initial metric card explorer commit

README.md

@@ -1,13 +1,12 @@
 ---
-title: Metric Explorer
-emoji: 🚀
-colorFrom: red
-colorTo: blue
-sdk: streamlit
-sdk_version: 1.2.0
+title: Accuracy_metric
+emoji: 📈
+colorFrom: green
+colorTo: red
+sdk: gradio
+sdk_version: 2.8.13
 app_file: app.py
 pinned: false
-license: cc-by-sa-4.0
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference

app.py

@@ -0,0 +1,69 @@
+from bleu import Bleu
+from rouge import Rouge
+from datasets import load_metric
+from pathlib import Path
+import streamlit as st
+import streamlit.components.v1 as components
+#from .nmt_bleu import compute_bleu  # From: https://github.com/tensorflow/nmt/blob/master/nmt/scripts/bleu.py
+
+rouge = Rouge()
+bleu = Bleu()
+
+def read_markdown_file(markdown_file):
+    return Path(markdown_file).read_text()
+
+
+def compute(data):
+    return metric.compute(predictions=data["predictions"], references=data["references"])["accuracy"]
+
+with st.sidebar.expander("Metric", expanded=True):
+    metrics= ['rouge','bleu']
+    metric_name = st.selectbox(
+        f"Choose metric to explore:",
+        metrics)
+
+metric = load_metric(metric_name)
+st.markdown("# You chose " + metric_name.upper())
+
+st.markdown("## You can test it out below:")
+
+reference = st.text_input('Input a reference sentence here:')
+prediction = st.text_input('Input a prediction sentence here:')
+
+predictions = []
+predictions.append(prediction.split())
+#print(predictions)
+references = []
+references.append(reference.split())
+#print(references)
+
+if metric_name == "bleu":
+    score = metric.compute(predictions=predictions, references=[references])
+    col1, col2, col3 = st.columns(3)
+    col1.metric("BLEU", score['bleu'])
+    col2.metric("Brevity penalty", score['brevity_penalty'])
+    col3.metric('Length Ratio', score['length_ratio'])
+
+
+if metric_name == "rouge":
+    score = metric.compute(predictions=predictions, references=references)
+    #print(score)
+    col1, col2, col3 = st.columns(3)
+    col1.metric("Rouge 1 Precision", score['rouge1'].mid.precision)
+    col2.metric("Rouge 1 Recall", score['rouge1'].mid.recall)
+    col3.metric("Rouge 1 FMeasure", score['rouge1'].mid.fmeasure)
+
+    col4, col5, col6 = st.columns(3)
+    col4.metric("Rouge 2 Precision", score['rouge2'].mid.precision)
+    col5.metric("Rouge 2 Recall", score['rouge2'].mid.recall)
+    col6.metric("Rouge 2 FMeasure", score['rouge2'].mid.fmeasure)
+
+
+#    col1.metric("BLEU", score['bleu'])
+#    col2.metric("Brevity penalty", score['brevity_penalty'])
+#    col3.metric('Length Ratio', score['length_ratio'])
+
+st.markdown('===================================================================================')
+#components.html("""<hr style="height:10px;border:none;color:#333;background-color:#333;" /> """)
+
+st.markdown(read_markdown_file(metric_name+"_metric_card.md"))

