initial wikitext commit

.gitattributes

@@ -35,3 +35,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.mp3 filter=lfs diff=lfs merge=lfs -text
 *.ogg filter=lfs diff=lfs merge=lfs -text
 *.wav filter=lfs diff=lfs merge=lfs -text
+wikitext103/ filter=lfs diff=lfs merge=lfs -text
+wikitext2 filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,317 @@
+---
+annotations_creators:
+- machine-generated
+language_creators:
+- machine-generated
+languages:
+- en
+licenses:
+- cc-by-sa-4.0
+multilinguality:
+- monolingual
+pretty_name: wikitext_linked
+size_categories:
+- 1M<n<10M
+source_datasets:
+- extended|wikitext
+task_categories:
+- fill-mask
+- token-classification
+- text-classification
+task_ids:
+- masked-language-modeling
+- named-entity-recognition
+- part-of-speech
+- lemmatization
+- parsing
+- entity-linking-classification
+---
+
+# Dataset Card for [Dataset Name]
+
+## Table of Contents
+- [Table of Contents](#table-of-contents)
+- [Dataset Description](#dataset-description)
+  - [Dataset Summary](#dataset-summary)
+  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
+  - [Languages](#languages)
+- [Dataset Structure](#dataset-structure)
+  - [Data Instances](#data-instances)
+  - [Data Fields](#data-fields)
+  - [Data Splits](#data-splits)
+- [Dataset Creation](#dataset-creation)
+  - [Curation Rationale](#curation-rationale)
+  - [Source Data](#source-data)
+  - [Annotations](#annotations)
+  - [Personal and Sensitive Information](#personal-and-sensitive-information)
+- [Considerations for Using the Data](#considerations-for-using-the-data)
+  - [Social Impact of Dataset](#social-impact-of-dataset)
+  - [Discussion of Biases](#discussion-of-biases)
+  - [Other Known Limitations](#other-known-limitations)
+- [Additional Information](#additional-information)
+  - [Dataset Curators](#dataset-curators)
+  - [Licensing Information](#licensing-information)
+  - [Citation Information](#citation-information)
+  - [Contributions](#contributions)
+
+## Dataset Description
+
+- **Homepage:** -
+- **Repository:** [https://github.com/GabrielKP/svo/](https://github.com/GabrielKP/svo/)
+- **Paper:** -
+- **Leaderboard:** -
+- **Point of Contact:** [[email protected]](mailto:[email protected])
+
+### Dataset Summary
+
+The WikiText language modeling dataset is a collection of over 100 million tokens extracted from
+the set of verified Good and Featured articles on Wikipedia. Dependency Relations, POS, NER tags
+are marked with [trankit](https://github.com/nlp-uoregon/trankit), entities are linked with
+[entity-fishing](https://nerd.readthedocs.io/en/latest/index.html), which also tags another field
+of NER tags. The dataset is available under the Creative Commons Attribution-ShareAlike License.
+
+Compared to the preprocessed version of Penn Treebank (PTB), WikiText-2 is over 2 times larger and
+WikiText-103 is over 110 times larger. The WikiText dataset also features a far larger vocabulary
+and retains the original case, punctuation and numbers - all of which are removed in PTB. As it is
+composed of full articles, the dataset is well suited for models that can take advantage of long
+term dependencies.
+
+### Supported Tasks and Leaderboards
+
+- masked-language-modeling
+- named-entity-recognition
+- part-of-speech
+- lemmatization
+- parsing
+- entity-linking-classification
+
+### Languages
+
+English.
+
+## Dataset Structure
+
+### Data Instances
+
+#### wikitext2
+
+- **Size of downloaded dataset files:** ?
+- **Size of the generated dataset:** ?
+- **Total amount of disk used:** 197.2 MB
+An example of 'validation' looks as follows.
+
+```
+{
+    'text': 'It is closely related to the American lobster , H. americanus .',
+    'original_id': 3,
+    'tok_span': [[0, 0], [0, 2], [3, 5], [6, 13], [14, 21], [22, 24], [25, 28], [29, 37], [38, 45], [46, 47], [48, 50], [51, 61], [62, 63]],
+    'tok_upos': ['root', 'PRON', 'AUX', 'ADV', 'ADJ', 'ADP', 'DET', 'ADJ', 'NOUN', 'PUNCT', 'PROPN', 'PROPN', 'PUNCT'],
