Remember to twist your head to the right! I hear that’s good for range.

Do you know if any of the results from this have ever been published? The video on the procedures was cool, but it didn’t actually explain much of what they saw in Will’s throat that surprised them so.

(Obviously neither of these ended up being the case)
Also there weren’t as many high-cost “I Win” cards. You couldn’t just use Dream Halls to play Enter the Infinite and then goldfish your opponent.

I love the concept of DIY feeds, but we need tools for (paid) feed building because everybody setting up their own hosting is expensive and difficult.

A lot of scientists came here, but they check it less than old Twitter.

Blocking is bad for individuals, but is overall healthy for the network.

The last season had a lot of problems, which I largely attribute to compressed pacing and the showrunners knowing how to write characters, not worlds.

That being said, S8E2 was one of the best episodes of character-driven television I’ve ever seen.

Tools:
The RNA helix was generated using oxView (oxview.org), a oxDNA viewer and structure editor.

QRNA (github.com/sunandanmukherjee/QRNAS) for relaxation, specifically of the backbones.

All-atom forcefield OL3 (kfc.upol.cz/ff_ol/index.php), currently the best out-of-the-box RNA forcefield.

They can be re-purposed for research usage, they’re the same kind of GPUs used in things like weather prediction, drug design, and data analysis. But that requires the capital to buy up the hardware when the AI companies fall apart, which research institutes are currently struggling with.

This really is part of the joy of science! I love hunting down papers from the early days of biochemistry or molecular biology and seeing the methods, the thought processes, and feeling the continuity with great names from the past.

Oof, this is gnarly! Thanks for sharing.

When you put it that way, I suppose what I’m asking for isn’t so different than what attention does with the tokens themselves.

Has there been any research on ‘fuzzy’ tokenizers? The reason we can mentally combine ‘going to’ and ‘gonna’ is because of this map between words and sound which lets those co-exist, while ‘boing to’ would be different. Probably too much processing power in the tokenizer layer for that to work tho.

Also, I added you to the Molecular Programming feed posting list. If you include a 🧬 emoji in your posts, they should get picked up by the feed:

bsky.app/profile/did:...

This is great! Would love to see if this method could be used to get the stacking energies of the inverted A|A stack in the kissing loops used in RNA origami.

Anyway, learning another language is good for you because you die a thousand deaths of embarrassment and then you come out more humble and even more aware of your various human failings