SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Reorganize examples (#9010)

Browse Source 
* Reorganize example folder

* Continue reorganization

* Change requirements for tests

* Final cleanup

* Finish regroup with tests all passing

* Copyright

* Requirements and readme

* Make a full link for the documentation

* Address review comments

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Add symlink

* Reorg again

* Apply suggestions from code review

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Adapt title

* Update to new strucutre

* Remove test

* Update READMEs

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>
This commit is contained in:
Sylvain Gugger
2020-12-11 10:07:02 -05:00
committed by GitHub
parent 86896de064
commit 783d7d2629
215 changed files with 4454 additions and 1193 deletions
Download Patch File Download Diff File Expand all files Collapse all files

198
examples/token-classification/README.md
View File

@@ -1,3 +1,19 @@
<!---
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.
-->
## Token classification
Fine-tuning the library models for token classification task such as Named Entity Recognition (NER) or Parts-of-speech
@@ -32,10 +48,16 @@ python run_ner.py \
--do_eval
```
**Note:** This script only works with models that have a fast tokenizer (backed by the 🤗 Tokenizers library) as it
uses special features of those tokenizers. You can check if your favorite model has a fast tokenizer in
[this table](https://huggingface.co/transformers/index.html#bigtable), if it doesn't you can still use the old version
of the script.
## Old version of the script
Based on the scripts [`run_ner_old.py`](https://github.com/huggingface/transformers/blob/master/examples/token-classification/run_ner_old.py) for Pytorch and
[`run_tf_ner.py`](https://github.com/huggingface/transformers/blob/master/examples/token-classification/run_tf_ner.py) for Tensorflow 2.
You can find the old version of the PyTorch script [here](https://github.com/huggingface/transformers/blob/master/examples/contrib/legacy/token-classification/run_ner_old.py).
### TensorFlow version
The following examples are covered in this section:
@@ -98,79 +120,6 @@ export SAVE_STEPS=750
export SEED=1
```
#### Run the Pytorch version
To start training, just run:
```bash
python3 run_ner_old.py --data_dir ./ \
--labels ./labels.txt \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--per_device_train_batch_size $BATCH_SIZE \
--save_steps $SAVE_STEPS \
--seed $SEED \
--do_train \
--do_eval \
--do_predict
```
If your GPU supports half-precision training, just add the `--fp16` flag. After training, the model will be both evaluated on development and test datasets.
#### JSON-based configuration file
Instead of passing all parameters via commandline arguments, the `run_ner_old.py` script also supports reading parameters from a json-based configuration file:
```json
{
"data_dir": ".",
"labels": "./labels.txt",
"model_name_or_path": "bert-base-multilingual-cased",
"output_dir": "germeval-model",
"max_seq_length": 128,
"num_train_epochs": 3,
"per_device_train_batch_size": 32,
"save_steps": 750,
"seed": 1,
"do_train": true,
"do_eval": true,
"do_predict": true
}
```
It must be saved with a `.json` extension and can be used by running `python3 run_ner_old.py config.json`.
#### Evaluation
Evaluation on development dataset outputs the following for our example:
```bash
10/04/2019 00:42:06 - INFO - __main__ - ***** Eval results *****
10/04/2019 00:42:06 - INFO - __main__ - f1 = 0.8623348017621146
10/04/2019 00:42:06 - INFO - __main__ - loss = 0.07183869666975543
10/04/2019 00:42:06 - INFO - __main__ - precision = 0.8467916366258111
10/04/2019 00:42:06 - INFO - __main__ - recall = 0.8784592370979806
```
On the test dataset the following results could be achieved:
```bash
10/04/2019 00:42:42 - INFO - __main__ - ***** Eval results *****
10/04/2019 00:42:42 - INFO - __main__ - f1 = 0.8614389652384803
10/04/2019 00:42:42 - INFO - __main__ - loss = 0.07064602487454782
10/04/2019 00:42:42 - INFO - __main__ - precision = 0.8604651162790697
10/04/2019 00:42:42 - INFO - __main__ - recall = 0.8624150210424085
```
#### Run PyTorch version using PyTorch-Lightning
Run `bash run_pl.sh` from the `ner` directory. This would also install `pytorch-lightning` and the `examples/requirements.txt`. It is a shell pipeline which would automatically download, pre-process the data and run the models in `germeval-model` directory. Logs are saved in `lightning_logs` directory.
Pass `--gpus` flag to change the number of GPUs. Default uses 1. At the end, the expected results are: `TEST RESULTS {'val_loss': tensor(0.0707), 'precision': 0.852427800698191, 'recall': 0.869537067011978, 'f1': 0.8608974358974358}`
#### Run the Tensorflow 2 version
To start training, just run:
@@ -235,102 +184,3 @@ On the test dataset the following results could be achieved:
micro avg 0.8722 0.8774 0.8748 13869
macro avg 0.8712 0.8774 0.8740 13869
```
### Emerging and Rare Entities task: WNUT’17 (English NER) dataset
Description of the WNUT’17 task from the [shared task website](http://noisy-text.github.io/2017/index.html):
> The WNUT’17 shared task focuses on identifying unusual, previously-unseen entities in the context of emerging discussions.
> Named entities form the basis of many modern approaches to other tasks (like event clustering and summarization), but recall on
> them is a real problem in noisy text - even among annotators. This drop tends to be due to novel entities and surface forms.
Six labels are available in the dataset. An overview can be found on this [page](http://noisy-text.github.io/2017/files/).
