MétaCan
Menu
Back to cohort
Record W4405907376 · doi:10.1109/tbdata.2024.3524101

Is Split Learning Privacy-Preserving for Fine-Tuning Large Language Models?

2024· article· en· W4405907376 on OpenAlexaff
Dixi Yao, Baochun Li

Bibliographic record

VenueIEEE Transactions on Big Data · 2024
Typearticle
Languageen
FieldComputer Science
TopicPrivacy-Preserving Technologies in Data
Canadian institutionsUniversity of Toronto
Fundersnot available
KeywordsComputer scienceTheoretical computer scienceArtificial intelligence

Abstract

fetched live from OpenAlex

With the success of pre-trained large language models in various tasks, users, individuals and enterprises alike, may need to fine-tune these models with their own datasets. Split learning was proposed to divide the model and place a portion on each user's own device, and intermediate results in each iteration of training will be sent to the server to complete the forward pass. There were concerns in the literature about whether private data can be leaked by sending such intermediate results from the training process. In this paper, we conduct empirical studies on typical large language models, such as GPT-2, OPT, Llama, and Qwen, to show that in most situations, an honest-but-curious server is not able to reconstruct private data using such intermediate results. To find out the reason why large language models preserve data privacy better in these situations, we present our theoretical analyses on these empirical observations. In one special case, where a state-of-the-art existing attack can reconstruct data in the first iteration, we show that it can be easily defended with a simple but effective solution leveraging publicly accessible data.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.016
metaresearch head score (Gemma)0.066
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.016
Threshold uncertainty score0.086

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0160.066
Meta-epidemiology (narrow)0.0010.001
Meta-epidemiology (broad)0.0020.002
Bibliometrics0.0010.001
Science and technology studies0.0020.006
Scholarly communication0.0040.018
Open science0.0040.008
Research integrity0.0030.007
Insufficient payload (model declined to judge)0.0030.001

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.120
GPT teacher head0.329
Teacher spread0.209 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designSimulation or modeling
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations5
Published2024
Admission routes1
Has abstractyes

Explore more

Same venueIEEE Transactions on Big DataSame topicPrivacy-Preserving Technologies in DataFrench-language works237,207