Fine-tuning: Token Classification with W-NUT Emerging Entities

Issue

I’d like to run the sample code for Token Classification with W-NUT Emerging Entities on Google Colaboratory, but I cannot run it both CPU and GPU environment.

How can I check default values of Trainer for each pre-trained model?

Errors

I didn’t set Target on my code.

Where can I fix it and what number is appropriate in this case?

CPU

Target 12 is out of bounds.

GPU
If there are any solutions to fix below, I also hope to hear your experiences.

RuntimeError: CUDA error: device-side assert triggered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

Code

The code is almost same as the original one, but I’m trying to run on the note, custom_datasets.ipynb which can be opened from web browsers.

# Transformers installation
! pip install transformers datasets
! wget http://noisy-text.github.io/2017/files/wnut17train.conll

from pathlib import Path
import re

def read_wnut(file_path):
    file_path = Path(file_path)

    raw_text = file_path.read_text().strip()
    raw_docs = re.split(r'\n\t?\n', raw_text)
    token_docs = []
    tag_docs = []
    for doc in raw_docs:
        tokens = []
        tags = []
        for line in doc.split('\n'):
            token, tag = line.split('\t')
            tokens.append(token)
            tags.append(tag)
        token_docs.append(tokens)
        tag_docs.append(tags)

    return token_docs, tag_docs

texts, tags = read_wnut('wnut17train.conll')
from sklearn.model_selection import train_test_split
train_texts, val_texts, train_tags, val_tags = train_test_split(texts, tags, test_size=.2)
unique_tags = set(tag for doc in tags for tag in doc)
tag2id = {tag: id for id, tag in enumerate(unique_tags)}
id2tag = {id: tag for tag, id in tag2id.items()}

from transformers import DistilBertTokenizerFast
tokenizer = DistilBertTokenizerFast.from_pretrained('distilbert-base-cased')
train_encodings = tokenizer(train_texts, is_split_into_words=True, return_offsets_mapping=True, padding=True, truncation=True)
val_encodings = tokenizer(val_texts, is_split_into_words=True, return_offsets_mapping=True, padding=True, truncation=True)
import numpy as np

def encode_tags(tags, encodings):
    labels = [[tag2id[tag] for tag in doc] for doc in tags]
    encoded_labels = []
    for doc_labels, doc_offset in zip(labels, encodings.offset_mapping):
        # create an empty array of -100
        doc_enc_labels = np.ones(len(doc_offset),dtype=int) * -100
        arr_offset = np.array(doc_offset)

        # set labels whose first offset position is 0 and the second is not 0
        doc_enc_labels[(arr_offset[:,0] == 0) & (arr_offset[:,1] != 0)] = doc_labels
        encoded_labels.append(doc_enc_labels.tolist())

    return encoded_labels

train_labels = encode_tags(train_tags, train_encodings)
val_labels = encode_tags(val_tags, val_encodings)
## PYTORCH CODE
import torch

class WNUTDataset(torch.utils.data.Dataset):
    def __init__(self, encodings, labels):
        self.encodings = encodings
        self.labels = labels

    def __getitem__(self, idx):
        item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}
        item['labels'] = torch.tensor(self.labels[idx])
        return item

    def __len__(self):
        return len(self.labels)

train_encodings.pop("offset_mapping") # we don't want to pass this to the model
val_encodings.pop("offset_mapping")
train_dataset = WNUTDataset(train_encodings, train_labels)
val_dataset = WNUTDataset(val_encodings, val_labels)

## PYTORCH CODE
from transformers import DistilBertForTokenClassification
model = DistilBertForTokenClassification.from_pretrained('distilbert-base-cased', num_labels=len(unique_tags))
## PYTORCH CODE
from transformers import DistilBertForTokenClassification, Trainer, TrainingArguments, 

training_args = TrainingArguments(
    output_dir='./results',          # output directory
    num_train_epochs=3,              # total number of training epochs
    per_device_train_batch_size=16,  # batch size per device during training
    per_device_eval_batch_size=64,   # batch size for evaluation
    warmup_steps=500,                # number of warmup steps for learning rate scheduler
    weight_decay=0.01,               # strength of weight decay
    logging_dir='./logs',            # directory for storing logs
    logging_steps=10,
)

model = DistilBertForTokenClassification.from_pretrained("distilbert-base-uncased")

trainer = Trainer(
    model=model,                         # the instantiated 🤗 Transformers model to be trained
    args=training_args,                  # training arguments, defined above
    train_dataset=train_dataset,         # training dataset
    eval_dataset=val_dataset             # evaluation dataset
)

trainer.train()

What I changed

On Trainer, I adjust the function name for Token Classification with W-NUT Emerging Entities rather than the sample code on Hugging Face’s Fine-tuning with Trainer page.

DistilBertForSequenceClassification → DistilBertForTokenClassification

Did you get any solution for this? I am also facing same issue

I see you’re instantiating the model twice, one time providing num_labels, one time not. Note that any model in HuggingFace Transformers will have num_labels = 2 if not specified.

In any case, you should initialize the model only once, as follows:

from transformers import DistilBertForTokenClassification

model = DistilBertForTokenClassification.from_pretrained('distilbert-base-cased', num_labels=len(unique_tags))

Different question about the same tutorial - but im following tensoflow version.
I managed to follow and train the model, first using example data then with my own sentences and tags.

But I cant figure out how to run model.predict on arbitrary string - how to format the input ?

You can check out this thread regarding performing inference with NER models: Decoding the predicted output array in distilbertbase uncased model for NER - #2 by nielsr