tatsu-lab/linguistic_calibration
收藏资源简介:
--- license: cc-by-nc-4.0 configs: - config_name: trivia_qa_paragraph_generation data_files: - split: sft path: "trivia_qa_paragraph_generation/sft_10k.csv" - split: reward_model path: "trivia_qa_paragraph_generation/reward_model_20k.csv" - split: prompt_validation path: "trivia_qa_paragraph_generation/prompt_validation_1k.csv" - split: ppo path: "trivia_qa_paragraph_generation/ppo_40k.csv" - split: ppo_validation path: "trivia_qa_paragraph_generation/ppo_validation_1k.csv" - split: validation path: "trivia_qa_paragraph_generation/validation_1k.csv" - split: test path: "trivia_qa_paragraph_generation/test_11k.csv" - config_name: jeopardy_paragraph_generation data_files: - split: test path: "jeopardy_paragraph_generation/test.csv" - config_name: sft_training data_files: - split: train path: "sft_training.csv" - config_name: reward_model_training data_files: - split: train path: "reward_model_training.csv" - config_name: sciq_paragraph_generation data_files: - split: test path: sciq_paragraph_generation/test.csv - config_name: bioasq_paragraph_generation data_files: - split: test path: bioasq_paragraph_generation/test.csv --- This Datasets repo contains training and evaluation datasets for the paper "Linguistic Calibration of Long-Form Generations". Please refer to our GitHub repo at https://github.com/tatsu-lab/linguistic_calibration for more information, and check out our paper for our research findings: https://arxiv.org/abs/2404.00474
数据集概述
数据集名称及配置
-
trivia_qa_paragraph_generation
- sft:
trivia_qa_paragraph_generation/sft_10k.csv - reward_model:
trivia_qa_paragraph_generation/reward_model_20k.csv - prompt_validation:
trivia_qa_paragraph_generation/prompt_validation_1k.csv - ppo:
trivia_qa_paragraph_generation/ppo_40k.csv - ppo_validation:
trivia_qa_paragraph_generation/ppo_validation_1k.csv - validation:
trivia_qa_paragraph_generation/validation_1k.csv - test:
trivia_qa_paragraph_generation/test_11k.csv
- sft:
-
jeopardy_paragraph_generation
- test:
jeopardy_paragraph_generation/test.csv
- test:
-
sft_training
- train:
sft_training.csv
- train:
-
reward_model_training
- train:
reward_model_training.csv
- train:
-
sciq_paragraph_generation
- test:
sciq_paragraph_generation/test.csv
- test:
-
bioasq_paragraph_generation
- test:
bioasq_paragraph_generation/test.csv
- test:
许可证
- cc-by-nc-4.0



