The paper introduces PTH (Probe The Harness), a set of checks designed to expose hidden details in experimental setups that can alter the ranking of stale-data reinforcement learning methods for language models. By applying PTH to a comparison between SAN and truncated importance sampling (TIS), the authors demonstrate that subtle harness configurations—such as how PPO ratios are computed, data seeding, replay queue reuse, and loss normalisation—can reverse the observed performance order. The study provides a detailed signature of each influencing factor, reference results for TIS and uncorrected GRPO, and a checklist to ensure fair comparisons.
By Taiheng Pan
arXiv:2605.24155v2 Announce Type: replace-cross
Abstract: Effective skills-aware talent recommendation must balance behavioral transition patterns, trajectory-sensitive adaptation, and inspectable oc...
By \"Ozkan Canay
arXiv:2608.29179v1 Announce Type: cross
Abstract: Generative recommendation encodes items as hierarchical semantic identifiers (SIDs) and retrieves the next item through autoregressive decoding. Stan...
By Lianjie Li, Zhiying Tu, Dianhui Chu, Hongliang Sun
UpgradeBench is a decision‑centric longitudinal benchmark that evaluates how fine‑tuned language‑model specialists should be handled when new base‑model releases occur. It covers four consecutive Qwen releases, a continuation checkpoint, six tasks, two model sizes, and OLMo checkpoints with known training lineage, and examines whether retraining, adapter transfer, or other recovery strategies improve specialist performance. The benchmark reveals that upgrade gains vary by task and release interval, that direct adapter copying is sensitive to pretraining distance, and that teacher relabeling can recover specialists without new annotations.
"whyItMatters":"The study provides actionable insights into the cost‑effective management of specialist models across model releases, showing how to balance retraining effort with performance gains."
By Ye Chen, Weining Zhang
arXiv:2604. 02183v3 Announce Type: replace Abstract: Multimodal recommendation systems (MRS) jointly model user-item interaction graphs and rich item content, but this tight coupling makes user data difficult to remove once learned.
By Zhanting Zhou, KaHou Tam, Ziqiang Zheng, Zeyu Ma, Yang Yang
arXiv:2608.24079v1 Announce Type: cross
Abstract: A shared search-and-recommendation index must score new items from features alone because search has no exploration slot. In a public log covering bo...
By Theodore Rogers, Joe Standerfer, Dmitrii Timoshenko, Haoxue Li, Zuhaib Akhtar, Soyoung Yang
The paper examines LLM-based recommendation rerankers that are often evaluated under an oracle protocol, which guarantees the ground-truth item is present in the scored set. Across Amazon datasets, this protocol overestimates realistic NDCG@10 by 92–95% because realistic retrieval only covers 2–19% of relevant items at K=100, creating a recall ceiling that limits any closed-candidate reranker's top‑k NDCG. The authors find that various optimisation strategies—including prompt engineering, model scaling, sequential models, supervised neural rerankers, LoRA fine‑tuning, hybrid retrieval, score‑aware prompting, and LLM+CF fusion—do not significantly improve over a collaborative‑filtering baseline under realistic retrieval, and they propose a Recall‑Aware Evaluation Protocol (RAEP) to better assess rerankers in low‑recall regimes.
By Zhaohui Wang
The paper introduces TrustSwap, a counterfactual test that swaps or removes source reliability labels while keeping evidence text constant, to evaluate how retrieval‑augmented fact‑checking models respond across verdict, confidence, and search decisions. Experiments on untrained and RL‑trained models show that confidence and search largely follow labels, yet label changes can flip a significant portion of verdicts, especially in larger models. The authors propose trust‑swap augmentation (TSA) to mitigate this shortcut, demonstrating reduced verdict flip rates and maintained accuracy in several settings, though its effectiveness diminishes at larger model scales.
By Jianchang Su, Yiwei Yang, Wei Zhang
The paper investigates the trade‑off of using a shared search‑and‑recommendation index that scores new items purely from features, thereby keeping the index open to unseen items. Experiments on public logs show that a feature‑based tower can match warm‑item performance (Recall@20 0.9595 vs 0.9510) and a lexical baseline, while a full‑catalog check is inconclusive. The study also quantifies the cost of this openness on recommendation quality across several baselines, revealing that exact full‑softmax training improves recall but is impractical at catalog scale.
By Theodore Rogers, Joe Standerfer, Dmitrii Timoshenko, Haoxue Li, Zuhaib Akhtar, Soyoung Yang
arXiv:2610.31786v1 Announce Type: new
Abstract: Federated unlearning (FU) aims to remove a designated target client's influence from a trained model without the prohibitive cost of retraining from sc...
By Thanh Linh Nguyen, Quoc-Viet Pham
The paper introduces REPAIR, a method that corrects lossy user preference states in personalization encoders by comparing cached representations with the current state in a learned coordinate space. REPAIR selectively aggregates corrective evidence from past interactions and adds it to the state before the task head, enabling encoder–host repair without re‑encoding history. Experiments on MovieLens, PENS, MIND, and Amazon Reviews 2023 show that training only REPAIR improves MRR and nDCG@10 across all twelve recommendation hosts, while head‑only finetuning yields smaller gains.
By Parthiv Chatterjee, Dhiraj Golhar, Ummesalma Diwan, Sourish Dasgupta, Manjunath Joshi, Tanmoy Chakraborty
arXiv:2607. 07719v1 Announce Type: cross Abstract: Parameter-efficient fine-tuning adapts a large language model to one task cheaply, but across a task sequence LoRA-style methods keep stacking low-rank updates on the same frozen weight, so each new task tends to overwrite the previous ones.
By Wentao Lu