bleu.py

@@ -0,0 +1,111 @@
+
+""" BLEU metric. """
+
+import datasets
+
+_CITATION = """\
+@INPROCEEDINGS{Papineni02bleu:a,
+    author = {Kishore Papineni and Salim Roukos and Todd Ward and Wei-jing Zhu},
+    title = {BLEU: a Method for Automatic Evaluation of Machine Translation},
+    booktitle = {},
+    year = {2002},
+    pages = {311--318}
+}
+@inproceedings{lin-och-2004-orange,
+    title = "{ORANGE}: a Method for Evaluating Automatic Evaluation Metrics for Machine Translation",
+    author = "Lin, Chin-Yew  and
+      Och, Franz Josef",
+    booktitle = "{COLING} 2004: Proceedings of the 20th International Conference on Computational Linguistics",
+    month = "aug 23{--}aug 27",
+    year = "2004",
+    address = "Geneva, Switzerland",
+    publisher = "COLING",
+    url = "https://www.aclweb.org/anthology/C04-1072",
+    pages = "501--507",
+}
+"""
+
+_DESCRIPTION = """\
+BLEU (bilingual evaluation understudy) is an algorithm for evaluating the quality of text which has been machine-translated from one natural language to another.
+Quality is considered to be the correspondence between a machine's output and that of a human: "the closer a machine translation is to a professional human translation,
+the better it is" – this is the central idea behind BLEU. BLEU was one of the first metrics to claim a high correlation with human judgements of quality, and
+remains one of the most popular automated and inexpensive metrics.
+
+Scores are calculated for individual translated segments—generally sentences—by comparing them with a set of good quality reference translations.
+Those scores are then averaged over the whole corpus to reach an estimate of the translation's overall quality. Intelligibility or grammatical correctness
+are not taken into account[citation needed].
+
+BLEU's output is always a number between 0 and 1. This value indicates how similar the candidate text is to the reference texts, with values closer to 1
+representing more similar texts. Few human translations will attain a score of 1, since this would indicate that the candidate is identical to one of the
+reference translations. For this reason, it is not necessary to attain a score of 1. Because there are more opportunities to match, adding additional
+reference translations will increase the BLEU score.
+"""
+
+_KWARGS_DESCRIPTION = """
+Computes BLEU score of translated segments against one or more references.
+Args:
+    predictions: list of translations to score.
+        Each translation should be tokenized into a list of tokens.
+    references: list of lists of references for each translation.
+        Each reference should be tokenized into a list of tokens.
+    max_order: Maximum n-gram order to use when computing BLEU score.
+    smooth: Whether or not to apply Lin et al. 2004 smoothing.
+Returns:
+    'bleu': bleu score,
+    'precisions': geometric mean of n-gram precisions,
+    'brevity_penalty': brevity penalty,
+    'length_ratio': ratio of lengths,
+    'translation_length': translation_length,
+    'reference_length': reference_length
+Examples:
+
+    >>> predictions = [
+    ...     ["hello", "there", "general", "kenobi"],                             # tokenized prediction of the first sample
+    ...     ["foo", "bar", "foobar"]                                             # tokenized prediction of the second sample
+    ... ]
+    >>> references = [
+    ...     [["hello", "there", "general", "kenobi"], ["hello", "there", "!"]],  # tokenized references for the first sample (2 references)
+    ...     [["foo", "bar", "foobar"]]                                           # tokenized references for the second sample (1 reference)
+    ... ]
+    >>> bleu = datasets.load_metric("bleu")
+    >>> results = bleu.compute(predictions=predictions, references=references)
+    >>> print(results["bleu"])
+    1.0
+"""
+
+
+@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
+class Bleu(datasets.Metric):
+    def _info(self):
+        return datasets.MetricInfo(
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            inputs_description=_KWARGS_DESCRIPTION,
+            features=datasets.Features(
+                {
+                    "predictions": datasets.Sequence(datasets.Value("string", id="token"), id="sequence"),
+                    "references": datasets.Sequence(
+                        datasets.Sequence(datasets.Value("string", id="token"), id="sequence"), id="references"
+                    ),
+                }
+            ),
+            codebase_urls=["https://github.com/tensorflow/nmt/blob/master/nmt/scripts/bleu.py"],
+            reference_urls=[
+                "https://en.wikipedia.org/wiki/BLEU",
+                "https://towardsdatascience.com/evaluating-text-output-in-nlp-bleu-at-your-own-risk-e8609665a213",
+            ],
+        )
+
+    def _compute(self, predictions, references, max_order=4, smooth=False):
+        score = compute_bleu(
+            reference_corpus=references, translation_corpus=predictions, max_order=max_order, smooth=smooth
+        )
+        (bleu, precisions, bp, ratio, translation_length, reference_length) = score
+        return {
+            "bleu": bleu,
+            "precisions": precisions,
+            "brevity_penalty": bp,
+            "length_ratio": ratio,
+            "translation_length": translation_length,
+            "reference_length": reference_length,
+        }

