EngineX-Cambricon/enginex-mlu370-any2any
9
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init

Browse Source
This commit is contained in:
luopingyi
2025-10-09 16:47:16 +08:00
parent c8feb4deb5
commit e27e3f16bb
5248 changed files with 1778505 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

196
transformers/examples/pytorch/summarization/README.md Normal file
View File

@@ -0,0 +1,196 @@
<!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
## Summarization
This directory contains examples for finetuning and evaluating transformers on summarization tasks.
Please tag @patil-suraj with any issues/unexpected behaviors, or send a PR!
For deprecated `bertabs` instructions, see https://github.com/huggingface/transformers-research-projects/blob/main/bertabs/README.md.
For the old `finetune_trainer.py` and related utils, see [`examples/legacy/seq2seq`](https://github.com/huggingface/transformers/blob/main/examples/legacy/seq2seq).
### Supported Architectures
- `BartForConditionalGeneration`
- `FSMTForConditionalGeneration` (translation only)
- `MBartForConditionalGeneration`
- `MarianMTModel`
- `PegasusForConditionalGeneration`
- `T5ForConditionalGeneration`
- `MT5ForConditionalGeneration`
`run_summarization.py` is a lightweight example of how to download and preprocess a dataset from the [🤗 Datasets](https://github.com/huggingface/datasets) library or use your own files (jsonlines or csv), then fine-tune one of the architectures above on it.
For custom datasets in `jsonlines` format please see: https://huggingface.co/docs/datasets/loading_datasets#json-files
and you also will find examples of these below.
## With Trainer
Here is an example on a summarization task:
```bash
python run_summarization.py \
--model_name_or_path google-t5/t5-small \
--do_train \
--do_eval \
--dataset_name cnn_dailymail \
--dataset_config "3.0.0" \
--source_prefix "summarize: " \
--output_dir /tmp/tst-summarization \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
Only T5 models `google-t5/t5-small`, `google-t5/t5-base`, `google-t5/t5-large`, `google-t5/t5-3b` and `google-t5/t5-11b` must use an additional argument: `--source_prefix "summarize: "`.
We used CNN/DailyMail dataset in this example as `google-t5/t5-small` was trained on it and one can get good scores even when pre-training with a very small sample.
Extreme Summarization (XSum) Dataset is another commonly used dataset for the task of summarization. To use it replace `--dataset_name cnn_dailymail --dataset_config "3.0.0"` with `--dataset_name xsum`.
And here is how you would use it on your own files, after adjusting the values for the arguments
`--train_file`, `--validation_file`, `--text_column` and `--summary_column` to match your setup:
```bash
python run_summarization.py \
--model_name_or_path google-t5/t5-small \
--do_train \
--do_eval \
--train_file path_to_csv_or_jsonlines_file \
--validation_file path_to_csv_or_jsonlines_file \
--source_prefix "summarize: " \
--output_dir /tmp/tst-summarization \
--overwrite_output_dir \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--predict_with_generate
```
The task of summarization supports custom CSV and JSONLINES formats.
#### Custom CSV Files
If it's a csv file the training and validation files should have a column for the inputs texts and a column for the summaries.
If the csv file has just two columns as in the following example:
```csv
text,summary
"I'm sitting here in a boring room. It's just another rainy Sunday afternoon. I'm wasting my time I got nothing to do. I'm hanging around I'm waiting for you. But nothing ever happens. And I wonder","I'm sitting in a room where I'm waiting for something to happen"
"I see trees so green, red roses too. I see them bloom for me and you. And I think to myself what a wonderful world. I see skies so blue and clouds so white. The bright blessed day, the dark sacred night. And I think to myself what a wonderful world.","I'm a gardener and I'm a big fan of flowers."
"Christmas time is here. Happiness and cheer. Fun for all that children call. Their favorite time of the year. Snowflakes in the air. Carols everywhere. Olden times and ancient rhymes. Of love and dreams to share","It's that time of year again."
```
The first column is assumed to be for `text` and the second is for summary.
If the csv file has multiple columns, you can then specify the names of the columns to use:
```bash
--text_column text_column_name \
--summary_column summary_column_name \
```
For example if the columns were:
```csv
id,date,text,summary
```
and you wanted to select only `text` and `summary`, then you'd pass these additional arguments:
```bash
--text_column text \
--summary_column summary \
```
#### Custom JSONLINES Files
The second supported format is jsonlines. Here is an example of a jsonlines custom data file.
```json
{"text": "I'm sitting here in a boring room. It's just another rainy Sunday afternoon. I'm wasting my time I got nothing to do. I'm hanging around I'm waiting for you. But nothing ever happens. And I wonder", "summary": "I'm sitting in a room where I'm waiting for something to happen"}
{"text": "I see trees so green, red roses too. I see them bloom for me and you. And I think to myself what a wonderful world. I see skies so blue and clouds so white. The bright blessed day, the dark sacred night. And I think to myself what a wonderful world.", "summary": "I'm a gardener and I'm a big fan of flowers."}
{"text": "Christmas time is here. Happiness and cheer. Fun for all that children call. Their favorite time of the year. Snowflakes in the air. Carols everywhere. Olden times and ancient rhymes. Of love and dreams to share", "summary": "It's that time of year again."}
```
Same as with the CSV files, by default the first value will be used as the text record and the second as the summary record. Therefore you can use any key names for the entries, in this example `text` and `summary` were used.
And as with the CSV files, you can specify which values to select from the file, by explicitly specifying the corresponding key names. In our example this again would be:
```bash
--text_column text \
--summary_column summary \
```
## With Accelerate
Based on the script [`run_summarization_no_trainer.py`](https://github.com/huggingface/transformers/blob/main/examples/pytorch/summarization/run_summarization_no_trainer.py).
Like `run_summarization.py`, this script allows you to fine-tune any of the models supported on a
summarization task, the main difference is that this
script exposes the bare training loop, to allow you to quickly experiment and add any customization you would like.
