Heterogeneous Low-Bandwidth Pre-Training of LLMs.

Paul Janson

Yazan Obeidi Amir Sarfi Joel Lidin Paul Janson Eugene Belilovsky Heterogeneous Low-Bandwidth Pre-Training of LLMs. 2026 January abs/2601.02360 CoRR https://doi.org/10.48550/arXiv.2601.02360 db/journals/corr/corr2601.html#abs-2601-02360 streams/journals/corr

Paul Janson Edouard Oyallon Eugene Belilovsky Stabilizing Native Low-Rank LLM Pretraining. 2026 February abs/2602.12429 CoRR https://doi.org/10.48550/arXiv.2602.12429 db/journals/corr/corr2602.html#abs-2602-12429 streams/journals/corr

Vaibhav Singh Paul Janson Paria Mehrbod Adam Ibrahim Irina Rish Eugene Belilovsky Benjamin Thérien Beyond Cosine Decay: On the effectiveness of Infinite Learning Rate Schedule for Continual Pre-training. 2025 March abs/2503.02844 CoRR https://doi.org/10.48550/arXiv.2503.02844 db/journals/corr/corr2503.html#abs-2503-02844 streams/journals/corr

Paul Janson Benjamin Thérien Quentin Anthony Xiaolong Huang Abhinav Moudgil Eugene Belilovsky PyLO: Towards Accessible Learned Optimizers in PyTorch. 2025 June abs/2506.10315 CoRR https://doi.org/10.48550/arXiv.2506.10315 db/journals/corr/corr2506.html#abs-2506-10315 streams/journals/corr

Wenxuan Zhang 0001 Paul Janson Rahaf Aljundi Mohamed Elhoseiny Overcoming Generic Knowledge Loss with Selective Parameter Update. 24046-24056 2024 CVPR https://doi.org/10.1109/CVPR52733.2024.02270 conf/cvpr/2024 db/conf/cvpr/cvpr2024.html#ZhangJAE24 Paul Janson Tiberiu Popa Eugene Belilovsky Towards Motion from Video Diffusion Models. 415-426 2024 ECCV Workshops (12) https://doi.org/10.1007/978-3-031-92591-7_27 conf/eccv/2024-w12 db/conf/eccv/eccv2024-w12.html#JansonPB24 streams/conf/eccv

Piraveen Sivakumar Paul Janson Jathushan Rajasegaran Thanuja D. Ambegoda FewShotNeRF: Meta-Learning-based Novel View Synthesis for Rapid Scene-Specific Adaptation. 2024 abs/2408.04803 CoRR https://doi.org/10.48550/arXiv.2408.04803 db/journals/corr/corr2408.html#abs-2408-04803 streams/journals/corr

Paul Janson Tiberiu Popa Eugene Belilovsky Towards motion from video diffusion models. 2024 abs/2411.12831 CoRR https://doi.org/10.48550/arXiv.2411.12831 db/journals/corr/corr2411.html#abs-2411-12831 streams/journals/corr

Wenxuan Zhang 0001 Paul Janson Kai Yi Ivan Skorokhodov Mohamed Elhoseiny Continual Zero-Shot Learning through Semantically Guided Generative Random Walks. 11540-11551 2023 ICCV https://doi.org/10.1109/ICCV51070.2023.01063 conf/iccv/2023 db/conf/iccv/iccv2023.html#ZhangJYSE23

Wenxuan Zhang 0001 Paul Janson Kai Yi Ivan Skorokhodov Mohamed Elhoseiny Continual Zero-Shot Learning through Semantically Guided Generative Random Walks. 2023 abs/2308.12366 CoRR https://doi.org/10.48550/arXiv.2308.12366 db/journals/corr/corr2308.html#abs-2308-12366

Wenxuan Zhang 0001 Paul Janson Rahaf Aljundi Mohamed Elhoseiny Overcoming General Knowledge Loss with Selective Parameter Finetuning. 2023 abs/2308.12462 CoRR https://doi.org/10.48550/arXiv.2308.12462 db/journals/corr/corr2308.html#abs-2308-12462

Paul Janson Wenxuan Zhang 0001 Rahaf Aljundi Mohamed Elhoseiny A Simple Baseline that Questions the Use of Pretrained-Models in Continual Learning. 2022 abs/2210.04428 CoRR https://doi.org/10.48550/arXiv.2210.04428 db/journals/corr/corr2210.html#abs-2210-04428

Rahaf Aljundi Thanuja D. Ambegoda Quentin Anthony Eugene Belilovsky Mohamed Elhoseiny Xiaolong Huang Adam Ibrahim Joel Lidin Paria Mehrbod Abhinav Moudgil Yazan Obeidi Edouard Oyallon Tiberiu Popa Jathushan Rajasegaran Irina Rish Amir Sarfi Vaibhav Singh Piraveen Sivakumar Ivan Skorokhodov Benjamin Thérien Kai Yi Wenxuan Zhang 0001