bleu_metric_card.md

@@ -0,0 +1,123 @@
+# Metric Card for BLEU
+
+
+## Metric Description
+BLEU (Bilingual Evaluation Understudy) is an algorithm for evaluating the quality of text which has been machine-translated from one natural language to another. Quality is considered to be the correspondence between a machine's output and that of a human: "the closer a machine translation is to a professional human translation, the better it is" – this is the central idea behind BLEU. BLEU was one of the first metrics to claim a high correlation with human judgements of quality, and remains one of the most popular automated and inexpensive metrics.
+
+Scores are calculated for individual translated segments—generally sentences—by comparing them with a set of good quality reference translations. Those scores are then averaged over the whole corpus to reach an estimate of the translation's overall quality. Neither intelligibility nor grammatical correctness are not taken into account.
+
+## Intended Uses
+BLEU and BLEU-derived metrics are most often used for machine translation.
+
+## How to Use
+
+This metric takes as input lists of predicted sentences and reference sentences:
+
+```python
+>>> predictions = [
+...     ["hello", "there", "general", "kenobi",
+...     ["foo", "bar" "foobar"]
+... ]
+>>> references = [
+...     [["hello", "there", "general", "kenobi"]],
+...     [["foo", "bar", "foobar"]]
+... ]
+>>> bleu = datasets.load_metric("bleu")
+>>> results = bleu.compute(predictions=predictions, references=references)
+>>> print(results)
+{'bleu': 0.6370964381207871, 'precisions': [0.8333333333333334, 0.75, 1.0, 1.0], 'brevity_penalty': 0.7165313105737893, 'length_ratio': 0.75, 'translation_length': 6, 'reference_length': 8}
+```
+
+### Inputs
+- **predictions** (`list` of `list`s): Translations to score. Each translation should be tokenized into a list of tokens.
+- **references** (`list` of `list`s): references for each translation. Each reference should be tokenized into a list of tokens.
+- **max_order** (`int`): Maximum n-gram order to use when computing BLEU score. Defaults to `4`.
+- **smooth** (`boolean`): Whether or not to apply Lin et al. 2004 smoothing. Defaults to `False`.
+
+### Output Values
+- **bleu** (`float`): bleu score
+- **precisions** (`list` of `float`s): geometric mean of n-gram precisions,
+- **brevity_penalty** (`float`): brevity penalty,
+- **length_ratio** (`float`): ratio of lengths,
+- **translation_length** (`int`): translation_length,
+- **reference_length** (`int`): reference_length
+
+Output Example:
+```python
+{'bleu': 1.0, 'precisions': [1.0, 1.0, 1.0, 1.0], 'brevity_penalty': 1.0, 'length_ratio': 1.167, 'translation_length': 7, 'reference_length': 6}
+```
+
+BLEU's output is always a number between 0 and 1. This value indicates how similar the candidate text is to the reference texts, with values closer to 1 representing more similar texts. Few human translations will attain a score of 1, since this would indicate that the candidate is identical to one of the reference translations. For this reason, it is not necessary to attain a score of 1. Because there are more opportunities to match, adding additional reference translations will increase the BLEU score.
+
+#### Values from Popular Papers
+The [original BLEU paper](https://aclanthology.org/P02-1040/) (Papineni et al. 2002) compares BLEU scores of five different models on the same 500-sentence corpus. These scores ranged from 0.0527 to 0.2571.
