🩵 Rikka Botan | Portfolio 🩵

logo

🩵 Japanese independent Researcher 🩵

My primary area of interest is the architecture of language models.
In particular, I study efficiency-enhancing techniques such as state space models, liquid time-constant networks, and Mixture of Experts (MoE).
I am also investigating novel architectures based on principles from probabilistic statistics and algebra.

🩵About me

Having shy and pampered personality.
Twin-tail hair is a charm point.
I love cute clothes in pastel colors, and frills and lace always make my heart skip a beat.
I enjoy baking, and baked cheesecakes are my specialty.
I also love tea, especially "Earl Grey" and "Arabian Night".

🩵Projects

SEA Model series projects

The goal is to develop high-accuracy, high-speed language models through original improvements to projection representations and model architectures.

SEAGIF

SEA Model series Op.0: Saint Iberis

SaintIberis

SEA Model series Op.1: Saint Lupinus

SaintLupinus

SEA Model series Op.2: Saint Plumeria

SaintPlumeria

🩵Articles

🩵GitHub Repositories

🩵Hugging Face Models

🩵Activities

2025.12

An exhibition at ローカルAIに向き合う展示会

I exhibited “LTCs: Explanation of convergence and system stability of linear dynamical systems in Liquid Time-Constant networks and applicability of LTCs”

Exhibition
2025.10

🥈Silver Award: Liquid AI Hackathon Series | Tokyo

Our team won the Silver Award in the Liquid AI Hackathon Series held in Tokyo. We developed a vision-language model and application specialized for casual, conversational interactions.

Competition
2025.09

An exhibition at 生成AIなんでも展示会 Vol.4

I exhibited “Projective Representations and Language Model Construction Based on Duality”

Exhibition

🩵Contact

If you're interested in talks, research collaborations, writing projects, or anything else, I'd love to hear from you.✿♡*.ᕱ⋈ᕱ

Rikka Botan's X(Twitter) account