+    'tok_xpos': ['root', 'PRP', 'VBZ', 'RB', 'JJ', 'IN', 'DT', 'JJ', 'NN', ',', 'NNP', 'NNP', '.'],
+    'tok_dephead': [0, 4, 4, 4, 0, 8, 8, 8, 4, 8, 8, 10, 4],
+    'tok_deprel': ['root', 'nsubj', 'cop', 'advmod', 'root', 'case', 'det', 'amod', 'obl', 'punct', 'appos', 'flat', 'punct'],
+    'tok_lemma': [None, 'it', 'be', 'closely', 'related', 'to', 'the', 'american', 'lobster', ',', 'H.', 'americanus', '.'],
+    'tok_ner': [None, 'O', 'O', 'O', 'O', 'O', 'O', 'S-MISC', 'O', 'O', 'O', 'O', 'O'],
+    'ent_span': [[29, 45]],
+    'ent_wikipedia_external_ref': ['377397'],
+    'ent_ner': [None],
+    'ent_domains': [['Enterprise']],
+}
+```
+
+#### wikitext103
+
+- **Size of downloaded dataset files:** ?
+- **Size of the generated dataset:** ?
+- **Total amount of disk used:** 7.82 GB
+An example of 'train' looks as follows.
+```
+{
+    'text': 'Vision for the PlayStation Portable .',
+    'original_id': 3,
+    'tok_span': [[0, 0], [0, 6], [7, 10], [11, 14], [15, 26], [27, 35], [36, 37]],
+    'tok_upos': ['root', 'NOUN', 'ADP', 'DET', 'PROPN', 'PROPN', 'PUNCT'],
+    'tok_xpos': ['root', 'NN', 'IN', 'DT', 'NNP', 'NNP', '.'],
+    'tok_dephead': [0, 0, 5, 5, 5, 1, 1],
+    'tok_deprel': ['root', 'root', 'case', 'det', 'compound', 'nmod', 'punct'],
+    'tok_lemma': [None, 'vision', 'for', 'the', 'PlayStation', 'Portable', '.'],
+    'tok_ner': [None, 'O', 'O', 'O', 'B-MISC', 'E-MISC', 'O'],
+    'ent_span': [[15, 35]],
+    'ent_wikipedia_external_ref': ['619009'],
+    'ent_ner': [None],
+    'ent_domains': [['Electronics', 'Computer_Science']]
+}
+```
+
+Use following code to print the examples nicely:
+```py
+def print_tokens_entities(example):
+    text = example['text']
+    print(
+        "Text:\n"
+        f"   {text}"
+        "\nOrig-Id: "
+        f"{example['original_id']}"
+        "\nTokens:"
+    )
+    iterator = enumerate(zip(
+        example["tok_span"],
+        example["tok_upos"],
+        example["tok_xpos"],
+        example["tok_ner"],
+        example["tok_dephead"],
+        example["tok_deprel"],
+        example["tok_lemma"],
+    ))
+    print(f"    Id  | {'token':12} | {'upos':8} | {'xpos':8} | {'ner':8}  | {'deph':4} | {'deprel':9} | {'lemma':12} | Id")
+    print("---------------------------------------------------------------------------------------------------")
+    for idx, (tok_span, upos, xpos, ner, dephead, deprel, lemma) in iterator:
+        print(f"    {idx:3} | {text[tok_span[0]:tok_span[1]]:12} | {upos:8} | {xpos:8} | {str(ner):8}  | {str(dephead):4} | {deprel:9} | {str(lemma):12} | {idx}")
+
+    iterator = list(enumerate(zip(
+        example.get("ent_span", []),
+        example.get("ent_wikipedia_external_ref", []),
+        example.get("ent_ner", []),
+        example.get("ent_domains", []),
+    )))
+    if len(iterator) > 0:
+        print("Entities")
+        print(f"    Id  | {'entity':21} | {'wiki_ref':7} | {'ner':7} | domains")
+        print("--------------------------------------------------------------------")
+        for idx, ((start, end), wiki_ref, ent_ner, ent_domains) in iterator:
+            print(f"    {idx:3} | {text[start:end]:21} | {str(wiki_ref):7} | {str(ent_ner):7} | {ent_domains}")
+```
+
+### Data Fields
+
+The data fields are the same among all splits.
+
+* text: string feature.
+* original_id: int feature. Mapping to index within original wikitext dataset.
+* tok_span: sequence of (int, int) tuples. Denotes token spans (start inclusive, end exclusive)
+within each sentence.
+**Note that each sentence includes an artificial root node to align dependency relations.**
+* tok_upos: string feature. [Universal Dependency POS tag](https://universaldependencies.org/)
+tags. Aligned with tok_span. Root node has tag "root".
+* tok_xpos: string geature. [XPOS POS tag](https://trankit.readthedocs.io/en/latest/overview.html#token-list).
+Aligned with tok_span. Root node has tag "root".
+* tok_dephead: int feature.