+
+The [Attention is All you Need paper](https://proceedings.neurips.cc/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf) (Vaswani et al. 2017) got a BLEU score of 0.284 on the WMT 2014 English-to-German translation task, and 0.41 on the WMT 2014 English-to-French translation task.
+
+### Examples
+
+Example where each sample has 1 reference:
+```python
+>>> predictions = [
+...     ["hello", "there", "general", "kenobi",
+...     ["foo", "bar" "foobar"]
+... ]
+>>> references = [
+...     [["hello", "there", "general", "kenobi"]],
+...     [["foo", "bar", "foobar"]]
+... ]
+>>> bleu = datasets.load_metric("bleu")
+>>> results = bleu.compute(predictions=predictions, references=references)
+>>> print(results)
+{'bleu': 0.6370964381207871, 'precisions': [0.8333333333333334, 0.75, 1.0, 1.0], 'brevity_penalty': 0.7165313105737893, 'length_ratio': 0.75, 'translation_length': 6, 'reference_length': 8}
+```
+
+Example where the second sample has 2 references:
+```python
+>>> predictions = [
+...     ["hello", "there", "general", "kenobi",
+...     ["foo", "bar" "foobar"]
+... ]
+>>> references = [
+...     [["hello", "there", "general", "kenobi"], ["hello", "there", "!"]],
+...     [["foo", "bar", "foobar"]]
+... ]
+>>> bleu = datasets.load_metric("bleu")
+>>> results = bleu.compute(predictions=predictions, references=references)
+>>> print(results)
+{'bleu': 1.0, 'precisions': [1.0, 1.0, 1.0, 1.0], 'brevity_penalty': 1.0, 'length_ratio': 1.1666666666666667, 'translation_length': 7, 'reference_length': 6}
+```
+
+## Limitations and Bias
+This metric hase multiple known limitations and biases:
+- BLEU compares overlap in tokens from the predictions and references, instead of comparing meaning. This can lead to discrepencies between BLEU scores and human ratings.
+- BLEU scores are not comparable across different datasets, nor are they comparable across different languages.
+- BLEU scores can vary greatly depending on which parameters are used to generate the scores, especially when different tokenization and normalization techniques are used. It is therefore not possible to compare BLEU scores generated using different parameters, or when these parameters are unknown.
+- Shorter predicted translations achieve higher scores than longer ones, simply due to how the score is calculated. A brevity penalty is introduced to attempt to counteract this.
+
+
+## Citation
+```bibtex
+@INPROCEEDINGS{Papineni02bleu:a,
+    author = {Kishore Papineni and Salim Roukos and Todd Ward and Wei-jing Zhu},
+    title = {BLEU: a Method for Automatic Evaluation of Machine Translation},
+    booktitle = {},
+    year = {2002},
+    pages = {311--318}
+}
+@inproceedings{lin-och-2004-orange,
+    title = "{ORANGE}: a Method for Evaluating Automatic Evaluation Metrics for Machine Translation",
+    author = "Lin, Chin-Yew  and
+      Och, Franz Josef",
+    booktitle = "{COLING} 2004: Proceedings of the 20th International Conference on Computational Linguistics",
+    month = "aug 23{--}aug 27",
+    year = "2004",
+    address = "Geneva, Switzerland",
+    publisher = "COLING",
+    url = "https://www.aclweb.org/anthology/C04-1072",
+    pages = "501--507",
+}
+```
+
+## Further References
+- This Hugging Face implementation uses [this Tensorflow implementation](https://github.com/tensorflow/nmt/blob/master/nmt/scripts/bleu.py)

