sumset

youtu.be/HBMy-y2wb4I Father featuring Travis Scott directed by Bianca Censori

@apoorvcodes @Hi_Mrinal just read the post. holy😭 folder structure as architecture is insane.

@Hi_Mrinal @sumir_30 dekhra hai bhai kya kr rha hai ye

Some conversations which I think should be publicly available 1/n > I don't want to sound dumb but like, even just organizing code into folders counts as architecture right ?? >> That is what you call modularity, when you add folders you are just trying to separate concerns which is also a good habit. But there's the thing it's only a small piece of architecture. Modularity is more of a coding practice thing than a full architecture decision. What it does help with is debugging, testing, and keeping your code easy to maintain.

BULLY ON THE WAY NO AI

@Muskanjain0401 @smallest_AI @apoorvcodes @apoorvcodes bhai wo sf...

This 19 year old is an AI Researcher at @smallest_AI and he’s already done what most people don’t in a decade: > @apoorvcodes Started coding at 14 by building a Discord bot and scaled it to 500K+ users in just a few months > Won major hackathons + science fairs in India, built real-world ML/robotics projects, and even got invited to meet PM Modi > Built viral AI products as a teen founder, got attention from top founders/VCs, and is now moving to SF Had him on the podcast today, and his journey is unrea!.

New Uttar Pradesh is embracing the power of Artificial Intelligence. A ₹25,000 Crore MoU with Puch AI will bring AI Parks, large-scale data center infrastructure, AI Commons, and an AI University to the state. This initiative will strengthen governance, drive innovation, and create future-ready opportunities for our youth.

@neet_sol @apoorvcodes

they’re calling me the unemployee of the month

@PGC1a_RB excited to follow this. maybe do it somewhere where it's easier to track along and not be at the mercy of the algo to show your posts. wdyt?

Pre-bed supplementation [for sleep/sleep 'optimization'] tonight will be nothing and for a few days, we'll see No melatonin, epitalon, pinealon, magnesium, glycine, taurine etc, supplements used for getting to sleep easier (relaxation, GABA, etc) or fixing things associated with sleep (circadian rhythm) My usual pre-bed supplementation is alpha lipoic acid, ubiquinol, nattokinase, quercetin, berberine. These are more for the improvement of certain metabolic processes and related blood flow, not directly, maybe very minorly indirectly related to getting to sleep

@HansAmato GUT. Appreciate it :)

After 100+ personalized health reports, one pattern keeps showing up: Most gut problems start with inflammation, not bacterial imbalance. Calm the inflammation first → everything else stabilizes So I built the Gut Stabilization Reset Guide showing you exactly how to lower inflammation to fix your gut - starting with the new year. Want it? Like this post + comment "GUT". Must follow & RT so I can DM you.

@apoorvcodes 😭

What a tiresome Sunday 😭🙏

@luo_fuli14427 amazing breakdown and incredible work!!

MiMo-V2-Flash is live. It’s just step 2 on our AGI roadmap, but I wanted to dump some notes on the engineering choices that actually moved the needle. Architecture: We settled on a Hybrid SWA. It’s simple, elegant, and in our internal benchmarks, it outperformed other Linear Attention variants on long context reasoning. Plus, a fixed KV cache just plays way nicer with current infra. Note: Window size 128 turned out to be the magic number (512 actually degraded performance). Also, sink values are non-negotiable—don't skip them. MTP (Multi-Token Prediction): This is underrated for efficient RL. Aside from the first layer, it needs surprisingly little fine-tuning to hit high accept length. With a 3-layer MTP, we're seeing >3 accept length and ~2.5x speedup in coding tasks. It effectively solves the GPU idle time from long-tail samples in small-batch On-Policy RL. We didn't get to squeeze it into the RL loop this time due to deadlines, but it’s a perfect fit. We open-sourced the 3-layer MTPs so you can develop with it. Posttrain with MOPD: We adopted On-Policy-Distillation from Thinking Machine to merge multiple RL models, and the efficiency gains were wild. We matched the teacher model's performance using less than 1/50th the compute of a standard SFT+RL pipeline. There’s a clear path here for a self-reinforcing loop where the student evolves into a stronger teacher. Huge props to my team. They sculpted these ideas from scratch into production in just a few months. Full breakdown is in the tech report. If this kind of pragmatic engineering resonates with you, we should talk.

@trulightwaves @BLUECOW009 ai is such a general term for this

But how do you forget that you forgot? Catastrophic forgetfulness.

@BLUECOW009 heeeyyy

who want to help me add thje people who lliked this post

@BLUECOW009 hey🫡

im starting a group to talking about my new project the holographic memory and more, like this or react to be added

@b1tesnails not without a deficiency

is magnesium the answer

@kadirnardev hi kadir, can you please dm me. i can't as a non premium user. appreciate it.

Please stop playing with all the video models and read this article. They've shared amazing insights, especially about data. I'm eager to apply the techniques they used for Speech2Speech models to TTS models 👨‍💻

@shri_shobhit "So slow that I even realized it pretty late in life :)" wow

Sadly, I am a slow learner. It takes me weeks to understand something that my friends grasp in days. OTOH, once I put the effort, I "really" get it! So slow that I even realized it pretty late in life :) This realisation, and planning for it, helped me a lot!

@FashionCartier there is no music

Why does his music never leak

@TheKhelIndia same pattern in the last fast time control event. incredible day ones followed by poor days.

So it confirmed, Gukesh will finish at the last place in Clutch Chess: Champions Showdown! After Incredible Day 1, Gukesh had poor Day 2 & 3 Still need to work more on the shorter format! COMEBACK MUCH STRONGER CHAMP 🇮🇳💙

excited to be a part of the team, try luna at heypixa.ai/chat

@sparsh_17 🙌 finally lfg

Where is India's AI? Every WhatsApp group, every conference hallway, every founder call asks the same question Today, we’re sharing the answer Meet Luna, world’s first speech-to-speech foundational AI model to unify audio, music and speech.