Pinned Tweet
yoo guys this is my first post on X.
- didn't make it last year but
- got accepted in gsoc this year lessgoo🚀
much more ahead!
would love to connect with folks interested in oss, web3, ai, full stack.
#gsoc
9
23
2,691
X premium free for a month, thank me later!
Expires on: Monday, November 30, 2026 at 10:30:00 PM
Redeem At: grok.com/plans/referral/?sku…
Readers added context they thought people might want to know
The post's claim of "X premium free for a month" via referral is inaccurate. Official program offers eligible new users a 14-day trial requiring a credit card at checkout with auto-renewal unless cancelled; referrers must meet US eligibility rules and receive usage resets.
x.ai/legal/referral…
docs.x.ai/x-subscription…
i'm waiting for @Alibaba_Qwen to drop qwen 4 27b
which local model do you use, any recs?
1
336
Khushal Agarwal retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
Khushal Agarwal retweeted
software engineers are going to have to tackle a new set of problems to unlock powerful multiagent systems to solve the world's hardest problems - right now is one of the most interesting moments in history to be an engineer.
finally!!
btw if you find me chronically active, that's because i want to make every penny i spent on premium worth it 🥹
1
175
fyi: claude haiku 5.5 (high) is 6.6× cheaper than kimi k3 and 9.4× cheaper than gpt-5.6 sol xhigh.
$0.44 vs $2.92 and $4.14 per task.
it still scores 4.44% net improvement—3.2× kimi’s 1.40% and way above deepseek v4 pro’s 0.03%.
151
9,980?
chat, the best way to learn something is to fuck around and find out. you learn more by breaking things than any overpriced course could ever teach you.
play around with stuff, read engineering blogs written by senior devs, read books, touch grass and save your money.
Launch of sys0.in was incredible 🎉
Got thousands of registrations, people are loving it 💛
Except few on x - for you new video is coming soon 👌
4
155
5 more to go! hoping to reach 100 by oct end.
if you're into oss, llm's and cloud native infra let's connect :)
1
94
Out of curiosity, I decided to give the red dot on my laptop a try, and honestly, it’s actually super fun to use! @Lenovo
1
3
343
