Ken Osumi
  • 日本語
  • English
Categories
All (5)
binder-design (1)
esmc (2)
interpretability (2)
optimization (1)
reproduction (2)
tnfa (1)

Ken Osumi

Reading the internal representations of protein language models to generate residue-level mechanistic hypotheses.

About · 日本語 · GitHub

Package model docs

This document describes the kinetic models used to fit the binding curves in this package: for each model, the reaction scheme, the binding response, the system of…
9 min

Data — I Beat Claude on TNFα for $677

Both arms of both targets, the per-engine scores behind the selection, and the binding kinetics from Adaptyv Bio. Released under ODC-BY.
2 min

I Beat Claude on TNFα for $677

Seven of ten designs bound. Everything I used was public and free. The one step that mattered was written down in 2003.

Aug 21, 2026
8 min

Reproducing ESMC’s SAE features: I lost half a day to one wrong repo

Reproducing Figure S33 of the ESMC paper, solo. I picked the wrong one of two similarly-named SAEs, lost half a day, and came out the other side at the most interesting feature in the figure.

Jun 22, 2026
6 min

ESMC の SAE 図を一人で再現したら、repo の取り違えで半日溶かした話

ESMC 論文の Figure S33 を一人で再現した記録。よく似た名前の SAE を取り違えて半日ハマり、抜けた先で一番おもしろい特徴に行き着くまで。

Jun 22, 2026
2 min
No matching items
 

© 2026 Ken Osumi