Papers › Scaling Instruction-Finetuned Language Models

Scaling Instruction-Finetuned Language Models

20 Oct 2022arXiv:2210.11416archive 2025-07-28

Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Alex Castro-Ros, Marie Pellat, Kevin Robinson, Dasha Valter, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, Jason Wei

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.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 17 code samples harvested from 3 repositories linked to this paper: 8 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 7 where Syntology's instrument failed) · 9 unverified. Where Syntology's instrument failed: 7 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 17 samples from 3 repositories, 8 ran (1 with no instrument failure; 7 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

coastalcph/zeroshot_lexglue mentioned on GitHubSyntology: not harvested report
declare-lab/flan-alpaca mentioned on GitHubpytorchSyntology: 5 samples, 1 ran (0 instrument-failed) · 4 unverified report
formulamonks/llm-benchmarker-suite mentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (arXiv:2306.05685; that paper's own run record: 9 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified); nothing checks that they implement this paper's method. report
google-research/flan mentioned on GitHubtfApache-2.0Syntology: 10 samples, 5 ran (5 instrument-failed) · 5 unverified report
joelniklaus/lawinstruct mentioned on GitHubSyntology: not harvested report
kapllan/zeroshot_lexglue mentioned on GitHubSyntology: not harvested report
theoremone/llm-benchmarker-suite mentioned on GitHubpytorchMITSyntology: not harvested report
yli-z/ml4h_are_clinical_t5_models_better_for_clinical_text mentioned on GitHubSyntology: 2 samples, 2 ran (2 instrument-failed) · 0 unverified report
zchuz/timebench mentioned on GitHubMITSyntology: not harvested report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

17 samples harvested: 8 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 7 where Syntology's instrument failed) · 9 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

7ran · our draft was wrong
1ran
9unverified

Licence: 2 of the 17 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 3 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

declare-lab/flan-alpaca community (archive-listed) 5 samples, 1 ran (0 instrument-failed) · 4 unverified
TextToTextSample training.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · ce7574929468c029 · report fixture: returned TextToTextSample · level recorded 2026-09-22
LightningModel training.py unverified Apache-2.0 (permissive) · fd9be101d56df782 · report
TextToTextData training.py unverified Apache-2.0 (permissive) · 20789a28303d52df · report
TextToTextDataset training.py unverified Apache-2.0 (permissive) · 883e0c27ae1147a1 · report
TokensLengthAnalyzer training.py unverified Apache-2.0 (permissive) · 826cda9a9517d6b1 · report
google-research/flan community (archive-listed) 10 samples, 5 ran (5 instrument-failed) · 5 unverified
example_list_to_batch flan/v2/preprocessors.py ran · our draft was wrong Apache-2.0 (permissive) · 09b037bcbd40904b · report fixture: signature example_list:list/1/dict · returned dict · level recorded 2026-09-22
format_from_feature_dictionary flan/v2/preprocessors.py ran · our draft was wrong Apache-2.0 (permissive) · 348e3fae4483828f · report fixture: signature format_string:str/0/str, feature_dictionary:dict/0/? · returned EagerTensor · level recorded 2026-09-22
get_training_keys flan/v2/preprocessors.py ran · our draft was wrong Apache-2.0 (permissive) · d9dfc4a47b45d445 · report fixture: signature patterns_list:list/1/list, add_keys:list/1/str · returned set · level recorded 2026-09-22
remove_unbatchable_items_ds flan/v2/preprocessors.py ran · our draft was wrong Apache-2.0 (permissive) · 945f9e10849aba5f · report fixture: signature ds:Dataset/0/?, training_keys:list/1/str · returned Dataset · level recorded 2026-09-22
remove_unbatchable_items_ex flan/v2/preprocessors.py ran · our draft was wrong Apache-2.0 (permissive) · aed287a99a6a9c29 · report fixture: signature ex:dict/0/?, training_keys:list/1/str · returned dict · level recorded 2026-09-22
batch_apply_template flan/v2/preprocessors.py unverified Apache-2.0 (permissive) · e4eeb164dc20068a · report
example_batch_to_list flan/v2/preprocessors.py unverified Apache-2.0 (permissive) · 69440900d762a595 · report
get_batch_formatter flan/v2/preprocessors.py unverified Apache-2.0 (permissive) · da56567fc45532d9 · report
reformat_batched_example flan/v2/preprocessors.py unverified Apache-2.0 (permissive) · 1a288d0d4efefd6a · report
reformat_single_example flan/v2/preprocessors.py unverified Apache-2.0 (permissive) · 8c6076b9e23aff01 · report
yli-z/ml4h_are_clinical_t5_models_better_for_clinical_text community (archive-listed) 2 samples, 2 ran (2 instrument-failed) · 0 unverified
format_single_prompted_example src/finetune_bc5cdr.py ran · our draft was wrong fingerprinted no licence file found · pointer only · 965f08c21d5cd016 · report fixture: signature sentence:str/0/str · returned str · level recorded 2026-09-22
preprocess_function in 29 papers src/finetune_bc5cdr.py ran · our draft was wrong no licence file found · pointer only · 298d3ae7382f4e0b · report fixture: signature examples:dict/0/?, tokenizer:callable/0/?, max_seq_length:int/0/int, is_instruction_finetuned:bool/0/bool · returned dict · level recorded 2026-09-22