It offers less options than the script with `Trainer` (for instance you can easily change the options for the optimizer
or the dataloaders directly in the script) but still run in a distributed setup, on TPU and supports mixed precision by
the mean of the [🤗 `Accelerate`](https://github.com/huggingface/accelerate) library. You can use the script normally
after installing it:
```bash
pip install git+https://github.com/huggingface/accelerate
```
then
```bash
python run_summarization_no_trainer.py \
--model_name_or_path google-t5/t5-small \
--dataset_name cnn_dailymail \
--dataset_config "3.0.0" \
--source_prefix "summarize: " \
--output_dir ~/tmp/tst-summarization
```
You can then use your usual launchers to run in it in a distributed environment, but the easiest way is to run
```bash
accelerate config
```
and reply to the questions asked. Then
```bash
accelerate test
```
that will check everything is ready for training. Finally, you can launch training with
```bash
accelerate launch run_summarization_no_trainer.py \
--model_name_or_path google-t5/t5-small \
--dataset_name cnn_dailymail \
--dataset_config "3.0.0" \
--source_prefix "summarize: " \
--output_dir ~/tmp/tst-summarization
```
This command is the same and will work for:
- a CPU-only setup
- a setup with one GPU
- a distributed training with several GPUs (single or multi node)
- a training on TPUs
Note that this library is in alpha release so your feedback is more than welcome if you encounter any problem using it.

9
transformers/examples/pytorch/summarization/requirements.txt Normal file
View File

@@ -0,0 +1,9 @@
accelerate >= 0.12.0
datasets >= 1.8.0
sentencepiece != 0.1.92
protobuf
rouge-score
nltk
py7zr
torch >= 1.3
evaluate

788
transformers/examples/pytorch/summarization/run_summarization.py Executable file
View File

@@ -0,0 +1,788 @@
#!/usr/bin/env python
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# /// script
# dependencies = [
# "transformers @ git+https://github.com/huggingface/transformers.git",
# "accelerate >= 0.12.0",
# "datasets >= 1.8.0",
# "sentencepiece != 0.1.92",
# "protobuf",
# "rouge-score",
# "nltk",
# "py7zr",
# "torch >= 1.3",
# "evaluate",
# ]
# ///
"""
Fine-tuning the library models for sequence to sequence.
"""
# You can also adapt this script on your own sequence to sequence task. Pointers for this are left as comments.
import logging
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
import datasets
import evaluate
import nltk # Here to have a nice missing dependency error message early on
import numpy as np
from datasets import load_dataset
from filelock import FileLock
import transformers
from transformers import (
AutoConfig,
AutoModelForSeq2SeqLM,
AutoTokenizer,
DataCollatorForSeq2Seq,
HfArgumentParser,
MBart50Tokenizer,
MBart50TokenizerFast,
MBartTokenizer,
MBartTokenizerFast,
Seq2SeqTrainer,
Seq2SeqTrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version, is_offline_mode, send_example_telemetry
from transformers.utils.versions import require_version
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.57.0.dev0")
require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
logger = logging.getLogger(__name__)
try:
nltk.data.find("tokenizers/punkt")
except (LookupError, OSError):
if is_offline_mode():
raise LookupError(
"Offline mode: run this script without TRANSFORMERS_OFFLINE first to download nltk data files"
)
with FileLock(".lock") as lock:
nltk.download("punkt", quiet=True)
# A list of all multilingual tokenizer which require lang attribute.
MULTILINGUAL_TOKENIZERS = [MBartTokenizer, MBartTokenizerFast, MBart50Tokenizer, MBart50TokenizerFast]
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
)
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 to store the pretrained models downloaded from huggingface.co"},
)
use_fast_tokenizer: bool = field(
default=True,
metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
token: str = field(
default=None,
metadata={
"help": (
"The token to use as HTTP bearer authorization for remote files. If not specified, will use the token "
"generated when running `hf auth login` (stored in `~/.huggingface`)."
)
},
)
trust_remote_code: bool = field(
default=False,
metadata={
"help": (
"Whether to trust the execution of code from datasets/models defined on the Hub."
" This option should only be set to `True` for repositories you trust and in which you have read the"
" code, as it will execute code present on the Hub on your local machine."
)
},
)
resize_position_embeddings: Optional[bool] = field(
default=None,
metadata={
"help": (
"Whether to automatically resize the position embeddings if `max_source_length` exceeds "
"the model's position embeddings."
)
},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
lang: Optional[str] = field(default=None, metadata={"help": "Language id for summarization."})
dataset_name: Optional[str] = field(
default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
text_column: Optional[str] = field(
default=None,
metadata={"help": "The name of the column in the datasets containing the full texts (for summarization)."},
)
summary_column: Optional[str] = field(
default=None,
metadata={"help": "The name of the column in the datasets containing the summaries (for summarization)."},
)
train_file: Optional[str] = field(
default=None, metadata={"help": "The input training data file (a jsonlines or csv file)."}
)
validation_file: Optional[str] = field(
default=None,
metadata={
"help": (
"An optional input evaluation data file to evaluate the metrics (rouge) on (a jsonlines or csv file)."
)
},
)
test_file: Optional[str] = field(
default=None,
metadata={
"help": "An optional input test data file to evaluate the metrics (rouge) on (a jsonlines or csv file)."
},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
preprocessing_num_workers: Optional[int] = field(
default=None,
metadata={"help": "The number of processes to use for the preprocessing."},
)
max_source_length: Optional[int] = field(
default=1024,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
max_target_length: Optional[int] = field(
default=128,
metadata={
"help": (
"The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
val_max_target_length: Optional[int] = field(
default=None,
metadata={
"help": (
"The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`. "
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": (
"Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
num_beams: Optional[int] = field(
default=1,
metadata={
"help": (
"Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
)
},
)
ignore_pad_token_for_loss: bool = field(
default=True,
metadata={
"help": "Whether to ignore the tokens corresponding to padded labels in the loss computation or not."
},
)
source_prefix: Optional[str] = field(
default=None, metadata={"help": "A prefix to add before every source text (useful for T5 models)."}
)
forced_bos_token: Optional[str] = field(
default=None,
metadata={
"help": (
"The token to force as the first generated token after the decoder_start_token_id. "
"Useful for multilingual models like mBART where the first generated token"
"needs to be the target language token (Usually it is the target language token)"
)
},
)
def __post_init__(self):
if (
self.dataset_name is None
and self.train_file is None
and self.validation_file is None
and self.test_file is None
):
raise ValueError("Need either a dataset name or a training, validation, or test file.")
else:
if self.train_file is not None:
extension = self.train_file.split(".")[-1]
assert extension in ["csv", "json"], "`train_file` should be a csv or a json file."
