thebes
@vgel.me
ꙮ surfed on by the information superhighway
ꙮ 💕 @linneaisaac.bsky.social
ꙮ she/they 🏳️⚧️
ꙮ blog posts and games @ https://vgel.me
ꙮ still mostly active on twitter https://x.com/voooooogel
ꙮ 💕 @linneaisaac.bsky.social
ꙮ she/they 🏳️⚧️
ꙮ blog posts and games @ https://vgel.me
ꙮ still mostly active on twitter https://x.com/voooooogel
Pinned
new blog post! can small, open-source models also introspect, detecting when foreign concepts have been injected into their activations? yes! (thread, or full post here: vgel.me/posts/qwen-i...)
protip if you have an old enameled dutch oven that'a kinda brown and nasty on the bottom no matter what you do, the residue is probably acidic. boil some water in it and add a couple tablespoons of baking soda, then scrape the bottom with a wooden spoon as it boils. this LC is 10 years old!
January 8, 2026 at 8:05 AM
protip if you have an old enameled dutch oven that'a kinda brown and nasty on the bottom no matter what you do, the residue is probably acidic. boil some water in it and add a couple tablespoons of baking soda, then scrape the bottom with a wooden spoon as it boils. this LC is 10 years old!
Reposted by thebes
neo-thebes too cheap to meter arrives from the future 27 dot com
January 5, 2026 at 5:05 PM
neo-thebes too cheap to meter arrives from the future 27 dot com
manifold released a wordlelike, p fun!
---
Predictle #1
✅✅✅
✅✅✅
❌❌✅
❌✅✅
❌❌✅
Play at manifold.markets/predictle?r=...
---
Predictle #1
✅✅✅
✅✅✅
❌❌✅
❌✅✅
❌❌✅
Play at manifold.markets/predictle?r=...
Predictle
A daily game where you arrange prediction markets by probability.
manifold.markets
January 7, 2026 at 1:34 AM
manifold released a wordlelike, p fun!
---
Predictle #1
✅✅✅
✅✅✅
❌❌✅
❌✅✅
❌❌✅
Play at manifold.markets/predictle?r=...
---
Predictle #1
✅✅✅
✅✅✅
❌❌✅
❌✅✅
❌❌✅
Play at manifold.markets/predictle?r=...
it's just wild to me how useful llm computer use is. i should've done this a long time ago, but in the past i would've had to make a complex harness for this. (in fact i did, but never automated running it over my site.) now it's just claude code orchestrating subagents for a few minutes, ~$free
January 2, 2026 at 7:35 PM
it's just wild to me how useful llm computer use is. i should've done this a long time ago, but in the past i would've had to make a complex harness for this. (in fact i did, but never automated running it over my site.) now it's just claude code orchestrating subagents for a few minutes, ~$free
kind of hilarious how right after the hyperbanger central principle of the torah in leviticus 19:18 comes the archetypal confusing "why is this a law"
December 25, 2025 at 1:27 AM
kind of hilarious how right after the hyperbanger central principle of the torah in leviticus 19:18 comes the archetypal confusing "why is this a law"
Reposted by thebes
This is a really cool and surprising result on model introspection! For me, this raises two big questions:
1. Why do these models believe (or at least report) that they’re unable to do something that they demonstrably can do?
2. What else can models do that they aren’t aware of?
1. Why do these models believe (or at least report) that they’re unable to do something that they demonstrably can do?
2. What else can models do that they aren’t aware of?
new blog post! can small, open-source models also introspect, detecting when foreign concepts have been injected into their activations? yes! (thread, or full post here: vgel.me/posts/qwen-i...)
December 21, 2025 at 12:44 AM
This is a really cool and surprising result on model introspection! For me, this raises two big questions:
1. Why do these models believe (or at least report) that they’re unable to do something that they demonstrably can do?
2. What else can models do that they aren’t aware of?
1. Why do these models believe (or at least report) that they’re unable to do something that they demonstrably can do?
2. What else can models do that they aren’t aware of?
new blog post! can small, open-source models also introspect, detecting when foreign concepts have been injected into their activations? yes! (thread, or full post here: vgel.me/posts/qwen-i...)
December 21, 2025 at 12:14 AM
new blog post! can small, open-source models also introspect, detecting when foreign concepts have been injected into their activations? yes! (thread, or full post here: vgel.me/posts/qwen-i...)
Reposted by thebes
@godoglyness.bsky.social as resident coin dream expert, care to weigh in?
December 20, 2025 at 8:17 PM
@godoglyness.bsky.social as resident coin dream expert, care to weigh in?
Reposted by thebes
I recently had occasion to review some of the akrasia tricks I’ve found on LessWrong, and it occurred to me that I can will what is right, but I cannot do it… Wretched man that I am! Who will deliver me from this body of death?
December 15, 2025 at 2:41 AM
I recently had occasion to review some of the akrasia tricks I’ve found on LessWrong, and it occurred to me that I can will what is right, but I cannot do it… Wretched man that I am! Who will deliver me from this body of death?