Ondrej Dusek tuetschek

## lm2heads.py
import torch
import transformers
import tqdm
import copy
import numpy as np
from logzero import logger

# some tiny data -- sentiment classification + LM
DATA = [
    [{'text': 'This is good . [CLS]',

## dataset_info.json
{
  "citation": "@inproceedings{dusek_neural_2019,\n        author = {Du\u0161ek, Ond\u0159ej and Jur\u010d\u00ed\u010dek, Filip},\n        title = {Neural {Generation} for {Czech}: {Data} and {Baselines}},\n        shorttitle = {Neural   {Generation} for {Czech}},\n        url = {https://www.aclweb.org/anthology/W19-8670/},\n        urldate = {2019-10-18},\n        booktitle = {Proceedings of the 12th {International} {Conference} on {Natural} {Language} {Generation}     ({INLG} 2019)},\n        month = oct,\n        address = {Tokyo, Japan},\n        year = {2019},\n        pages = {563--574},\n        abstract = {We present the first dataset targeted at end-to-end NLG in Czech in the restaurant        domain, along with several strong baseline models using the sequence-to-sequence approach. While non-English NLG is under-explored in general, Czech, as a morphologically rich language, makes the task even harder: Since Czech requires   inflecting named entities, delexicalization or copy mechanisms do not work
	import torch
	import transformers
	import tqdm
	import copy
	import numpy as np
	from logzero import logger

	# some tiny data -- sentiment classification + LM
	DATA = [
	[{'text': 'This is good . [CLS]',