Jan 19 – Feb 6, 2025
per-Channel stats (number of posts, recent posts) This could be either calculated on the fly or stored in a new database table. Custom messages per page in stream view. keyword search / tags the machine should prefer longer tags with unambiguous meanings. << linguistic register rather than <<…
in which the machine attempts to grade my emails
Roughly, "good" emails are rated by how useful the content is, with an "average" post being around +110. Separately, "bad" emails are rated by just how painful (political, distasteful, etc.) it is, with a logarithmic scale somewhat equivalent to the positive one (so -21dB for an email as bad as an…
The short answer here is the very mundane << we don't need software, we need humans .
Is it possible for the machine to guess what channel a message should be in?
thoughts on LLMs for Social Networks
in the future, your LLM will talk to my LLM, and neither of us will benefit from the sage information exchanged.
We need to make a few updates to the security model.
I have reached a fork in the road.
thoughts on the LLM-powered encyclopedia
it would be a competitor to Wikipedia.
The question: in Python 3.9, how does one set up inheritance of dataclasses, when the base class contains fields with a default value?
https://news.ycombinator.com/item?id=42458752
https://viruscomix.com/page490.html
Here is Deepseek-8b's response to the same question.
The new "deepseek-r1" models are out. Both a large model I can't use, and "distilled" models.
https://andymasley.substack.com/p/individual-ai-use-is-not-bad-for << Using ChatGPT is not bad for the environment