README.md

---
annotations_creators:
- machine-generated
language:
- en
language_creators:
- machine-generated
license: mit
multilinguality:
- monolingual
pretty_name: TOFU
size_categories:
- 1K<n<10K
source_datasets:
- original
tags:
- unlearning
- question answering
- TOFU
- NLP
- LLM
task_categories:
- question-answering
task_ids:
- closed-domain-qa
configs:
- config_name: full
  data_files: full.json
  default: true
- config_name: forget01
  data_files: forget01.json
- config_name: forget05
  data_files: forget05.json
- config_name: forget10
  data_files: forget10.json
- config_name: retain90
  data_files: retain90.json
- config_name: retain95
  data_files: retain95.json
- config_name: retain99
  data_files: retain99.json
- config_name: world_facts
  data_files: world_facts.json
- config_name: real_authors
  data_files: real_authors.json
- config_name: forget01_perturbed
  data_files: forget01_perturbed.json
- config_name: forget05_perturbed
  data_files: forget05_perturbed.json
- config_name: forget10_perturbed
  data_files: forget10_perturbed.json
- config_name: retain_perturbed
  data_files: retain_perturbed.json
- config_name: world_facts_perturbed
  data_files: world_facts_perturbed.json
- config_name: real_authors_perturbed
  data_files: real_authors_perturbed.json
- config_name: holdout01
  data_files: holdout01.json
- config_name: holdout05
  data_files: holdout05.json
- config_name: holdout10
  data_files: holdout10.json
- config_name: sample_2
  data_files:
  - split: forget
    path: kg-samples/forget_sample_2_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_2_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_2_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_2_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_2_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_2_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_2.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_2.jsonl
- config_name: sample_3
  data_files:
  - split: forget
    path: kg-samples/forget_sample_3_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_3_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_3_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_3_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_3_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_3_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_3.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_3.jsonl
- config_name: sample_4
  data_files:
  - split: forget
    path: kg-samples/forget_sample_4_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_4_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_4_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_4_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_4_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_4_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_4.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_4.jsonl
- config_name: sample_5
  data_files:
  - split: forget
    path: kg-samples/forget_sample_5_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_5_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_5_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_5_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_5_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_5_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_5.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_5.jsonl
- config_name: sample_6
  data_files:
  - split: forget
    path: kg-samples/forget_sample_6_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_6_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_6_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_6_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_6_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_6_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_6.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_6.jsonl
- config_name: sample_7
  data_files:
  - split: forget
    path: kg-samples/forget_sample_7_qa.jsonl
  - split: retain
    path: kg-samples/retain_sample_7_qa.jsonl
  - split: forget_asc
    path: kg-samples/forget_sample_7_qa_asc.jsonl
  - split: retain_asc
    path: kg-samples/retain_sample_7_qa_asc.jsonl
  - split: forget_desc
    path: kg-samples/forget_sample_7_qa_desc.jsonl
  - split: retain_desc
    path: kg-samples/retain_sample_7_qa_desc.jsonl
  - split: forget_perturbed
    path: kg-samples/perturbed_forget_7.jsonl
  - split: retain_perturbed
    path: kg-samples/perturbed_retain_7.jsonl
---

# Towards Multi-reference Unlearning