initlayers

@Neilblaze007 Very much

@initlayers Missing childhood already?

At this point I am working on Sundays too. Every day, every week, every month. It just keeps going.

After going back and forth between Claude and Codex, I finally spent ₹2300 on Claude. And this might be one of the best purchases I've made. Forget Claude Code, even Claude Chat feels incredibly intuitive. The explanations are clear, the visualizations actually help, and things just click faster. Didn't expect to like it this much.

@brutebomby @_vmlops @real_deep_ml @cneuralnetwork @lochan_twt @saurabhtwq @TheEigenNerd @neural_avb @hkproj @aladdin_persson @yacinelearning @mrdbourke @yellow_void22 @herooffjustice @NeuralNine I don't think so. Learn as it comes to you.

@initlayers @_vmlops @real_deep_ml @cneuralnetwork @lochan_twt @saurabhtwq @TheEigenNerd @neural_avb @hkproj @aladdin_persson @yacinelearning @mrdbourke @yellow_void22 @herooffjustice @NeuralNine yeah ill be starting in 2-3 days after I go over for a quick revision on the math required. Any tips or beginner advice type thingy?

@OjasSharma276 Bro you are inspiring. Let's connect.

Tech Twitter posts in a nutshell: 50%: AI will eat your jobs 20%: Dev vs DSA debates 15%: Meet this guy… threads 10%: Striver/Harkirat reach farming 5%: Let’s connect posts

Working two jobs feels like doing one properly and just surviving the other.

@_notapenguin @datavorous_ I referred him, he knows me 🥲

@datavorous_ @initlayers

Is there anyone here working at the language research lab at IIIT H? I'd like to have a chat. I will be doing gpu inference work there, this coming summer.

I have a meeting tomorrow morning, but I still went out to watch Dhurandhar: The Revenge. Honestly, the best risk I've taken in a while. Enjoyed it to the core. Now let's just hope I don't mess up the meeting tomorrow.

@Abde_Mustafa0 I don't like their reports for some reason. It feels very niche and irrelevant most of the time.

@initlayers Gemini deep search is still better

Tried Deep Research on ChatGPT again after a long time. Still not there for me.

@RobertJoo2 I am not sure

@initlayers better than before though

@RobertJoo2 You can say that

@lifeofDuggal Thanks. I will check it.

@initlayers Hey, I have DM'ed you my resume

We're looking for research assistants or early-career research engineers at the Language Technologies Research Center, IIITH. If you're genuinely motivated to learn and build, this could be you. The work involves designing workflows for data ingestion, working with multiple ML/AI models, and building search and indexing pipelines. Familiarity with cloud infrastructure or building optimized standalone applications is a strong plus. You don't need to know everything already. If you're serious and willing to put in the work, you'll learn on the job. There's proper support, mentorship, and technical training to help you grow while you build. If this sounds like you, reach out.

If you've noticed, I've been posting less lately, the reason is: I'm currently working with two labs, and there's a meeting every 1-2 days. Most of the work moves around deadlines, so the pace can get pretty intense. Usually I only get 2 relatively less hectic days in a week. Yes, I work on weekends as well.

@Dojo__0 Thanks. I'll check these

@initlayers If you're interested in local attention do read these papers one is "Hiera: Hierarchical transformers without the bells & whistle" then another paper I belive it was window attention is bugged

It's a Saturday afternoon and I'm preparing a presentation to explain to my university what I'm doing during my internship. The funny part is most of the technical details probably won't make much sense to them anyway. Still have to make the slides, simplify everything, and explain months of work in a few minutes. Sometimes that's just how these things go.

@ThedatagGuy I read the abstract, introduction and directly skip to the architecture diagram and see what they are doing. I then try to break things down. Understand methodology.

@initlayers What's ur approach on reading papers?

Reading the Restormer paper today. The more papers you read, the easier it gets.

@zephanylh That's a brutal take. But true.

@initlayers Something I’ve come to realise is that intelligent capable people always can tell when someone has real skill vs when someone is faking it. Competence recognises competence, so if you find that you’re getting by by faking, it usually just means you’re surrounded by idiots.

It's very hard to fake real skill. Maybe you can get through an interview, but imagine having to sit in meetings two or three times every week and explain your work. If you don't actually understand what you're doing, it shows very quickly. Confidence usually comes from clarity, and clarity takes time. At some point in your career you have to sit down and spend those six months really learning something properly. There's no shortcut around that.