Police arrested over 40 people outside the Trump International Hotel in New York City
as hundreds gathered for a peaceful action led by Jewish leaders
calling for the end to Israel’s starvation and ethnic cleansing of Gaza.
Democracy Now! was at the demonstration and spoke to some of the protesters,
including Motaz Azaiza, renowned photojournalist from Gaza,
and Rabbi Ari Lev Fornari, who was arrested.
“We’re here to say, ‘Let Gaza live,’ to risk everyt…
Für die breite Verwendung von #KI, speziell im Kontext #Schule, muss sichergestellt sein, dass #LLMs user:innen nicht zu selbstgefährdendem Verhalten animieren.
Das Nonprofit Transluce arbeitet an verschie…
For the first time this year, I've been trying out keeping a basic list in a spreadsheet of TV and movie premieres (and I started a new tab for ones coming out next year, too) -- just to keep up with things. This is the last half of the 2025 now, so here's everything I have from yesterday til the end of the year.
I won't necessarily watch ALL of this, but it's things I might be interested in, and I find that it's helped to have a basic guide as to what comes out wh…
I'm not even putting a politics warning here, the shitheads in the U.S. government made The Corporation for Public Broadcasting (PBS) shut down due to lack of funding and NPR will face similar cuts, etc, really making some rural areas in the USA lose access to news. I grew up loving Sesame Street and Mr. Rogers, and this is beyond fucked.
I hope the useless garbage assholes who allowed this to happen are happy with themselves. Fuck everything that led to this.
Research Challenges and Progress in the End-to-End V2X Cooperative Autonomous Driving Competition
Ruiyang Hao, Haibao Yu, Jiaru Zhong, Chuanye Wang, Jiahao Wang, Yiming Kan, Wenxian Yang, Siqi Fan, Huilin Yin, Jianing Qiu, Yao Mu, Jiankai Sun, Li Chen, Walter Zimmer, Dandan Zhang, Shanghang Zhang, Mac Schwager, Wei Huang, Xiaobo Zhang, Ping Luo, Zaiqing Nie
htt…
Been meaning to post about this for sometime now. Our bass player got one of these, and it absolutely kills. Perfect 80s synth bass sounds; super quick, accurate tracking. He has tried everything else, and this is the best thing he's ever had on his board.
#bass #BassGuitar
In one respect, many leftists/liberals are no different from the right: their reflexive, fundamental opposition to everything that comes from the other camp.
Forcing an end to Iran's nuclear program and, ideally, laying the groundwork for the Iranian people to send the mullahs to hell shouldn't be a 'controversial' issue, imo.
But some rather defend a fascist-clerical regime that executes hundreds of people every year and oppresses its population.
Dynamic Chunking for End-to-End Hierarchical Sequence Modeling
Sukjun Hwang, Brandon Wang, Albert Gu
https://arxiv.org/abs/2507.07955 https://arxiv.org/pdf/2507.07955 https://arxiv.org/html/2507.07955
arXiv:2507.07955v1 Announce Type: new
Abstract: Despite incredible progress in language models (LMs) in recent years, largely resulting from moving away from specialized models designed for specific tasks to general models based on powerful architectures (e.g. the Transformer) that learn everything from raw data, pre-processing steps such as tokenization remain a barrier to true end-to-end foundation models. We introduce a collection of new techniques that enable a dynamic chunking mechanism which automatically learns content -- and context -- dependent segmentation strategies learned jointly with the rest of the model. Incorporating this into an explicit hierarchical network (H-Net) allows replacing the (implicitly hierarchical) tokenization-LM-detokenization pipeline with a single model learned fully end-to-end. When compute- and data- matched, an H-Net with one stage of hierarchy operating at the byte level outperforms a strong Transformer language model operating over BPE tokens. Iterating the hierarchy to multiple stages further increases its performance by modeling multiple levels of abstraction, demonstrating significantly better scaling with data and matching a token-based Transformer of twice its size. H-Nets pretrained on English show significantly increased character-level robustness, and qualitatively learn meaningful data-dependent chunking strategies without any heuristics or explicit supervision. Finally, the H-Net's improvement over tokenized pipelines is further increased in languages and modalities with weaker tokenization heuristics, such as Chinese and code, or DNA sequences (nearly 4x improvement in data efficiency over baselines), showing the potential of true end-to-end models that learn and scale better from unprocessed data.
toXiv_bot_toot
Being on the spectrum myself, I’ve noticed just how much easier it is to be authentic around autistic females, they genuinely help me unmask and feel like myself more and more every day.
But when I’m around guys on the spectrum, it’s the opposite: I end up masking even harder, feeling embarrassed and on edge, like everything just gets more complicated.
Sometimes I’m just hit with these realizations, and it really does feel like autism is full of surprises, never really stops catc…