Fixed-SAE Track is introduced, a framework that trains one shared SAE per considered layer on activations pooled across the base model and all RL checkpoints, holding every feature direction fixed so that representation shifts are rigorously defined through the activations of interpretable SAE latents, including the de...
Ling-Heng Du, Yi-Ming Tang, Xufeng Duan et al.· 1 citation
T-SMART is presented, a neurosymbolic framework that separates these roles: a frozen LLM interprets questions and selects operations, deterministic tools perform numerical computation, and structured perception is invoked only when needed to provide a controlled framework for analyzing component contributions in neuros...
I. Delgado, Himansi Gupta, B. Khatri et al.· 0 citations
Predictive likelihood ratios that average over uncertain probability deficits and residual-tail distributions are constructed, showing robustness across the evaluated alternatives, not uniform power guarantees or resistance to arbitrary text edits.
Mechanistic interpretability has become the dominant lens for understanding frontier language models, as their inner workings are complex and inherently black boxes. To gain insights into these models, interpretable replacement networks (IRNs) are trained at all layers, exposing interpretable features through sparsely...
Tobias Ladner, Matthias Althoff· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Using food package images to monitor sodium and salt content against South Africa's R214 sodium limits is challenging when screening decisions require product identity, nutrition facts panel evidence, reporting basis, and category-specific thresholds. This study presents a conservative image-based workflow that combine...
Mayimunah Nagayi, A. Khan, Tamryn Frank et al.· 1 citation
This work investigates prompt echoing small language models from different families and shows that echoing prompts are likely to have partial overlap with the training dataset but the phenomenon is primarily driven by the model's induction heads.
Inez Okulska, B. Naskrȩcki, Jan Piotrowski et al.· 0 citations
This work proposes a simple yet effective adaptive sampling strategy that assigns trainable sampling probabilities to each language and employs an inner-outer optimization loop with a small overhead, leading to consistent performance improvements and, more importantly, distributing the gains across languages.
Quang Phuoc Nguyen, Félix Gaschi, David Anugraha et al.· 0 citations
This work presents TriCalRAG, a benchmark evaluating open-weight LLMs served locally via vLLM on a single high-memory workstation GPU against a classical LSTM-based log anomaly detector (DeepLog), across four real, publicly available log datasets (BGL, HDFS, Thun-derbird, OpenStack).
Rohit Patel, S. K. Mohanty, Jeenal Chaudhary· 0 citations
Results show the feasibility of delegating adaptive reasoning to the cloud while retaining execution control at the edge, and closed-loop repair improves application success from 0/6 to 6/6 over one-shot generation, validation gating prevents all three observed target-scale failures, and skill conditioning improves suc...
A corpus of 10,000 Bangla sentences, manually annotated into four functional categories, namely declarative, interrogative, imperative, and exclamatory is introduced, and Experimental results show that TF-IDF consistently outperforms Word2Vec, likely due to its ability to emphasize discriminative lexical cues associate...
Swapnil Kundu Argha, Abdullah Al Shafi, Rowzatul Zannat et al.· 0 citations
ReWeight is introduced, a framework that incorporates human data into VLA post-training through demonstration-level retrieval and sample-level weighting and learns a cross-embodiment visuomotor representation that combines visual observations with future actions to measure behavioral similarity between human and robot...
Chen-Wei Wang, Dian-Ye Huang, Match W. L. Ko et al.· 0 citations
Negation remains a longstanding challenge for both language models (LMs) and large language models (LLMs). Prior work mainly focuses on a small set of high-frequency single-word negation cues, such as not and never, with limited exploration of broader negation types and modern LLMs. To address this gap, we construct Ne...
