flancian

14 ms·

91.
by flancian 7y ago
Sam Harris's: Making Sense (used to be called Waking Up). https://samharris.org/ I don't agree with him on everything -- he's a bit to the center/right of me in some areas, particularly around what he ca
92.
by flancian 7y ago
Super simple project of course, but I realized I had not done something "simple and fun" for a long time outside of the day job. I'd like to keep improving it and at some point add phrase generation; train a generative ML mod
93.
I made a bot that tweets lines from Virginia Woolf's works(twitter.com)
1 points by flancian 7y ago|1 comments
94.
by flancian 7y ago
I understand why you think that, and you might be right. But I also know there are a lot of good people at Google, and I dream of the company waking up as a group again and focusing on its mission responsibly. Doing the right thing, openl
95.
by flancian 7y ago
This is what I want as well. I feel as if this is the kind of thing Google might still be able to get right; "to organize the world's information and make it universally accessible and useful" is still the company's miss
96.
by flancian 7y ago
I'm glad that you found something that works for you! FWIW, that description seems compatible with meditation technique. Meditation can be non judgmental in that way, and just be a matter of experiencing the state of your mind.
97.
I'm starting an entity for the pursuit of knowledge and improvement of the world(flancia.org)
2 points by flancian 7y ago|0 comments
98.
by flancian 7y ago
I'm interested on this as well. I have been trying to fine-tune GPT-2 on genre fiction to work as a sort of "fiction replicator". Stylistically it actually seems to do quite reasonably, but it lacks narrative cohesion. This p
99.
by flancian 7y ago
Author here. I'm going through the process of getting back into writing, so posting some of my own material sometimes to try to get an audience -- which I know is a bit awkward, but oh well. Hopefully it can be interesting for HN. Plea
100.
Meditation for Programmers(flancia.org)
2 points by flancian 7y ago|1 comments
101.
by flancian 7y ago
This is one in a series of quick thought experiments I plan to write up and post -- I hope someone finds them interesting. Let me know if you have any questions or comments, they'd be very much appreciated!
102.
Embracing Piracy(flancia.org)
3 points by flancian 7y ago|1 comments
103.
by flancian 7y ago
Yes, I agree. I should have written "never" in quotes. One way to think about this is: which companies would you never work for without feeling you have relinquished some part of your identity? I'd risk saying that most peopl
104.
by flancian 7y ago
To get this started: Palantir. I know the world, and myself, may change; but I choose to believe that I would never work for "defense" or the military-industrial complex. It could become inconsistent as a position in the long term
105.
Which companies would you never work for, and why?
19 points by flancian 7y ago|29 comments
106.
by flancian 7y ago
The article tells us to prefer some phrases: - I want to speak with you (instead of: talk to you). - Is there something else you'd like to talk about? (instead of: anything else). To ellicit a better response from your interlocutors. T
107.
Program generates bitmaps that are locally similar to the input bitmap(github.com)
3 points by flancian 7y ago|0 comments
108.
by flancian 7y ago
I agree. It seems that his overall position could be stronger if he just started with the requisite of having some taste. Which everybody has. It'd still all ring true, and be more general, in the sense that one's taste is linke
109.
by flancian 7y ago
This is great, thanks! Unsolicited advice: perhaps consider making the bot pin the tweet about the most interesting encounter of the last N days, for some definition. Likely something like the nearest miss.
110.
by flancian 7y ago
The "processes" defined here remind me of Minsky's "agents" (which came later), and the "hierarchy" of his Society of Mind -- which also happens to be hierarchycally organized, like most (all?) human socie
111.
by flancian 7y ago
I call these arbitrash. I hate them with a passion, partly because I understand (deep down) why they continue to exist: they sometimes work on me, in the sense that they arouse the basest form of curiosity and I get the impulse of clicking
112.
by flancian 7y ago
It resonated with me deeply too, which I guess is what often happens when you are at the just the right time in some intellectual development -- you've read the right things, and you already think or believe (is there a difference?) so
113.
by flancian 8y ago
Thanks for this comment -- I'll put the second strategy in effect (the first one perhaps I'll leave for later as I currently work in an open plan office). I wonder if this "comment driven" strategy has potential for appl
114.
by flancian 8y ago
This is intended to be the second in a series of posts about possible applications of NLP for the purpose aiding writing and other creative activities. My models are very entry-level right now, except for this excursion to the land of basic
115.
GPT-2 and Philip K. Dick(flancia.org)
3 points by flancian 8y ago|1 comments
116.
by flancian 8y ago
Author here. This is my first relatively serious try at writing an article about current events in the field of ML/AI, which is something I've wanted to do for some time now. All feedback is welcome, although I know attention is a
117.
About unsupervised language models, or unicorns in South America(flancia.org)
2 points by flancian 8y ago|1 comments
118.
by flancian 8y ago
Previously: https://news.ycombinator.com/item?id=7557964 https://news.ycombinator.com/item?id=9814114 But not a lot of discussion over there. The visualizations are great, and this basically blew my mind. I
119.
Neural Networks, Manifolds, and Topology (2014)(colah.github.io)
129 points by flancian 8y ago|25 comments
120.
How to beat procrastination(flancia.org)
3 points by flancian 8y ago|0 comments

More ›