if self.validation_file is not None:
extension = self.validation_file.split(".")[-1]
assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
if self.test_file is not None:
extension = self.test_file.split(".")[-1]
assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
if self.val_max_target_length is None:
self.val_max_target_length = self.max_target_length
summarization_name_mapping = {
"amazon_reviews_multi": ("review_body", "review_title"),
"big_patent": ("description", "abstract"),
"cnn_dailymail": ("article", "highlights"),
"orange_sum": ("text", "summary"),
"pn_summary": ("article", "summary"),
"psc": ("extract_text", "summary_text"),
"samsum": ("dialogue", "summary"),
"thaisum": ("body", "summary"),
"xglue": ("news_body", "news_title"),
"xsum": ("document", "summary"),
"wiki_summary": ("article", "highlights"),
"multi_news": ("document", "summary"),
}
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, Seq2SeqTrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_summarization", model_args, data_args)
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
handlers=[logging.StreamHandler(sys.stdout)],
)
if training_args.should_log:
# The default of training_args.log_level is passive, so we set log level at info here to have that default.
transformers.utils.logging.set_verbosity_info()
log_level = training_args.get_process_log_level()
logger.setLevel(log_level)
datasets.utils.logging.set_verbosity(log_level)
transformers.utils.logging.set_verbosity(log_level)
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Log on each process the small summary:
logger.warning(
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}, "
+ f"distributed training: {training_args.parallel_mode.value == 'distributed'}, 16-bits training: {training_args.fp16}"
)
logger.info(f"Training/evaluation parameters {training_args}")
if data_args.source_prefix is None and model_args.model_name_or_path in [
"google-t5/t5-small",
"google-t5/t5-base",
"google-t5/t5-large",
"google-t5/t5-3b",
"google-t5/t5-11b",
]:
logger.warning(
"You're running a t5 model but didn't provide a source prefix, which is the expected, e.g. with "
"`--source_prefix 'summarize: ' `"
)
# Detecting last checkpoint.
last_checkpoint = None
if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
last_checkpoint = get_last_checkpoint(training_args.output_dir)
if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
"Use --overwrite_output_dir to overcome."
)
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
logger.info(
f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
"the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
)
# Set seed before initializing model.
set_seed(training_args.seed)
# Get the datasets: you can either provide your own CSV/JSON training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files this script will use the first column for the full texts and the second column for the
# summaries (unless you specify column names for this with the `text_column` and `summary_column` arguments).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
raw_datasets = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
cache_dir=model_args.cache_dir,
token=model_args.token,
trust_remote_code=model_args.trust_remote_code,
)
else:
data_files = {}
if data_args.train_file is not None:
data_files["train"] = data_args.train_file
extension = data_args.train_file.split(".")[-1]
if data_args.validation_file is not None:
data_files["validation"] = data_args.validation_file
extension = data_args.validation_file.split(".")[-1]
if data_args.test_file is not None:
data_files["test"] = data_args.test_file
extension = data_args.test_file.split(".")[-1]
raw_datasets = load_dataset(
extension,
data_files=data_files,
cache_dir=model_args.cache_dir,
token=model_args.token,
)
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
config = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
token=model_args.token,
trust_remote_code=model_args.trust_remote_code,
)
tokenizer = AutoTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
use_fast=model_args.use_fast_tokenizer,
revision=model_args.model_revision,
token=model_args.token,
trust_remote_code=model_args.trust_remote_code,
)
model = AutoModelForSeq2SeqLM.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,
revision=model_args.model_revision,
token=model_args.token,
trust_remote_code=model_args.trust_remote_code,
)
# We resize the embeddings only when necessary to avoid index errors. If you are creating a model from scratch
# on a small vocab and want a smaller embedding size, remove this test.
embedding_size = model.get_input_embeddings().weight.shape[0]
if len(tokenizer) > embedding_size:
model.resize_token_embeddings(len(tokenizer))
if model.config.decoder_start_token_id is None and isinstance(tokenizer, (MBartTokenizer, MBartTokenizerFast)):
if isinstance(tokenizer, MBartTokenizer):
model.config.decoder_start_token_id = tokenizer.lang_code_to_id[data_args.lang]
else:
model.config.decoder_start_token_id = tokenizer.convert_tokens_to_ids(data_args.lang)
if model.config.decoder_start_token_id is None:
raise ValueError("Make sure that `config.decoder_start_token_id` is correctly defined")
if (
hasattr(model.config, "max_position_embeddings")
and model.config.max_position_embeddings < data_args.max_source_length
):
if model_args.resize_position_embeddings is None:
logger.warning(
"Increasing the model's number of position embedding vectors from"
f" {model.config.max_position_embeddings} to {data_args.max_source_length}."
)
model.resize_position_embeddings(data_args.max_source_length)
elif model_args.resize_position_embeddings:
model.resize_position_embeddings(data_args.max_source_length)
else:
raise ValueError(
f"`--max_source_length` is set to {data_args.max_source_length}, but the model only has"
f" {model.config.max_position_embeddings} position encodings. Consider either reducing"
f" `--max_source_length` to {model.config.max_position_embeddings} or to automatically resize the"
" model's position encodings by passing `--resize_position_embeddings`."
)
prefix = data_args.source_prefix if data_args.source_prefix is not None else ""
# Preprocessing the datasets.
# We need to tokenize inputs and targets.
if training_args.do_train:
if "train" not in raw_datasets:
raise ValueError("--do_train requires a train dataset")
column_names = raw_datasets["train"].column_names
elif training_args.do_eval:
if "validation" not in raw_datasets:
raise ValueError("--do_eval requires a validation dataset")
column_names = raw_datasets["validation"].column_names
elif training_args.do_predict:
if "test" not in raw_datasets:
raise ValueError("--do_predict requires a test dataset")
column_names = raw_datasets["test"].column_names
else:
logger.info("There is nothing to do. Please pass `do_train`, `do_eval` and/or `do_predict`.")
return
if isinstance(tokenizer, tuple(MULTILINGUAL_TOKENIZERS)):
assert data_args.lang is not None, (
f"{tokenizer.__class__.__name__} is a multilingual tokenizer which requires --lang argument"
)
tokenizer.src_lang = data_args.lang
tokenizer.tgt_lang = data_args.lang
# For multilingual translation models like mBART-50 and M2M100 we need to force the target language token
# as the first generated token. We ask the user to explicitly provide this as --forced_bos_token argument.
forced_bos_token_id = (
tokenizer.lang_code_to_id[data_args.forced_bos_token] if data_args.forced_bos_token is not None else None
)
model.config.forced_bos_token_id = forced_bos_token_id
# Get the column names for input/target.
dataset_columns = summarization_name_mapping.get(data_args.dataset_name, None)
if data_args.text_column is None:
text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
else:
text_column = data_args.text_column
if text_column not in column_names:
raise ValueError(
f"--text_column' value '{data_args.text_column}' needs to be one of: {', '.join(column_names)}"
)
if data_args.summary_column is None:
summary_column = dataset_columns[1] if dataset_columns is not None else column_names[1]
else:
summary_column = data_args.summary_column
if summary_column not in column_names:
raise ValueError(
f"--summary_column' value '{data_args.summary_column}' needs to be one of: {', '.join(column_names)}"
)
