Can I drag an email directly onto a Kanban or a Todo list, and prioritize it like a task, and then click on the card or task to go directly to the mail message, in the context of its thread?
No, and probably won't be. Each tool is intentionally standalone. You can link to things manually but there's no cross-tool wiring. I'd rather keep the codebase simple and each tool easy to understand on its own.
It defaults to being a wrapper around git when it's not custom implemented, and it's recommended that you alias nit as git so the agent can work the way it normally would, just faster and cheaper.
Look into how much current war costs already. And there is no end in sight yet. Loosing one F35 is insignificant (well maybe it will hurt someone's ego but that is about it).
Apologies if I missed this in the article (or in the first article in the series) - what happens if you add two copies of the layer set? Does performance improve over adding one copy of the layer set?
Author here: That was done in this blog post, in the beam search. I started with the best re-layer configs, and iteratively added more blocks, including the same multiple times, during a long beam search.
It turns out this does not help (somewhat surprisingly).
Actually not surprised.
I guess this is for the same reason “say it twice” [1] is working. Because LLm are trained as causal language model, past token cannot attend to future token.
One copy of the layer set solve this.
[1]https://arxiv.org/html/2512.14982v1
It's possible that the gains are despite the noise the coarse process introduces. After two repetitions the noise may overwhelm the advantage.
The residual connections resemble the Euler method (this observation led to Neural ODE's IIRC) which isn't known to be exactly clean. If the model has been trained to be a particular number of layers, adding more layers will also add a lot of noise.
Ultimately, the LLM will need to be fine tuned with the loops or a looped architecture trained from scratch, such as: <https://ouro-llm.github.io> unfortunately they made the mistake of looping the entire LLM rather than just the center portion.
If you look at convolutional neural nets used in image processing, it's super common for the first layer or so to learn a family of wavelet basis functions. Later layers then do recognition in wavelet space, without that space ever being explained or communicated to the training algorithm.
This work here is obviously more complex than that, but suggests something similar is going on with early layers transforming to some sort of generalized basis functions defining a universal language representation.
For all its flaws, TSA (at least under previous administrations) did a lot of design thinking work around how to streamline flows through airports, minimize travel stress and conflict, and optimize to minimize traveler complaints while continuing to maintain security.
Bringing in shock theater optimized staff is a particularly poor fit for a scenario that will impact a disproportionately voting and bipartisan pool of citizens.
There's a reason advertising in airports is generally targeted at corporate leaders and decision makers.
They continued to maintain the illusion of security. The underwear bomber and shoe bomber had no problem smuggling explosives onto an aircraft directly under their noses.
Their idea of "security" is to get you into a scanner so they can stare at and save images of your naked body. Or to buy really expensive "sniffer" robots that don't work in one of the most corrupt government contracts recently known about.
Meanwhile, cockpit doors still have several functional deficiencies that make pilots vulnerable to the original attack that led to the creation of this derelict agency.
If your goal is to intimidate and frighten people into submitting to you, then sending armed, masked "shock theater" thugs in should accomplish that goal.
Counterpoint: these patches are used specifically for slow, controlled absorption of the medication in order to provide symptom relief over long durations. And even then, many drug patches use microneedles to overcome the skin's natural defense against absorption. Only the smallest of molecules can naturally pass through unbroken skin, and the fact that absorption is slow is the primary benefit of medicinal skin patches. They're used for things like hormonal therapeutic treatment or nicotine replacement because for those use cases, the slow, controlled absorption rate is beneficial for long-term relief.
I figured there was quite a big surface here for misuse. But I've done my best to curb that. You're right, that if this scales, the compute to oversee moderation would be inaccessible for a solo hobbyist like me. I don't have a huge audience, so I think for now, at least, the moderation is in scope.
Also MQ is one of my all-time favourite shows (mainly first seasons, it kinda lost steam in the later ones).
reply