The paper investigates which parts of a low‑rank adaptation (LoRA) update actually drive behavior in language models. It shows that behaviorally effective LoRA writes are sparse, structured, and highly concentrated, with only a few module‑wise components carrying most of the signal. Using a learned‑basis continuation recipe, the authors demonstrate that these sparse, structured writes can be recovered with minimal error and that they remain robust across different training continuations and evaluation tasks.
By Haruto Sato, Yuki Tanaka, Ren Nakamura, Aoi Kobayashi, Mei Ito
Federated fine-tuning of large pre-trained models increasingly relies on Low-Rank Adaptation (LoRA) to reduce communication and computation, but heterogeneous clients can make adapter aggregation unstable. We identify the data-parameter interference as a geometric source of this instability.
arXiv:2607. 14367v1 Announce Type: new Abstract: Federated fine-tuning of large pre-trained models increasingly relies on Low-Rank Adaptation (LoRA) to reduce communication and computation, but heterogeneous clients can make adapter aggregation unstable.
By Haobo Zhang, Jiankun Wang, Suraj Rajendran, Weishen Pan, Lam Tsoi, Yong Chen, Fei Wang, Jiayu Zhou
The paper introduces READ, a method for composing low‑rank adapters (LoRA) in large language models. By rewriting each adapter into a balanced canonical form and enforcing a one‑directional coupling, READ allows new skills to read but never write into the output subspaces of existing skills, eliminating interference. Experiments on four benchmark suites and two model families show that READ consistently outperforms existing baselines, improving SuperGLUE scores by over twenty points and domain suite scores by more than seven points.
By Zeyan Li, Panqi Yang, Qirong Guo, Shengda Zhuo, SIyuan Qiu, Hu Xu, Chun Li, Jianfeng Xu
Delta N-gram introduces a forward‑only write path into a language model’s native hashed N‑gram table, preserving the pretrained table while adding evidence through a sparse patch Δ. The method uses a 4.1 M‑parameter writer that emits gated targets per event and applies a block‑averaged delta rule to keep each row a convex combination of its old value and the new targets. Experiments on a 0.6 B prototype and Qwen3.8‑Flash‑Next show that the delta writer maintains more document‑specific memory and reduces drift compared to online gradient writes, achieving stable performance over thousands of writes.
whyItMatters":"The approach allows continual updates to a language model’s memory without altering its pretrained parameters, enabling efficient, stable, and document‑specific knowledge retention during inference."
By Hao Wu
arXiv:2610.01133v1 Announce Type: cross
Abstract: Scaling reasoning typically spends more compute on reinforcement learning (RL) or on inference. We show that a completed RL training history can yiel...
By Bangji Yang, Jiajun Fan, Hongba Ma, Ruihan Guo, Ge Liu
arXiv:2601. 13020v2 Announce Type: replace-cross Abstract: Continual instruction tuning (CIT) requires multimodal large language models (MLLMs) to adapt to a stream of tasks without forgetting prior capabilities.
By Zhiyan Hou, Haiyun Guo, Haokai Ma, Yandu Sun, Yonghui Yang, Jinqiao Wang
arXiv:2605. 07284v2 Announce Type: replace Abstract: A late-layer change learned during post-training may work on the base model's earlier state, or it may depend on earlier computation learned with it.
By Yifan Zhou
The paper introduces a Fisher-guided submodular data selection method for continual pre‑training of large language models, addressing the forgetting problem that arises when a target‑domain corpus overwrites pretrained knowledge. By decomposing candidate gradients into anchor and frontier components and optimizing a log‑determinant submodular objective, the method selects data that both improves target‑domain performance and limits forgetting. Experiments on TinyLlama‑1.1B and Llama‑3.1‑8B show that the selector achieves superior adaptation and forgetting control while using ten times fewer tokens than traditional replay strategies.
By Zhenghao Zhao, Gaowen Liu, Zhiling Lan, Yan Yan
arXiv:2606. 13767v1 Announce Type: cross Abstract: Low-rank adaptation (LoRA) and its variants provide a memory- and compute-efficient alternative to full fine-tuning of pre-trained models.
By Elijah Cadenhead, Cristian McGee, Xin Li, El Houcine Bergou, Aritra Dutta
The paper investigates how pretraining and alignment stages of transformer training affect weight updates in different matrix groups. By examining the residual stream’s shared coordinate system, it finds that alignment updates to query and key matrices align with the principal input subspace, whereas updates to output matrices remain largely random relative to the prediction subspace. This pattern persists across objectives and training stages, indicating that the geometry of updates is determined by the training phase rather than the specific objective.
By Valeria Ruscio, Keiran Thompson
The paper introduces Activation Boundary Matching for Low‑Rank Adaptation (ABM‑LoRA), a task‑informed initialization strategy that uses the signs of layer‑wise pre‑activations from a brief probe adapter as targets for a fresh adapter. By training with a margin‑based hinge objective on these activation boundaries, ABM‑LoRA captures useful adaptation directions that standard LoRA initializers miss, while requiring only a few forward passes. Experiments show that ABM‑LoRA outperforms or matches existing LoRA, SVD, and gradient‑based initializers across multiple models and benchmarks, including T5‑base/GLUE, ConvNeXt‑T, Swin‑T, Qwen2.5‑1.5B, and LLaMA2‑7B.
By Dongha Lee, Jinhee Park, Minjun Kim, Junseok Kwon