#### Data (Download and pre-processing steps)
The dataset can be downloaded from the [official GitHub](https://github.com/leondz/emerging_entities_17) repository.
The following commands show how to prepare the dataset for fine-tuning:
```bash
mkdir -p data_wnut_17
curl -L 'https://github.com/leondz/emerging_entities_17/raw/master/wnut17train.conll' | tr '\t' ' ' > data_wnut_17/train.txt.tmp
curl -L 'https://github.com/leondz/emerging_entities_17/raw/master/emerging.dev.conll' | tr '\t' ' ' > data_wnut_17/dev.txt.tmp
curl -L 'https://raw.githubusercontent.com/leondz/emerging_entities_17/master/emerging.test.annotated' | tr '\t' ' ' > data_wnut_17/test.txt.tmp
```
Let's define some variables that we need for further pre-processing steps:
```bash
export MAX_LENGTH=128
export BERT_MODEL=bert-large-cased
```
Here we use the English BERT large model for fine-tuning.
The `preprocess.py` scripts splits longer sentences into smaller ones (once the max. subtoken length is reached):
```bash
python3 scripts/preprocess.py data_wnut_17/train.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/train.txt
python3 scripts/preprocess.py data_wnut_17/dev.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/dev.txt
python3 scripts/preprocess.py data_wnut_17/test.txt.tmp $BERT_MODEL $MAX_LENGTH > data_wnut_17/test.txt
```
In the last pre-processing step, the `labels.txt` file needs to be generated. This file contains all available labels:
```bash
cat data_wnut_17/train.txt data_wnut_17/dev.txt data_wnut_17/test.txt | cut -d " " -f 2 | grep -v "^$"| sort | uniq > data_wnut_17/labels.txt
```
#### Run the Pytorch version
Fine-tuning with the PyTorch version can be started using the `run_ner_old.py` script. In this example we use a JSON-based configuration file.
This configuration file looks like:
```json
{
"data_dir": "./data_wnut_17",
"labels": "./data_wnut_17/labels.txt",
"model_name_or_path": "bert-large-cased",
"output_dir": "wnut-17-model-1",
"max_seq_length": 128,
"num_train_epochs": 3,
"per_device_train_batch_size": 32,
"save_steps": 425,
"seed": 1,
"do_train": true,
"do_eval": true,
"do_predict": true,
"fp16": false
}
```
If your GPU supports half-precision training, please set `fp16` to `true`.
Save this JSON-based configuration under `wnut_17.json`. The fine-tuning can be started with `python3 run_ner_old.py wnut_17.json`.
#### Evaluation
Evaluation on development dataset outputs the following:
```bash
05/29/2020 23:33:44 - INFO - __main__ - ***** Eval results *****
05/29/2020 23:33:44 - INFO - __main__ - eval_loss = 0.26505235286212275
05/29/2020 23:33:44 - INFO - __main__ - eval_precision = 0.7008264462809918
05/29/2020 23:33:44 - INFO - __main__ - eval_recall = 0.507177033492823
05/29/2020 23:33:44 - INFO - __main__ - eval_f1 = 0.5884802220680084
05/29/2020 23:33:44 - INFO - __main__ - epoch = 3.0
```
On the test dataset the following results could be achieved:
```bash
05/29/2020 23:33:44 - INFO - transformers.trainer - ***** Running Prediction *****
05/29/2020 23:34:02 - INFO - __main__ - eval_loss = 0.30948806500973547
05/29/2020 23:34:02 - INFO - __main__ - eval_precision = 0.5840108401084011
05/29/2020 23:34:02 - INFO - __main__ - eval_recall = 0.3994439295644115
05/29/2020 23:34:02 - INFO - __main__ - eval_f1 = 0.47440836543753434
```
WNUT’17 is a very difficult task. Current state-of-the-art results on this dataset can be found [here](http://nlpprogress.com/english/named_entity_recognition.html).

2
examples/token-classification/requirements.txt Normal file
View File

@@ -0,0 +1,2 @@
seqeval
datasets >= 1.1.3

14
examples/token-classification/run.sh
View File

@@ -1,3 +1,17 @@
# 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.
python3 run_ner.py \
--model_name_or_path bert-base-uncased \
--dataset_name conll2003 \

37
examples/token-classification/run_chunk.sh
View File

@@ -1,37 +0,0 @@
if ! [ -f ./dev.txt ]; then
echo "Downloading CONLL2003 dev dataset...."
curl -L -o ./dev.txt 'https://github.com/davidsbatista/NER-datasets/raw/master/CONLL2003/valid.txt'
fi
if ! [ -f ./test.txt ]; then
echo "Downloading CONLL2003 test dataset...."
curl -L -o ./test.txt 'https://github.com/davidsbatista/NER-datasets/raw/master/CONLL2003/test.txt'
fi
if ! [ -f ./train.txt ]; then
echo "Downloading CONLL2003 train dataset...."
curl -L -o ./train.txt 'https://github.com/davidsbatista/NER-datasets/raw/master/CONLL2003/train.txt'
fi
export MAX_LENGTH=200
export BERT_MODEL=bert-base-uncased
export OUTPUT_DIR=chunker-model
export BATCH_SIZE=32
export NUM_EPOCHS=3
export SAVE_STEPS=750
export SEED=1
python3 run_ner_old.py \
--task_type Chunk \
--data_dir . \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--per_gpu_train_batch_size $BATCH_SIZE \
--save_steps $SAVE_STEPS \
--seed $SEED \
--do_train \
--do_eval \
--do_predict

316
examples/token-classification/run_ner_old.py
View File

@@ -1,316 +0,0 @@
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. 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.
""" Fine-tuning the library models for named entity recognition on CoNLL-2003. """
import logging
import os
import sys
from dataclasses import dataclass, field
from importlib import import_module
from typing import Dict, List, Optional, Tuple
import numpy as np
from seqeval.metrics import accuracy_score, f1_score, precision_score, recall_score
from torch import nn
import transformers
from transformers import (
AutoConfig,
AutoModelForTokenClassification,
AutoTokenizer,
EvalPrediction,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import is_main_process