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.

3 assemblies on this paper: 3 harvested, 0 generated, listed in 2 groups (siblings lifted in one batch form one group).

2 sibling assemblies for Flan-T5 (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Flan-T5 (graph method: Flan-T5). 2 bricks; individually on a fixture: 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. preprocess_function entry · level 1 · ran · our draft was wrong · 298d3ae7382f4e0b
  2. format_single_prompted_example data_transform · level 1 · ran · our draft was wrong · 965f08c21d5cd016

harvested Assembly for Flan-T5 (graph method: Flan-T5). 5 bricks; individually on a fixture: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 4 unverified; the one sample that ran constructed an object rather than computing a result. The assembly has not been run end-to-end.

one of 2 siblings

  1. LightningModel entry · level 0 · unverified · fd9be101d56df782
  2. TextToTextDataset helper · level 0 · unverified · 883e0c27ae1147a1
  3. TextToTextData helper · level 0 · unverified · 20789a28303d52df
  4. TextToTextSample helper · level 1 · ran · ce7574929468c029
  5. TokensLengthAnalyzer helper · level 0 · unverified · 826cda9a9517d6b1

harvested Assembly for Flan-PaLM (graph method: Flan-PaLM). 10 bricks; individually on a fixture: 5 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 5 where Syntology's instrument failed) · 5 unverified. The assembly has not been run end-to-end.

  1. get_batch_formatter entry · level 0 · unverified · da56567fc45532d9
  2. batch_apply_template helper · level 0 · unverified · e4eeb164dc20068a
  3. get_training_keys helper · level 1 · ran · our draft was wrong · d9dfc4a47b45d445
  4. reformat_batched_example helper · level 0 · unverified · 1a288d0d4efefd6a
  5. remove_unbatchable_items_ds helper · level 1 · ran · our draft was wrong · 945f9e10849aba5f
  6. example_batch_to_list helper · level 0 · unverified · 69440900d762a595
  7. example_list_to_batch helper · level 1 · ran · our draft was wrong · 09b037bcbd40904b
  8. reformat_single_example helper · level 0 · unverified · 8c6076b9e23aff01
  9. remove_unbatchable_items_ex helper · level 1 · ran · our draft was wrong · aed287a99a6a9c29
  10. format_from_feature_dictionary data_transform · level 1 · ran · our draft was wrong · 348e3fae4483828f

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2210.11416") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Coreference ResolutionCross-Lingual Question AnsweringMMLUMulti-task Language UnderstandingParaphrase IdentificationQuestion Answering