requirements.txt

@@ -0,0 +1,2 @@
+datasets
+sklearn

rouge.py

@@ -0,0 +1,130 @@
+# Copyright 2020 The HuggingFace Datasets Authors.
+#
+# 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.
+""" ROUGE metric from Google Research github repo. """
+
+# The dependencies in https://github.com/google-research/google-research/blob/master/rouge/requirements.txt
+import absl  # Here to have a nice missing dependency error message early on
+import nltk  # Here to have a nice missing dependency error message early on
+import numpy  # Here to have a nice missing dependency error message early on
+import six  # Here to have a nice missing dependency error message early on
+from rouge_score import rouge_scorer, scoring
+
+import datasets
+
+
+_CITATION = """\
+@inproceedings{lin-2004-rouge,
+    title = "{ROUGE}: A Package for Automatic Evaluation of Summaries",
+    author = "Lin, Chin-Yew",
+    booktitle = "Text Summarization Branches Out",
+    month = jul,
+    year = "2004",
+    address = "Barcelona, Spain",
+    publisher = "Association for Computational Linguistics",
+    url = "https://www.aclweb.org/anthology/W04-1013",
+    pages = "74--81",
+}
+"""
+
+_DESCRIPTION = """\
+ROUGE, or Recall-Oriented Understudy for Gisting Evaluation, is a set of metrics and a software package used for
+evaluating automatic summarization and machine translation software in natural language processing.
+The metrics compare an automatically produced summary or translation against a reference or a set of references (human-produced) summary or translation.
+
+Note that ROUGE is case insensitive, meaning that upper case letters are treated the same way as lower case letters.
+
+This metrics is a wrapper around Google Research reimplementation of ROUGE:
+https://github.com/google-research/google-research/tree/master/rouge
+"""
+
+_KWARGS_DESCRIPTION = """
+Calculates average rouge scores for a list of hypotheses and references
+Args:
+    predictions: list of predictions to score. Each predictions
+        should be a string with tokens separated by spaces.
+    references: list of reference for each prediction. Each
+        reference should be a string with tokens separated by spaces.
+    rouge_types: A list of rouge types to calculate.
+        Valid names:
+        `"rouge{n}"` (e.g. `"rouge1"`, `"rouge2"`) where: {n} is the n-gram based scoring,
+        `"rougeL"`: Longest common subsequence based scoring.
+        `"rougeLSum"`: rougeLsum splits text using `"\n"`.
+        See details in https://github.com/huggingface/datasets/issues/617
+    use_stemmer: Bool indicating whether Porter stemmer should be used to strip word suffixes.
+    use_agregator: Return aggregates if this is set to True
+Returns:
+    rouge1: rouge_1 (precision, recall, f1),
+    rouge2: rouge_2 (precision, recall, f1),
+    rougeL: rouge_l (precision, recall, f1),
+    rougeLsum: rouge_lsum (precision, recall, f1)
+Examples:
+
+    >>> rouge = datasets.load_metric('rouge')
+    >>> predictions = ["hello there", "general kenobi"]
+    >>> references = ["hello there", "general kenobi"]
+    >>> results = rouge.compute(predictions=predictions, references=references)
+    >>> print(list(results.keys()))
+    ['rouge1', 'rouge2', 'rougeL', 'rougeLsum']
+    >>> print(results["rouge1"])
+    AggregateScore(low=Score(precision=1.0, recall=1.0, fmeasure=1.0), mid=Score(precision=1.0, recall=1.0, fmeasure=1.0), high=Score(precision=1.0, recall=1.0, fmeasure=1.0))
+    >>> print(results["rouge1"].mid.fmeasure)
+    1.0
+"""
+
+
+@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
+class Rouge(datasets.Metric):
+    def _info(self):
+        return datasets.MetricInfo(
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            inputs_description=_KWARGS_DESCRIPTION,
+            features=datasets.Features(
+                {
+                    "predictions": datasets.Value("string", id="sequence"),
+                    "references": datasets.Value("string", id="sequence"),
+                }
+            ),
+            codebase_urls=["https://github.com/google-research/google-research/tree/master/rouge"],
+            reference_urls=[
+                "https://en.wikipedia.org/wiki/ROUGE_(metric)",
+                "https://github.com/google-research/google-research/tree/master/rouge",
+            ],
+        )
+
+    def _compute(self, predictions, references, rouge_types=None, use_agregator=True, use_stemmer=False):
+        if rouge_types is None:
+            rouge_types = ["rouge1", "rouge2", "rougeL", "rougeLsum"]
+
+        scorer = rouge_scorer.RougeScorer(rouge_types=rouge_types, use_stemmer=use_stemmer)
+        if use_agregator:
+            aggregator = scoring.BootstrapAggregator()
+        else:
+            scores = []
+
+        for ref, pred in zip(references, predictions):
+            score = scorer.score(ref, pred)
+            if use_agregator:
+                aggregator.add_scores(score)
+            else:
+                scores.append(score)
+
+        if use_agregator:
+            result = aggregator.aggregate()
+        else:
+            result = {}
+            for key in scores[0]:
+                result[key] = list(score[key] for score in scores)
+
+        return result

