The paper investigates the components of group attention used in the Chronos-2 time‑series forecasting model, which attends over variates of a group for both multivariate (MV) and in‑context learning (ICL) forecasting. By editing the attention matrix at inference, the authors separate the head into a V/O pathway that projects a weighted summary of the group and a Q/K pathway that determines the weights. They find that uniform pooling (V/O without Q/K weighting) improves performance in most sensor‑network configurations, while learned weighting helps MV but degrades ICL in many cases; applying uniform weighting in the first block alone benefits all tested ICL settings.
arXiv:2609.36448v1 Announce Type: new Abstract: Transformers have demonstrated remarkable in-context learning (ICL) capabilities, enabling them to perform new tasks without additional fine-tuning. Ho...
The paper investigates post‑training quantization of transformer attention blocks by optimizing a joint loss over the Q, K, V projections rather than individual weight matrices. Using this joint attention‑based objective (JAB), the authors achieve significant compression on Mistral‑7B, recovering 77‑90% of the performance gap at 3 bits, but the method fails when MLP layers are included. A role‑aware offset rule that ignores sensitivity estimates outperforms JAB on GPT‑2 and full Mistral‑7B, demonstrating that the matrix a weight belongs to is more critical than sensitivity metrics.
arXiv:2605. 13181v2 Announce Type: replace-cross Abstract: Precipitation nowcasting remains challenging due to the highly localized, rapidly evolving, and heterogeneous nature of atmospheric dynamics.
The paper introduces Declarative Attention (DA), a protocol that lets language models explicitly declare which parts of their context to focus on during generation. By partitioning decoding into full-context, region-specific, and recent-output-only modes, the inference engine can skip large portions of the KV cache, dramatically reducing attended tokens. Experiments on 15 long-context tasks with off-the-shelf models show significant savings (52.0% and 31.1% reductions) with only modest accuracy drops that diminish as model size increases.
arXiv:2607. 18363v1 Announce Type: cross Abstract: Feed-forward networks hold two thirds of a transformer's non-embedding parameters, yet the architecture has not received a necessity test that controls parameters, compute, and depth at once.
arXiv:2609.01231v1 Announce Type: new Abstract: We prove that a multi-head scaled dot product attention can be viewed as a parameter identification strategy. The ratio of unidentified parameters to t...
arXiv:2607. 21353v1 Announce Type: new Abstract: Machine unlearning aims to remove the influence of specific training data while preserving model utility.
arXiv:2405. 16440v2 Announce Type: replace-cross Abstract: In recent years, Transformers have become the de-facto architecture for long-term time series forecasting (LTSF), yet they face challenges associated with the self-attention mechanism, including quadratic complexity and permutation-invariant bias.
arXiv:2606. 27791v1 Announce Type: cross Abstract: Hybrid attention models that mix full and sliding-window attention across layers offer a promising approach to efficient long-context inference, but the critical question of \emph{which layers} should retain full attention remains unsolved.
The paper investigates the training dynamics of attention mechanisms in high-dimensional settings, focusing on attention-indexed models that encompass multi-layer and multi-head architectures. It shows that while the loss landscape can be described by a finite set of trace order parameters, the online stochastic gradient descent dynamics involve an infinite hierarchy of matrix moments that can be accurately approximated by a finite truncated system. The study further reveals that the choice of attention parameterization acts as an implicit bias: untied attention can get trapped in uninformative states, whereas tied attention induces symmetry breaking and enables weak recovery with θ(d² log d) samples, and untied attention exhibits a fast-slow dynamic leading to weak recovery when symmetry is broken.
Machine unlearning aims to remove the influence of specific training data while preserving model utility. Many state-of-the-art approaches pursue this goal by restricting the forgetting update to a subset of parameters selected through gradient-based saliency.
The paper reports that chain‑of‑thought (CoT) supervised fine‑tuning (SFT) improves reasoning but systematically harms long‑context recall in hybrid linear‑attention models such as HypeNet and Jet‑Nemotron. Retrieval performance on the Needle‑In‑A‑Haystack benchmark drops dramatically after CoT‑SFT, especially with harder settings and longer contexts. The authors introduce QK‑Restore, a training‑free method that reinstates the query‑key projection matrices from the pre‑SFT checkpoint, which recovers long‑range recall while preserving reasoning gains.