add mutual

lm_eval/tasks/mutual/README.md

+# MuTual
+
+### Paper
+
+Title: `MuTual: A Dataset for Multi-Turn Dialogue Reasoning`
+
+Abstract: https://www.aclweb.org/anthology/2020.acl-main.130/
+
+MuTual is a retrieval-based dataset for multi-turn dialogue reasoning, which is
+modified from Chinese high school English listening comprehension test data.
+
+Homepage: https://github.com/Nealcly/MuTual
+
+### Citation
+
+```
+@inproceedings{mutual,
+    title = "MuTual: A Dataset for Multi-Turn Dialogue Reasoning",
+    author = "Cui, Leyang  and Wu, Yu and Liu, Shujie and Zhang, Yue and Zhou, Ming" ,
+    booktitle = "Proceedings of the 58th Conference of the Association for Computational Linguistics",
+    year = "2020",
+    publisher = "Association for Computational Linguistics",
+}
+```
+
+### Groups and Tasks
+
+#### Groups
+
+* Not part of a group yet.
+
+#### Tasks
+
+* `mutual`
+* `mutual_plus`
+
+### Checklist
+
+For adding novel benchmarks/datasets to the library:
+* [ ] Is the task an existing benchmark in the literature?
+  * [ ] Have you referenced the original paper that introduced the task?
+  * [ ] If yes, does the original paper provide a reference implementation? If so, have you checked against the reference implementation and documented how to run such a test?
+
+
+If other tasks on this dataset are already supported:
+* [ ] Is the "Main" variant of this task clearly denoted?
+* [ ] Have you provided a short sentence in a README on what each new variant adds / evaluates?
+* [ ] Have you noted which, if any, published evaluation setups are matched by this variant?

lm_eval/tasks/mutual/multual_plus.yaml

+include: mutual.yaml
+task: mutual_plus
+dataset_name: mutual_plus

lm_eval/tasks/mutual/mutual.yaml

+task: mutual
+dataset_path: "EleutherAI/mutual"
+dataset_name: mutual
+output_type: multiple_choice
+training_split: train
+validation_split: validation
+doc_to_text: "{{article}}"
+doc_to_target: "{{['A', 'B', 'C', 'D'].index(answers)}}"
+doc_to_choice: "{{options}}"
+process_docs: !function utils.process_docs
+process_results: !function utils.process_results
+should_decontaminate: true
+doc_to_decontamination_query: "{{article}}"
+metric_list:
+  - metric: r@1
+    aggregation: mean
+    higher_is_better: true
+  - metric: r@2
+    aggregation: mean
+    higher_is_better: true
+  - metric: mrr
+    aggregation: mean
+    higher_is_better: true

lm_eval/tasks/mutual/utils.py

+import numpy as np
+
+def process_docs(dataset):
+
+    def _detokenize(text):
+        text = text.replace(" '", "'")
+        text = text.replace(" \n", "\n")
+        text = text.replace("\n ", "\n")
+        text = text.replace(" n't", "n't")
+        text = text.replace("`` ", '"')
+        text = text.replace("''", '"')
+        # punctuation
+        text = text.replace(" :", ":")
+        text = text.replace(" ;", ";")
+        text = text.replace(" !", "!")
+        text = text.replace(" ?", "?")
+        text = text.replace(" ,", ",")
+        text = text.replace(" .", ".")
+        return text
+
+    def _process(doc):
+        return {
+            "article": _detokenize(doc["article"]),
+            "options": [_detokenize(option) for option in doc["options"]],
+        }
+
+    return dataset.map(_process)
+
+
+def process_results(doc, results):
+    gold = ["A", "B", "C", "D"].index(doc["answers"])
+    r4_1 = np.argmax(results) == gold  # r4_1 = accuracy
+    ranks = sorted(results, reverse=True)
+    r4_2 = (ranks.index(results[gold]) == 1) + r4_1
+    mrr = 1.0 / (ranks.index(results[gold]) + 1)  # `+ 1` for index offset
+    return {"r@1": r4_1, "r@2": r4_2, "mrr": mrr}