6 papers
ARCHead: Activation-Metric Residual Correction for Large Language Model Output Heads
Şuayp Talha Kocabay, Talha Rüzgar Akkuş, Kamer Ali Yuksel
Weight-only quantization substantially reduces the storage of large language model (LLM) transformer blocks, but practical backends often retain the final language-modeling head (L…
Optimism as a Vulnerability: Deceptive Stackelberg Control of UCB Bandit Followers
Şuayp Talha Kocabay, Kerem Yalçın, Talha Rüzgar Akkuş
Upper Confidence Bound (UCB) algorithms guarantee sublinear regret for agents learning unknown stochastic environments, yet the same principle that makes them statistically efficie…
Sample Complexity of Scientific Discovery: PAC Learnability of Compositional Function Trees
Şuayp Talha Kocabay, Talha Rüzgar Akkuş, Kerem Yalçın
Scientific discovery via symbolic regression is often viewed as statistically and computationally intractable because the hypothesis space of expressions grows combinatorially with…
Selectivity and Shape in the Design of Forward-Forward Goodness Functions
Talha Ruzgar Akkus, Suayp Talha Kocabay, Kamer Ali Yuksel +1
The Forward-Forward (FF) algorithm trains networks layer-by-layer using a local "goodness function," yet sum-of-squares (SoS) has remained the only choice studied. We systematicall…
Diffutron: A Masked Diffusion Language Model for Turkish Language
Şuayp Talha Kocabay, Talha Rüzgar Akkuş
Masked Diffusion Language Models (MDLMs) have emerged as a compelling non-autoregressive alternative to standard large language models; however, their application to morphologicall…
Enhancing Human-Like Responses in Large Language Models
Ethem Yağız Çalık, Talha Rüzgar Akkuş
This paper explores the advancements in making large language models (LLMs) more human-like. We focus on techniques that enhance natural language understanding, conversational cohe…