Abdelaziz Bounhar

BounharAbdelaziz

AI & ML interests

Deep Learning, Reinforcement Learning, AI Agents, Generative Modeling, NLP, Information Theory, Security of Machine Learning, ...etc

Recent Activity

liked a Space about 15 hours ago

aminediroHF/trainer-generator-bf16-mismatch

liked a dataset 2 days ago

TeraflopAI/SEC-EDGAR

published a dataset 10 days ago

BounharAbdelaziz/Morocco-Darija-Speech-35h-Fixed

View all activity

Organizations

upvoted a paper 23 days ago

Voxtral TTS

Paper • 2603.25551 • Published 25 days ago • 59

upvoted an article about 2 months ago

Article

Mixture of Experts (MoEs) in Transformers

Feb 26

•

155

upvoted a paper about 2 months ago

VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training

Paper • 2602.10693 • Published Feb 11 • 220

upvoted an article about 2 months ago

Article

Did GPT 5.2 make a breakthrough discovery in theoretical physics?

Feb 19

•

upvoted a paper 3 months ago

YaPO: Learnable Sparse Activation Steering Vectors for Domain Adaptation

Paper • 2601.08441 • Published Jan 13 • 8

upvoted a collection 3 months ago

AfriqueLLM

Collection

Best open African LLM • 9 items • Updated about 6 hours ago • 22

upvoted 2 articles 4 months ago

Article

Shadow AI - Where are the CIOs?

Dec 19, 2025

•

Article

CircleGuardBench: New Standard for Evaluating AI Moderation Models

May 7, 2025

•

upvoted a collection 4 months ago

NeMo Gym

Collection

Collection of RL verifiable data for NeMo Gym • 22 items • Updated 5 days ago • 57

upvoted a paper 4 months ago

Entropy Ratio Clipping as a Soft Global Constraint for Stable Reinforcement Learning

Paper • 2512.05591 • Published Dec 5, 2025 • 17

upvoted an article 4 months ago

Article

We Got Claude to Fine-Tune an Open Source LLM

Dec 4, 2025

•

622

upvoted a paper 5 months ago

DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

Paper • 2512.02556 • Published Dec 2, 2025 • 265

upvoted an article 5 months ago

Article

Continuous batching from first principles

Nov 25, 2025

•

361

upvoted 2 collections 5 months ago

Frugal-Thinking

Collection

13 items • Updated Mar 12 • 3

Holo2

Collection

Holo2 - Cost-Efficient Models for Cross-Platform Computer-Use Agents • 4 items • Updated Feb 2 • 27

upvoted 2 papers 6 months ago

Shorter but not Worse: Frugal Reasoning via Easy Samples as Length Regularizers in Math RLVR

Paper • 2511.01937 • Published Nov 2, 2025 • 16

Surfer 2: The Next Generation of Cross-Platform Computer Use Agents

Paper • 2510.19949 • Published Oct 22, 2025 • 38

upvoted an article 7 months ago

Article

Smol2Operator: Post-Training GUI Agents for Computer Use

Sep 23, 2025

•

138

upvoted 2 collections 8 months ago

Reward Models 06-2025

Collection

Nemotron reward models. For use in RLHF pipelines and LLM-as-a-Judge • 8 items • Updated 5 days ago • 24

Nemotron-Pre-Training-Datasets

Collection

Large scale pre-training datasets used in the Nemotron family of models. • 12 items • Updated 5 days ago • 141

Abdelaziz Bounhar

AI & ML interests

Recent Activity

Organizations

BounharAbdelaziz's activity

Mixture of Experts (MoEs) in Transformers

Did GPT 5.2 make a breakthrough discovery in theoretical physics?

Shadow AI - Where are the CIOs?

CircleGuardBench: New Standard for Evaluating AI Moderation Models

We Got Claude to Fine-Tune an Open Source LLM

Continuous batching from first principles

Smol2Operator: Post-Training GUI Agents for Computer Use