Pith. sign in

REVIEW 3 cited by

A Survey on Prompt Tuning

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2507.06085 v2 pith:OAPJMIBU submitted 2025-07-08 cs.CL

A Survey on Prompt Tuning

Zongqian Li , Yixuan Su , Nigel Collier This is my paper
classification cs.CL
keywords learningmethodspromptapproachestransferdecompositiondirectencoder-based
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

This survey reviews prompt tuning, a parameter-efficient approach for adapting language models by prepending trainable continuous vectors while keeping the model frozen. We classify existing approaches into two categories: direct prompt learning and transfer learning. Direct prompt learning methods include: general optimization approaches, encoder-based methods, decomposition strategies, and mixture-of-experts frameworks. Transfer learning methods consist of: general transfer approaches, encoder-based methods, and decomposition strategies. For each method, we analyze method designs, innovations, insights, advantages, and disadvantages, with illustrative visualizations comparing different frameworks. We identify challenges in computational efficiency and training stability, and discuss future directions in improving training robustness and broadening application scope.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. AutoFed: Personalized Federated Traffic Prediction via Adaptive Prompt

    cs.LG 2025-12 unverdicted novelty 6.0

    AutoFed introduces a prompt-based personalized federated learning method that achieves superior traffic prediction performance on real-world datasets by sharing compact prompt matrices while preserving local specificity.

  2. DynaGraph: Lightweight Multi-Model Interaction Framework via Dynamic Topological Reconfiguration

    cs.MA 2026-05 unverdicted novelty 5.0

    DynaGraph is a multi-model framework that multiplexes PEFT adapters on a shared base model with evaluator-driven dynamic topology reconfiguration and hierarchical self-healing to achieve near-72B performance on reason...

  3. AutoFed: Personalized Federated Traffic Prediction via Adaptive Prompt

    cs.LG 2025-12 conditional novelty 5.0

    AutoFed generates a globally shared prompt from local traffic data via an autoencoder, graph encoder, and FedBN adapter, and uses it to condition a personalized AGCRN predictor, reporting SOTA results on demand and fl...