The paper introduces SPD (Single Forward Pass), a decoding strategy that efficiently produces ranked lists from large language models by emitting only the ordinal positions of items. SPD extracts an item‑position score matrix from the model’s hidden states, then solves a bipartite assignment problem via the Hungarian algorithm to generate a valid permutation in a single forward pass. Experiments show that LoRA‑based fine‑tuning combined with autoregressive ranking distillation achieves 28 ms end‑to‑end inference, a 64× speed‑up while preserving ranking quality.
By Emil Laftchiev, Prachi Agrawal, Moe Kayali, Bixing Yan, Qi Xu, Zijie Lei, Chen Qiu, Zhi Hua, Ke Li, Luke Simon
arXiv:2510. 00192v3 Announce Type: replace Abstract: Low-rank adaptation (LoRA) has become a widely used paradigm for parameter-efficient fine-tuning of large language models, yet its representational capacity often lags behind full fine-tuning.
By Xin Yu, Cong Xie, Xunmei Liu, Tiantian Fan, Lingzhou Xue, Zhi Zhang
arXiv:2607. 01170v1 Announce Type: cross Abstract: Generative reasoning re-rankers achieve strong recommendation accuracy by emitting a chain-of-thought before re-ordering a candidate list, but they are slow at inference: an autoregressive (AR) decoder spends one sequential forward pass per reasoning token, and the reasoning trace far exceeds the ranking it produces.
By Zhuoxuan Zhang (Yang), Kangqi Ni (Yang), Yuhang Chen (Yang), Mingfu Liang (Yang), Xiaohan Wei (Yang), Yunchen Pu (Yang), Fei Tian (Yang), Chonglin Sun (Yang), Frank Shyu (Yang), Adam (Yang), Song, Sandeep Pandey, Luke Simon, Tianlong Chen, Xi Liu
arXiv:2609.05885v1 Announce Type: new
Abstract: Low-rank adaptation (LoRA) has become the standard for parameter-efficient fine-tuning of large language models. Most LoRA variants follow a uniform-LR...
By Huiyi Wang, Daijiao Liu, Lina Yao, Dong Gong
ChainDoRA is a new parameter‑efficient fine‑tuning framework for large language models that replaces the dense low‑rank factorization of LoRA with a connected Tensor‑Train (TT) chain. By separating weight magnitude and direction and using a TT rank to control representation capacity, ChainDoRA achieves higher average accuracy on seven commonsense reasoning benchmarks while dramatically reducing trainable parameters—down to 5.35 M versus 56 M for LoRA and DoRA. Ablation studies show that the TT parameterization offers controllable trade‑offs between parameter cost and accuracy.
By Ashfak Yeafi, Mehedi Hasan, Md Khairul Islam
arXiv:2607. 22251v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) is a widely used parameter-efficient fine-tuning method for large language models, but its performance depends strongly on how a fixed rank budget is distributed across Transformer modules.
By Wei Zhang, Xinwu Liu, Yihang Cheng
arXiv:2410.02343v2 Announce Type: replace
Abstract: Large language models (LLMs) routinely fail to output the correct option in multiple-choice question answering (MCQA) while encoding the answer int...
By Eduard Tulchinskii, Kristian Kuznetsov, Laida Kushnareva, Anastasia Voznyuk, Andrei Andriiainen, Irina Piontkovskaya, Evgeny Burnaev, Serguei Barannikov
arXiv:2601. 16991v3 Announce Type: replace-cross Abstract: Adapting large pre-trained language models to downstream tasks often entails fine-tuning millions of parameters or deploying costly dense weight updates, which hinders their use in resource-constrained environments.
By Longteng Zhang, Sen Wu, Shuai Hou, Zhengyu Qing, Zhuo Zheng, Danning Ke, Qihong Lin, Qiang Wang, Shaohuai Shi, Xiaowen Chu
arXiv:2608. 06111v1 Announce Type: cross Abstract: Positional embeddings (PE) in Transformers encode token distance and order but are largely agnostic to \textit{syntactic structure}.
By Haris Riaz, Hyungji Kim, Mihai Surdeanu
The paper investigates how temporal aggregation and ranking preservation affect aggressive decoding-time KV cache eviction. It shows that exponential‑moving‑average (EMA) aggregation makes many scorer modifications indistinguishable at the eviction‑set level, while certain variants (KeyDiff, key norm, recency, learned scorer) alter rankings and degrade performance. Building on this, the authors propose InertiaKV and its lazy variant, achieving 1.34–1.46× faster decoding, and also explore a score‑free decoding approach that slightly improves quality while eliminating further scoring.
By Bo Zeng, Yu Zhao, Yefeng Liu, Zhihong Lu, Xuanfan Ni, Xintong Wang
LOCUS is a post‑training technique that selects a task‑aware low‑rank adaptation subspace to reduce token‑cost while preserving utility. By updating only a tiny fraction of parameters (0.24–0.28 %) on 3 B‑parameter backbones, LOCUS cuts continuation length by up to 39.84 % on Pythia‑2.8B and 14.87–17.58 % on Qwen2.5‑3B, without materially affecting the internal preference diagnostic.
DeGRe is a dense‑supervised generative reranking framework designed to improve multi‑stage recommender systems by addressing label bias and credit assignment issues. It uses an offline Lookahead Evaluator with beam search to generate dense supervision signals, which are distilled into a lightweight Online Generator that can perform efficient greedy decoding at inference time. Experiments show that DeGRe outperforms baselines on public benchmarks and industrial datasets, and it has been successfully deployed on Taobao Flash Shopping to enhance online recommendations.
By Chaotian Song, Jingyao Zhang, Chenghao Chen, Zisen Sang, Dehai Zhao, Guodong Cao, Boxi Wu, Deng Cai, Jia Jia