# Temporarily set max_target_length for training.
max_target_length = data_args.max_target_length
padding = "max_length" if data_args.pad_to_max_length else False
if training_args.label_smoothing_factor > 0 and not hasattr(model, "prepare_decoder_input_ids_from_labels"):
logger.warning(
"label_smoothing is enabled but the `prepare_decoder_input_ids_from_labels` method is not defined for "
f"`{model.__class__.__name__}`. This will lead to loss being calculated twice and will take up more memory"
)
def preprocess_function(examples):
# remove pairs where at least one record is None
inputs, targets = [], []
for i in range(len(examples[text_column])):
if examples[text_column][i] and examples[summary_column][i]:
inputs.append(examples[text_column][i])
targets.append(examples[summary_column][i])
inputs = [prefix + inp for inp in inputs]
model_inputs = tokenizer(inputs, max_length=data_args.max_source_length, padding=padding, truncation=True)
# Tokenize targets with the `text_target` keyword argument
labels = tokenizer(text_target=targets, max_length=max_target_length, padding=padding, truncation=True)
# If we are padding here, replace all tokenizer.pad_token_id in the labels by -100 when we want to ignore
# padding in the loss.
if padding == "max_length" and data_args.ignore_pad_token_for_loss:
labels["input_ids"] = [
[(l if l != tokenizer.pad_token_id else -100) for l in label] for label in labels["input_ids"]
]
model_inputs["labels"] = labels["input_ids"]
return model_inputs
if training_args.do_train:
train_dataset = raw_datasets["train"]
if data_args.max_train_samples is not None:
max_train_samples = min(len(train_dataset), data_args.max_train_samples)
train_dataset = train_dataset.select(range(max_train_samples))
with training_args.main_process_first(desc="train dataset map pre-processing"):
train_dataset = train_dataset.map(
preprocess_function,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
desc="Running tokenizer on train dataset",
)
if training_args.do_eval:
max_target_length = data_args.val_max_target_length
eval_dataset = raw_datasets["validation"]
if data_args.max_eval_samples is not None:
max_eval_samples = min(len(eval_dataset), data_args.max_eval_samples)
eval_dataset = eval_dataset.select(range(max_eval_samples))
with training_args.main_process_first(desc="validation dataset map pre-processing"):
eval_dataset = eval_dataset.map(
preprocess_function,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
desc="Running tokenizer on validation dataset",
)
if training_args.do_predict:
max_target_length = data_args.val_max_target_length
predict_dataset = raw_datasets["test"]
if data_args.max_predict_samples is not None:
max_predict_samples = min(len(predict_dataset), data_args.max_predict_samples)
predict_dataset = predict_dataset.select(range(max_predict_samples))
with training_args.main_process_first(desc="prediction dataset map pre-processing"):
predict_dataset = predict_dataset.map(
preprocess_function,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
desc="Running tokenizer on prediction dataset",
)
# Data collator
label_pad_token_id = -100 if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id
data_collator = DataCollatorForSeq2Seq(
tokenizer,
model=model,
label_pad_token_id=label_pad_token_id,
pad_to_multiple_of=8 if training_args.fp16 else None,
)
# Metric
metric = evaluate.load("rouge", cache_dir=model_args.cache_dir)
def postprocess_text(preds, labels):
preds = [pred.strip() for pred in preds]
labels = [label.strip() for label in labels]
# rougeLSum expects newline after each sentence
preds = ["\n".join(nltk.sent_tokenize(pred)) for pred in preds]
labels = ["\n".join(nltk.sent_tokenize(label)) for label in labels]
return preds, labels
def compute_metrics(eval_preds):
preds, labels = eval_preds
if isinstance(preds, tuple):
preds = preds[0]
# Replace -100s used for padding as we can't decode them
preds = np.where(preds != -100, preds, tokenizer.pad_token_id)
decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
# Some simple post-processing
decoded_preds, decoded_labels = postprocess_text(decoded_preds, decoded_labels)
result = metric.compute(predictions=decoded_preds, references=decoded_labels, use_stemmer=True)
result = {k: round(v * 100, 4) for k, v in result.items()}
prediction_lens = [np.count_nonzero(pred != tokenizer.pad_token_id) for pred in preds]
result["gen_len"] = np.mean(prediction_lens)
return result
# Override the decoding parameters of Seq2SeqTrainer
training_args.generation_max_length = (
training_args.generation_max_length
if training_args.generation_max_length is not None
else data_args.val_max_target_length
)
training_args.generation_num_beams = (
data_args.num_beams if data_args.num_beams is not None else training_args.generation_num_beams
)
# Initialize our Trainer
trainer = Seq2SeqTrainer(
model=model,
args=training_args,
train_dataset=train_dataset if training_args.do_train else None,
eval_dataset=eval_dataset if training_args.do_eval else None,
processing_class=tokenizer,
data_collator=data_collator,
compute_metrics=compute_metrics if training_args.predict_with_generate else None,
)
# Training
if training_args.do_train:
checkpoint = None
if training_args.resume_from_checkpoint is not None:
checkpoint = training_args.resume_from_checkpoint
elif last_checkpoint is not None:
checkpoint = last_checkpoint
train_result = trainer.train(resume_from_checkpoint=checkpoint)
trainer.save_model() # Saves the tokenizer too for easy upload
metrics = train_result.metrics
max_train_samples = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(train_dataset)
)
metrics["train_samples"] = min(max_train_samples, len(train_dataset))
trainer.log_metrics("train", metrics)
trainer.save_metrics("train", metrics)
trainer.save_state()
# Evaluation
results = {}
if training_args.do_eval:
logger.info("*** Evaluate ***")
if isinstance(eval_dataset, dict):
metrics = {}
for eval_ds_name, eval_ds in eval_dataset.items():
dataset_metrics = trainer.evaluate(eval_dataset=eval_ds, metric_key_prefix=f"eval_{eval_ds_name}")
metrics.update(dataset_metrics)
else:
metrics = trainer.evaluate(metric_key_prefix="eval")
max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))
trainer.log_metrics("eval", metrics)
trainer.save_metrics("eval", metrics)
if training_args.do_predict:
logger.info("*** Predict ***")
predict_results = trainer.predict(predict_dataset, metric_key_prefix="predict")
metrics = predict_results.metrics
max_predict_samples = (
data_args.max_predict_samples if data_args.max_predict_samples is not None else len(predict_dataset)
)
metrics["predict_samples"] = min(max_predict_samples, len(predict_dataset))
trainer.log_metrics("predict", metrics)
trainer.save_metrics("predict", metrics)
if trainer.is_world_process_zero():
if training_args.predict_with_generate:
predictions = predict_results.predictions
predictions = np.where(predictions != -100, predictions, tokenizer.pad_token_id)
predictions = tokenizer.batch_decode(
predictions, skip_special_tokens=True, clean_up_tokenization_spaces=True
)
predictions = [pred.strip() for pred in predictions]
output_prediction_file = os.path.join(training_args.output_dir, "generated_predictions.txt")
with open(output_prediction_file, "w") as writer:
writer.write("\n".join(predictions))
kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "summarization"}
if data_args.dataset_name is not None:
kwargs["dataset_tags"] = data_args.dataset_name
if data_args.dataset_config_name is not None:
kwargs["dataset_args"] = data_args.dataset_config_name
kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
else:
kwargs["dataset"] = data_args.dataset_name
if data_args.lang is not None:
kwargs["language"] = data_args.lang
if training_args.push_to_hub:
trainer.push_to_hub(**kwargs)
else:
trainer.create_model_card(**kwargs)
return results
def _mp_fn(index):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()

816
transformers/examples/pytorch/summarization/run_summarization_no_trainer.py Normal file
View File

@@ -0,0 +1,816 @@
#!/usr/bin/env python
# Copyright The HuggingFace Team and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# /// script
# dependencies = [
# "transformers @ git+https://github.com/huggingface/transformers.git",
# "accelerate >= 0.12.0",
# "datasets >= 1.8.0",
# "sentencepiece != 0.1.92",
# "protobuf",
# "rouge-score",
# "nltk",
# "py7zr",
# "torch >= 1.3",
# "evaluate",
# ]
# ///
"""
Fine-tuning a 🤗 Transformers model on summarization.
