pama

9 ms·

151.
by pama 10mo ago
The article explains that there are other betting systems/organizations where you bet against the house, especially in early times of a new type of bet, before there is market information. These organizations try to eliminate/red
152.
by pama 10mo ago
Sure. However, the hiccup that salesforce faces will affect slack usage.
153.
by pama 10mo ago
Agree on broad strokes, but slack is a useful product.
154.
by pama 10mo ago
You asked for specific feedback, but here is generic feedback: a new github account coupled to a new HN account does not inspire any sense of added infra safety. I would rather use modern pytorch/safetensors and tools that dont allow e
155.
by pama 10mo ago
I’d be curious about this global replication rate as a function of age.
156.
by pama 10mo ago
Yeah I like the simplicity and power of it. You might want to tackle the math and write it up.
157.
by pama 10mo ago
Even without a formal proof you could test it empirically if you generate a large number of samples and run regular tests for pseudorandom number generators. [Edit: a quick test on a million samples and relatively simple RNG tests suggests
158.
by pama 10mo ago
Maybe you haven’t carefully checked yet the correctness of automatic tools or of the associated metadata. Zotero is certainly not bug free. Even authors themselves have miss-cited their own past work on occasion, and author lists have had
159.
by pama 10mo ago
Given how many errors I have seen in my years as a reviewer from well before the time of AI tools, it would be very surprizing if 99.75% of the ~20,000 submitted papers to didnt have such errors. If the 300 sample they used was truly rando
160.
by pama 10mo ago
The simple explanation is that the causal agent for the excees of the non-covid deaths is the same SARS-CoV2 virus, but death comes later and not at the acute phase of the disease.
161.
by pama 10mo ago
The table is confusing. It is not clear what is known and what is predicted (and how it is predicted). Why not measure the missing pieces instead of predicting—is it too expensive or is the tooling missing?
162.
by pama 11mo ago
LLMs have the ability to learn certain classes of algorithms from their datasets in order to reduce errors when compressing their pretraining data. If you are technically inclined, read the reference: https://arxiv.org/abs&#
163.
by pama 11mo ago
Personal antigravity hack: add a GPL license to every file, so google filters them before training to avoid legal complications. IANAL.
164.
by pama 11mo ago
Agreed. I keep almost all things same between terminal and graphical UI, so I would never use it, but I can see the appeal of having the ability for handling keyboard events differently than character inputs in some cases in the GUI.
165.
by pama 11mo ago
If you want all keybindings, C-h b typically helps, and you can search within the single buffer it returns. Every key in Emacs is bound to something, but a plain modifier key event like Ctrl will not be sent from a terminal to any command t
166.
by pama 11mo ago
You mean something like which-key? It existed for a long time as an external package and was added to main emacs recently. https://github.com/emacs-mirror/emacs/commit/fa4203300fde682...
167.
by pama 11mo ago
The US electricity grid is a complex problem at the interface of state and local governments and cannot scale quickly. There are no major concerns with scaling up the local electricity production however, other than the installation time (a
168.
by pama 11mo ago
The sources for some of the future data centers will be local and not necessarily influence the US grid. Consider also that cement production uses about 3000 TWh per year worldwide, and aluminium smelting uses about 1000 TWh per year.
169.
by pama 11mo ago
Which has no relation to the word gamified. So the editorialized title is misleading. Editorialized here simply means applying editor-level changes to the title of the website to express an opinion. No deceit is implied. It is against HN g
170.
by pama 11mo ago
Table 1: https://arxiv.org/html/2412.19437v2
171.
by pama 11mo ago
You start an emacs server on each machine you might use. Then ssh and emacsclient -nw. You need a decent terminal but life is simple.
172.
by pama 11mo ago
It is not clear that a simple/small model with inference running on home hardware is energy or cost efficient compared to the scaled up inference of a large model with batch processing. There are dozens of optimizations possible when
173.
by pama 11mo ago
> only uses 1/18th of the total parameters per-query. only uses 1/18th of the total parameters per token. It may use the large fraction of them in a single query.
174.
by pama 1y ago
Many people still use package managers though and these run such checks and disclose infosec events. And I still know enough other people who also look at the checksums of individual downloads when needed. Piping curl to bash, especially a
175.
by pama 1y ago
Package managers or old school download from a website (gnu) provide a separate public checksum and GPG signature in multiple mirrored locations. Even if their server was compromised I can still be safe.
176.
by pama 1y ago
I love uv. But the post starts with a simple install using a oneliner curl piping to sh, which is such a big attack surface area… I would much rather have a much longer one liner that increases safety.
177.
by pama 1y ago
I would test how much of the total flop capability of the hardware you are using. Take the first order terms of your model and estimate how many flops you need per data point (a good guide is 6*param for training if you mostly have large mu
178.
by pama 1y ago
Thankfully they are unconnected by default. Unfortunately the sidebar can get triggered accidentally with keystrokes from muscle memory (Emacs) and then one has to deal with the side bar. It would have been nicer if by default nothing exist
179.
by pama 1y ago
I have been out of it for a while and dont miss it much, but I thought that the rule of zero, not five, was the modern goal.
180.
by pama 1y ago
I guess I should have added “periodic” structures, explained self assemble, ... oh well. So, No, nature didnt figure out anything close to these constructs until it evolved humans who did it. Zeolites exist in nature but they are not made

More ›