Showing cs.LGShow all
3 papers · 1 filter
cs.LG2026
HPO: Hysteretic Policy Optimization for Stable and Efficient Training under Sparse-Reward Regime
Mohamed Sana, Nicola Piovesan, Antonio De Domenico +2
We investigate a narrow but common failure mode of GRPO-style reinforcement learning in the context of sparse verifiable rewards: early updates contain more responses with negative…
cs.LG2025
KVCompose: Efficient Structured KV Cache Compression with Composite Tokens
Dmitry Akulov, Mohamed Sana, Antonio De Domenico +3
Large language models (LLMs) rely on key-value (KV) caches for efficient autoregressive decoding; however, cache size grows linearly with context length and model depth, becoming a…
cs.LG2025
Goal-Oriented Time-Series Forecasting: Foundation Framework Design
Luca-Andrei Fechete, Mohamed Sana, Fadhel Ayed +4
Conventional time-series forecasting methods typically aim to minimize overall prediction error, without accounting for the varying importance of different forecast ranges in downs…