"""
# You can also adapt this script on your own summarization task. Pointers for this are left as comments.
import argparse
import json
import logging
import math
import os
import random
from pathlib import Path
import datasets
import evaluate
import nltk
import numpy as np
import torch
from accelerate import Accelerator
from accelerate.logging import get_logger
from accelerate.utils import set_seed
from datasets import load_dataset
from filelock import FileLock
from huggingface_hub import HfApi
from torch.utils.data import DataLoader
from tqdm.auto import tqdm
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_MAPPING,
AutoConfig,
AutoModelForSeq2SeqLM,
AutoTokenizer,
DataCollatorForSeq2Seq,
SchedulerType,
get_scheduler,
)
from transformers.utils import check_min_version, is_offline_mode, send_example_telemetry
from transformers.utils.versions import require_version
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.57.0.dev0")
logger = get_logger(__name__)
require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
# You should update this to your particular problem to have better documentation of `model_type`
MODEL_CONFIG_CLASSES = list(MODEL_MAPPING.keys())
MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
try:
nltk.data.find("tokenizers/punkt")
except (LookupError, OSError):
if is_offline_mode():
raise LookupError(
"Offline mode: run this script without TRANSFORMERS_OFFLINE first to download nltk data files"
)
with FileLock(".lock") as lock:
nltk.download("punkt", quiet=True)
summarization_name_mapping = {
"amazon_reviews_multi": ("review_body", "review_title"),
"big_patent": ("description", "abstract"),
"cnn_dailymail": ("article", "highlights"),
"orange_sum": ("text", "summary"),
"pn_summary": ("article", "summary"),
"psc": ("extract_text", "summary_text"),
"samsum": ("dialogue", "summary"),
"thaisum": ("body", "summary"),
"xglue": ("news_body", "news_title"),
"xsum": ("document", "summary"),
"wiki_summary": ("article", "highlights"),
}
def parse_args():
parser = argparse.ArgumentParser(description="Finetune a transformers model on a summarization task")
parser.add_argument(
"--dataset_name",
type=str,
default=None,
help="The name of the dataset to use (via the datasets library).",
)
parser.add_argument(
"--dataset_config_name",
type=str,
default=None,
help="The configuration name of the dataset to use (via the datasets library).",
)
parser.add_argument(
"--train_file", type=str, default=None, help="A csv or a json file containing the training data."
)
parser.add_argument(
"--validation_file", type=str, default=None, help="A csv or a json file containing the validation data."
)
parser.add_argument(
"--ignore_pad_token_for_loss",
type=bool,
default=True,
help="Whether to ignore the tokens corresponding to padded labels in the loss computation or not.",
)
parser.add_argument(
"--max_source_length",
type=int,
default=1024,
help=(
"The maximum total input sequence length after "
"tokenization.Sequences longer than this will be truncated, sequences shorter will be padded."
),
)
parser.add_argument(
"--source_prefix",
type=str,
default=None,
help="A prefix to add before every source text (useful for T5 models).",
)
parser.add_argument(
"--preprocessing_num_workers",
type=int,
default=None,
help="The number of processes to use for the preprocessing.",
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
parser.add_argument(
"--max_target_length",
type=int,
default=128,
help=(
"The maximum total sequence length for target text after "
"tokenization. Sequences longer than this will be truncated, sequences shorter will be padded. "
"during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--val_max_target_length",
type=int,
default=None,
help=(
"The maximum total sequence length for validation "
"target text after tokenization.Sequences longer than this will be truncated, sequences shorter will be "
"padded. Will default to `max_target_length`.This argument is also used to override the ``max_length`` "
"param of ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--num_beams",
type=int,
default=None,
help=(
"Number of beams to use for evaluation. This argument will be "
"passed to ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--pad_to_max_length",
action="store_true",
help="If passed, pad all samples to `max_length`. Otherwise, dynamic padding is used.",
)
parser.add_argument(
"--model_name_or_path",
type=str,
help="Path to pretrained model or model identifier from huggingface.co/models.",
required=False,
)
parser.add_argument(
"--config_name",
type=str,
default=None,
help="Pretrained config name or path if not the same as model_name",
)
parser.add_argument(
"--tokenizer_name",
type=str,
default=None,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--text_column",
type=str,
default=None,
help="The name of the column in the datasets containing the full texts (for summarization).",
)
parser.add_argument(
"--summary_column",
type=str,
default=None,
help="The name of the column in the datasets containing the summaries (for summarization).",
)
parser.add_argument(
"--use_slow_tokenizer",
action="store_true",
help="If passed, will use a slow tokenizer (not backed by the 🤗 Tokenizers library).",
)
parser.add_argument(
"--per_device_train_batch_size",
type=int,
default=8,
help="Batch size (per device) for the training dataloader.",
)
parser.add_argument(
"--per_device_eval_batch_size",
type=int,
default=8,
help="Batch size (per device) for the evaluation dataloader.",
)
parser.add_argument(
"--learning_rate",
type=float,
default=5e-5,
help="Initial learning rate (after the potential warmup period) to use.",
)
parser.add_argument("--weight_decay", type=float, default=0.0, help="Weight decay to use.")
