Divyansh Chaurasia

2.6K posts

Divyansh Chaurasia banner
Divyansh Chaurasia

Divyansh Chaurasia

@AsDivyansh

building. mostly llms, genai, and mlops

nan Katılım Haziran 2018
1.2K Takip Edilen1.5K Takipçiler
Raghav Chadha
Raghav Chadha@raghav_chadha·
1 extra kg of baggage and airline bills you exorbitantly. But 4 hours of flight delay and airline owes you nothing. Airlines track your baggage weight to the gram. Why don’t they value your time to the hour? If they can charge you by the gram, they must compensate you by the hour. Today in Parliament, I asked the Civil Aviation Minister whether they are working towards a compensation framework for passengers affected by significant flight delays!
English
995
4.8K
22.1K
660.3K
Divyansh Chaurasia
Divyansh Chaurasia@AsDivyansh·
Chatgpt prompt: Give me 50 pain points that I can talk about in parliament. Make no mistake, give me list it should feel like I care for people. 😤
Raghav Chadha@raghav_chadha

1 extra kg of baggage and airline bills you exorbitantly. But 4 hours of flight delay and airline owes you nothing. Airlines track your baggage weight to the gram. Why don’t they value your time to the hour? If they can charge you by the gram, they must compensate you by the hour. Today in Parliament, I asked the Civil Aviation Minister whether they are working towards a compensation framework for passengers affected by significant flight delays!

English
0
0
2
140
Guohao Li 🐫
Guohao Li 🐫@guohao_li·
Just arrived in SF as employee #001 at Eigent US Fun facts: > Turned down an offer from Meta to join a startup > Cancelled an offer from Stanford and started a company in London > Missed SF twice… now finally here building 🇺🇸 Folks interested in chatting about agent environments, RL, and self-evolving agents - let’s catch up!
English
34
2
169
10K
Hubert Thieblot
Hubert Thieblot@hthieblot·
Only incredible founders can reply to this tweet
English
447
1
508
44.3K
Divyansh Chaurasia retweetledi
Jen Zhu
Jen Zhu@jenzhuscott·
When I was consulting for @HBO Silicon Valley, zero-loss compression was the holy grail Richard Hendricks chases that perfect middle-out algo could shrink everything w/out breaking a single bit. Google just did something even more practical for the AI era: TurboQuant compresses LLM key-value caches down to 3 bits per value using random orthogonal rotation + PolarQuant scalar quantization & optional 1-bit QJL residual correction. =>> 6× memory reduction, up to 8× faster attention (on H100), & 0 degradation on LongBench, Needle-in-a-Haystack, and RULER for models like Gemma. No retraining, no calibration needed. Fiction just got out-engineered by reality. 😅💚💚
Jen Zhu tweet media
Google Research@GoogleResearch

Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI

English
160
701
8.8K
1.2M
Divyansh Chaurasia
Divyansh Chaurasia@AsDivyansh·
Chatgpt's thinking mode is so useless now. It feels like the openai is running short on compute, so they degraded the models and added thinking to make it work normal. @sama or is it by design? If it's by design then please reconsider.
English
0
0
0
73
Divyansh Chaurasia
Divyansh Chaurasia@AsDivyansh·
it makes sense why so much of the world’s innovation happens here.
English
0
0
1
68
Divyansh Chaurasia
Divyansh Chaurasia@AsDivyansh·
@nikitabier - can we have events directory somewhere in X? Like where I can filter what events are happening in my area and sign up for it?
English
0
0
0
76
Divyansh Chaurasia
Divyansh Chaurasia@AsDivyansh·
We’re hosting an in-person AI Agents meetup next week in Menlo Park, CA with @p0 and @Snowflake.✨ If you’re interested in joining, you can register here: luma.com/z2qae6kx See you there next week!
English
1
1
4
229

Keşfet

@Fried_rice @raghav_chadha @guohao_li @agent_wrapper @hthieblot @HBO @sama @NHAI_Official