+[Universal Dependency Head Node](https://universaldependencies.org/introduction.html). Int refers
+to tokens in tok_span. Root node has head `0` (itself).
+* tok_deprel: [Universal Dependency Relation Description](https://universaldependencies.org/introduction.html).
+Refers to the relation between this token and head token. Aligned with tok_span. Root node has
+dependency relation "root" to itself.
+* tok_lemma: string feature. Lemma of token. Aligend with tok_span.
+* tok_ner: string feature. NER tag of token. Marked in BIOS schema (e.g. S-MISC, B-LOC, ...)
+Aligned with tok_span. Root node has NER tag `None`.
+* ent_span: sequence of (int, int) tuples. Denotes entities found by entity-fishing
+(start inclusive, end exclusive).
+* ent_wikipedia_external_ref: string feature. External Reference to wikipedia page. You can
+access the wikipedia page via the url `https://en.wikipedia.org/wiki?curid=<ent_wikipedia_external_ref>`.
+Aligend with ent_span. All entities either have this field, or the `ent_ner` field, but not both.
+An empty field is denoted by the string `None`. Aligned with ent_span.
+* ent_ner: string feature. Denotes NER tags. An empty field is denoted by the string `None`.
+Aligned with ent_span.
+"ent_domains": sequence of string. Denotes domains of entity. Can be empty sequence. Aligned with
+ent_span.
+
+### Data Splits
+
+|       name        | train |validation| test|
+|-------------------|------:|---------:|----:|
+|wikitext103        |4076530|      8607|10062|
+|wikitext2          |  82649|      8606|10062|
+
+## Dataset Creation
+
+### Curation Rationale
+
+[More Information Needed]
+
+### Source Data
+
+#### Initial Data Collection and Normalization
+
+[https://huggingface.co/datasets/wikitext](https://huggingface.co/datasets/wikitext)
+
+#### Who are the source language producers?
+
+[More Information Needed]
+
+### Annotations
+
+#### Annotation process
+
+1. Started with `wikitext2-raw-v1` and `wikitext103-raw-v1` from [wikitext](https://huggingface.co/datasets/wikitext)
+2. Ran datasets through Trankit. Marked all fields starting with `tok`.
+
+In this step, the texts have been split into sentences. To retain the original text sections
+you can accumulate over `original_id` (examples are in order).
+
+3. Ran datasets through entity-fishing. Marked all fields starting with `ent`.
+
+#### Who are the annotators?
+
+Machines powered by [DFKI](https://www.dfki.de/web).
+
+### Personal and Sensitive Information
+
+[More Information Needed]
+
+## Considerations for Using the Data
+
+### Social Impact of Dataset
+
+[More Information Needed]
+
+### Discussion of Biases
+
+[More Information Needed]
+
+### Other Known Limitations
+
+[More Information Needed]
+
+## Additional Information
+
+### Dataset Curators
+
+[More Information Needed]
+
+### Licensing Information
+
+Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
+
+### Citation Information
+
+Please cite the original creators of wikitext, and the great people
+developing trankit and entity-fishing.
+```
+@misc{merity2016pointer,
+      title={Pointer Sentinel Mixture Models},
+      author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},
+      year={2016},
+      eprint={1609.07843},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL}
+}
+
+@inproceedings{nguyen2021trankit,
+      title={Trankit: A Light-Weight Transformer-based Toolkit for Multilingual Natural Language Processing},
+      author={Nguyen, Minh Van and Lai, Viet Dac and Veyseh, Amir Pouran Ben and Nguyen, Thien Huu},
+      booktitle="Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations",
+      year={2021}
+}
+
+@misc{entity-fishing,
+    title = {entity-fishing},
+    howpublished = {\\url{https://github.com/kermitt2/entity-fishing}},
+    publisher = {GitHub},
+    year = {2016--2022},
+    archivePrefix = {swh},
+    eprint = {1:dir:cb0ba3379413db12b0018b7c3af8d0d2d864139c}
+}
+```
+
+### Contributions
+
+Thanks to [@GabrielKP](https://github.com/GabrielKP) for adding this dataset.

