Entropy Controllable Direct Preference Optimization

no code implementations • 12 Nov 2024 • Motoki Omura, Yasuhiro Fujita, Toshiki Kataoka

In the post-training of large language models (LLMs), Reinforcement Learning from Human Feedback (RLHF) is an effective approach to achieve generation aligned with human preferences.

PLaMo-100B: A Ground-Up Language Model Designed for Japanese Proficiency

no code implementations • 10 Oct 2024 • Preferred Elements, :, Kenshin Abe, Kaizaburo Chubachi, Yasuhiro Fujita, Yuta Hirokawa, Kentaro Imajo, Toshiki Kataoka, Hiroyoshi Komatsu, Hiroaki Mikami, Tsuguo Mogami, Shogo Murai, Kosuke Nakago, Daisuke Nishino, Toru Ogawa, Daisuke Okanohara, Yoshihiko Ozaki, Shotaro Sano, Shuji Suzuki, Tianqi Xu, Toshihiko Yanase

We introduce PLaMo-100B, a large-scale language model designed for Japanese proficiency.

Language Modelling

ChainerRL: A Deep Reinforcement Learning Library

1 code implementation • 9 Dec 2019 • Yasuhiro Fujita, Prabhat Nagarajan, Toshiki Kataoka, Takahiro Ishikawa

In this paper, we introduce ChainerRL, an open-source deep reinforcement learning (DRL) library built using Python and the Chainer deep learning framework.

Deep Reinforcement Learning reinforcement-learning +1

Spectral Normalization for Generative Adversarial Networks

38 code implementations • ICLR 2018 • Takeru Miyato, Toshiki Kataoka, Masanori Koyama, Yuichi Yoshida

One of the challenges in the study of generative adversarial networks is the instability of its training.

Image Generation