Datasetpublic

Harvard-DCML/tis-dolci-random-unbalanced

A Critical Look at Targeted Instruction Selection This repository contains the pre-computed random unbalanced subsets used as baselines in the paper "A Critical Look at Targeted Instruction Selection: Disentangling What Matters (and What Doesn't)". Paper: https://huggingface.co/papers/2602.14696 GitHub Repository: https://github.com/dcml-lab/targeted-instruction-selection Description Instruction fine-tuning of large language models (LLMs) often involves… See the full description on the dataset page: https://huggingface.co/datasets/Harvard-DCML/tis-dolci-random-unbalanced.

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes39downloads

Dataset Card

A Critical Look at Targeted Instruction Selection

This repository contains the pre-computed random unbalanced subsets used as baselines in the paper "A Critical Look at Targeted Instruction Selection: Disentangling What Matters (and What Doesn't)".

—Paper: https://huggingface.co/papers/2602.14696
—GitHub Repository: https://github.com/dcml-lab/targeted-instruction-selection

Description

Instruction fine-tuning of large language models (LLMs) often involves selecting a subset of instruction training data from a large candidate pool. This dataset provides random unbalanced subsets of 10,000 examples across different seeds (0, 1, 2) to serve as a controlled comparison point for various selection algorithms and data representations.

The subsets are part of a broader framework designed to systematically analyze core ingredients in instruction selection, helping provide actionable guidance for practitioners fine-tuning LLMs on target tasks.

Citation

bibtex

@article{nayak2026critical,
  title={A Critical Look at Targeted Instruction Selection: Disentangling What Matters (and What Doesn't)},
  author={Nayak, Nihal V and Rodriguez-Diaz, Paula and Hulkund, Neha and Beery, Sara and Alvarez-Melis, David},
  journal={arXiv preprint arXiv:2602.14696},
  year={2026}
}