benob

9 ms·

61.
by benob 1y ago
This work is a good argument against memorization of information seen less than 250 times during training.
62.
by benob 1y ago
I wonder if one could store only the binary representation at training and sample a floating point representation (both weights and gradient) during backprop.
63.
by benob 1y ago
There was a time when people would estimate n-gram probabilities with feed-forward neural networks [1,2]. We just improved that with the (multilayer) attention mechanism which allows for better factoring over individual tokens. It also allo
64.
by benob 1y ago
Like it or not, LLMs are effectively high-order Markov chains
65.
by benob 1y ago
These techniques are limited to structures that can be checked with bounded history or bounded memory (that can be checked with a grammar or FSA). What about more complex structures that don't factor easily?
66.
by benob 1y ago
My understanding was that the Alpaca data was a distillation from text-davinci-003
67.
by benob 1y ago
a good benchmark for video understanding in IA
68.
by benob 1y ago
It's funny because automation is the only thing you can expect from AI
69.
by benob 1y ago
> YouTube views seem to have fallen off a cliff recently So they started discounting AI data collection bots?
70.
by benob 1y ago
Stop thinking about text being the data. There are so many other sources, even some that you can generate. https://arxiv.org/pdf/2506.20057
71.
by benob 1y ago
Even if it's not on topic, that post is quite interesting.
72.
by benob 1y ago
How is it different from mere syntactic sugar over the same programming concepts? What does it bring that C++ cannot do? Isn't it just a way of controlling the language vs using normative bodies?
73.
by benob 1y ago
Are there efforts to include the neccessary context in compilers to autovectorize?
74.
by benob 1y ago
I wonder how difficult it would be to bias a model so that it subtly corrupts election results when performing OCR.
75.
by benob 1y ago
You can generate 1.0 programs with 3.0 programs. But can you generate 2.0 programs the same way?
76.
by benob 1y ago
Cray1 should be compared to nowadays raspberry pi pico 2 / rp2350 which has similar specs (using external ram).
77.
by benob 1y ago
> This is not an advertisment for Claude Code. It's just the agent I use at the moment. What else is there? Alternatives that are similar in their user experiences are OpenCode, goose, Codex and many others. There is also Devin and
78.
by benob 1y ago
Watermarking is easily disabled in the code. I a wondering when they will release model weights with embedded watermarking.
79.
by benob 1y ago
The gemini web (smolweb) has no effective search engine, and therefore links also play a crutial role in content discovery...
80.
by benob 1y ago
I am seeing different prices everywhere: - $199 https://shop.vassarrobotics.com/ - $219 https://shop.vassarrobotics.com/products/navrim-robot-that-l... - $599 https://www.youtube.com/wa
81.
by benob 1y ago
Isn't it time for disallowing browsers to connect to anything outside same origin pages except for actual navigation? Servers can do all the hard work of gathering content from here and there.
82.
by benob 1y ago
It looks pretty and featureful, but I'd love a cheaper version, maybe without a display.
83.
by benob 1y ago
I wonder how to import markdown into Notes
84.
by benob 1y ago
I guess autoregressive llms can be finetuned (or continual-pretrained) to do inference using diffusion. We've seen a recent paper (which I don't remember) training from scratch, but it seems overkill. Do Google say how they did it
85.
by benob 1y ago
Ok, one of my email addresses is in a bunch of leaks. What is interesting is that most services on this list I have never used. How did they get my email in the first place? What is the accuracy of that whole business?
86.
by benob 1y ago
It looks like it is an extension of 3d model generation techniques.
87.
by benob 1y ago
I wonder whether they used a mirror or simple recordings as baseline
88.
by benob 1y ago
> If you ever remove the interconnection PCB, make sure to put it back with the same orientation. It will fit just fine when rotated 180 degrees but the scope won’t work anymore! I remember pulling a 486 out of its socket in the 1990s an
89.
by benob 1y ago
I think you are mixing things. Quaternions are in the same category as complex numbers. They can be represented as matrices, and there are probably nice uses of matrices where the element is a quaternion (such as QDNNs) instead of a real nu
90.
by benob 1y ago
> Parts of France also appear to be affected, according to Spanish media reports, which said Seville, Barcelona and Valencia were hit by the outage. The wording in the article makes it look like Seville, Barcelona and Valencia are in Fra

More ›