from utils_ner import Split, TokenClassificationDataset, TokenClassificationTask
logger = logging.getLogger(__name__)
@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"}
)
task_type: Optional[str] = field(
default="NER", metadata={"help": "Task type to fine tune in training (e.g. NER, POS, etc)"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
use_fast: bool = field(default=False, metadata={"help": "Set this flag to use fast tokenization."})
# If you want to tweak more attributes on your tokenizer, you should do it in a distinct script,
# or just modify its tokenizer_config.json.
cache_dir: Optional[str] = field(
default=None,
metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
data_dir: str = field(
metadata={"help": "The input data dir. Should contain the .txt files for a CoNLL-2003-formatted task."}
)
labels: Optional[str] = field(
default=None,
metadata={"help": "Path to a file containing all labels. If not specified, CoNLL-2003 labels are used."},
)
max_seq_length: int = field(
default=128,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
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))
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()
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."
)
module = import_module("tasks")
try:
token_classification_task_clazz = getattr(module, model_args.task_type)
token_classification_task: TokenClassificationTask = token_classification_task_clazz()
except AttributeError:
raise ValueError(
f"Task {model_args.task_type} needs to be defined as a TokenClassificationTask subclass in {module}. "
f"Available tasks classes are: {TokenClassificationTask.__subclasses__()}"
)
# 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,
)
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info("Training/evaluation parameters %s", training_args)
# Set seed
set_seed(training_args.seed)
# Prepare CONLL-2003 task
labels = token_classification_task.get_labels(data_args.labels)
label_map: Dict[int, str] = {i: label for i, label in enumerate(labels)}
num_labels = len(labels)
# 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,
num_labels=num_labels,
id2label=label_map,
label2id={label: i for i, label in enumerate(labels)},
cache_dir=model_args.cache_dir,
)
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,
)
model = AutoModelForTokenClassification.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,
)
# Get datasets
train_dataset = (
TokenClassificationDataset(
token_classification_task=token_classification_task,
data_dir=data_args.data_dir,
tokenizer=tokenizer,
labels=labels,
model_type=config.model_type,
max_seq_length=data_args.max_seq_length,
overwrite_cache=data_args.overwrite_cache,
mode=Split.train,
)
if training_args.do_train
else None
)
eval_dataset = (
TokenClassificationDataset(
token_classification_task=token_classification_task,
data_dir=data_args.data_dir,
tokenizer=tokenizer,
labels=labels,
model_type=config.model_type,
max_seq_length=data_args.max_seq_length,
overwrite_cache=data_args.overwrite_cache,
mode=Split.dev,
)
if training_args.do_eval
else None
)
def align_predictions(predictions: np.ndarray, label_ids: np.ndarray) -> Tuple[List[int], List[int]]:
preds = np.argmax(predictions, axis=2)
batch_size, seq_len = preds.shape
out_label_list = [[] for _ in range(batch_size)]
preds_list = [[] for _ in range(batch_size)]
for i in range(batch_size):
for j in range(seq_len):
if label_ids[i, j] != nn.CrossEntropyLoss().ignore_index:
out_label_list[i].append(label_map[label_ids[i][j]])
preds_list[i].append(label_map[preds[i][j]])
return preds_list, out_label_list
def compute_metrics(p: EvalPrediction) -> Dict:
preds_list, out_label_list = align_predictions(p.predictions, p.label_ids)
return {
"accuracy_score": accuracy_score(out_label_list, preds_list),
"precision": precision_score(out_label_list, preds_list),
"recall": recall_score(out_label_list, preds_list),
"f1": f1_score(out_label_list, preds_list),
}
# Initialize our Trainer
trainer = Trainer(
model=model,
args=training_args,
train_dataset=train_dataset,
eval_dataset=eval_dataset,
compute_metrics=compute_metrics,
)
# Training
if training_args.do_train:
trainer.train(
model_path=model_args.model_name_or_path if os.path.isdir(model_args.model_name_or_path) else None
)
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_process_zero():
tokenizer.save_pretrained(training_args.output_dir)
# Evaluation
results = {}
if training_args.do_eval:
logger.info("*** Evaluate ***")
result = trainer.evaluate()
output_eval_file = os.path.join(training_args.output_dir, "eval_results.txt")
if trainer.is_world_process_zero():
with open(output_eval_file, "w") as writer:
logger.info("***** Eval results *****")
for key, value in result.items():
logger.info(" %s = %s", key, value)
writer.write("%s = %s\n" % (key, value))
results.update(result)
# Predict
if training_args.do_predict:
test_dataset = TokenClassificationDataset(
token_classification_task=token_classification_task,
data_dir=data_args.data_dir,
tokenizer=tokenizer,
labels=labels,
model_type=config.model_type,
max_seq_length=data_args.max_seq_length,
overwrite_cache=data_args.overwrite_cache,
mode=Split.test,
)
predictions, label_ids, metrics = trainer.predict(test_dataset)
preds_list, _ = align_predictions(predictions, label_ids)
output_test_results_file = os.path.join(training_args.output_dir, "test_results.txt")
if trainer.is_world_process_zero():
with open(output_test_results_file, "w") as writer:
for key, value in metrics.items():
logger.info(" %s = %s", key, value)
writer.write("%s = %s\n" % (key, value))