parser.add_argument("--num_train_epochs", type=int, default=3, help="Total number of training epochs to perform.")
parser.add_argument(
"--max_train_steps",
type=int,
default=None,
help="Total number of training steps to perform. If provided, overrides num_train_epochs.",
)
parser.add_argument(
"--gradient_accumulation_steps",
type=int,
default=1,
help="Number of updates steps to accumulate before performing a backward/update pass.",
)
parser.add_argument(
"--lr_scheduler_type",
type=SchedulerType,
default="linear",
help="The scheduler type to use.",
choices=["linear", "cosine", "cosine_with_restarts", "polynomial", "constant", "constant_with_warmup"],
)
parser.add_argument(
"--num_warmup_steps", type=int, default=0, help="Number of steps for the warmup in the lr scheduler."
)
parser.add_argument("--output_dir", type=str, default=None, help="Where to store the final model.")
parser.add_argument("--seed", type=int, default=None, help="A seed for reproducible training.")
parser.add_argument(
"--model_type",
type=str,
default=None,
help="Model type to use if training from scratch.",
choices=MODEL_TYPES,
)
parser.add_argument("--push_to_hub", action="store_true", help="Whether or not to push the model to the Hub.")
parser.add_argument(
"--hub_model_id", type=str, help="The name of the repository to keep in sync with the local `output_dir`."
)
parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
parser.add_argument(
"--trust_remote_code",
action="store_true",
help=(
"Whether to trust the execution of code from datasets/models defined on the Hub."
" This option should only be set to `True` for repositories you trust and in which you have read the"
" code, as it will execute code present on the Hub on your local machine."
),
)
parser.add_argument(
"--checkpointing_steps",
type=str,
default=None,
help="Whether the various states should be saved at the end of every n steps, or 'epoch' for each epoch.",
)
parser.add_argument(
"--resume_from_checkpoint",
type=str,
default=None,
help="If the training should continue from a checkpoint folder.",
)
parser.add_argument(
"--with_tracking",
action="store_true",
help="Whether to enable experiment trackers for logging.",
)
parser.add_argument(
"--report_to",
type=str,
default="all",
help=(
'The integration to report the results and logs to. Supported platforms are `"tensorboard"`,'
' `"wandb"`, `"comet_ml"` and `"clearml"`. Use `"all"` (default) to report to all integrations. '
"Only applicable when `--with_tracking` is passed."
),
)
args = parser.parse_args()
# Sanity checks
if args.dataset_name is None and args.train_file is None and args.validation_file is None:
raise ValueError("Need either a dataset name or a training/validation file.")
else:
if args.train_file is not None:
extension = args.train_file.split(".")[-1]
assert extension in ["csv", "json"], "`train_file` should be a csv or a json file."
if args.validation_file is not None:
extension = args.validation_file.split(".")[-1]
assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
if args.push_to_hub:
assert args.output_dir is not None, "Need an `output_dir` to create a repo when `--push_to_hub` is passed."
return args
def main():
args = parse_args()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_summarization_no_trainer", args)
# Initialize the accelerator. We will let the accelerator handle device placement for us in this example.
# If we're using tracking, we also need to initialize it here and it will by default pick up all supported trackers
# in the environment
accelerator_log_kwargs = {}
if args.with_tracking:
accelerator_log_kwargs["log_with"] = args.report_to
accelerator_log_kwargs["project_dir"] = args.output_dir
accelerator = Accelerator(gradient_accumulation_steps=args.gradient_accumulation_steps, **accelerator_log_kwargs)
if args.source_prefix is None and args.model_name_or_path in [
"google-t5/t5-small",
"google-t5/t5-base",
"google-t5/t5-large",
"google-t5/t5-3b",
"google-t5/t5-11b",
]:
logger.warning(
"You're running a t5 model but didn't provide a source prefix, which is the expected, e.g. with "
"`--source_prefix 'summarize: ' `"
)
# Make one log on every process with the configuration for debugging.
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO,
)
logger.info(accelerator.state, main_process_only=False)
if accelerator.is_local_main_process:
datasets.utils.logging.set_verbosity_warning()
transformers.utils.logging.set_verbosity_info()
else:
datasets.utils.logging.set_verbosity_error()
transformers.utils.logging.set_verbosity_error()
# If passed along, set the training seed now.
if args.seed is not None:
set_seed(args.seed)
# Handle the repository creation
if accelerator.is_main_process:
if args.push_to_hub:
# Retrieve of infer repo_name
repo_name = args.hub_model_id
if repo_name is None:
repo_name = Path(args.output_dir).absolute().name
# Create repo and retrieve repo_id
api = HfApi()
repo_id = api.create_repo(repo_name, exist_ok=True, token=args.hub_token).repo_id
with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
if "step_*" not in gitignore:
gitignore.write("step_*\n")
if "epoch_*" not in gitignore:
gitignore.write("epoch_*\n")
elif args.output_dir is not None:
os.makedirs(args.output_dir, exist_ok=True)
accelerator.wait_for_everyone()
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
raw_datasets = load_dataset(
args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
)
else:
data_files = {}
if args.train_file is not None:
data_files["train"] = args.train_file
extension = args.train_file.split(".")[-1]
if args.validation_file is not None:
data_files["validation"] = args.validation_file
extension = args.validation_file.split(".")[-1]
raw_datasets = load_dataset(extension, data_files=data_files)
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.
# Load pretrained model and tokenizer
#
# In distributed training, the .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
if args.config_name:
config = AutoConfig.from_pretrained(args.config_name, trust_remote_code=args.trust_remote_code)
elif args.model_name_or_path:
config = AutoConfig.from_pretrained(args.model_name_or_path, trust_remote_code=args.trust_remote_code)
else:
config = CONFIG_MAPPING[args.model_type]()
logger.warning("You are instantiating a new config instance from scratch.")
if args.tokenizer_name:
tokenizer = AutoTokenizer.from_pretrained(
args.tokenizer_name, use_fast=not args.use_slow_tokenizer, trust_remote_code=args.trust_remote_code
)
elif args.model_name_or_path:
tokenizer = AutoTokenizer.from_pretrained(
args.model_name_or_path, use_fast=not args.use_slow_tokenizer, trust_remote_code=args.trust_remote_code
)
else:
raise ValueError(
"You are instantiating a new tokenizer from scratch. This is not supported by this script. "
"You can do it from another script, save it, and load it from here, using --tokenizer_name."
