Pith. sign in

REVIEW 21 cited by

AdapterFusion: Non-Destructive Task Composition for Transfer Learning

Not yet reviewed by Pith; the record is open.

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 2005.00247 v3 pith:MNNWJO3Y submitted 2020-05-01 cs.CL

classification cs.CL
keywords knowledgelearningtasksadapterfusionadapterscompositionmultipleeffectively
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Sequential fine-tuning and multi-task learning are methods aiming to incorporate knowledge from multiple tasks; however, they suffer from catastrophic forgetting and difficulties in dataset balancing. To address these shortcomings, we propose AdapterFusion, a new two stage learning algorithm that leverages knowledge from multiple tasks. First, in the knowledge extraction stage we learn task specific parameters called adapters, that encapsulate the task-specific information. We then combine the adapters in a separate knowledge composition step. We show that by separating the two stages, i.e., knowledge extraction and knowledge composition, the classifier can effectively exploit the representations learned from multiple tasks in a non-destructive manner. We empirically evaluate AdapterFusion on 16 diverse NLU tasks, and find that it effectively combines various types of knowledge at different layers of the model. We show that our approach outperforms traditional strategies such as full fine-tuning as well as multi-task learning. Our code and adapters are available at AdapterHub.ml.

Discussion (0). Sign in to comment.

Forward citations

Cited by 21 Pith papers

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

  1. FlexOlmo: Open Language Models for Flexible Data Use

    cs.CL 2025-07 conditional novelty 7.0 of 10

    FlexOlmo merges independently trained language-model experts, trained on private data, into a single mixture-of-experts model without joint training.

  2. Test-Time Scaling via Error Localization

    cs.LG 2026-07 conditional novelty 6.0 of 10

    TTEL uses feedback-induced token probability drops to localize the first error in a failed reasoning trace and branch a new generation from that prefix, improving pass@k per token on coding and math benchmarks.

  3. CrossEarth-Gate: Fisher-Guided Adaptive Tuning Engine for Efficient Adaptation of Cross-Domain Remote Sensing Semantic Segmentation

    cs.CV 2025-11 conditional novelty 6.0 of 10

    A Fisher-information-guided dynamic selection over a toolbox of LoRA, adapter, and frequency-adapter modules improves cross-domain remote sensing segmentation over static PEFT methods.

  4. DivMerge: A divergence-based model merging method for multi-tasking

    cs.LG 2025-09 conditional novelty 6.0 of 10

    DivMerge learns task-arithmetic merging weights by minimizing Jensen-Shannon divergence between each specialist model and the merged model, improving multi-task performance and scalability.

  5. Time Series Foundation Models for Multivariate Financial Time Series Forecasting

    q-fin.GN 2025-07 reject novelty 6.0 of 10

    Pretrained TTM shows large transfer and sample-efficiency gains in three financial forecasting tasks relative to training from scratch, but methodological flaws including possible look-ahead bias weaken the quantitati...

  6. CaptionSmiths: Flexibly Controlling Language Pattern in Image Captioning

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A single LLaVA-based captioning model continuously controls caption length, descriptiveness, and word uniqueness by interpolating between learned endpoint conditioning vectors.

  7. Video-Guided Text-to-Music Generation Using Public Domain Movie Collections

    cs.SD 2025-06 conditional novelty 6.0 of 10

    OSSL is the first self-hosted, mood-annotated video-music dataset, and a video adapter on MusicGen-Medium improves film music generation over text-only baselines.

  8. GaRA-SAM: Robustifying Segment Anything Model with Gated-Rank Adaptation

    cs.CV 2025-06 conditional novelty 6.0 of 10

    GaRA-SAM improves SAM's robustness to image corruption by using input-dependent gating to adjust the effective rank of low-rank adapters, beating prior methods on robust segmentation benchmarks.

  9. SOS-LoRA: Static Orthogonal-Subspace Low-Rank Adaptation with Fixed Multi-Scale Scaling

    cs.LG 2026-06 conditional novelty 5.0 of 10

    A LoRA update split into several fixed, differently-scaled low-rank experts with orthogonal input directions improves fine-tuning accuracy at the same parameter count.

