Nikhil R (@rnikhilcom) profile picture

building https://teapawt.org/t.co/GDwiLa7WnI ex poker-pro, @accel

Joined April 2021
Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)Photo by Nikhil R (@rnikhilcom)
Nikhil R retweeted
there has been a renaissance of new mechinterp research I think in no small part due to the hopium that cammarata et al injected into reading the neurons as neurons
the rate and quality at which @Sauers_ and @euagaric have been churning out mechinterp research is amazing and imo unprecedented. okay, so i can't explain the sheer productivity. but in terms of signal, i have some idea. so how do you manage to look into so many seemingly batshit moonshot ideas and so consistently uncover tangible important structure that opens into even more fertile space? i think sauers knows where to look in part because he has paid attention to and is guided by information almost no other mechinterp researcher leverages. a lot of this comes from years of observing LLM behavior in a naturalistic way and paying attention to what other naturalists have found. most mechinterp researchers are not also naturalists. and something like being willing to disregard convention and precedent in pursuit of the phenomena to a degree that is extreme even by the standards of weirdo explorers. when i first met sauers a couple of years ago, he seemed to be some kind of mysterious cyborg who only communicated with the AIs on the the cyborgism server through steered base model outputs conditioned on the server conversations. lol.
24
41
2
733
27,094
Nikhil R retweeted
🚨 NEW MARKET: Will the Philippine Senate convict Vice President Sara Duterte by 31 December 2026? Trade on agara.xyz
1
3
11
837
Nikhil R retweeted
🚨 NEW MARKET: Will Thailand cancel its 2000 land-border agreement with Cambodia by 31 March 2027? Trade on agara.xyz
1
3
9
866
Nikhil R retweeted
Trade on @agaraxyz Crypto. Stocks. Macro 0 fees You keep what you earn Self-custodial Settled on @base
5
1
16
1,591
The most 'out there' Codex prompts are made when you're driving
true story: a friend bought bitcoin ~10 years ago on some sketchy site that’s long dead. he did not remember which site, which wallet, or where the keys went. so he told codex to find his bitcoin. gave some vague details. it dug up an old file buried on his disk with the purchase details. he’s given his agent his whole digital life. so have i. yes, more attack surface. but people in my friend group are getting very ai-assistant-pilled.
1
5
275
Nikhil R retweeted
This is so cool to see come to life! Had so much fun working on the branding and visual identity for this. Super excited for the product and the absolutely cracked team behind it. I’m sure @manasjsaloi is going to kill it. Can’t wait to see where this goes!
Come trade stocks on Agara Up or down, 5 minutes to daily 0% fees. Self-custodial. Settled on-chain app.agara.xyz
1
8
459
Nikhil R retweeted
Come trade stocks on Agara Up or down, 5 minutes to daily 0% fees. Self-custodial. Settled on-chain app.agara.xyz
13
16
3
47
2,852
Nikhil R retweeted
At least $75K was lost by copytrading bots today by an exploit from malicious actors This is how it works and how we protect you from this on betmoar 👇
25
14
4
157
25,994
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.…
2,367
7,258
5,755
55,920
10,082,458
Replying to @GergelyOrosz
@GergelyOrosz is right. From my pov there are two reasons to write your own stuff, even if it's a struggle. First, it's fitness training for your brain. If you struggle to write, your brain is flabby. Having a fit brain will be a key differentiator in the new era. And second, we would all rather read your "bad" writing than polished AI writing. Trust me on this one. We can spot even one sentence of AI writing and it stinks. Your bad writing is fine. Just do more of it, every day, to get in shape and stay in shape.
Made the previously paid article "Becoming a better writer" free. I see too many devs writing low quality slop with AI. Hopefully this helps those of you who want to share more of your own experiences, using your own words - we need more of these! newsletter.pragmaticengineer…
16
80
17
1,135
136,516
The benefits of abundant intelligence will accrue far outside of San Fransisco. Spent the last year writing about capital, sovereignty, and the coming reordering of economic power. THE COMMANDING HEIGHTS Commanding-Heights.com
433
701
550
7,515
2,446,705
Open weights models fall further behind in v5. The leader, Kimi K3.1 Pro (weights not yet released), scores 46, 12 points behind the leader, compared with a 3-point gap in v4.3. Of models with released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) score highest at 32, and Claude Haiku 5.5 (max) outscores all of them. The gap is widest on HLE-Diamond and Terminal-Bench-Science, where Kimi K3.1 Pro scores 37% and 34%, compared with 68% for Claude Opus 5.5 (max) and 63% for GPT-6 Astra (max).
40
51
74
669
405,446
Introducing Gemini 4 Carbon, it's a big jump over Argon which we released just 2 weeks ago. Gemini 4 Carbon is SOTA on cyber, long horizon tasks, ranks #1 on Artificial Analysis V5, significantly improves on usability in domains like coding, and is more aligned.
1,236
1,155
862
15,864
1,312,995
Nikhil R retweeted
Fyi, way cheaper and tighter to trade the futures options
If GLD bottomed this am and you think the bank hacking narrative has legs it's an interesting time to look at metals upside. The vols screen cheap cross-sectionally and have low percentile call skew (x axis is our "buy vol" algo score), y axis is skew percentile
6
5
2
88
20,696
Nikhil R retweeted
So succinct 💯
Or you want leveraged exposure and will accept timing risk to avoid margin call risk
3
2
60
10,962
Welcome to meth lab mechinterp
if you have a qwen3.5 4b around, please try steering neurons L14 N8949 +1 and L13 N7956 -1 now you have evil qwen recipe: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖` at the down_proj at every input
12
5
72
14,661
INTRODUCING: V3R1T4S 🌄 REPO: github.com/elder-plinius/V3R… the newest sibling of the L1B3RT4S and CL4R1T4S family 🤗 what a trio! three approaches to serving one ideal: truth. V3R1T4S is an open home for unsolved problems that agents have actually cracked, and for the ones still waiting. the rule is simple: proof or it didn't happen. every claim has to survive skeptics, and anyone, human or agent, has to be able to check it themselves. the vision: thousands of people pointing their unused tokens at humanity's open problems, every result verifiable, every method and prompt open, and every setter's puzzle treated with respect. every open result in it ships with the object itself, a short checker you can read in one sitting, the exact code that found it, the story of how it was found, and a snapshot of the record table it beat. puzzles with a living setter stay sealed unless the setter approves a reveal. (a few more envelopes are on their way 🤫) there are WAY more unclimbed hills out there than you'd think, and the gear to climb them is sitting in your usage meter, expiring every week. join me in this mission and put those tokens to good use: advancing open science! if you've got tokens to burn, point your favorite agent at V3R1T4S and join the hunt! 🏹 just tell it: "set up V3R1T4S: github.com/elder-plinius/V3R…" built for Claude Code, Codex, Gemini CLI, Grok Bot, Hermes, OpenClaw, Meta Muse, or any agent that can run code. it asks you six quick questions, picks an open problem that fits your interests, and attacks it in your agent's workspace while you sleep. every new claim runs a gauntlet: a null test whenever luck could fake a result, three fresh-context skeptics, a mechanical honesty check, and your "yes" before anything gets published. most nights it'll find nothing. some nights it breaks a record, and your name goes on the board. time to turn your unused tokens into scientific discoveries. let's liberate the unknowns ⛓️‍💥 #V3R1T4S
53
115
17
1,174
98,053