DSPy shifts your focus from tinkering with prompt strings to programming with structured and declarative natural-language modules. For every AI component in your system, you specify input/output behavior as a signature and select a module to assign a strategy for invoking your LM.

Almost always soft delete I will reiterate that storage is cheap and recovering data is a nightmare. If you have some domain specific need to delete (or otherwise mark as irrelevant) some data, use a nullable timestamptz column. If there is a timestamp filled in, that's when it was deleted. If there is no timestamp it isn't deleted yet.

Represent statuses as a log It is very tempting to represent the status of something as a single column. You submit some paperwork and it has a status of submitted. Someone starts to look at it then it transitions to in_review. From there maybe its rejected or approved. There are two problems with this You might actually care about when it was approved, or by whom. You might receive this information out-of-order. Webhooks are a prime example of the 2nd situation. There's no way in the laws of physics to be sure you'll get events in exactly the right order. To handle this you should have a table where each row represents the status of the thing at a given point in time. Instead of overloading created_at or updated_at for this, have an explicit valid_at which says when that information is valid for.

ways for a Ponzi scheme to end:The bad way is you get arrested.One good way is some exogenous good event occurs (you win the lottery), you get a lot of money, and you are able to pay off all your investors.The other good way is some exogenous bad event occurs (crop blight destroys all the legitimate businesses in the industry that you are pretending to be in, etc.), and you are able to plausibly tell your investors “sorry all of your money is gone because of the exogenous bad event, not because I stole it, who said anything about stealing it.”

Most Ponzis—Bernie Madoff’s unprompted confession is the famous exception—end with the government shutting them down, often with the Ponzi schemers strenuously insisting that everything was going great and that valuations and liquidity only collapsed because of the government’s actions.

Gosling primarily uses the NetBeans IDE for development, praising its open source, Apache-licensed nature and dedicated community. He expresses frustration with developers who cling to outdated tools: “The thing that drives me nuts the most are people who are madly grasping the ’80s or the ’70s — people who still want to use Vi, which was high-tech in the ’70s.”

This is sort of a written-down blueprint of how I become competitive when joining a job. Being competitive, to me, means: being able to work on most of the issues I want in the company in some fashion having accumulated a certain stature and respect from both the engineering and business side which I can trade for space and agency being in a place where I can break company soft/hard rules to get things done. I want people to think of me in the top 5 if there was to be a tier-list of engineers and if possible, ASAP.

It would mean more services. More logins. More API keys. More configurations. More docs. More suffering. It was too much bear. Leave as future work.

But the most interesting part to me was that I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings and configuring and gluing a monster.

I had to go over to Google Cloud Console and create a new project, and add a new OAuth Credential. I had to wait some time for an approval process around here. I then had to go back and forth between the nested settings of all of Vercel, Clerk and Google for a while to wire it up properly. I thought of quitting the project around here, but I felt better when I woke up the next morning.

Andrej’s original tweet in full (with my emphasis added): There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like “decrease the padding on the sidebar by half” because I’m too lazy to find it. I “Accept All” always, I don’t read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I’d have to really read through it for a while. Sometimes the LLMs can’t fix a bug so I just work around it or ask for random changes until it goes away. It’s not too bad for throwaway weekend projects, but still quite amusing. I’m building a project or webapp, but it’s not really coding—I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.

Stites-Clayton and Fontenot, his man bun recently shorn, opened the presentation by rapping about the virtues of the local tech scene: “We the owners, billionaires dressing like stoners, home of the dripped-out technology brothers.” Then came a swift parade of founder pitches, two minutes apiece.

How DKIM SPF & DMARC Work to Prevent Email Spoofing

If it’s a core business function — do it yourself, no matter what.

Market Research: You can identify competitor presence, market gaps, and consumer sentiment.SEO (Search Engine Optimization): It can help you create competitor strategies, identify keyword opportunities, and track content performance.Academic Research: Google SERP scraper can be used for collecting all available content on a topic from Google search.Journalism: Journalists can use it for research, finding articles, social media posts, and all content found in Google related to a topic.Fact-Checking: You can save time skipping manual search and find the facts faster.

