Tweets
2022-11-09T04:29:29+00:00 | 🔗
Not all roses, ... struggles a lot with larger compounds like this peptide. Seems to start teasing out a structure but quickly runs out of space and becomes ugly AF https://t.co/bgTw8VnC3O
2022-11-09T03:15:41+00:00 | 🔗
@McdarghZach Pre-defined, I think we have some idea for making that (and the atoms themselves) generated via Diffusion. But, this was my first diffusion project so I just wanted to get something working! It's hard to generate everything without bumping into V^2 problems
2022-11-09T01:15:51+00:00 | 🔗
@bertonearnshaw @RecursionPharma @omendezlucio @ProjectTDC Super cool! Are code/pretrained weights available?
2022-11-08T20:40:11+00:00 | 🔗
RT @samplchallenges: It looks like we will shortly be able to launch a water-toluene logP challenge with new experimental measurements for…
2022-11-08T20:22:02+00:00 | 🔗
Wow! The side project turned out somewhat nicely... 2D coordinate generation of molecules from pure noise! Fairly accessible project for exploring diffusion models + GNNs + molecules! https://t.co/rtpOEWPVKB
2022-11-08T18:56:03+00:00 | 🔗
This figure was such a journey to create. Manual database scraping -> Excel -> RStudio to generate individual SVGs -> Illustrator to highlight bar + annotation + layout… I had to pull out a calculator to lay everything out https://t.co/vryAd1jkjk
2022-11-08T03:47:40+00:00 | 🔗
@__mharrison__ Task manager or force quit
2022-11-08T03:43:20+00:00 | 🔗
It de-noises 2D Depictions! Kind of... it definitely looks a bit tangled, but at least it's doing something? This problem doesn't seem like it should be this hard though, and I've spent quite a bit of time training/tuning... #AI #CompChem https://t.co/j7oVgoAdJp
2022-11-07T14:11:54+00:00 | 🔗
The loss is dependent on noise level, but oddly enough is the denoiser is more accurate at higher noise levels? does this ever make sense #AI #DiffusionModels #WackyLoss... https://t.co/pLMkUKYD0B
