electroglyph

7 ms·

61.
by electroglyph 8mo ago
because all the shit is locked down and the corpos can use state violence to stop you from doing so if you manage to succeed
62.
by electroglyph 8mo ago
i would read that write-up!
63.
by electroglyph 8mo ago
damn this is old. how relevant is this today?
64.
by electroglyph 8mo ago
that response is not comforting
65.
by electroglyph 8mo ago
comes to comments before reading story reads something from the comments instead never reads original article submission leaves satisfied
66.
by electroglyph 8mo ago
Microsoft's rwlock implementation was borked up until sometime last year iirc. this stuff is difficult to do correctly
67.
by electroglyph 8mo ago
some models are capable of metacognition. i've seen Anthropic's research replicated.
68.
by electroglyph 9mo ago
well, first the model has to actually follow the instructions in agents.md =)
69.
by electroglyph 9mo ago
Treating prisoners humanely and preparing them to resume a productive life after prison helps us all.
70.
1.8-3.3x faster Embedding finetuning now in Unsloth(unsloth.ai)
3 points by electroglyph 9mo ago|3 comments
71.
by electroglyph 9mo ago
see also: https://www.reddit.com/r/LocalLLaMA/comments/1qk9vmv/1833x_f...
72.
by electroglyph 9mo ago
simple lil setup with qdrant
73.
by electroglyph 9mo ago
the Nokia hardware was pretty great, too!
74.
by electroglyph 9mo ago
time to train a classifier!
75.
by electroglyph 9mo ago
> Once people will understand that even a tiny amount of data can slightly change models and still greatly change their behaviour, there will be a shift in AI security. i think the subliminal learning paper clued a lot of people into tha
76.
by electroglyph 9mo ago
unrelated in any way? that's not normal. have you tested the model to make sure you have sane output? unless you're using sentence-transformers (which is pretty foolproof) you have to be careful about how you pool the raw output v
77.
by electroglyph 9mo ago
there are tons of models released still. even some non-Qwen ones!
78.
by electroglyph 9mo ago
there's actually quite a bit of research in this field, here's a couple: "ExpertPrompting: Instructing Large Language Models to be Distinguished Experts" https://arxiv.org/abs/2305.14688 "Perso
79.
by electroglyph 9mo ago
did it tell you how smart your questions were, too?
80.
by electroglyph 9mo ago
check out Forsaken Lands: https://wiki.theforsakenlands.com/
81.
by electroglyph 9mo ago
Cloudflare worker would work, too
82.
by electroglyph 9mo ago
Great band!
83.
by electroglyph 9mo ago
they just steal different things
84.
by electroglyph 9mo ago
yep! reversing Ultima Online clients started it all for me
85.
by electroglyph 9mo ago
i started with x86 intel asm. i just derped around with ollydbg a lot while i referenced the intel instruction set pdf that i had. after i was able to do basic patches i learned C# (then 1.0) so that i could load and patch processes in Wind
86.
by electroglyph 9mo ago
i really wanted "mimtipled" to be a word =)
87.
by electroglyph 9mo ago
you think so? i think making a good TUI is a pain in the ass
88.
by electroglyph 9mo ago
those jerks put a zoo in the desert!? =(
89.
by electroglyph 9mo ago
https://en.wikipedia.org/wiki/Gang_stalking this sums it up nicely: "A study from Australia and the United Kingdom by Lorraine Sheridan and David James compared 128 self-defined victims of gang stalking with a ran
90.
by electroglyph 10mo ago
the beauty of open source is there's nothing stopping you! this might be your calling. best of luck

More ›