Paper models Transformer components as stochastic geometry and tests five architectures

Zhihua Liang · hf · 2026-07-21

Continuous geometric framework for Transformers

This paper recasts core Transformer components — RMSNorm, RoPE, softmax attention, FFN, residual streams, SGD, and weight decay — as parts of an integro-differential equation on a semantic fiber bundle.

What it claims

Experiments

The authors run a six-part campaign across five architectures — Qwen3, LLaMA-3.1, Gemma-3, GPT-2, and Mistral — covering 124M to 8B parameters.

They report quantitative agreement with several geometric predictions, including:

Takeaway

The paper argues that continuous stochastic differential geometry can provide a predictive vocabulary for Transformer stability limits, context bounds, and optimization dynamics.

Original post →

More from Research

Research channel →