Ethan Sunshine

[e chi @echinaceous]

@quadrillion_ai is hosting a social at #icml2026, tuesday 7/7 at 7pm! https://t.co/8PnE5CHYig

we've booked out a wonderful bar minutes away from COEX. come hang out with us and chat about everything autoresearch and recursive self-improvement. we'll have drinks, wonderful food, and great people!

[Ethan Sunshine @EthanSunshine_]

@namankatyal14 I apologize for being a little fast and loose with the term “accurate”. I just meant that 8% MAE seems quite small for a baseline method. However, I am not convinced that this metric is the one we should be using for this task anyway.

[Ethan Sunshine @EthanSunshine_]

@namankatyal14 Hi Naman! Glad you enjoyed this paper! We included the SACD plot because it gives more context and helps the reader understand some of the advantages and disadvantages of our model. The fact that the super cheap method is already 92% accurate is also quite interesting.

[Greg Hartland @greg_hartland]

Most Read @JPhysChem 2/4 Graph Neural Network predictions of chemical properties

https://t.co/Uj99log1SL

[Profess.or Profess.and @johnkitchin]

New #OpenAccess #publication in collab with @zackulissi and @OpenCatalyst on predicting electron density with graph neural networks in @JPhysChem. https://t.co/ALtkvTToWw Congratulations to Ethan Sunshine, first author!