dataset_infos.json

@@ -0,0 +1 @@
+{"wikitext2": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. Dependency Relations, POS, NER tags are marked with trankit and\n entities are linked with entity-fishing.\n The dataset is available under the Creative Commons Attribution-ShareAlike License.\n", "citation": "@misc{merity2016pointer,\n      title={Pointer Sentinel Mixture Models},\n      author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n      year={2016},\n      eprint={1609.07843},\n      archivePrefix={arXiv},\n      primaryClass={cs.CL}\n}\n\n@inproceedings{nguyen2021trankit,\n      title={Trankit: A Light-Weight Transformer-based Toolkit for Multilingual Natural Language Processing},\n      author={Nguyen, Minh Van and Lai, Viet Dac and Veyseh, Amir Pouran Ben and Nguyen, Thien Huu},\n      booktitle=\"Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations\",\n      year={2021}\n}\n\n@misc{entity-fishing,\n    title = {entity-fishing},\n    howpublished = {\\url{https://github.com/kermitt2/entity-fishing}},\n    publisher = {GitHub},\n    year = {2016--2022},\n    archivePrefix = {swh},\n    eprint = {1:dir:cb0ba3379413db12b0018b7c3af8d0d2d864139c}\n}\n", "homepage": "", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "original_id": {"dtype": "int64", "id": null, "_type": "Value"}, "tok_span": {"feature": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_upos": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_xpos": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_dephead": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_deprel": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_lemma": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_ner": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_span": {"feature": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_wikipedia_external_ref": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_ner": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_domains": {"feature": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext_linked", "config_name": "wikitext2", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 168970698, "num_examples": 82649, "dataset_name": "wikitext_linked"}, "validation": {"name": "validation", "num_bytes": 17702375, "num_examples": 8606, "dataset_name": "wikitext_linked"}, "test": {"name": "test", "num_bytes": 20013719, "num_examples": 10062, "dataset_name": "wikitext_linked"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 206686792, "size_in_bytes": 206686792}, "wikitext103": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. Dependency Relations, POS, NER tags are marked with trankit and\n entities are linked with entity-fishing.\n The dataset is available under the Creative Commons Attribution-ShareAlike License.\n", "citation": "@misc{merity2016pointer,\n      title={Pointer Sentinel Mixture Models},\n      author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n      year={2016},\n      eprint={1609.07843},\n      archivePrefix={arXiv},\n      primaryClass={cs.CL}\n}\n\n@inproceedings{nguyen2021trankit,\n      title={Trankit: A Light-Weight Transformer-based Toolkit for Multilingual Natural Language Processing},\n      author={Nguyen, Minh Van and Lai, Viet Dac and Veyseh, Amir Pouran Ben and Nguyen, Thien Huu},\n      booktitle=\"Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations\",\n      year={2021}\n}\n\n@misc{entity-fishing,\n    title = {entity-fishing},\n    howpublished = {\\url{https://github.com/kermitt2/entity-fishing}},\n    publisher = {GitHub},\n    year = {2016--2022},\n    archivePrefix = {swh},\n    eprint = {1:dir:cb0ba3379413db12b0018b7c3af8d0d2d864139c}\n}\n", "homepage": "", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "original_id": {"dtype": "int64", "id": null, "_type": "Value"}, "tok_span": {"feature": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_upos": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_xpos": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_dephead": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_deprel": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_lemma": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "tok_ner": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_span": {"feature": {"feature": {"dtype": "int64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_wikipedia_external_ref": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_ner": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ent_domains": {"feature": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext_linked", "config_name": "wikitext103", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 8356828955, "num_examples": 4076530, "dataset_name": "wikitext_linked"}, "validation": {"name": "validation", "num_bytes": 17702461, "num_examples": 8607, "dataset_name": "wikitext_linked"}, "test": {"name": "test", "num_bytes": 20013609, "num_examples": 10062, "dataset_name": "wikitext_linked"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 8394545025, "size_in_bytes": 8394545025}}

