Hasan Abed Al Kader Hammoud

TAPS: Task Aware Proposal Distributions for Speculative Sampling

Mohamad Zbib, Mohamad Bazzi, Ammar Mohanna, Hasan Abed Al Kader Hammoud et autres

Speculative decoding accelerates autoregressive generation by letting a lightweight draft model propose future tokens that a larger target model then verifies in parallel. In practice, however, draft models are usually trained on broad generic corpora, which leaves it unclear how much speculative …

TAPS: Task Aware Proposal Distributions for Speculative Sampling

Mohamad Zbib, Mohamad Bazzi, Ammar Mohanna, Hasan Abed Al Kader Hammoud et autres

Speculative decoding accelerates autoregressive generation by letting a lightweight draft model propose future tokens that a larger target model then verifies in parallel. In practice, however, draft models are usually trained on broad generic corpora, which leaves it unclear how much speculative …

sa, lb (code pays fourni par la source)

QuanBench+: A Unified Multi-Framework Benchmark for LLM-Based Quantum Code Generation

Ali Slim, Haydar Hamieh, Jawad Kotaich, Yehya Ghosn et autres

Large Language Models (LLMs) are increasingly used for code generation, yet quantum code generation is still evaluated mostly within single frameworks, making it difficult to separate quantum reasoning from framework familiarity. We introduce QuanBench+, a unified benchmark spanning Qiskit, PennyLane, and Cirq, …

QuanBench+: A Unified Multi-Framework Benchmark for LLM-Based Quantum Code Generation

Ali Slim, Haydar Hamieh, Jawad Kotaich, Yehya Ghosn et autres

Large Language Models (LLMs) are increasingly used for code generation, yet quantum code generation is still evaluated mostly within single frameworks, making it difficult to separate quantum reasoning from framework familiarity. We introduce QuanBench+, a unified benchmark spanning Qiskit, PennyLane, and Cirq, …

lb, sa (code pays fourni par la source)

Hala Technical Report Building Arabic-Centric Instruction & Translation Models at Scale

Association for Computational Linguistics 2026, Bernard Ghanem, Hasan Abed Al Kader Hammoud, Mohamad Zbib

We present HALA, a family of Arabic-centric instruction and translation models built with our translate-and-tune pipeline. We first compress a strong AR↔EN teacher to FP8 (yielding ~2× higher throughput with no quality loss) and use it to create high-fidelity bilingual supervision. A …

AraLingBench: A Human-Annotated Benchmark for Evaluating Arabic Linguistic Capabilities of Large Language Models

Association for Computational Linguistics 2026, Bernard Ghanem, Hasan Abed Al Kader Hammoud, Fatima Karnib et autres

We present AraLingBench, a fully human annotated benchmark for evaluating the Arabic linguistic competence of large language mod- els (LLMs). The benchmark spans five core categories: grammar, morphology, spelling, reading comprehension, and syntax, through 150 expert designed multiple choice questions that directly …

Hala Technical Report Building Arabic-Centric Instruction & Translation Models at Scale

Hasan Abed Al Kader Hammoud, Mohamad Bilal Zbib, Bernard Ghanem

We present HALA, a family of Arabic-centric instruction and translation models built with our translate-and-tune pipeline.We first compress a strong AR↔EN teacher to FP8 (yielding ∼2× higher throughput with no quality loss) and use it to create high-fidelity bilingual supervision.A lightweight language …

sa (code pays fourni par la source)

AraLingBench: A Human-Annotated Benchmark for Evaluating Arabic Linguistic Capabilities of Large Language Models

Mohamad Bilal Zbib, Hasan Abed Al Kader Hammoud, Ammar Mohanna, Nadine Rizk et autres

Mohamad Bilal Zbib, Hasan Abed Al Kader Hammoud, Ammar Mohanna, Nadine Rizk, Fatima Karnib, Sina Moukaled, Bernard Ghanem. Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script. 2026.

sa, lb (code pays fourni par la source)

The Three Pillars of Practical Machine Learning

Hasan Abed Al Kader Hammoud

The proliferation of large scale foundation models has marked a new era in machine learning, but practical deployment remains limited by adaptation, resource, and reliability challenges. This thesis argues that practical machine learning rests on three pillars: Adaptation, the ability to learn …

sa (code pays fourni par la source)

Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning

Lama Alssum, Hani Itani, Hasan Abed Al Kader Hammoud, Philip H. S. Torr et autres

The safety alignment of large language models (LLMs) is becoming increasingly important with their democratization. In this paper, we study the safety degradation that comes with adapting LLMs to new tasks. We attribute this safety compromise to catastrophic forgetting and frame the …

Forget Less, Retain More: A Lightweight Regularizer for Rehearsal-Based Continual Learning

Lama Alssum, Hasan Abed Al Kader Hammoud, Motasem Alfarra, Juan C Leon Alcazar et autres

Deep neural networks suffer from catastrophic forgetting, where performance on previous tasks degrades after training on a new task. This issue arises due to the model's tendency to overwrite previously acquired knowledge with new information. We present a novel approach to address …

Hala Technical Report: Building Arabic-Centric Instruction & Translation Models at Scale

Hasan Abed Al Kader Hammoud, Mohammad Zbeeb, Bernard Ghanem

We present Hala, a family of Arabic-centric instruction and translation models built with our translate-and-tune pipeline. We first compress a strong AR$\leftrightarrow$EN teacher to FP8 (yielding $\sim$2$\times$ higher throughput with no quality loss) and use it to create high-fidelity bilingual supervision. A …