# Save predictions
output_test_predictions_file = os.path.join(training_args.output_dir, "test_predictions.txt")
if trainer.is_world_process_zero():
with open(output_test_predictions_file, "w") as writer:
with open(os.path.join(data_args.data_dir, "test.txt"), "r") as f:
token_classification_task.write_predictions_to_file(writer, f, preds_list)
return results
def _mp_fn(index):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()

36
examples/token-classification/run_old.sh
View File

@@ -1,36 +0,0 @@
## The relevant files are currently on a shared Google
## drive at https://drive.google.com/drive/folders/1kC0I2UGl2ltrluI9NqDjaQJGw5iliw_J
## Monitor for changes and eventually migrate to nlp dataset
curl -L 'https://drive.google.com/uc?export=download&id=1Jjhbal535VVz2ap4v4r_rN1UEHTdLK5P' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > train.txt.tmp
curl -L 'https://drive.google.com/uc?export=download&id=1ZfRcQThdtAR5PPRjIDtrVP7BtXSCUBbm' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > dev.txt.tmp
curl -L 'https://drive.google.com/uc?export=download&id=1u9mb7kNJHWQCWyweMDRMuTFoOHOfeBTH' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > test.txt.tmp
export MAX_LENGTH=128
export BERT_MODEL=bert-base-multilingual-cased
python3 scripts/preprocess.py train.txt.tmp $BERT_MODEL $MAX_LENGTH > train.txt
python3 scripts/preprocess.py dev.txt.tmp $BERT_MODEL $MAX_LENGTH > dev.txt
python3 scripts/preprocess.py test.txt.tmp $BERT_MODEL $MAX_LENGTH > test.txt
cat train.txt dev.txt test.txt | cut -d " " -f 2 | grep -v "^$"| sort | uniq > labels.txt
export OUTPUT_DIR=germeval-model
export BATCH_SIZE=32
export NUM_EPOCHS=3
export SAVE_STEPS=750
export SEED=1
python3 run_ner_old.py \
--task_type NER \
--data_dir . \
--labels ./labels.txt \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--per_gpu_train_batch_size $BATCH_SIZE \
--save_steps $SAVE_STEPS \
--seed $SEED \
--do_train \
--do_eval \
--do_predict

44
examples/token-classification/run_pl.sh
View File

@@ -1,44 +0,0 @@
#!/usr/bin/env bash
# for seqeval metrics import
pip install -r ../requirements.txt
## The relevant files are currently on a shared Google
## drive at https://drive.google.com/drive/folders/1kC0I2UGl2ltrluI9NqDjaQJGw5iliw_J
## Monitor for changes and eventually migrate to nlp dataset
curl -L 'https://drive.google.com/uc?export=download&id=1Jjhbal535VVz2ap4v4r_rN1UEHTdLK5P' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > train.txt.tmp
curl -L 'https://drive.google.com/uc?export=download&id=1ZfRcQThdtAR5PPRjIDtrVP7BtXSCUBbm' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > dev.txt.tmp
curl -L 'https://drive.google.com/uc?export=download&id=1u9mb7kNJHWQCWyweMDRMuTFoOHOfeBTH' \
| grep -v "^#" | cut -f 2,3 | tr '\t' ' ' > test.txt.tmp
export MAX_LENGTH=128
export BERT_MODEL=bert-base-multilingual-cased
python3 scripts/preprocess.py train.txt.tmp $BERT_MODEL $MAX_LENGTH > train.txt
python3 scripts/preprocess.py dev.txt.tmp $BERT_MODEL $MAX_LENGTH > dev.txt
python3 scripts/preprocess.py test.txt.tmp $BERT_MODEL $MAX_LENGTH > test.txt
cat train.txt dev.txt test.txt | cut -d " " -f 2 | grep -v "^$"| sort | uniq > labels.txt
export BATCH_SIZE=32
export NUM_EPOCHS=3
export SEED=1
export OUTPUT_DIR_NAME=germeval-model
export CURRENT_DIR=${PWD}
export OUTPUT_DIR=${CURRENT_DIR}/${OUTPUT_DIR_NAME}
mkdir -p $OUTPUT_DIR
# Add parent directory to python path to access lightning_base.py
export PYTHONPATH="../":"${PYTHONPATH}"
python3 run_pl_ner.py --data_dir ./ \
--labels ./labels.txt \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--train_batch_size $BATCH_SIZE \
--seed $SEED \
--gpus 1 \
--do_train \
--do_predict

215
examples/token-classification/run_pl_ner.py
View File

@@ -1,215 +0,0 @@
import argparse
import glob
import logging
import os
from argparse import Namespace
from importlib import import_module
import numpy as np
import torch
from seqeval.metrics import accuracy_score, f1_score, precision_score, recall_score
from torch.nn import CrossEntropyLoss
from torch.utils.data import DataLoader, TensorDataset
from lightning_base import BaseTransformer, add_generic_args, generic_train
from utils_ner import TokenClassificationTask
logger = logging.getLogger(__name__)
class NERTransformer(BaseTransformer):
"""
A training module for NER. See BaseTransformer for the core options.
"""
mode = "token-classification"
def __init__(self, hparams):
if type(hparams) == dict:
hparams = Namespace(**hparams)
module = import_module("tasks")
try:
token_classification_task_clazz = getattr(module, hparams.task_type)
self.token_classification_task: TokenClassificationTask = token_classification_task_clazz()
except AttributeError:
raise ValueError(
f"Task {hparams.task_type} needs to be defined as a TokenClassificationTask subclass in {module}. "
f"Available tasks classes are: {TokenClassificationTask.__subclasses__()}"
)
self.labels = self.token_classification_task.get_labels(hparams.labels)
self.pad_token_label_id = CrossEntropyLoss().ignore_index
super().__init__(hparams, len(self.labels), self.mode)
def forward(self, **inputs):
return self.model(**inputs)
def training_step(self, batch, batch_num):
"Compute loss and log."
inputs = {"input_ids": batch[0], "attention_mask": batch[1], "labels": batch[3]}
if self.config.model_type != "distilbert":
inputs["token_type_ids"] = (
batch[2] if self.config.model_type in ["bert", "xlnet"] else None
) # XLM and RoBERTa don"t use token_type_ids
outputs = self(**inputs)
loss = outputs[0]
# tensorboard_logs = {"loss": loss, "rate": self.lr_scheduler.get_last_lr()[-1]}
return {"loss": loss}
def prepare_data(self):