wikitext103/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:791c364b0c84bfff37738a8a12c6c08374d354cc2dc19d8c2fb540bcd9690143
+size 3121651

wikitext103/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f73231cf4fbf2ce77a7762dc8662656baaecebbe06ef0a9eaf60ac0c5646d9c
+size 1313130334

wikitext103/validation.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08199a4d18e9ce663b52948d54f64f6e65207471a1d2c7ac917a1c26642c0b9e
+size 2755917

wikitext2/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4674d94eff80195b563cb9a2a1bc97aa25ef7a6b500bf2257036c14bdd0ae306
+size 3127331

wikitext2/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:722ecc0b545203de71e6891b3b8b92f1f1cec301b95f68380adef222a5da3777
+size 26645424

wikitext2/validation.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f48f826bfe84b977f6d983f37ee08fab0dce48c3ec187387836661b242e807ff
+size 2763767

wikitext_linked.py

@@ -0,0 +1,163 @@
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# 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.
+"""Datasets loading script for wikitext_linked"""
+
+import os
+
+import datasets
+import pyarrow as pa
+import pyarrow.parquet as pq
+
+
+logger = datasets.utils.logging.get_logger(__name__)
+
+
+_CITATION = """\
+@misc{merity2016pointer,
+      title={Pointer Sentinel Mixture Models},
+      author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},
+      year={2016},
+      eprint={1609.07843},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL}
+}
+
+@inproceedings{nguyen2021trankit,
+      title={Trankit: A Light-Weight Transformer-based Toolkit for Multilingual Natural Language Processing},
+      author={Nguyen, Minh Van and Lai, Viet Dac and Veyseh, Amir Pouran Ben and Nguyen, Thien Huu},
+      booktitle="Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations",
+      year={2021}
+}
+
+@misc{entity-fishing,
+    title = {entity-fishing},
+    howpublished = {\\url{https://github.com/kermitt2/entity-fishing}},
+    publisher = {GitHub},
+    year = {2016--2022},
+    archivePrefix = {swh},
+    eprint = {1:dir:cb0ba3379413db12b0018b7c3af8d0d2d864139c}
+}
+"""
+
+_DESCRIPTION = """\
+ The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified
+ Good and Featured articles on Wikipedia. Dependency Relations, POS, NER tags are marked with trankit and
+ entities are linked with entity-fishing.
+ The dataset is available under the Creative Commons Attribution-ShareAlike License.
+"""
+
+_HOMEPAGE = "https://github.com/GabrielKP/svo/"
+
+_LICENSE = "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)"
+
+
+FEATURES = datasets.Features(
+    {
+        "text": datasets.Value("string"),
+        "original_id": datasets.Value("int64"),
+        "tok_span": datasets.Sequence(feature=datasets.Sequence(feature=datasets.Value("int64"))),
+        "tok_upos": datasets.Sequence(feature=datasets.Value("string")),
+        "tok_xpos": datasets.Sequence(feature=datasets.Value("string")),
+        "tok_dephead": datasets.Sequence(feature=datasets.Value("int64")),
+        "tok_deprel": datasets.Sequence(feature=datasets.Value("string")),
+        "tok_lemma": datasets.Sequence(feature=datasets.Value("string")),
+        "tok_ner": datasets.Sequence(feature=datasets.Value("string")),
