Scaling Instruction-Finetuned Language Models

Code

• github.com/google-research/flan
  tf★ 1,559
• github.com/declare-lab/flan-alpaca
  pytorch★ 357
• github.com/formulamonks/llm-benchmarker-suite
  pytorch★ 48
• github.com/theoremone/llm-benchmarker-suite
  pytorch★ 48
• github.com/zchuz/timebench
  none★ 34
• github.com/coastalcph/zeroshot_lexglue
  none★ 29
• github.com/joelniklaus/lawinstruct
  none★ 26
• github.com/kapllan/zeroshot_lexglue
  none★ 0
• github.com/yli-z/ml4h_are_clinical_t5_models_better_for_clinical_text
  none★ 0

Abstract

Finetuning language models on a collection of datasets phrased as instructions has been shown to improve model performance and generalization to unseen tasks. In this paper we explore instruction finetuning with a particular focus on (1) scaling the number of tasks, (2) scaling the model size, and (3) finetuning on chain-of-thought data. We find that instruction finetuning with the above aspects dramatically improves performance on a variety of model classes (PaLM, T5, U-PaLM), prompting setups (zero-shot, few-shot, CoT), and evaluation benchmarks (MMLU, BBH, TyDiQA, MGSM, open-ended generation). For instance, Flan-PaLM 540B instruction-finetuned on 1.8K tasks outperforms PALM 540B by a large margin (+9.4% on average). Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints, which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.

Tasks

Benchmark Results

Reproductions