"Called to initialize data. Use the call to construct features"
args = self.hparams
for mode in ["train", "dev", "test"]:
cached_features_file = self._feature_file(mode)
if os.path.exists(cached_features_file) and not args.overwrite_cache:
logger.info("Loading features from cached file %s", cached_features_file)
features = torch.load(cached_features_file)
else:
logger.info("Creating features from dataset file at %s", args.data_dir)
examples = self.token_classification_task.read_examples_from_file(args.data_dir, mode)
features = self.token_classification_task.convert_examples_to_features(
examples,
self.labels,
args.max_seq_length,
self.tokenizer,
cls_token_at_end=bool(self.config.model_type in ["xlnet"]),
cls_token=self.tokenizer.cls_token,
cls_token_segment_id=2 if self.config.model_type in ["xlnet"] else 0,
sep_token=self.tokenizer.sep_token,
sep_token_extra=False,
pad_on_left=bool(self.config.model_type in ["xlnet"]),
pad_token=self.tokenizer.pad_token_id,
pad_token_segment_id=self.tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
logger.info("Saving features into cached file %s", cached_features_file)
torch.save(features, cached_features_file)
def get_dataloader(self, mode: int, batch_size: int, shuffle: bool = False) -> DataLoader:
"Load datasets. Called after prepare data."
cached_features_file = self._feature_file(mode)
logger.info("Loading features from cached file %s", cached_features_file)
features = torch.load(cached_features_file)
all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long)
all_attention_mask = torch.tensor([f.attention_mask for f in features], dtype=torch.long)
if features[0].token_type_ids is not None:
all_token_type_ids = torch.tensor([f.token_type_ids for f in features], dtype=torch.long)
else:
all_token_type_ids = torch.tensor([0 for f in features], dtype=torch.long)
# HACK(we will not use this anymore soon)
all_label_ids = torch.tensor([f.label_ids for f in features], dtype=torch.long)
return DataLoader(
TensorDataset(all_input_ids, all_attention_mask, all_token_type_ids, all_label_ids), batch_size=batch_size
)
def validation_step(self, batch, batch_nb):
"""Compute validation""" ""
inputs = {"input_ids": batch[0], "attention_mask": batch[1], "labels": batch[3]}
if self.config.model_type != "distilbert":
inputs["token_type_ids"] = (
batch[2] if self.config.model_type in ["bert", "xlnet"] else None
) # XLM and RoBERTa don"t use token_type_ids
outputs = self(**inputs)
tmp_eval_loss, logits = outputs[:2]
preds = logits.detach().cpu().numpy()
out_label_ids = inputs["labels"].detach().cpu().numpy()
return {"val_loss": tmp_eval_loss.detach().cpu(), "pred": preds, "target": out_label_ids}
def _eval_end(self, outputs):
"Evaluation called for both Val and Test"
val_loss_mean = torch.stack([x["val_loss"] for x in outputs]).mean()
preds = np.concatenate([x["pred"] for x in outputs], axis=0)
preds = np.argmax(preds, axis=2)
out_label_ids = np.concatenate([x["target"] for x in outputs], axis=0)
label_map = {i: label for i, label in enumerate(self.labels)}
out_label_list = [[] for _ in range(out_label_ids.shape[0])]
preds_list = [[] for _ in range(out_label_ids.shape[0])]
for i in range(out_label_ids.shape[0]):
for j in range(out_label_ids.shape[1]):
if out_label_ids[i, j] != self.pad_token_label_id:
out_label_list[i].append(label_map[out_label_ids[i][j]])
preds_list[i].append(label_map[preds[i][j]])
results = {
"val_loss": val_loss_mean,
"accuracy_score": accuracy_score(out_label_list, preds_list),
"precision": precision_score(out_label_list, preds_list),
"recall": recall_score(out_label_list, preds_list),
"f1": f1_score(out_label_list, preds_list),
}
ret = {k: v for k, v in results.items()}
ret["log"] = results
return ret, preds_list, out_label_list
def validation_epoch_end(self, outputs):
# when stable
ret, preds, targets = self._eval_end(outputs)
logs = ret["log"]
return {"val_loss": logs["val_loss"], "log": logs, "progress_bar": logs}
def test_epoch_end(self, outputs):
# updating to test_epoch_end instead of deprecated test_end
ret, predictions, targets = self._eval_end(outputs)
# Converting to the dict required by pl
# https://github.com/PyTorchLightning/pytorch-lightning/blob/master/\
# pytorch_lightning/trainer/logging.py#L139
logs = ret["log"]
# `val_loss` is the key returned by `self._eval_end()` but actually refers to `test_loss`
return {"avg_test_loss": logs["val_loss"], "log": logs, "progress_bar": logs}
@staticmethod
def add_model_specific_args(parser, root_dir):
# Add NER specific options
BaseTransformer.add_model_specific_args(parser, root_dir)
parser.add_argument(
"--task_type", default="NER", type=str, help="Task type to fine tune in training (e.g. NER, POS, etc)"
)
parser.add_argument(
"--max_seq_length",
default=128,
type=int,
help="The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded.",
)
parser.add_argument(
"--labels",
default="",
type=str,
help="Path to a file containing all labels. If not specified, CoNLL-2003 labels are used.",
)
parser.add_argument(
"--gpus",
default=0,
type=int,
help="The number of GPUs allocated for this, it is by default 0 meaning none",
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
return parser
if __name__ == "__main__":
parser = argparse.ArgumentParser()
add_generic_args(parser, os.getcwd())
parser = NERTransformer.add_model_specific_args(parser, os.getcwd())
args = parser.parse_args()
model = NERTransformer(args)
trainer = generic_train(model, args)
if args.do_predict:
# See https://github.com/huggingface/transformers/issues/3159
# pl use this default format to create a checkpoint:
# https://github.com/PyTorchLightning/pytorch-lightning/blob/master\
# /pytorch_lightning/callbacks/model_checkpoint.py#L322
checkpoints = list(sorted(glob.glob(os.path.join(args.output_dir, "checkpoint-epoch=*.ckpt"), recursive=True)))
model = model.load_from_checkpoint(checkpoints[-1])
trainer.test(model)

37
examples/token-classification/run_pos.sh
View File

@@ -1,37 +0,0 @@
if ! [ -f ./dev.txt ]; then
echo "Download dev dataset...."
curl -L -o ./dev.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-dev.conllu'
fi
if ! [ -f ./test.txt ]; then
echo "Download test dataset...."
curl -L -o ./test.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-test.conllu'
fi
if ! [ -f ./train.txt ]; then
echo "Download train dataset...."
curl -L -o ./train.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-train.conllu'
fi
export MAX_LENGTH=200
export BERT_MODEL=bert-base-uncased
export OUTPUT_DIR=postagger-model
export BATCH_SIZE=32
export NUM_EPOCHS=3
export SAVE_STEPS=750
export SEED=1
python3 run_ner_old.py \
--task_type POS \
--data_dir . \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--per_gpu_train_batch_size $BATCH_SIZE \
--save_steps $SAVE_STEPS \
--seed $SEED \
--do_train \
--do_eval \
--do_predict

39
examples/token-classification/run_pos_pl.sh
View File

@@ -1,39 +0,0 @@
#!/usr/bin/env bash
if ! [ -f ./dev.txt ]; then
echo "Download dev dataset...."
curl -L -o ./dev.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-dev.conllu'
fi
if ! [ -f ./test.txt ]; then
echo "Download test dataset...."
curl -L -o ./test.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-test.conllu'
fi
if ! [ -f ./train.txt ]; then
echo "Download train dataset...."
curl -L -o ./train.txt 'https://github.com/UniversalDependencies/UD_English-EWT/raw/master/en_ewt-ud-train.conllu'
fi
export MAX_LENGTH=200
export BERT_MODEL=bert-base-uncased
export OUTPUT_DIR=postagger-model
export BATCH_SIZE=32
export NUM_EPOCHS=3
export SAVE_STEPS=750
export SEED=1
# Add parent directory to python path to access lightning_base.py
export PYTHONPATH="../":"${PYTHONPATH}"
python3 run_pl_ner.py --data_dir ./ \
--task_type POS \
--model_name_or_path $BERT_MODEL \
--output_dir $OUTPUT_DIR \
--max_seq_length $MAX_LENGTH \
--num_train_epochs $NUM_EPOCHS \
--train_batch_size $BATCH_SIZE \
--seed $SEED \
--gpus 1 \
--do_train \
--do_predict

41
examples/token-classification/scripts/preprocess.py
View File

@@ -1,41 +0,0 @@
import sys
from transformers import AutoTokenizer
dataset = sys.argv[1]
model_name_or_path = sys.argv[2]
max_len = int(sys.argv[3])
subword_len_counter = 0
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
max_len -= tokenizer.num_special_tokens_to_add()
with open(dataset, "rt") as f_p:
for line in f_p:
line = line.rstrip()
if not line:
print(line)
subword_len_counter = 0
continue
token = line.split()[0]
current_subwords_len = len(tokenizer.tokenize(token))
# Token contains strange control characters like \x96 or \x95
# Just filter out the complete line
if current_subwords_len == 0:
continue
if (subword_len_counter + current_subwords_len) > max_len:
print("")
print(line)
subword_len_counter = current_subwords_len
continue
subword_len_counter += current_subwords_len
print(line)

163
examples/token-classification/tasks.py
View File

@@ -1,163 +0,0 @@
import logging
import os
from typing import List, TextIO, Union
from conllu import parse_incr
from utils_ner import InputExample, Split, TokenClassificationTask
logger = logging.getLogger(__name__)
class NER(TokenClassificationTask):
def __init__(self, label_idx=-1):
# in NER datasets, the last column is usually reserved for NER label
self.label_idx = label_idx
def read_examples_from_file(self, data_dir, mode: Union[Split, str]) -> List[InputExample]:
if isinstance(mode, Split):
mode = mode.value
file_path = os.path.join(data_dir, f"{mode}.txt")
guid_index = 1
examples = []
with open(file_path, encoding="utf-8") as f:
words = []
labels = []
for line in f:
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
if words:
examples.append(InputExample(guid=f"{mode}-{guid_index}", words=words, labels=labels))
guid_index += 1
words = []
labels = []
else:
splits = line.split(" ")
words.append(splits[0])
if len(splits) > 1:
labels.append(splits[self.label_idx].replace("\n", ""))
else:
# Examples could have no label for mode = "test"
labels.append("O")
if words:
examples.append(InputExample(guid=f"{mode}-{guid_index}", words=words, labels=labels))
return examples
def write_predictions_to_file(self, writer: TextIO, test_input_reader: TextIO, preds_list: List):
example_id = 0
for line in test_input_reader:
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
writer.write(line)
if not preds_list[example_id]:
example_id += 1
elif preds_list[example_id]:
output_line = line.split()[0] + " " + preds_list[example_id].pop(0) + "\n"
writer.write(output_line)
else:
logger.warning("Maximum sequence length exceeded: No prediction for '%s'.", line.split()[0])
def get_labels(self, path: str) -> List[str]:
if path:
with open(path, "r") as f:
labels = f.read().splitlines()
if "O" not in labels:
labels = ["O"] + labels
return labels
else:
return ["O", "B-MISC", "I-MISC", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"]
class Chunk(NER):
def __init__(self):
# in CONLL2003 dataset chunk column is second-to-last
super().__init__(label_idx=-2)
def get_labels(self, path: str) -> List[str]:
if path:
with open(path, "r") as f:
labels = f.read().splitlines()
if "O" not in labels:
labels = ["O"] + labels
return labels
else:
return [
"O",
"B-ADVP",
"B-INTJ",
"B-LST",
"B-PRT",
"B-NP",
"B-SBAR",
"B-VP",
"B-ADJP",
"B-CONJP",