+        "ent_span": datasets.Sequence(feature=datasets.Sequence(feature=datasets.Value("int64"))),
+        "ent_wikipedia_external_ref": datasets.Sequence(feature=datasets.Value("string")),
+        "ent_ner": datasets.Sequence(feature=datasets.Value("string")),
+        "ent_domains": datasets.Sequence(
+            feature=datasets.Sequence(feature=datasets.Value("string"))
+        ),
+    }
+)
+
+
+class WikitextLinked(datasets.ArrowBasedBuilder):
+    """wikitext_linked is an annotated and linked version from wikitext. Wikitext is a
+    collection of over 100 million tokens extracted from the set of verified Good and
+    Featured articles on Wikipedia.
+    """
+
+    VERSION = datasets.Version("1.0.0")
+
+    BUILDER_CONFIGS = [
+        datasets.BuilderConfig(
+            name="wikitext2",
+            version=VERSION,
+            description="The small version",
+            data_dir="wikitext2",
+        ),
+        datasets.BuilderConfig(
+            name="wikitext103",
+            version=VERSION,
+            description="The big version",
+            data_dir="wikitext103",
+        ),
+    ]
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            license=_LICENSE,
+            features=FEATURES,
+            version=self.VERSION,
+            homepage=_HOMEPAGE,
+        )
+
+    def _split_generators(self, dl_manager):
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                # These kwargs will be passed to _generate_examples
+                gen_kwargs={
+                    "filepath": os.path.join(self.config.data_dir, "train.parquet"),
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.VALIDATION,
+                # These kwargs will be passed to _generate_examples
+                gen_kwargs={
+                    "filepath": os.path.join(self.config.data_dir, "validation.parquet"),
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.TEST,
+                # These kwargs will be passed to _generate_examples
+                gen_kwargs={
+                    "filepath": os.path.join(self.config.data_dir, "test.parquet"),
+                },
+            ),
+        ]
+
+    # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
+    def _generate_tables(self, filepath):
+        schema = pa.schema(FEATURES.type)
+        with open(filepath, "rb") as f:
+            parquet_file = pq.ParquetFile(f)
+            try:
+                for batch_idx, record_batch in enumerate(
+                    parquet_file.iter_batches(batch_size=10000, columns=None)
+                ):
+                    pa_table = pa.Table.from_batches([record_batch])
+                    pa_table = pa.Table.from_arrays(
+                        [pa_table[field.name] for field in schema], schema=schema
+                    )
+                    # Uncomment for debugging (will print the Arrow table size and elements)
+                    # logger.warning(f"pa_table: {pa_table} num rows: {pa_table.num_rows}")
+                    # logger.warning('\n'.join(str(pa_table.slice(i, 1).to_pydict()) for i in range(pa_table.num_rows)))
+                    yield f"{batch_idx}", pa_table
+            except ValueError as e:
+                logger.error(f"Failed to read file '{filepath}' with error {type(e)}: {e}")
+                raise