)
if args.model_name_or_path:
model = AutoModelForSeq2SeqLM.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
trust_remote_code=args.trust_remote_code,
)
else:
logger.info("Training new model from scratch")
model = AutoModelForSeq2SeqLM.from_config(config, trust_remote_code=args.trust_remote_code)
# We resize the embeddings only when necessary to avoid index errors. If you are creating a model from scratch
# on a small vocab and want a smaller embedding size, remove this test.
embedding_size = model.get_input_embeddings().weight.shape[0]
if len(tokenizer) > embedding_size:
model.resize_token_embeddings(len(tokenizer))
if model.config.decoder_start_token_id is None:
raise ValueError("Make sure that `config.decoder_start_token_id` is correctly defined")
prefix = args.source_prefix if args.source_prefix is not None else ""
# Preprocessing the datasets.
# First we tokenize all the texts.
column_names = raw_datasets["train"].column_names
# Get the column names for input/target.
dataset_columns = summarization_name_mapping.get(args.dataset_name, None)
if args.text_column is None:
text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
else:
text_column = args.text_column
if text_column not in column_names:
raise ValueError(
f"--text_column' value '{args.text_column}' needs to be one of: {', '.join(column_names)}"
)
if args.summary_column is None:
summary_column = dataset_columns[1] if dataset_columns is not None else column_names[1]
else:
summary_column = args.summary_column
if summary_column not in column_names:
raise ValueError(
f"--summary_column' value '{args.summary_column}' needs to be one of: {', '.join(column_names)}"
)
if args.val_max_target_length is None:
args.val_max_target_length = args.max_target_length
# Temporarily set max_target_length for training.
max_target_length = args.max_target_length
padding = "max_length" if args.pad_to_max_length else False
def preprocess_function(examples):
inputs = examples[text_column]
targets = examples[summary_column]
inputs = [prefix + inp for inp in inputs]
model_inputs = tokenizer(inputs, max_length=args.max_source_length, padding=padding, truncation=True)
# Tokenize targets with the `text_target` keyword argument
labels = tokenizer(text_target=targets, max_length=max_target_length, padding=padding, truncation=True)
# If we are padding here, replace all tokenizer.pad_token_id in the labels by -100 when we want to ignore
# padding in the loss.
if padding == "max_length" and args.ignore_pad_token_for_loss:
labels["input_ids"] = [
[(l if l != tokenizer.pad_token_id else -100) for l in label] for label in labels["input_ids"]
]
model_inputs["labels"] = labels["input_ids"]
return model_inputs
with accelerator.main_process_first():
train_dataset = raw_datasets["train"].map(
preprocess_function,
batched=True,
num_proc=args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not args.overwrite_cache,
desc="Running tokenizer on dataset",
)
# Temporarily set max_target_length for validation.
max_target_length = args.val_max_target_length
eval_dataset = raw_datasets["validation"].map(
preprocess_function,
batched=True,
num_proc=args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not args.overwrite_cache,
desc="Running tokenizer on dataset",
)
# Log a few random samples from the training set:
for index in random.sample(range(len(train_dataset)), 1):
logger.info(f"Sample {index} of the training set: {train_dataset[index]}.")
label_pad_token_id = -100 if args.ignore_pad_token_for_loss else tokenizer.pad_token_id
if accelerator.mixed_precision == "fp8":
pad_to_multiple_of = 16
elif accelerator.mixed_precision != "no":
pad_to_multiple_of = 8
else:
pad_to_multiple_of = None
data_collator = DataCollatorForSeq2Seq(
tokenizer,
model=model,
label_pad_token_id=label_pad_token_id,
pad_to_multiple_of=pad_to_multiple_of,
)
def postprocess_text(preds, labels):
preds = [pred.strip() for pred in preds]
labels = [label.strip() for label in labels]
# rougeLSum expects newline after each sentence
preds = ["\n".join(nltk.sent_tokenize(pred)) for pred in preds]
labels = ["\n".join(nltk.sent_tokenize(label)) for label in labels]
return preds, labels
train_dataloader = DataLoader(
train_dataset, shuffle=True, collate_fn=data_collator, batch_size=args.per_device_train_batch_size
)
eval_dataloader = DataLoader(eval_dataset, collate_fn=data_collator, batch_size=args.per_device_eval_batch_size)
# Optimizer
# Split weights in two groups, one with weight decay and the other not.
no_decay = ["bias", "LayerNorm.weight", "layer_norm.weight"]
optimizer_grouped_parameters = [
{
"params": [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)],
"weight_decay": args.weight_decay,
},
{
"params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)],
"weight_decay": 0.0,
},
]
optimizer = torch.optim.AdamW(optimizer_grouped_parameters, lr=args.learning_rate)
# Scheduler and math around the number of training steps.
overrode_max_train_steps = False
num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
if args.max_train_steps is None:
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
overrode_max_train_steps = True
lr_scheduler = get_scheduler(
name=args.lr_scheduler_type,
optimizer=optimizer,
num_warmup_steps=args.num_warmup_steps * accelerator.num_processes,
num_training_steps=args.max_train_steps
if overrode_max_train_steps
else args.max_train_steps * accelerator.num_processes,
)
# Prepare everything with our `accelerator`.
model, optimizer, train_dataloader, eval_dataloader, lr_scheduler = accelerator.prepare(
model, optimizer, train_dataloader, eval_dataloader, lr_scheduler
)
# We need to recalculate our total training steps as the size of the training dataloader may have changed.
num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
if overrode_max_train_steps:
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
# Afterwards we recalculate our number of training epochs
args.num_train_epochs = math.ceil(args.max_train_steps / num_update_steps_per_epoch)
# Figure out how many steps we should save the Accelerator states
checkpointing_steps = args.checkpointing_steps
if checkpointing_steps is not None and checkpointing_steps.isdigit():
checkpointing_steps = int(checkpointing_steps)
