Clyde Johnson
269 posts
Clyde Johnson
@BULLSTAT1
Military | AI | East Asia
Beijing Katılım Temmuz 2021
166 Takip Edilen32 Takipçiler
@BULLSTAT1 @SakanaAILabs yeah, Japan likes to create their own version of existing service and restrict access from overseas
English
🐟 Sakana Chat 公開 🐟
Sakana AIは、Sakana Chatを無料公開しました。
chat.sakana.ai
Web検索機能と高速レスポンスを備えたAIチャットです。日本国内から、どなたでもお使いいただけます。ぜひ、お試しください。
GIF
日本語
@FauzKhalid Guess somebody is still living in a sweet dream. wake up yo
English
Iran attacks US base in Kuwait.
Iran was underestimated and this arrogance is now backfiring.
English
@danushman this is better than secret ads and bullshit content
i am giving what worked for me and added a mode ( a system prompt to turn llm behaviour) and made it useful.
English
🚨BREAKING: Claude has a secret mode called "First Principles Breakdown."
It strips any complex topic down to its raw fundamentals like Elon Musk thinks through problems.
Here's how to activate it:
English
English
Feel free to #askmeanything about @Microsoft @MicrosoftAI in terms of roles in the Superintelligence Lab, org, interviews or anything of interest..
#ai #hiring
English
🚨BREAKING: Someone just solved Claude Code's biggest problem.
It's called Claude-Mem and it gives Claude persistent memory across sessions.
- You can use up to 95% fewer tokens each time.
- Make 20 times more tool calls before reaching limits.
100% Opensource.
English
Goodfire makes employees sign a "permanent non-disparagement agreement" on joining? 🤨
Liv@livgorton
Just a heads up: I signed a permanent non-disparagement agreement when I joined Goodfire. To keep things simple, I won't be commenting on the company, its research, etc.
English
@livgorton @GoodfireAI Same here. What I want to is to make AI safe to mankind, and save future victims. Thank you for sharing Liv!
English
1/ I've been thinking about the "non-linear features" objection to interpretability work. But it can be hard to reason about, since the space of possible non-linear features is so large.
Here's an attempt to untangle it: livgorton.com/non-linear-fea…
English
A challenge to the mechanistic interpretability community: fully interpret our 432-parameter RNN. (Thread)
English
Clyde Johnson retweetledi
@Zai_org @jietang Some people are arguing that @upstageai's Solar-Open-100B is derived from GLM-4.5-Air. What is your opinion? It seems like a valid claim to me.
English
@aiwithmayank It’s been dead for at least 100 times since 2024.
English
CHAIN OF THOUGHT IS OFFICIALLY DEAD ☠️
I’ve been testing Atom of Thought prompting and it’s casually boosting reasoning accuracy by 30-40% on hard problems.
This is how people will use ChatGPT and Claude going forward.
Here's how to use it:
English
Clyde Johnson retweetledi
Kirk Cousins and multiple other Buccaneers players pray after the game.
English
@GoodfireAI @EricBigelow @danielwurgaft @EkdeepL I would like to call this "D&S system" (Detect and Steer)
English
@GoodfireAI @EricBigelow @danielwurgaft @EkdeepL It would be highly recommended to implement SAE probing, so that we can steer the model when it is only needed.
SAE detects jailbreaking -> Activation steering by adding vector in the direction of safe prompt.
English
New research: are prompting and activation steering just two sides of the same coin?
@EricBigelow @danielwurgaft @EkdeepL and coauthors argue they are: ICL and steering have formally equivalent effects. (1/4)
English
@samhogan @laion_ai @Wyndlabs_ai @AmarSVS @0xdrej @bonham_sol @francescodvirga @ChristophSchuh6 @inference_net Great visualization. Definitely will fork it.
English
We're introducing Project AELLA, in partnership with @laion_ai & @wyndlabs_ai
AELLA is an open-science initiative to make scientific research accessible via structured summaries created by LLMs
Available now:
- Dataset of 100K summaries
- 2 fine-tuned LLMs
- 3d visualizer
👇
English
@NeelNanda5 You are the reason I started off AI Safety journey. Props to your work on SAE, truly inspiring.
English
It's my 27th birthday today!
I do a lot of public facing work, and it's often hard to tell if any of it actually matters, or if I'm just talking into the void. If anything I've done has impacted you, it would really make my day to hear about it!
English