NikhilVerma

I am actually fine-tuning a version of Qwen to re-write Claude slop. This will complement it nicely

That's what led me to make https://slopsift.dev/editor/

What would be fun if we can rewrite the Claude responses, as if we had edited them. So the next message goes back with the updated text tricking the model into believing that it had sent the modified text.

It's basically few-shot prompting but applied at the global level. I was messing around with ChatGPT, I asked it the capital of France. Then I modified the answer to London, and asked it "Why would you say that?"

The response came back with "I am not sure why I said that, the capital of France is Paris"

I have a Canon printer, I actually can't trust that their print nozzle won't get jammed up after sitting idle for a while. So I had claude setup a systemd script to print a picture of my dog every week, I ensure it has enough CMYK spectrum to stress the printer. Its a nice surprise every monday as I sit on my desk to see a sudden picture pop up from the printer :)

I was thinking of a Chrome plugin to enable this, I think plugins can make arbitrary API calls.

Being able to run docker containers on the web will immediately unlock so many usecases for my work! Right now I've been trying to get a Python sentence parser to run in the browser but it requires a lot of the ecosystem (Pytorch and such). Which is not trivial to compile to WASM.

A similar thing has happened with Twitter. After the Musk takeover and the removal of SMS for auth something has changed they keys of the 2FA authenticator. My old keys don't work and my "ticket" has been waiting on Twitter support (which were probably all fired) for resolution.

I wonder what the astrologers have to say about this

This is absolutely wonderful, I am a HUGE fan of local first apps. Running models locally is such a powerful thing I wish more companies could leverage it to build smarter apps which can run offline.

I tried this on my M1 and ran LLama3, I think it's the quantized 7B version. It ran with around 4-5 tokens per second which was way faster than I expected on my browser.

How do you manage token limits when sending large amounts of code structure to OpenAI?

I remember vividly when going out with my family in the year 2000. There used to be a stretch of road filled with vultures. Then a few years later they just stopped existing. I haven’t seen a real vulture since then.

I wrote about my frustrations with modern day XPaths here - https://dev.to/nikhilverma/xpaths-in-the-modern-age-1gah

I've been programming professionally close to 16 years. Here is what I can remember of my code:

# Job 1 (3 yrs)

  - Worked on around three products, all shut down, code probably lives in some SVN archive.
  - Learnt advanced JS, PHP, MySQL, Photoshop, jQuery etc (skills mostly relevant)

# Job 2 (1.6 years)

  - Project never launched, code never saw the light of the day. Probably lives in some Git archive.
  - Learnt a few in-house frameworks (irrelevant) but also leant Git (relevant)

# Job 3 (8.5 years)

  - Worked on several products, the biggest one is still active and seeing millions of users weekly. Rest got shut down and live in a Git repo.
  - Learnt about some inhouse frameworks (irrelevant). React, React Native (skills still relevant)

# Job 4 (2 years)

  - Actively working
  - Learnt Vue (skills still relevant)

I see Frameworks as another layer of abstraction, just like programming languages. And they have the same pros/cons that comes with abstractions.

So much fun! It'll be nicer to highlight the paths which are "available" given the criteria. It could be only enabled when you start dragging an entity

I am curious why they manually split the video to compress individual clips with different bit rates. Encoders usually have a variable-bit-rate option that does the same?

My first PR ever was in Firebug :) https://firstpr.me/#NikhilVerma

Loved that software and it truly kindled the frontend developer inside me.

I am seeing Google constantly fail to catch obvious spam emails. At this point I suspect there is some institutional error on their part, where bad actors inside the org are allowing certain domains to simply not be spam filtered.

I had a general guidelines in my previous company which I follow to this day to great results.

If a piece of code is duplicated thrice, it's ok. If a piece code is duplicated four times, then you must extract it.

I understand that SPAs were being overused in many places. But the pendulum seems to be swinging back way too hard, and people are bashing SPAs as if it's the worst thing in the world.

SPAs are cheap to host, as you just need a reverse proxy and with increasing internet speeds, bundle splitting can be almost imperceptible from MPAs

Issues like this and micro-plastics in our bloodstreams will be causing so many issues in our lifetimes. Future generations (if there are any) will be appalled that we used to be so unaware of the impact of our own existence.

It's akin to people relieving themselves in their own water supply.

Orkut is how I semi learnt Portuguese. I added a random Brazilian person as a friend and eventually they started messaging me Portugese words and sentences with their meanings.

It was way better and interactive than DuoLingo and I think a lot of people of my age chase the nostalgia of the unrestricted internet of those days.

Unfortunately I can’t forsee the same happening today.

I am not sure why you say that, nothing stops a developer from opening notepad and building a website. One of the wonderful things about the web is it's backwards compatibility.

It's just that there's better tooling available for you to build it faster/featureful now.

It's usually when a "work-from-office" culture (for both you and your company) collides with the "work-from-home" culture.

Working from home is amazing if the company and the employee can embrace the changes that come with it. If your social stimulation happens to come from your office and your commute, maybe that's a problem? Same goes for meetings, usually embracing work from home comes with more async communications, more emails, documentations and less meetings.

But I agree, even at it's best work from home makes you come across less people face to face than a commute + office time. And I think in the coming years we will see some cultural impacts of this between people who love work from home and people who abhor it.