rouge_metric_card.md

@@ -0,0 +1,121 @@
+# Metric Card for ROUGE
+
+## Metric Description
+ROUGE, or Recall-Oriented Understudy for Gisting Evaluation, is a set of metrics and a software package used for evaluating automatic summarization and machine translation software in natural language processing. The metrics compare an automatically produced summary or translation against a reference or a set of references (human-produced) summary or translation.
+
+Note that ROUGE is case insensitive, meaning that upper case letters are treated the same way as lower case letters.
+
+This metrics is a wrapper around the [Google Research reimplementation of ROUGE](https://github.com/google-research/google-research/tree/master/rouge)
+
+## How to Use
+At minimum, this metric takes as input a list of predictions and a list of references:
+```python
+>>> rouge = datasets.load_metric('rouge')
+>>> predictions = ["hello there", "general kenobi"]
+>>> references = ["hello there", "general kenobi"]
+>>> results = rouge.compute(predictions=predictions,
+...                         references=references)
+>>> print(list(results.keys()))
+['rouge1', 'rouge2', 'rougeL', 'rougeLsum']
+>>> print(results["rouge1"])
+AggregateScore(low=Score(precision=1.0, recall=1.0, fmeasure=1.0), mid=Score(precision=1.0, recall=1.0, fmeasure=1.0), high=Score(precision=1.0, recall=1.0, fmeasure=1.0))
+>>> print(results["rouge1"].mid.fmeasure)
+1.0
+```
+
+### Inputs
+- **predictions** (`list`): list of predictions to score. Each prediction
+        should be a string with tokens separated by spaces.
+- **references** (`list`): list of reference for each prediction. Each
+        reference should be a string with tokens separated by spaces.
+- **rouge_types** (`list`): A list of rouge types to calculate. Defaults to `['rouge1', 'rouge2', 'rougeL', 'rougeLsum']`.
+    - Valid rouge types:
+        - `"rouge1"`: unigram (1-gram) based scoring
+        - `"rouge2"`: bigram (2-gram) based scoring
+        - `"rougeL"`: Longest common subsequence based scoring.
+        - `"rougeLSum"`: splits text using `"\n"`
+        - See [here](https://github.com/huggingface/datasets/issues/617) for more information
+- **use_aggregator** (`boolean`): If True, returns aggregates. Defaults to `True`.
+- **use_stemmer** (`boolean`): If `True`, uses Porter stemmer to strip word suffixes. Defaults to `False`.
+
+### Output Values
+The output is a dictionary with one entry for each rouge type in the input list `rouge_types`. If `use_aggregator=False`, each dictionary entry is a list of Score objects, with one score for each sentence. Each Score object includes the `precision`, `recall`, and `fmeasure`. E.g. if `rouge_types=['rouge1', 'rouge2']` and `use_aggregator=False`, the output is:
+
+```python
+{'rouge1': [Score(precision=1.0, recall=0.5, fmeasure=0.6666666666666666), Score(precision=1.0, recall=1.0, fmeasure=1.0)], 'rouge2': [Score(precision=0.0, recall=0.0, fmeasure=0.0), Score(precision=1.0, recall=1.0, fmeasure=1.0)]}
+```
+
+If `rouge_types=['rouge1', 'rouge2']` and `use_aggregator=True`, the output is of the following format:
