AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Composing Parameter-Efficient Modules with Arithmetic Operation

Jinghan Zhang, Shiqi Chen, Junteng Liu, Junxian He

NeurIPS 2023 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 5 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
merge two LoRA modules trained on different subsetsLoRA RoBERTa-base · GLUE RTEaccuracy: +5.2Improved
merge two (IA)3 modules trained on different subsets(IA)3 RoBERTa-base · GLUE RTEaccuracy: +4.0Improved
merge two LoRA modulesLoRA RoBERTa-base · GLUE SST-2accuracy: +0.2Improved
negate toxic LoRA (lambda=1) instead of toxic LoRAGPT-2 large · Civil Comments toxicitytoxicity score: 0.43 -> 0.01Improved
negate toxic (IA)3GPT-2 large · toxicitytoxic generation (%): 20.5 -> 0.9Improved
negate toxic LoRAGPT-2 large · WikiText-103perplexity: 16.44 -> 16.67Hurt
simply negate all LoRA parameters (ablation) (ablation)LoRA GPT-2 · toxicitytoxicity score: 0.43 -> 0.43No clear effect
simply negate all (IA)3 parameters (lambda=1) (ablation)(IA)3 GPT-2 · WikiText-103perplexity: 16.44 -> 843.83Hurt
merge LoRA on multi-task MNLI+RTELoRA T5-base · MNLI/RTEindividual task accuracyHurt
domain transfer by arithmetic merge for LoRALoRA T5-small · Yelp testaccuracy: 94.76 -> 95.83Improved
domain transfer by arithmetic merge for LoRALoRA T5-small · Amazon testaccuracy: 91.03 -> 91.94Improved
domain transfer by arithmetic merge for LoRALoRA T5-base · Amazon testaccuracy: 94.50 -> 94.50No clear effect

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