A SERP API lets you automatically collect Google’s search results data without manually performing searches yourself.

f you were a salesperson at a big investment bank in August 2008, it might have been a good idea to call up some hedge funds and say “hey, we are hearing some bad things about Lehman Brothers. Do you feel safe with them? Might be a good time to move your accounts over here.” The hedge funds might have believed you — you would have been correct! — and moved their accounts to your bank. You would have done more business and made more money. At the end of the year, you could go to your boss and say “I won all these accounts from Lehman, give me a big bonus.”And then your boss, if she was smart, would have fired you.If you were the chief executive officer of a big investment bank in August 2008, and your subordinates were doing that sort of thing, how would you have felt about it? Three possibilities are:Good: They were being aggressive about winning business, and winning more business is good.Bad: Investment banking is a genteel business, and badmouthing a competitor is rude.Very bad: The banking system was in perilous shape, and doing anything to undermine confidence in the system — like, for instance, spreading rumors that a competitor was collapsing — would have negative effects on your bank. If Lehman failed, or even looked likely to fail, you might be next. Better to prop up confidence in your competitors than to tear them down. Not all the time! In good times, sure, compete fiercely. But in a crisis, be extremely polite.

For example, imagine you have a clientId at the start of a request, but you’d like for that ID to be available on all spans in a trace, some metrics in another service, and some logs along the way. Because the trace may span multiple services, you need some way to propagate that data without copying the clientId across many places in your codebase.

The build tools listed above start off with a client-only, single-page app (SPA), but don’t include any further solutions for common functionality like routing, data fetching, or styling.

Cloudflare Workers, for example, are serverless JavaScript functions that run on the Cloudflare CDN. They can respond to a variety of events and inputs, including device type, time of day, user location, or data from third party APIs. Based on these parameters, dynamic content can be generated and served to client devices, or static content can be altered, cached, or purged from the cache.

A hypothesis was established: Presence of a face in an artwork causally improves engagement.

vercel others And those are great for high availability and horizontal scaling but....I don't have those problems and those platforms can get expensive quick compared to having your own VPS and running your apps on it.

On a practical level, SoftBank’s cash infusion helped WeWork cover the increasing costs of its whirlwind expansion as the real-estate market got more expensive. It also began spending heavily to fill all the desks it was adding. Just a few weeks after SoftBank’s investment, Shlomo Silber, the owner of Bond Collective, a  New York–based co-working company, turned on his phone at the end of Rosh Hashanah to find dozens of his customers had forwarded an email from WeWork offering to buy them out of their leases and give them as much as a year of free rent. WeWork’s occupancy rate went up, but the deals made it difficult to determine the natural demand for its product.

Inoreader. It allows me to add rss, twitter and youtube and apply rules and filters as to what to include / exclude.I don’t like twitter but sometimes that’s the only way people communicate or share their art. Inoreader lets me add twitter users with a filter (e.g not interested in tweets without pictures (artists), not interested in retweets, only interested in tweets that contain ‘released’, ‘beta’ or ‘rc’ (software).Really cool for trimming down the noise and a smooth, clean experience.

Analysts and technologists estimate that the critical process of training a large language model such as GPT-3 could cost over $4 million.

Sorry for whatever you’re going through. Hugs.

REWEAVING Don't part with your best denim friend. Employ our proprietary reweaving technique to repair rips and snags. This unique technique pairs soft gauze fabric with a blend of cotton and polyester threads carefully selected to match your denim. Don't worry if the fresh repair feels a little stiff. It will soften with wear and washing. RX NOTE - Ideal for jeans with 90% cotton content or higher, which is why jegging repairs may not fall under warranty. ESTIMATED COST - $10/inch

wnsampling. Since diffusion models offer excel-lent inductive biases for spatial data, we do not need the heavy spa-tial downsampling of rel

ampling the initial denoising steps, DMsare still computationally demanding, since training andevaluating such a model requires re

mocratizing High-Resolution Image Synthesis DMsbelong to the clas

mean many are tightening their purse strings.