paraphrasing
Long-Instruction-with-Paraphrasing
🔥 Updates
[2024.6.4] Add a slim version. The sample number is reduced from about 20k to 10k.
[2024.5.28]
The data format is converted from "chatml" to "messages", which is more convenient to use tokenizer.apply_chat_template. The old version has been moved to "legacy" branch.
The version without "Original text paraphrasing" is added.
📊 Long Context Instruction-tuning dataset with "Original text paraphrasing"
Paper
Github
consist of multiple tasks
Chinese and… See the full description on the dataset page: https://huggingface.co/datasets/yuyijiong/Long-Instruction-with-Paraphrasing.task045_miscellaneous_sentence_paraphrasing
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task045_miscellaneous_sentence_paraphrasing
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task045_miscellaneous_sentence_paraphrasing.paraphrasing-french
Attribution
MTEB-format derivative of ismailiismail/paraphrasing_french. Query = phrase; corpus = paraphrase.
task1288_glue_mrpc_paraphrasing
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task1288_glue_mrpc_paraphrasing
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks}… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task1288_glue_mrpc_paraphrasing.Long-Instruction-with-Paraphrasing-zhtwparsinlu_query_paraphrasingA Persian query paraphrasing task (paraphrase or not, given two questions).
The questions are partly mined using Google auto-complete, and partly translated from Quora paraphrasing dataset.