+```python
+{'rouge1': AggregateScore(low=Score(precision=1.0, recall=1.0, fmeasure=1.0), mid=Score(precision=1.0, recall=1.0, fmeasure=1.0), high=Score(precision=1.0, recall=1.0, fmeasure=1.0)), 'rouge2': AggregateScore(low=Score(precision=1.0, recall=1.0, fmeasure=1.0), mid=Score(precision=1.0, recall=1.0, fmeasure=1.0), high=Score(precision=1.0, recall=1.0, fmeasure=1.0))}
+```
+
+The `precision`, `recall`, and `fmeasure` values all have a range of 0 to 1.
+
+
+#### Values from Popular Papers
+
+
+### Examples
+An example without aggregation:
+```python
+>>> rouge = datasets.load_metric('rouge')
+>>> predictions = ["hello goodbye", "ankh morpork"]
+>>> references = ["goodbye", "general kenobi"]
+>>> results = rouge.compute(predictions=predictions,
+...                         references=references)
+>>> print(list(results.keys()))
+['rouge1', 'rouge2', 'rougeL', 'rougeLsum']
+>>> print(results["rouge1"])
+[Score(precision=0.5, recall=0.5, fmeasure=0.5), Score(precision=0.0, recall=0.0, fmeasure=0.0)]
+```
+
+The same example, but with aggregation:
+```python
+>>> rouge = datasets.load_metric('rouge')
+>>> predictions = ["hello goodbye", "ankh morpork"]
+>>> references = ["goodbye", "general kenobi"]
+>>> results = rouge.compute(predictions=predictions,
+...                         references=references,
+...                         use_aggregator=True)
+>>> print(list(results.keys()))
+['rouge1', 'rouge2', 'rougeL', 'rougeLsum']
+>>> print(results["rouge1"])
+AggregateScore(low=Score(precision=0.0, recall=0.0, fmeasure=0.0), mid=Score(precision=0.25, recall=0.25, fmeasure=0.25), high=Score(precision=0.5, recall=0.5, fmeasure=0.5))
+```
+
+The same example, but only calculating `rouge_1`:
+```python
+>>> rouge = datasets.load_metric('rouge')
+>>> predictions = ["hello goodbye", "ankh morpork"]
+>>> references = ["goodbye", "general kenobi"]
+>>> results = rouge.compute(predictions=predictions,
+...                         references=references,
+...                         rouge_types=['rouge_1'],
+...                         use_aggregator=True)
+>>> print(list(results.keys()))
+['rouge1']
+>>> print(results["rouge1"])
+AggregateScore(low=Score(precision=0.0, recall=0.0, fmeasure=0.0), mid=Score(precision=0.25, recall=0.25, fmeasure=0.25), high=Score(precision=0.5, recall=0.5, fmeasure=0.5))
+```
+
+## Limitations and Bias
+See [Schluter (2017)](https://aclanthology.org/E17-2007/) for an in-depth discussion of many of ROUGE's limits.
+
+## Citation
+```bibtex
+@inproceedings{lin-2004-rouge,
+    title = "{ROUGE}: A Package for Automatic Evaluation of Summaries",
+    author = "Lin, Chin-Yew",
+    booktitle = "Text Summarization Branches Out",
+    month = jul,
+    year = "2004",
+    address = "Barcelona, Spain",
+    publisher = "Association for Computational Linguistics",
+    url = "https://www.aclweb.org/anthology/W04-1013",
+    pages = "74--81",
+}
+```
+
+## Further References
+- This metrics is a wrapper around the [Google Research reimplementation of ROUGE](https://github.com/google-research/google-research/tree/master/rouge)