cheald

16 ms·

151.
by cheald 3y ago
Indeed. I've known that phenylepherine is worthless for a long while, and just make it a point to buy a pack of the real stuff every 6 months or so, so that I always have it on hand when I need it. In my state, it's controlled - y
152.
by cheald 3y ago
My guess is that it's not a very interesting problem because it's not particularly difficult to add spatial dimensions to arbitrary audio - after all, it is already commonly done in video games. All you have to do is manipulate th
153.
by cheald 3y ago
Dazzle camo and decoys seem like the obvious answer, and anything flying is still gonna have to contend with AA weapons platforms. It doesn't seem like it would be the hardest problem in the world to honeypot an autonomous drone into a
154.
by cheald 3y ago
It helps to think in terms of `any` first: any([], predicate) => false This is plainly obvious, because it is illogical to say "the predicate is true for at least one member of the empty set". If we can agree on that, th
155.
by cheald 3y ago
I'm pretty sure that it's just serially summing the network weights, which results in an accumulated offset to the self-attention layers of the transformer. It's not doing any kind of analysis of multiple networks prior to ap
156.
by cheald 3y ago
I'll bet that if you looked at calorie intake, you'd find a similar pattern. I'd be willing to bet that it's not because the person is eating more to heal their body (at least in the first world, where we have to work to
157.
by cheald 3y ago
I started by holding onto a stair bannister and squatting slooowly until I could actually get the full range of motion, then unsupported air squats, then I moved to an empty barbell. If you can't go straight to a barbell, or you have b
158.
by cheald 3y ago
It's definitely not magic, but it's on the 80 side of the 80/20 split, IMO. It isn't the right starter program for everyone, but then, nothing is. The best program for the beginner is whichever one keeps you consistently
159.
by cheald 3y ago
But you're still lifting, presumably. That sounds like success to me. I tried to start lifting multiple times. SL 5x5 is what actually made it click. Now that I know what I'm doing, I could write volumes on its flaws, but what I n
160.
by cheald 3y ago
This is a very silly argument to be made against a lifting program which has verifiably moved countless people off the couch and into resistance training. The reason people recommend Starting Strength is because it's what successful
161.
by cheald 3y ago
It still doesn't change the fact that a suboptimal program which keeps beginners engaged and moving forward is infinitely superior to a more complex one that they fall off of.
162.
by cheald 3y ago
I'm 40, I've had two knee surgeries, and have genetically terrible knees. Squatting incorrectly made them hurt, but I went to see a sports PT and he taught me how to squat correctly (I was overloading my quads and not using my pos
163.
by cheald 3y ago
Starting Strength is fine for a beginner. It's not an optimal program, but it's simple, easy to understand, and built on good foundational principles. As an on-ramp into resistance training, it's better than a complex program
164.
by cheald 3y ago
That's more correctly called an "entitlement" rather than a "right", IMO.
165.
by cheald 3y ago
Interestingly, the principle of single responsibility is specifically why I like GraphQL over REST. As your frontend complexity grows, your REST payloads tend to become carefully-balanced collections of complex shapes, and it becomes very d
166.
by cheald 3y ago
I feel like it's improved steadily over time. I used to find it severely lacking for anything code-related, but that's improved, too. It's not as good as Google was at its peak (and Google itself has degraded severely in qual
167.
by cheald 3y ago
I'm very glad for this! I've been using Brave search almost since it launched, and the standard results have gotten great, but it's always been a bummer to have to go out to Bing/Google for images/video. It's r
168.
by cheald 3y ago
People wealthy enough to utilize retirement accounts are also likely to be paying higher tax rates in general, though.
169.
by cheald 3y ago
On the other hand, GPTs often produce buggy code, and script kiddies are a lot less likely to be able to debug AI-generated code than they are to be successful running another human's hand-written exploit.
170.
by cheald 3y ago
Toxicity is defined by serum levels, and is at around 100 ng/mL. My last labs put me at 39 ng/mL, and I've been taking vitamin D for years. Under 30 is considered insufficient, under 20 is deficient. I mean, yeah, sure, don&#
171.
by cheald 3y ago
For fun, I took two shots of Sonia from the third TOTK trailer (who should be novel and not in the model), trained a LoRA for 400 steps on just them (based on the AbyssOrange anime model), then threw it into an entirely different model (D
172.
by cheald 3y ago
I get bloodwork done twice annually, so I have a very accurate idea of what my levels are. The dosage is the recommendation from my doctor based on my bloodwork. The goal isn't just staving off seasonal affective disorder, but actually
173.
by cheald 3y ago
I get what you mean, but my point is that fundamentally, what's happening in the LoRA and in the human artist is conceptually similar - each learns through examples what it is about a character that makes that character distinctive, an
174.
by cheald 3y ago
Nonsense. A neural net of sufficiently high parameter count can learn a function to reproduce any digital representation of information. This argument would suggest that anything which can be represented digitally is therefore of minimal va
175.
by cheald 3y ago
By the same token, you had no ability to conceive of any original character before it was shown to you; why would an AI art model be any different? A fan artist must observe a character's official depiction before they can produce thei
176.
by cheald 3y ago
A lot of people in the first world are heavily deficient in vitamin D, and deficiency is linked to a whole host of health issues and vulnerabilities. It's dirt cheap and essentially completely safe below toxicity levels (which are quit
177.
by cheald 3y ago
Cialis has been linked to mitigation of cognitive decline: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6705107/ https://www.nature.com/articles/s41598-019-53136-y And Viagra has been id
178.
by cheald 3y ago
A lot of people are having success by adding extra networks (lora is the most common) which are trained on the type of image you're looking for. It's still a raster image, of course, but you can produce images which look very much
179.
by cheald 3y ago
Is there a trick to getting pytorch+cu121 and xformers to play nicely together? All the xformers packages I can find are torch==2.01+cu118. Edit: After a bit more research it looks like scaled dot product attention in Pytorch 2 provides muc
180.
by cheald 3y ago
You can torrent a copy of Wikipedia, including article history. Locally, you can go back to any revision of any article you want. I keep a copy locally just because it seems something valuable to have.

More ›