"B-PP",
"I-ADVP",
"I-INTJ",
"I-LST",
"I-PRT",
"I-NP",
"I-SBAR",
"I-VP",
"I-ADJP",
"I-CONJP",
"I-PP",
]
class POS(TokenClassificationTask):
def read_examples_from_file(self, data_dir, mode: Union[Split, str]) -> List[InputExample]:
if isinstance(mode, Split):
mode = mode.value
file_path = os.path.join(data_dir, f"{mode}.txt")
guid_index = 1
examples = []
with open(file_path, encoding="utf-8") as f:
for sentence in parse_incr(f):
words = []
labels = []
for token in sentence:
words.append(token["form"])
labels.append(token["upos"])
assert len(words) == len(labels)
if words:
examples.append(InputExample(guid=f"{mode}-{guid_index}", words=words, labels=labels))
guid_index += 1
return examples
def write_predictions_to_file(self, writer: TextIO, test_input_reader: TextIO, preds_list: List):
example_id = 0
for sentence in parse_incr(test_input_reader):
s_p = preds_list[example_id]
out = ""
for token in sentence:
out += f'{token["form"]} ({token["upos"]}|{s_p.pop(0)}) '
out += "\n"
writer.write(out)
example_id += 1
def get_labels(self, path: str) -> List[str]:
if path:
with open(path, "r") as f:
return f.read().splitlines()
else:
return [
"ADJ",
"ADP",
"ADV",
"AUX",
"CCONJ",
"DET",
"INTJ",
"NOUN",
"NUM",
"PART",
"PRON",
"PROPN",
"PUNCT",
"SCONJ",
"SYM",
"VERB",
"X",
]

57
examples/token-classification/test_ner_examples.py
View File

@@ -1,57 +0,0 @@
import logging
import sys
import unittest
from unittest.mock import patch
import run_ner_old as run_ner
from transformers.testing_utils import require_torch_non_multi_gpu_but_fix_me, slow
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger()
class ExamplesTests(unittest.TestCase):
@slow
@require_torch_non_multi_gpu_but_fix_me
def test_run_ner(self):
stream_handler = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
testargs = """
--model_name distilbert-base-german-cased
--output_dir ./tests/fixtures/tests_samples/temp_dir
--overwrite_output_dir
--data_dir ./tests/fixtures/tests_samples/GermEval
--labels ./tests/fixtures/tests_samples/GermEval/labels.txt
--max_seq_length 128
--num_train_epochs 6
--logging_steps 1
--do_train
--do_eval
""".split()
with patch.object(sys, "argv", ["run.py"] + testargs):
result = run_ner.main()
self.assertLess(result["eval_loss"], 1.5)
@require_torch_non_multi_gpu_but_fix_me
def test_run_ner_pl(self):
stream_handler = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
testargs = """
--model_name distilbert-base-german-cased
--output_dir ./tests/fixtures/tests_samples/temp_dir
--overwrite_output_dir
--data_dir ./tests/fixtures/tests_samples/GermEval
--labels ./tests/fixtures/tests_samples/GermEval/labels.txt
--max_seq_length 128
--num_train_epochs 6
--logging_steps 1
--do_train
--do_eval
""".split()
with patch.object(sys, "argv", ["run.py"] + testargs):
result = run_ner.main()
self.assertLess(result["eval_loss"], 1.5)

372
examples/token-classification/utils_ner.py
View File

@@ -1,372 +0,0 @@
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. 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.
""" Named entity recognition fine-tuning: utilities to work with CoNLL-2003 task. """
import logging
import os
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional, Union
from filelock import FileLock
from transformers import PreTrainedTokenizer, is_tf_available, is_torch_available
logger = logging.getLogger(__name__)
@dataclass
class InputExample:
"""
A single training/test example for token classification.
Args:
guid: Unique id for the example.
words: list. The words of the sequence.
labels: (Optional) list. The labels for each word of the sequence. This should be
specified for train and dev examples, but not for test examples.
"""
guid: str
words: List[str]
labels: Optional[List[str]]
@dataclass
class InputFeatures:
"""
A single set of features of data.
Property names are the same names as the corresponding inputs to a model.
"""
input_ids: List[int]
attention_mask: List[int]
token_type_ids: Optional[List[int]] = None
label_ids: Optional[List[int]] = None
class Split(Enum):
train = "train"
dev = "dev"
test = "test"
class TokenClassificationTask:
@staticmethod
def read_examples_from_file(data_dir, mode: Union[Split, str]) -> List[InputExample]:
raise NotImplementedError
@staticmethod
def get_labels(path: str) -> List[str]:
raise NotImplementedError
@staticmethod
def convert_examples_to_features(
examples: List[InputExample],
label_list: List[str],
max_seq_length: int,
tokenizer: PreTrainedTokenizer,
cls_token_at_end=False,
cls_token="[CLS]",
cls_token_segment_id=1,
sep_token="[SEP]",
sep_token_extra=False,
pad_on_left=False,
pad_token=0,
pad_token_segment_id=0,
pad_token_label_id=-100,
sequence_a_segment_id=0,
mask_padding_with_zero=True,
) -> List[InputFeatures]:
"""Loads a data file into a list of `InputFeatures`
`cls_token_at_end` define the location of the CLS token:
- False (Default, BERT/XLM pattern): [CLS] + A + [SEP] + B + [SEP]
- True (XLNet/GPT pattern): A + [SEP] + B + [SEP] + [CLS]
`cls_token_segment_id` define the segment id associated to the CLS token (0 for BERT, 2 for XLNet)
"""
# TODO clean up all this to leverage built-in features of tokenizers
label_map = {label: i for i, label in enumerate(label_list)}
features = []
for (ex_index, example) in enumerate(examples):
if ex_index % 10_000 == 0:
logger.info("Writing example %d of %d", ex_index, len(examples))
tokens = []
label_ids = []
for word, label in zip(example.words, example.labels):
word_tokens = tokenizer.tokenize(word)
# bert-base-multilingual-cased sometimes output "nothing ([]) when calling tokenize with just a space.
if len(word_tokens) > 0:
