AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Fine-Tuning of Transformer models with Frames

Harshavardhan Adepu, Li Zhang, Sanjiv Kumar, Vikas Singh

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 0 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 per-layer random mask instead of shared random mask (ablation)FrameFT · GLUE avg (RoBERTa-base)avg score: 86.18 -> 86.26No clear effect
use gradient-based sparsity mask instead of random (ablation)FrameFT · GLUE avg (RoBERTa-base)avg score: 86.18 -> 74.83Hurt
use magnitude-based mask instead of random (ablation)FrameFT · GLUE avg (RoBERTa-base)avg score: 86.18 -> 85.60Hurt
increase number of non-zero coefficients (ablation)FrameFT · GLUE (RoBERTa-base)scoreImproved
increase subspace dimension rho from 2 to 256 (ablation)FrameFT · CoLA and MRPC (RoBERTa-base)scoreNo clear effect
vary scaling factor alpha from 10 to 600 (ablation)FrameFT · Alpaca Llama-2-7Bavg accuracy: 62.97 to 63.62No clear effect
replace LoRA with FrameFT (1K coefficients)Llama-2-7B adapter · inference throughputtokens/sec: 24.4k -> 37.9kImproved
replace LoRA with FrameFTLlama-2-7B · Alpaca training timeminutes: 238 -> 225Improved
replace LoRA with FrameFTViT-B · image classification (8 datasets) avgavg accuracy: 77.58 -> 85.80Improved
replace LoRA with FrameFTRoBERTa-base · GLUE avgavg score: 85.2 -> 86.1Improved
replace LoRA with FrameFTLlama-2-13B · Alpaca LM-eval avgavg accuracy: 66.42 -> 66.11Hurt
replace LoRA with FrameFTLlama-3.1-8B · Alpaca LM-eval avgavg accuracy: 68.56 -> 67.87Hurt

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