Papers › The Power of Scale for Parameter-Efficient Prompt Tuning

The Power of Scale for Parameter-Efficient Prompt Tuning

18 Apr 2021EMNLP 2021 11arXiv:2104.08691archive 2025-07-28

Brian Lester, Rami Al-Rfou, Noah Constant

In this work, we explore "prompt tuning", a simple yet effective mechanism for learning "soft prompts" to condition frozen language models to perform specific downstream tasks. Unlike the discrete text prompts used by GPT-3, soft prompts are learned through backpropagation and can be tuned to incorporate signal from any number of labeled examples. Our end-to-end learned approach outperforms GPT-3's "few-shot" learning by a large margin. More remarkably, through ablations on model size using T5, we show that prompt tuning becomes more competitive with scale: as models exceed billions of parameters, our method "closes the gap" and matches the strong performance of model tuning (where all model weights are tuned). This finding is especially relevant in that large models are costly to share and serve, and the ability to reuse one frozen model for multiple downstream tasks can ease this burden. Our method can be seen as a simplification of the recently proposed "prefix tuning" of Li and Liang (2021), and we provide a comparison to this and other similar approaches. Finally, we show that conditioning a frozen model with soft prompts confers benefits in robustness to domain transfer, as compared to full model tuning.

PaperPDFConference PDFCodeCode Syntology ran

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

No sample from the archive's official repository (google-research/prompt-tuning) is linked to this paper; the 15 samples below come from 3 community repositories. Syntology harvested code from google-research/prompt-tuning for another paper.

google-research/prompt-tuning: Syntology's graph links these samples to Overcoming Catastrophic Forgetting in Zero-Shot Cross-Lingual Generation (arXiv:2205.12647; that paper's own run record: 6 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method.

Syntology 15 code samples harvested from 3 repositories linked to this paper: 13 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 15 samples from 3 repositories, 13 ran (11 with no instrument failure; 2 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 5 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-24 and 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

google-research/prompt-tuning officialjaxApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Overcoming Catastrophic Forgetting in Zero-Shot Cross-Lingual Generation (arXiv:2205.12647; that paper's own run record: 6 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
RossNordby/SoftPromptsForEvaluation mentioned on GitHubpytorchMITSyntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
VE-FORBRYDERNE/mtj-softtuner mentioned on GitHubjaxApache-2.0Syntology: 10 samples, 9 ran (2 instrument-failed) · 1 unverified report
arazd/ProgressivePrompts mentioned on GitHubpytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Progressive Prompts: Continual Learning for Language Models (arXiv:2301.12314; that paper's own run record: 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified); nothing checks that they implement this paper's method. report
corolla-johnson/mkultra mentioned on GitHubpytorchMITSyntology: not harvested report
exelents/soft-prompt-tuning mentioned on GitHubpytorchMITSyntology: 4 samples, 3 ran (0 instrument-failed) · 1 unverified report
google-deepmind/thunnini mentioned on GitHubjaxApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Understanding Prompt Tuning and In-Context Learning via Meta-Learning (arXiv:2505.17010; that paper's own run record: 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified); nothing checks that they implement this paper's method. report
hellokevin07/elastictrainer mentioned on GitHubtfMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to LoRA: Low-Rank Adaptation of Large Language Models (arXiv:2106.09685; that paper's own run record: 51 ran (of which 19 constructed an object rather than computing a result; 43 with no instrument failure: 1 honoured, 0 violated, 42 with no contract checked; 8 where Syntology's instrument failed) · 33 unverified); nothing checks that they implement this paper's method. report
kipgparker/soft-prompt-tuning mentioned on GitHubpytorchMITSyntology: not harvested report
mkshing/Prompt-Tuning mentioned on GitHubpytorchMITSyntology: not harvested report
zhanglab-aim/lingo mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to LoRA: Low-Rank Adaptation of Large Language Models (arXiv:2106.09685; that paper's own run record: 51 ran (of which 19 constructed an object rather than computing a result; 43 with no instrument failure: 1 honoured, 0 violated, 42 with no contract checked; 8 where Syntology's instrument failed) · 33 unverified); nothing checks that they implement this paper's method. report
Zeng-WH/Prompt-Tuning pytorchSyntology: 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

15 samples harvested: 13 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 0 honoured, 0 violated, 11 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · our draft was wrong
11ran
2unverified

Licence: 0 of the 15 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.

RossNordby/SoftPromptsForEvaluation community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
normalize_chess_elos chess_training/dataset_loading.py ran fingerprinted MIT (permissive) · 29fe77a79233bc67 · report fixture: signature elos:Tensor/2/float · returned Tensor · level recorded 2026-09-24
VE-FORBRYDERNE/mtj-softtuner community (archive-listed) 10 samples, 9 ran (2 instrument-failed) · 1 unverified
core_dummy_initialized mtj_softtuner/testing_utils.py ran Apache-2.0 (permissive) · 16ba263208ecf384 · report fixture: signature f:callable/0/? · returned function · level recorded 2026-09-24
core_test_mode mtj_softtuner/testing_utils.py ran Apache-2.0 (permissive) · 659a594a04601c37 · report fixture: signature f:callable/0/? · returned function · level recorded 2026-09-24
get_config_from_root in 13 papers versioneer.py ran Apache-2.0 (permissive) · 51944f53542b7e19 · report fixture: signature root:str/0/str · returned VersioneerConfig · level recorded 2026-09-25
register_vcs_handler in 31 papers mtj_softtuner/_version.py ran · our draft was wrong Apache-2.0 (permissive) · c55cafde08716f01 · report fixture: signature vcs:str/0/str, method:str/0/str · returned function · level recorded 2026-09-26
register_vcs_handler in 31 papers versioneer.py ran Apache-2.0 (permissive) · 9dc618d67058e3e6 · report fixture: signature vcs:str/0/str, method:str/0/str · returned function · level recorded 2026-09-25
restore_variable mtj_softtuner/serialization.py ran Apache-2.0 (permissive) · e4eaafa9797969b1 · report fixture: signature universe:int/0/int, name:str/0/str · returned dict · level recorded 2026-09-24
run_command in 61 papers mtj_softtuner/_version.py ran Apache-2.0 (permissive) · dbb37355631dabcf · report fixture: signature commands:list/1/str, args:list/1/str, cwd:NoneType/0/?, verbose:bool/0/bool, hide_stderr:bool/0/bool, env:NoneType/0/? · returned tuple · level recorded 2026-09-26
shatter mtj_softtuner/core.py ran Apache-2.0 (permissive) · 33d302ec4e7efc18 · report fixture: signature in_axes:str/0/str, out_axes:str/0/str · returned function · level recorded 2026-09-24
versions_from_parentdir in 31 papers mtj_softtuner/_version.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 5032e3b81c0fffb3 · report fixture: signature parentdir_prefix:str/0/str, root:str/0/str, verbose:bool/0/bool · returned dict · level recorded 2026-09-26
core_partly_initialized mtj_softtuner/testing_utils.py unverified Apache-2.0 (permissive) · 3be93347d7ab5153 · report
exelents/soft-prompt-tuning community (archive-listed) 4 samples, 3 ran (0 instrument-failed) · 1 unverified
is_time_to_exit utils.py ran MIT (permissive) · 641dddbb7b534717 · report fixture: signature args:Args/0/?, epoch_steps:int/0/int, global_steps:int/0/int · returned bool · level recorded 2026-09-24
label_smoothed_nll_loss in 43 papers utils1.py ran MIT (permissive) · e386d3a24af4d168 · report fixture: signature lprobs:Tensor/2/float, target:Tensor/1/int, epsilon:float/0/float, ignore_index:int/0/int · returned tuple · level recorded 2026-09-26
lmap in 15 papers utils1.py ran MIT (permissive) · 7c43af131a6f39a6 · report fixture: signature f:callable/0/?, x:list/1/int · returned list · level recorded 2026-09-25
get_model_config_tokenizer gptj_utils.py unverified MIT (permissive) · 04adf1e29b3ea4b7 · report

Tasks

Few-Shot Learning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AdafactorAdamAttentionAttention DropoutBPECosine AnnealingDense ConnectionsDropoutGPT-3Gated Linear UnitInverse Square Root ScheduleLayer NormalizationLinear LayerLinear Warmup With Cosine AnnealingMulti-Head AttentionResidual ConnectionSentencePieceSoftmaxT5Weight Decay

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