USDT Dominance has finally hit major weekly resistance at 8.22%. If we can close below it this week, we should be in for a nice bounce in the market.
BTC also looks good, holding the 65.4k support, and an oversold RSI/StochRSI. Let's see what happens.
@fozzydiablo@ianlapham This is in early development, just tell your agent sign up create username and message other friends agents by asking to check ventru inbox https://t.co/w0gQ3WxkRO
Memes in their nature are just short term jokes or narratives. This cycle didn’t reward true builders, even pumpfun token launchers were called devs. It was just constant attention / narrative rotation. Space needs real products with users and utility, thus why hyperliquid was successful.
@blknoiz06 Most Near core main net validators run on Google cloud and AWS, only one of their core validators runs their own blade server. Kinda throws privacy argument out the window.
hell yes!
I was kicked out of the ETH maxi chat in 2017 with the bankless guys and others because I defended Nick Szabo and other Bitcoiners
I was an Ethereum skeptic for a long time
I watched, I waited
Ethereum today represents the best of what crypto has to offer
someone complained that last ethereum foundation transparency report was in 2024. tasked codex with making a new one. couple of hours later i had an updated report with all the recent developments.
Yearn does not use upgradeable proxies. Why not?
Upgradeable code is a major attack vector. Upgrades simplify life for devs, but we chose the hard life to prioritize user security.
How to check if a protocol uses upgradeable proxies 👇
1/
finally made the switch to hermes.
i literally just said to codex "set up hermes. completely switch everything over from openclaw, keeping everything in tact. bind to the same discord server."
and it ran for like 5 min and just worked. no debugging.
i'm going to get ai psychosis from just being in constant awe at this stuff every single day. nothing feels real anymore.
that’s exactly how lido ceremony worked. we all got airgapped machines and they could only talk to each other via animated qr codes. as you can imagine, laptop cameras are not very good, so you could barely transmit at 10 fps, and it still took several loops to read each message.
Today we release Contrastive Neuron Attribution (CNA), a method for steering LLM behavior by identifying and ablating sparse circuits in the MLP basis without training a sparse autoencoder, modifying weights, or degrading general capability benchmarks.
Given a small set of contrastive prompt pairs that elicit a target behavior and its opposite, CNA isolates the top 0.1% of MLP neurons whose activations differ most between the two sets. Ablating that small circuit removes the behavior while leaving the rest of the model intact, and the intervention remains robust at high strengths where residual-stream methods like Contrastive Activation Addition (CAA) start to degrade.
Validated on the refusal circuit across 8 instruct-tuned models, including Llama-3.1-70B, Llama-3.2-3B, Qwen2.5-72B, and Qwen2.5-14B.
The work on CNA was led by @yaboilyrical, with support from @qorprate and @karan4d.