  10. Sensitivity-LoRA: Low-Load Sensitivity-Based Fine-Tuning for Large Language Models

    cs.LG 2025-09 conditional novelty 5.0 of 10

    Sensitivity-LoRA allocates LoRA ranks across layers using Hessian-based sensitivity metrics, improving average GLUE score by 0.74 over AdaLoRA on RoBERTa-base.

  11. Tensorized Clustered LoRA Merging for Multi-Task Interference

    cs.LG 2025-08 unverdicted novelty 5.0 of 10

    Clustering training data by embedding similarity and jointly CP-decomposing LoRA adapters cuts multi-task merging interference: +1.4% on Phi-3 and +2.3% on Mistral-7B over SVD baselines.

  12. Few-Shot Object Detection via Spatial-Channel State Space Model

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A Mamba-based channel sequence model combined with spatial attention improves few-shot object detection on VOC and COCO.

  13. Exploring Adapter Design Tradeoffs for Low Resource Music Generation

    cs.SD 2025-06 conditional novelty 5.0 of 10

    Adapter placement, architecture, and size strongly change generation quality and cost for MusicGen and Mustango on two non-Western genres, with late-layer, mid-sized (40M) adapters reported as the best tradeoff.

  14. PrunePEFT: Iterative Hybrid Pruning for Parameter-Efficient Fine-tuning of LLMs

    cs.LG 2025-06 conditional novelty 5.0 of 10

    An iterative hybrid pruning method selects which PEFT modules to keep at each transformer layer, matching or improving fixed PEFT baselines on GLUE at 1% trainable parameters.

  15. TALL -- A Trainable Architecture for Enhancing LLM Performance in Low-Resource Languages

    cs.CL 2025-06 reject novelty 5.0 of 10

    A trainable pipeline of translation models and a frozen LLM improves Hebrew last-word prediction accuracy to 5.59%, about twice the best baseline.

  16. PFMBench: Protein Foundation Model Benchmark

    q-bio.BM 2025-06 conditional novelty 5.0 of 10

    A comprehensive benchmark of 17 protein foundation models across 38 tasks yields task correlations, a streamlined protocol, and identifies ProTrek as the strongest general performer.

  17. LARA: Lightweight Adapters in the Residual Stream for Composable Adaptation and Alignment

    cs.LG 2026-07 conditional novelty 4.0 of 10

    Low-rank residual-stream adapters match LoRA at near-equal parameter counts on code fine-tuning and DPO, and support inference-time scaling plus per-token routing of seven behaviors on one frozen 1.5B model.

  18. GEM: A Scale-Aware and Distribution-Sensitive Sparse Fine-Tuning Framework for Effective Downstream Adaptation

    cs.LG 2025-08 conditional novelty 4.0 of 10

    GEM selects fine-tuning parameters by gradient-to-weight ratio and distributes the budget by layer entropy, reaching 0.1% parameter updates with small accuracy gains on several NLP tasks.

  19. Weight Spectra Induced Efficient Model Adaptation

    cs.LG 2025-05 reject novelty 4.0 of 10

    Fine-tuning mostly amplifies and reorients the top singular directions of weight matrices, and SpecLoRA learns to rescale a top-left block plus LoRA to improve PEFT performance.

  20. MAP: Revisiting Weight Decomposition for Low-Rank Adaptation

    cs.LG 2025-05 conditional novelty 4.0 of 10

    MAP decouples a weight matrix's direction and magnitude by normalizing the whole matrix and the low-rank update by their Frobenius norms and scaling each with a learnable scalar.

  21. Structure-Learnable Adapter Fine-Tuning for Parameter-Efficient Large Language Models

    cs.CL 2025-09 reject novelty 2.0 of 10

    A gated-adapter method with a sparsity regularizer is reported to match or slightly beat full fine-tuning on MNLI and BoolQ using 1.4% of parameters, but lacks the experimental detail needed to verify the claim.

Pith tools