AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

ABBA-Adapters: Efficient and Expressive Fine-Tuning of Foundation Models

Raghav Singhal, Kaustubh Ponkshe, Rohit Vartak, Praneeth Vepakomma

ICLR 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use ABBA r=32 instead of LoRALlama-3.2 1B · CommonSense170K averageaccuracy: 71.49 -> 75.03Improved
use ABBA r=32 instead of HiRALlama-3.2 3B · CommonSense170K averageaccuracy: 82.56 -> 84.08Improved
use ABBA r=32 instead of LoRAMistral-7B · GSM8Kaccuracy: 61.94 -> 66.26Improved
use ABBA r=32 instead of full FTMistral-7B · GSM8K (20K)accuracy: 63.87 -> 66.26Improved
use ABBA instead of full FT with larger 40K dataMistral-7B · GSM8K (40K)accuracy: 66.28 -> 67.04No clear effect
use ABBA instead of LoRALlama-3.2 1B · HumanEvalPass@1: 20.73 -> 25.61Improved
use ABBA instead of HiRALlama-3.1 8B · HumanEvalPass@1: 48.17 -> 49.39Improved
initialize with SVD of W0 on one pair and LoRA init on the other instead of both-pairs sqrt(W0) SVD (ablation)ABBA · GSM8K (Mistral-7B)accuracy: 64.53 -> 66.26Improved
use SVD-plus-LoRA init instead of SVD-plus-ones init (ablation)ABBA · GSM8K (Mistral-7B)accuracy: 57.39 -> 66.26Improved
use four chained adapter pairs instead of two (ablation)ABBA · GSM8K (Mistral-7B)accuracy: 66.26 -> 64.84Hurt
raise alpha from 32 to 64 (ablation)ABBA · CommonSense170K average (Llama-3.2 3B)accuracy: 84.08 -> 76.36Hurt
raise alpha from 4 to 16 (ablation)ABBA · CommonSense170K average (Llama-3.2 3B)accuracy: 82.43 -> 84.04Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy