polish-roberta: A Python repository from Ermlab Software

Polish RoBERTa

This repository contains pre-trained RoBERTa models for Polish as well as evaluation code for several Polish linguistic tasks. The released models were trained using Fairseq toolkit in the National Information Processing Institute, Warsaw, Poland. We provide two models based on BERT base and BERT large architectures. Two versions of each model are available: one for Fairseq and one for Huggingface Transformers.

Model	L / H / A*	Batch size	Update steps	Corpus size	Final perplexity**	Fairseq	Transformers
RoBERTa (base)	12 / 768 / 12	8k	125k	~20GB	3.66	v0.9.0	v2.9
RoBERTa (large)	24 / 1024 / 16	30k	50k	~135GB	2.92	v0.9.0	v2.9

* L - the number of encoder blocks, H - hidden size, A - the number of attention heads
** Perplexity of the best checkpoint, computed on the validation split

More details are available in the paper Pre-training Polish Transformer-based Language Models at Scale.

@misc{dadas2020pretraining,
    title={Pre-training Polish Transformer-based Language Models at Scale},
    author={Sławomir Dadas and Michał Perełkiewicz and Rafał Poświata},
    year={2020},
    eprint={2006.04229},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

Getting started

How to use with Fairseq

import os
from fairseq.models.roberta import RobertaModel, RobertaHubInterface
from fairseq import hub_utils

model_path = "roberta_large_fairseq"
loaded = hub_utils.from_pretrained(
    model_name_or_path=model_path,
    data_name_or_path=model_path,
    bpe="sentencepiece",
    sentencepiece_vocab=os.path.join(model_path, "sentencepiece.bpe.model"),
    load_checkpoint_heads=True,
    archive_map=RobertaModel.hub_models(),
    cpu=True
)
roberta = RobertaHubInterface(loaded['args'], loaded['task'], loaded['models'][0])
roberta.eval()
input = roberta.encode("Zażółcić gęślą jaźń.")
output = roberta.extract_features(input)
print(output[0][1])

How to use with HuggingFace Transformers

import torch
from tokenizers import SentencePieceBPETokenizer
from tokenizers.processors import RobertaProcessing
from transformers import RobertaModel, AutoModel

model_dir = "roberta_large_transformers"
tokenizer = SentencePieceBPETokenizer(f"{model_dir}/vocab.json", f"{model_dir}/merges.txt")
getattr(tokenizer, "_tokenizer").post_processor = RobertaProcessing(sep=("</s>", 2), cls=("<s>", 0))
model: RobertaModel = AutoModel.from_pretrained(model_dir)

input = tokenizer.encode("Zażółcić gęślą jaźń.")
output = model(torch.tensor([input.ids]))[0]
print(output[0][1])

Evaluation

To replicate our experiments, first download the required datasets using download_data.py script:

python download_data.py

Next, run run_tasks.py script to prepare the data, fine-tune and evaluate the model. We used the following parameters for each task:

python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-NKJP --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-CDS-E --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-CDS-R --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 1 --tasks KLEJ-CBD --fp16 True --max-sentences 8 --update-freq 4 --resample 0:0.75,1:3
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-POLEMO-IN --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-POLEMO-OUT --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-DYK --fp16 True --max-sentences 8 --update-freq 4 --resample 0:1,1:3
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-PSC --fp16 True --max-sentences 8 --update-freq 4 --resample 0:1,1:3
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks KLEJ-ECR --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks 8TAGS --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks SICK-E --fp16 True --max-sentences 8 --update-freq 2
python run_tasks.py --arch roberta_base --model_dir roberta_base_fairseq --train-epochs 10 --tasks SICK-R --fp16 True --max-sentences 8 --update-freq 2

Evaluation results on KLEJ Benchmark

Below we show the evaluation results of our models on the tasks included in KLEJ Benchmark. We fine-tuned both models 5 times for each task. Detailed scores for each run and averaged scores are presented in Table 1 and Table 2.

Run	NKJP	CDSC‑E	CDSC‑R	CBD	PolEmo‑IN	PolEmo‑OUT	DYK	PSC	AR	Avg
1	93.15	93.30	94.26	66.67	91.97	78.74	66.86	98.63	87.75	85.70
2	93.93	94.20	93.94	68.16	91.83	75.91	65.93	98.77	87.93	85.62
3	94.22	94.20	94.04	69.23	90.17	76.92	65.69	99.24	87.76	85.72
4	93.97	94.70	93.98	63.81	90.44	76.32	65.18	99.39	87.58	85.04
5	93.63	94.00	93.96	65.95	90.58	74.09	65.92	98.48	87.08	84.85
Avg	93.78	94.08	94.04	66.77	91.00	76.40	65.92	98.90	87.62	85.39

Table 1. KLEJ results for RoBERTa base model

Run	NKJP	CDSC‑E	CDSC‑R	CBD	PolEmo‑IN	PolEmo‑OUT	DYK	PSC	AR	Avg
1	94.31	93.50	94.63	72.39	92.80	80.54	71.87	98.63	88.82	87.50
2	95.14	93.90	94.93	69.82	92.80	82.59	73.39	98.94	88.96	87.83
3	95.24	93.30	94.61	71.59	91.41	82.19	75.35	98.64	89.31	87.96
4	94.46	93.20	94.96	71.08	92.80	82.39	70.59	99.09	88.60	87.46
5	94.46	93.00	94.82	69.83	92.11	83.00	74.85	98.79	88.65	87.72
Avg	94.72	93.38	94.79	70.94	92.38	82.14	73.21	98.82	88.87	87.69

Table 2. KLEJ results for RoBERTa large model

Evaluation results on other tasks

Task	Task type	Metric	Base model	Large model
SICK-E	Textual entailment	Accuracy	86.13	87.67
SICK-R	Semantic relatedness	Spearman correlation	82.26	85.63
Poleval 2018 - NER	Named entity recognition	F1 score (exact match)	87.94	89.98
8TAGS	Multi class classification	Accuracy	77.22	80.84

Ermlab/polish-roberta