# We need to initialize the trackers we use, and also store our configuration.
# The trackers initializes automatically on the main process.
if args.with_tracking:
experiment_config = vars(args)
# TensorBoard cannot log Enums, need the raw value
experiment_config["lr_scheduler_type"] = experiment_config["lr_scheduler_type"].value
accelerator.init_trackers("summarization_no_trainer", experiment_config)
# Metric
metric = evaluate.load("rouge")
# Train!
total_batch_size = args.per_device_train_batch_size * accelerator.num_processes * args.gradient_accumulation_steps
logger.info("***** Running training *****")
logger.info(f" Num examples = {len(train_dataset)}")
logger.info(f" Num Epochs = {args.num_train_epochs}")
logger.info(f" Instantaneous batch size per device = {args.per_device_train_batch_size}")
logger.info(f" Total train batch size (w. parallel, distributed & accumulation) = {total_batch_size}")
logger.info(f" Gradient Accumulation steps = {args.gradient_accumulation_steps}")
logger.info(f" Total optimization steps = {args.max_train_steps}")
# Only show the progress bar once on each machine.
progress_bar = tqdm(range(args.max_train_steps), disable=not accelerator.is_local_main_process)
completed_steps = 0
starting_epoch = 0
# Potentially load in the weights and states from a previous save
if args.resume_from_checkpoint:
if args.resume_from_checkpoint is not None or args.resume_from_checkpoint != "":
checkpoint_path = args.resume_from_checkpoint
path = os.path.basename(args.resume_from_checkpoint)
else:
# Get the most recent checkpoint
dirs = [f.name for f in os.scandir(os.getcwd()) if f.is_dir()]
dirs.sort(key=os.path.getctime)
path = dirs[-1] # Sorts folders by date modified, most recent checkpoint is the last
checkpoint_path = path
path = os.path.basename(checkpoint_path)
accelerator.print(f"Resumed from checkpoint: {checkpoint_path}")
accelerator.load_state(checkpoint_path)
# Extract `epoch_{i}` or `step_{i}`
training_difference = os.path.splitext(path)[0]
if "epoch" in training_difference:
starting_epoch = int(training_difference.replace("epoch_", "")) + 1
resume_step = None
completed_steps = starting_epoch * num_update_steps_per_epoch
else:
# need to multiply `gradient_accumulation_steps` to reflect real steps
resume_step = int(training_difference.replace("step_", "")) * args.gradient_accumulation_steps
starting_epoch = resume_step // len(train_dataloader)
completed_steps = resume_step // args.gradient_accumulation_steps
resume_step -= starting_epoch * len(train_dataloader)
# update the progress_bar if load from checkpoint
progress_bar.update(completed_steps)
for epoch in range(starting_epoch, args.num_train_epochs):
model.train()
if args.with_tracking:
total_loss = 0
if args.resume_from_checkpoint and epoch == starting_epoch and resume_step is not None:
# We skip the first `n` batches in the dataloader when resuming from a checkpoint
active_dataloader = accelerator.skip_first_batches(train_dataloader, resume_step)
else:
active_dataloader = train_dataloader
for step, batch in enumerate(active_dataloader):
with accelerator.accumulate(model):
outputs = model(**batch)
loss = outputs.loss
# We keep track of the loss at each epoch
if args.with_tracking:
total_loss += loss.detach().float()
accelerator.backward(loss)
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
# Checks if the accelerator has performed an optimization step behind the scenes
if accelerator.sync_gradients:
progress_bar.update(1)
completed_steps += 1
if isinstance(checkpointing_steps, int):
if completed_steps % checkpointing_steps == 0 and accelerator.sync_gradients:
output_dir = f"step_{completed_steps}"
if args.output_dir is not None:
output_dir = os.path.join(args.output_dir, output_dir)
accelerator.save_state(output_dir)
if completed_steps >= args.max_train_steps:
break
model.eval()
gen_kwargs = {
"max_length": args.val_max_target_length,
"num_beams": args.num_beams,
}
for step, batch in enumerate(eval_dataloader):
with torch.no_grad():
generated_tokens = accelerator.unwrap_model(model).generate(
batch["input_ids"],
attention_mask=batch["attention_mask"],
**gen_kwargs,
)
generated_tokens = accelerator.pad_across_processes(
generated_tokens, dim=1, pad_index=tokenizer.pad_token_id
)
labels = batch["labels"]
if not args.pad_to_max_length:
# If we did not pad to max length, we need to pad the labels too
labels = accelerator.pad_across_processes(batch["labels"], dim=1, pad_index=tokenizer.pad_token_id)
generated_tokens, labels = accelerator.gather_for_metrics((generated_tokens, labels))
generated_tokens = generated_tokens.cpu().numpy()
labels = labels.cpu().numpy()
if args.ignore_pad_token_for_loss:
# Replace -100 in the labels as we can't decode them.
labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
if isinstance(generated_tokens, tuple):
generated_tokens = generated_tokens[0]
decoded_preds = tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
decoded_preds, decoded_labels = postprocess_text(decoded_preds, decoded_labels)
metric.add_batch(
predictions=decoded_preds,
references=decoded_labels,
)
result = metric.compute(use_stemmer=True)
result = {k: round(v * 100, 4) for k, v in result.items()}
logger.info(result)
if args.with_tracking:
result["train_loss"] = total_loss.item() / len(train_dataloader)
result["epoch"] = epoch
result["step"] = completed_steps
accelerator.log(result, step=completed_steps)
if args.push_to_hub and epoch < args.num_train_epochs - 1:
accelerator.wait_for_everyone()
unwrapped_model = accelerator.unwrap_model(model)
unwrapped_model.save_pretrained(
args.output_dir, is_main_process=accelerator.is_main_process, save_function=accelerator.save
)
if accelerator.is_main_process:
tokenizer.save_pretrained(args.output_dir)
api.upload_folder(
commit_message=f"Training in progress epoch {epoch}",
folder_path=args.output_dir,
repo_id=repo_id,
repo_type="model",
token=args.hub_token,
)
if args.checkpointing_steps == "epoch":
output_dir = f"epoch_{epoch}"
if args.output_dir is not None:
output_dir = os.path.join(args.output_dir, output_dir)
accelerator.save_state(output_dir)
if args.output_dir is not None:
accelerator.wait_for_everyone()
unwrapped_model = accelerator.unwrap_model(model)
unwrapped_model.save_pretrained(
args.output_dir, is_main_process=accelerator.is_main_process, save_function=accelerator.save
)
if accelerator.is_main_process:
tokenizer.save_pretrained(args.output_dir)
if args.push_to_hub:
api.upload_folder(
commit_message="End of training",
folder_path=args.output_dir,
repo_id=repo_id,
repo_type="model",
token=args.hub_token,
)
all_results = {f"eval_{k}": v for k, v in result.items()}
with open(os.path.join(args.output_dir, "all_results.json"), "w") as f:
json.dump(all_results, f)
if __name__ == "__main__":
main()