model_args.cache_dir)
    if training_args.do_eval
    else None
)
if config.model_type == "xlnet":
    data_collator = DataCollatorForPermutationLanguageModeling(
        tokenizer=tokenizer,
        plm_probability=data_args.plm_probability,
        max_span_length=data_args.max_span_length,
    )
else:
    data_collator = DataCollatorForLanguageModeling(
        tokenizer=tokenizer, mlm=data_args.mlm, mlm_probability=data_args.mlm_probability
    )

# Initialize our Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    data_collator=data_collator,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    prediction_loss_only=True,
)

# Training
if training_args.do_train:
    model_path = (
        model_args.model_name_or_path
        if model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path)
        else None
    )
    trainer.train(model_path=model_path)
    trainer.save_model()
    # For convenience, we also re-save the tokenizer to the same directory,
    # so that you can share your model easily on huggingface.co/models =)
    if trainer.is_world_master():
        tokenizer.save_pretrained(training_args.output_dir)

# Evaluation
results = {}
if training_args.do_eval:
    logger.info("*** Evaluate ***")

    eval_output = trainer.evaluate()

    perplexity = math.exp(eval_output["eval_loss"])
    result = {"perplexity": perplexity}

    output_eval_file = os.path.join(training_args.output_dir, "eval_results_lm.txt")
    if trainer.is_world_master():
        with open(output_eval_file, "w") as writer:
            logger.info("***** Eval results *****")
            for key in sorted(result.keys()):
                logger.info("  %s = %s", key, str(result[key]))
                writer.write("%s = %s\n" % (key, str(result[key])))

    results.update(result)

return results

def _mp_fn(index):

For xla_spawn (TPUs)

main()

if name == "main": main()


2. set torch-xla-nightly Conda & set env
3. run script from checkpoint (replace dataset, since I cannot upload 48 GB worth of arrow files)

XLA_USE_BF16=1 python3 examples/xla_spawn.py --num_cores 8 examples/language-modeling/run_language_modeling.py --output_dir=kogpt1 --model_type=gpt2 --do_train --train_data_file=/home/ksjcom0705_gmail_com/NEWS_ARROW --overwrite_output_dir --per_device_train_batch_size=6 --save_steps 10000 --num_train_epochs=1 --block_size 2048 --eval_steps 10000 --logging_steps=10000 --tokenizer_name /home/ksjcom0705_gmail_com/kotok --model_name_or_path=kogpt1/checkpoint-1000


The error is this:

Exception in device=TPU:3: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:5: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:6: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:1: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:0: don't know how to restore data location of torch.FloatStorage (tagged with xla:1) Exception in device=TPU:4: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:7: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Exception in device=TPU:2: don't know how to restore data location of torch.FloatStorage (tagged with xla:0) Traceback (most recent call last): File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 330, in _mp_start_fn _start_fn(index, pf_cfg, fn, args) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 324, in _start_fn fn(gindex, *args) File "/home/ksjcom0705_gmail_com/transformers/examples/language-modeling/run_language_modeling.py", line 332, in _mp_fn main() File "/home/ksjcom0705_gmail_com/transformers/examples/language-modeling/run_language_modeling.py", line 300, in main trainer.train(model_path=model_path) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/transformers/trainer.py", line 629, in train torch.load(os.path.join(model_path, "optimizer.pt"), map_location=self.args.device) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 592, in load return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 851, in _load result = unpickler.load() File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 843, in persistent_load load_tensor(data_type, size, key, _maybe_decode_ascii(location)) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 832, in load_tensor loaded_storages[key] = restore_location(storage, location) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 812, in restore_location return default_restore_location(storage, str(map_location)) File "/anaconda3/envs/torch-xla-nightly/lib/python3.6/site-packages/torch/serialization.py", line 180, in default_restore_location

Expected behavior

Run normally

LysandreJik commented 3 years ago

Pinging @sgugger