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Coreference Resolution Winograd Schema Challenge Flan-T5 XXL (zero -shot) Accuracy 89.82 #10 of 82 Archive leaderboard report
Cross-Lingual Question Answering TyDiQA-GoldP Flan-U-PaLM 540B (direct-prompting) EM 68.3 #3 of 11 Archive leaderboard report
Cross-Lingual Question Answering TyDiQA-GoldP Flan-PaLM 540B (direct-prompting) EM 67.8 #4 of 11 Archive leaderboard report
Multi-task Language Understanding BBH-alg Flan-PaLM 540B (3-shot, fine-tuned, CoT + SC) Average (%) 66.5 #2 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-alg PaLM 540B (CoT + self-consistency) Average (%) 62.2 #3 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-alg Flan-PaLM 540B (3-shot, fine-tuned, CoT) Average (%) 61.3 #4 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-alg PaLM 540B (CoT) Average (%) 57.6 #5 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-alg Flan-PaLM 540B (3-shot, fine-tuned) Average (%) 48.2 #6 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-alg PaLM 540B Average (%) 38.3 #7 of 7 Archive leaderboard report
Multi-task Language Understanding BBH-nlp Flan-PaLM 540B (3-shot, fine-tuned, CoT + SC) Average (%) 78.4 #7 of 15 Archive leaderboard report
Multi-task Language Understanding BBH-nlp PaLM 540B (CoT + self-consistency) Average (%) 78.2 #8 of 15 Archive leaderboard report
Multi-task Language Understanding BBH-nlp Flan-PaLM 540B (3-shot, fine-tuned, CoT) Average (%) 72.4 #10 of 15 Archive leaderboard report
Multi-task Language Understanding BBH-nlp PaLM 540B (CoT) Average (%) 71.2 #11 of 15 Archive leaderboard report
Multi-task Language Understanding BBH-nlp Flan-PaLM 540B (5-shot, finetuned) Average (%) 70.0 #12 of 15 Archive leaderboard report
Multi-task Language Understanding BBH-nlp PaLM 540B Average (%) 62.7 #13 of 15 Archive leaderboard report
Multi-task Language Understanding MGSM Flan-PaLM 540B (8-shot, fine-tuned, CoT + SC) Average (%) 72.0 #3 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM Flan-U-PaLM 540B (CoT) Average (%) 60.4 #4 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM Flan-PaLM 540B (8-shot, fine-tuned, CoT) Average (%) 57.0 #5 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM text-davinci-003 Average (%) 36 #8 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM code-davinci-002 Average (%) 35 #9 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM text-davinci-002 Average (%) 23.7 #10 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM Flan-PaLM 540B (8-shot, fine-tuned) Average (%) 21.2 #11 of 12 Archive leaderboard report
Multi-task Language Understanding MGSM GPT-3 Davinci 175B Average (%) 5.7 #12 of 12 Archive leaderboard report
Multi-task Language Understanding MML llama 2(65b) Average (%) 73.5 #9 of 44 Archive leaderboard report
Multi-task Language Understanding MML GPT-3 Davinci 175B (CoT) Average (%) 59.5 #19 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-XL 3B (CoT) Average (%) 45.5 #27 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-Large 780M Average (%) 45.1 #29 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-Large 780M (CoT) Average (%) 40.5 #31 of 44 Archive leaderboard report
Multi-task Language Understanding MML GPT-3 Davinci 175B (5-shot) Average (%) 39.7 #32 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-Base 250M Average (%) 35.9 #38 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-Base 250M (CoT) Average (%) 33.7 #39 of 44 Archive leaderboard report
Multi-task Language Understanding MML Flan-T5-Small 80M Average (%) 28.7 #43 of 44 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Introduced by this paper: Flan-T5

AdafactorAttentionAttention DropoutBPEDense ConnectionsDropoutFlan-T5Gated Linear UnitInverse Square Root ScheduleLayer NormalizationLinear LayerMulti-Head AttentionPaLMResidual ConnectionSentencePieceSoftmaxT5

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections