Tootfinder

I have my share of issues with Parkrose Permaculture, but she has a lot of things I do strongly agree with. I can't stress enough that you never dehumanize your enemies. You can respond appropriately to violence. You can defend yourself from them by any means necessary. But you do not dehumanize them. You always limit your response to the minimum necessary to defend yourself.
There are a number of former Nazi skins who became antifascists after realizing they were wrong. Those folks tend to be some of the most dedicated because they feel a debt, and some of the most knowledgeable because they were there. Coming out of these types of cults, police included, is hard and takes time. A lot of us don't have the ability to work with them. But some do.
By repeatedly humanizing your opponent, you can break some of them. The #Seattle Police Department was not defunded but saw a massive reduction in numbers because their morale was destroyed. Some people will never change. Some people are broken and feel like they need the power. But if you change one person's mind, even give them something to think about, it's a crack. If even one cop quits, that's one less trained gun pointed at you in the future.
The 18 year old marines and federalized national guard troops out there are literally kids. A lot of them came from poor communities. They are being used in a way they haven't been trained to do, doing things they (should) have been told are not legal. They joined to get out of poverty, to go to college, or to "defend the American people" (regardless of how misguided that is). Few, if any, of them joined to abuse people. They will be especially open to persuasion.
Remind those troops that they are carrying out illegal orders, that they are being called on to violate their oath to protect the constitution, that they are suppressing the free speech of the fellow Americans they swore to defend. Remind them that the people they could be illegally arresting now are just like their parents, their neighbors, their families, the friends who didn't join. Remind them that this is the first step. They will be called on to kill Americans if they let this keep going.
Remind them ICE sleeps in hotels while they sleep on the ground. Remind them that their drunk and incompetent leadership thinks of them as disposable tools. Remind them that some of these people are out protesting *for them* against cuts to the VA and other services. Remind them that the people they're defending refuse to make college free so they can recruit from poor schools. Remind them that they will always be welcome when they're ready to join the side of freedom and justice.
When you dehumanize your enemies, you unify them. When you humanize your enemies, you can divide them. There is no weapon available to us right now so powerful as compassion.
https://youtu.be/YtWOYUDMsBw

Also finished “Wild Dark Shore” by Charlotte McConaghy. A woman washes ashore on a cold and bleak island between Australia and Antarctica. A widowed father and his three children live there as caretakers of a seed bank, critically important in this time of further advanced climate change. They have a secret and she might too. Can they trust each other?
Great characters & a very visceral setting, but some cringy romantic moments.
4/5 stars ⭐️ ⭐️ ⭐️ ⭐️

Wild Dark Shore
A family on a remote island. A mysterious woman washed …

Regularizing Learnable Feature Extraction for Automatic Speech Recognition
Peter Vieting, Maximilian Kannen, Benedikt Hilmes, Ralf Schl\"uter, Hermann Ney
https://arxiv.org/abs/2506.09804

Regularizing Learnable Feature Extraction for Automatic Speech Recognition
Neural front-ends are an appealing alternative to traditional, fixed feature extraction pipelines for automatic speech recognition (ASR) systems since they can be directly trained to fit the acoustic model. However, their performance often falls short compared to classical methods, which we show is largely due to their increased susceptibility to overfitting. This work therefore investigates regularization methods for training ASR models with learnable feature extraction front-ends. First, we e…

This https://arxiv.org/abs/2506.04668 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCV_…

Feature-Based Lie Group Transformer for Real-World Applications
The main goal of representation learning is to acquire meaningful representations from real-world sensory inputs without supervision. Representation learning explains some aspects of human development. Various neural network (NN) models have been proposed that acquire empirically good representations. However, the formulation of a good representation has not been established. We recently proposed a method for categorizing changes between a pair of sensory inputs. A unique feature of this approa…

Prince William: How much are you paying in tax?
£22.9 million. That's the staggering amount Prince William raked in last year. But unlike his father, who revealed his tax contributions while he was Prince of Wales, Prince William is refusing to say how much tax he pays. What has he got to hide?
https://38d.gs/fd5e

Prince William: Publish your taxes
Prince William made £22.9M last year. But he won’t say how much tax he paid. What is he trying to hide? We fund the monarchy, so the least we deserve is transparency. Sign the petition if you agree:

Black hole photon ring beyond General Relativity: an integrable parametrization
Jibril Ben Achour, Eric Gourgoulhon, Hugo Roussille
https://arxiv.org/abs/2506.09882

Black hole photon ring beyond General Relativity: an integrable parametrization
In recent years, the shape of the photon ring in black holes images has been argued to provide a sharp test of the Kerr hypothesis for future black hole imaging missions. In this work, we confront this proposal to beyond Kerr geometries and investigate the degeneracy in the estimations of the black hole parameters using the circlipse shape proposed by Gralla and Lupsasca. To that end, we consider a model-independent parametrization of the deviations to the Kerr black hole geometry, dubbed Kerr …

A Cubic Regularization Method for Multiobjective Optimization
Douglas S. Gon\c{c}alves, Max L. N. Gon\c{c}alves, Jefferson G. Melo
https://arxiv.org/abs/2506.08181

A Cubic Regularization Method for Multiobjective Optimization
This work introduces a new cubic regularization method for nonconvex unconstrained multiobjective optimization problems. At each iteration of the method, a model associated with the cubic regularization of each component of the objective function is minimized. This model allows approximations for the first- and second-order derivatives, which must satisfy suitable error conditions. One interesting feature of the proposed algorithm is that the regularization parameter of the model and the accura…

This https://arxiv.org/abs/2407.19353 has been replaced.
initial toot: https://mastoxiv.page/@arX…

A spring-block theory of feature learning in deep neural networks
Feature-learning deep nets progressively collapse data to a regular low-dimensional geometry. How this emerges from the collective action of nonlinearity, noise, learning rate, and other factors, has eluded first-principles theories built from microscopic neuronal dynamics. We exhibit a noise-nonlinearity phase diagram that identifies regimes where shallow or deep layers learn more effectively and propose a macroscopic mechanical theory that reproduces the diagram and links feature learning acr…

Reinforcement Learning with Action Chunking
Qiyang Li, Zhiyuan Zhou, Sergey Levine
https://arxiv.org/abs/2507.07969 https://arxiv.org/pdf/2507.07969 https://arxiv.org/html/2507.07969
arXiv:2507.07969v1 Announce Type: new
Abstract: We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward tasks. Our recipe is designed for the offline-to-online RL setting, where the goal is to leverage an offline prior dataset to maximize the sample-efficiency of online learning. Effective exploration and sample-efficient learning remain central challenges in this setting, as it is not obvious how the offline data should be utilized to acquire a good exploratory policy. Our key insight is that action chunking, a technique popularized in imitation learning where sequences of future actions are predicted rather than a single action at each timestep, can be applied to temporal difference (TD)-based RL methods to mitigate the exploration challenge. Q-chunking adopts action chunking by directly running RL in a 'chunked' action space, enabling the agent to (1) leverage temporally consistent behaviors from offline data for more effective online exploration and (2) use unbiased $n$-step backups for more stable and efficient TD learning. Our experimental results demonstrate that Q-chunking exhibits strong offline performance and online sample efficiency, outperforming prior best offline-to-online methods on a range of long-horizon, sparse-reward manipulation tasks.
toXiv_bot_toot

Donald Trump just launched a strike on Iran — without Congressional authorization,
violating the War Powers Resolution.
This is an illegal act of war.
If Congress lets this slide,
it greenlights future unjust attacks.
Sign if you agree:
Impeach Trump now for this reckless, unconstitutional attack!