tokens.extend(word_tokens)
# Use the real label id for the first token of the word, and padding ids for the remaining tokens
label_ids.extend([label_map[label]] + [pad_token_label_id] * (len(word_tokens) - 1))
# Account for [CLS] and [SEP] with "- 2" and with "- 3" for RoBERTa.
special_tokens_count = tokenizer.num_special_tokens_to_add()
if len(tokens) > max_seq_length - special_tokens_count:
tokens = tokens[: (max_seq_length - special_tokens_count)]
label_ids = label_ids[: (max_seq_length - special_tokens_count)]
# The convention in BERT is:
# (a) For sequence pairs:
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
# (b) For single sequences:
# tokens: [CLS] the dog is hairy . [SEP]
# type_ids: 0 0 0 0 0 0 0
#
# Where "type_ids" are used to indicate whether this is the first
# sequence or the second sequence. The embedding vectors for `type=0` and
# `type=1` were learned during pre-training and are added to the wordpiece
# embedding vector (and position vector). This is not *strictly* necessary
# since the [SEP] token unambiguously separates the sequences, but it makes
# it easier for the model to learn the concept of sequences.
#
# For classification tasks, the first vector (corresponding to [CLS]) is
# used as as the "sentence vector". Note that this only makes sense because
# the entire model is fine-tuned.
tokens += [sep_token]
label_ids += [pad_token_label_id]
if sep_token_extra:
# roberta uses an extra separator b/w pairs of sentences
tokens += [sep_token]
label_ids += [pad_token_label_id]
segment_ids = [sequence_a_segment_id] * len(tokens)
if cls_token_at_end:
tokens += [cls_token]
label_ids += [pad_token_label_id]
segment_ids += [cls_token_segment_id]
else:
tokens = [cls_token] + tokens
label_ids = [pad_token_label_id] + label_ids
segment_ids = [cls_token_segment_id] + segment_ids
input_ids = tokenizer.convert_tokens_to_ids(tokens)
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
input_mask = [1 if mask_padding_with_zero else 0] * len(input_ids)
# Zero-pad up to the sequence length.
padding_length = max_seq_length - len(input_ids)
if pad_on_left:
input_ids = ([pad_token] * padding_length) + input_ids
input_mask = ([0 if mask_padding_with_zero else 1] * padding_length) + input_mask
segment_ids = ([pad_token_segment_id] * padding_length) + segment_ids
label_ids = ([pad_token_label_id] * padding_length) + label_ids
else:
input_ids += [pad_token] * padding_length
input_mask += [0 if mask_padding_with_zero else 1] * padding_length
segment_ids += [pad_token_segment_id] * padding_length
label_ids += [pad_token_label_id] * padding_length
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
assert len(label_ids) == max_seq_length
if ex_index < 5:
logger.info("*** Example ***")
logger.info("guid: %s", example.guid)
logger.info("tokens: %s", " ".join([str(x) for x in tokens]))
logger.info("input_ids: %s", " ".join([str(x) for x in input_ids]))
logger.info("input_mask: %s", " ".join([str(x) for x in input_mask]))
logger.info("segment_ids: %s", " ".join([str(x) for x in segment_ids]))
logger.info("label_ids: %s", " ".join([str(x) for x in label_ids]))
if "token_type_ids" not in tokenizer.model_input_names:
segment_ids = None
features.append(
InputFeatures(
input_ids=input_ids, attention_mask=input_mask, token_type_ids=segment_ids, label_ids=label_ids
)
)
return features
if is_torch_available():
import torch
from torch import nn
from torch.utils.data.dataset import Dataset
class TokenClassificationDataset(Dataset):
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = nn.CrossEntropyLoss().ignore_index
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
# Load data features from cache or dataset file
cached_features_file = os.path.join(
data_dir,
"cached_{}_{}_{}".format(mode.value, tokenizer.__class__.__name__, str(max_seq_length)),
)
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
lock_path = cached_features_file + ".lock"
with FileLock(lock_path):
if os.path.exists(cached_features_file) and not overwrite_cache:
logger.info(f"Loading features from cached file {cached_features_file}")
self.features = torch.load(cached_features_file)
else:
logger.info(f"Creating features from dataset file at {data_dir}")
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
logger.info(f"Saving features into cached file {cached_features_file}")
torch.save(self.features, cached_features_file)
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
if is_tf_available():
import tensorflow as tf
class TFTokenClassificationDataset:
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = -100
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
def gen():
for ex in self.features:
if ex.token_type_ids is None:
yield (
{"input_ids": ex.input_ids, "attention_mask": ex.attention_mask},
ex.label_ids,
)
else:
yield (
{
"input_ids": ex.input_ids,
"attention_mask": ex.attention_mask,
"token_type_ids": ex.token_type_ids,
},
ex.label_ids,
)
if "token_type_ids" not in tokenizer.model_input_names:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32}, tf.int64),
(
{"input_ids": tf.TensorShape([None]), "attention_mask": tf.TensorShape([None])},
tf.TensorShape([None]),
),
)
else:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32, "token_type_ids": tf.int32}, tf.int64),
(
{
"input_ids": tf.TensorShape([None]),
"attention_mask": tf.TensorShape([None]),
"token_type_ids": tf.TensorShape([None]),
},
tf.TensorShape([None]),
),
)
def get_dataset(self):
self.dataset = self.dataset.apply(tf.data.experimental.assert_cardinality(len(self.features)))
return self.dataset
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]