Hacker News: vjsrinivas

New comment by vjsrinivas in "US cities are axing Flock Safety surveillance technology"

vjsrinivas — Wed, 08 Apr 2026 19:46:53 +0000

These systems have provided incorrect data, and law enforcement often misuse that data to stalk, attack, and wrongfully arrest innocent people. Privacy matters to everyone - especially for ones who don't care about privacy.

New comment by vjsrinivas in "Beyond has dropped “meat” from its name and expanded its high-protein drink line"

vjsrinivas — Tue, 17 Mar 2026 23:58:37 +0000

Isn't that already being done to a certain extent with things like chili? They add TVP to supplement the meat, and people see it as a very negative thing.

New comment by vjsrinivas in "Beyond has dropped “meat” from its name and expanded its high-protein drink line"

vjsrinivas — Tue, 17 Mar 2026 23:56:12 +0000

What do you mean by overstating? It has really good protein numbers. The protein itself has an evenly balanced amino acid profile (or in other words - a "complete" protein). It has a good amount of calcium, iron, and low fat. You can technically make it yourself and there's numerous ways of cooking and flavoring it.

New comment by vjsrinivas in "Beyond has dropped “meat” from its name and expanded its high-protein drink line"

vjsrinivas — Tue, 17 Mar 2026 23:51:44 +0000

On top of that, there seems to be a pervasive misconception of the effectiveness of plant vs animal-based protein on things like muscle growth. Older studies showed that plant-based proteins had lower digestibility scores via metrics like PDCAAS. In turn, people interpolated that muscle growth would be lower. Some early studies comparing the two protein sources on muscle synthesis didn't do gram-for-gram comparisons and that increased the misconception. Newer studies are showing that, if you match the protein amount at or above the 1-1.6 g/kg for muscle growth, you will get the same level of muscle growth.

I feel like it'll take another 5 years for this "bio-availability" myth to die out.

New comment by vjsrinivas in "Show HN: How I topped the HuggingFace open LLM leaderboard on two gaming GPUs"

vjsrinivas — Fri, 13 Mar 2026 20:29:25 +0000

Great work and love the detailed breakdown. This is kind of tangential, but it reminded me of this work: https://arxiv.org/pdf/2310.12973 (Frozen Transformers in Language Models are Effective Visual Encoder Layers).

The paper puts out an interesting hypothesis that these LLM-derived transformer layers have the ability to "refine" any set of learned tokens, even in different modalities. I wonder if what you're seeing here is related?

New comment by vjsrinivas in "Ask HN: Share your personal website"

vjsrinivas — Thu, 15 Jan 2026 20:30:34 +0000

This is awesome! :]