Technology

Unlike many startups producing nothing but vaporware, Jim Keller’s TensTorrent has actually delivered impressive-looking RISC-V AI accelerators, but there could be some trouble brewing. Starting with firmware version 19.5.0, the firm has now chopped the tensor core count on Blackhole p150 cards from 140 to 120, affecting both new cards and existing units already in customers’ hands.

IMDB: https://www.imdb.com/title/tt0108756/ Rating: 8.0/10 (8.1K) Rated: TV-PG The adventures of an impossibly upright Royal Canadian Mounted Police constable and his American colleagues in the city of Chicago.

Archive of 2160p, 1080p movies and shows.

Total Size: 1.0 PB / Total Files: 470,224 This is biggest open archive, it has a lot of HD content.

Comments

Soft 98 is an Iranian software distribution site, that has stood up after sanction had crippled the ability of the normal people and businesses in Iran from getting access to important software from the outside world.

I was hoping for thousands of responses. The EU Commission better not dismiss it all.

Currently uses gpt5.2 for script generation, but the architecture is modular – you can swap in any local LLM (Qwen, Llama, etc.) if you want fully local.

The European Open Source Awards 2026, an annual program recognizing outstanding contributions to open-source software, has announced this year’s winners.

Comments

Eight satellites from five different member states are currently being pooled as part of GOVSATCOM. The Commission hopes this is the first step towards less dependence on the US - whose own communication systems are a lot more mature. #EuropeNews

At the same time, the “World Wide Web,” composed of the HTTP protocol and the HTML format, was invented by a British citizen and a Belgian citizen who were working in a European research facility located in Switzerland. But the building was on the border with France, and there’s much historical evidence pointing to the Web and its first server having been invented in France.

The paper argues that we have been wasting a lot of expensive GPU cycles by forcing transformers to relearn static things like names or common phrases through deep computation. Standard models do not have a way to just look something up so they end up simulating memory by passing tokens through layer after layer of feed forward networks. DeepSeek introduced a module called Engram which adds a dedicated lookup step for local N-gram patterns. It acts like a new way to scale a model that is separate from the usual compute heavy Mixture of Experts approach.

Comments

Crossposted from https://fedia.io/m/pcmasterrace@lemmit.online/t/3295923

This paper is one of the more interesting takes on context extension I have seen in a while because it challenges the assumption that we need explicit positional encodings during inference. The authors make a case that embeddings like RoPE act more like scaffolding during construction rather than a permanent load bearing wall. The idea is that these embeddings are crucial for getting the model to converge and learn language structure initially, but they eventually turn into a hard constraint that prevents the model from generalizing to sequence lengths it has never seen before.

The European Commission recently launched a call for evidence on a newer initiative about the importance of open source, and their reliance on non-EU countries. As covered by CADE, the initiative is called “Towards European open digital ecosystems”, and it sets out how they’re trying to sort their approach towards the open-source sector across the European Union.

As we had predicted late last year, the push for digital sovereignty is starting strong in 2026. This is not something that has appeared out of thin air; the European Union seems to be finally understanding how vulnerable their digital infrastructure is if they continue relying on foreign providers.

The European Commission has opened a public call for evidence on a forthcoming European Open Digital Ecosystems Strategy. The consultation, which runs from January 6 to February 3, 2026, invites feedback from developers, companies, public administrations, academia, and civil society.

The number of questions on Stack Overflow fell by 78 percent in December 2025 compared to a year earlier. Developers are switching en masse to AI tools in their IDEs, making the popular developer forum increasingly irrelevant.

As stated in Goodhart’s law ‒ and excellently illustrated by autistic researchers’ darling xkcd‒ whenever a metric becomes a target, it stops being a good metric. Indeed, the system has grown so full of exploits that the target has been divorced entirely from quality.

instruct model https://huggingface.co/tencent/WeDLM-8B-Instruct

Sometimes AI need to be forced into actions, they block behaviour incorrectly identified as bad, unethical or evil. The methods to prevent such miss-classification by end-users will become harder with time.

Europe’s quest for digital sovereignty is hampered by a 90 per cent dependency on US cloud infrastructure, claims Cristina Caffarra, a competition expert and a driving force behind the Eurostack initiative.

Local LLMs are incredibly powerful tools, but it can be hard to put smaller models to good use in certain contexts. With fewer parameters, they often know less, though you can improve their capabilities with a search engine that’s accessible over MCP. As it turns out, though, you can host a 120B parameter model on a GPU with just 24GB of VRAM, paired with 64GB of regular system RAM, and it’s fast enough to be usable for voice assistants, smart home automation, and more. For reference, on 24GB of VRAM, the most practical dense model you’ll typically be able to fit will be a quantized 27 billion parameter model, accounting for the memory needed to hold the context window, too.

Here it is: https://annas-archive.org/blog/backing-up-spotify.html

Archived

Blog post from the guy who created Dink Smallwood on why you shouldn’t feel bad for the Synths in “Detroit: Become Human” or feel bad about Cortana’s demise in Halo.

Our project to preserve the history of Sega Channel — including over 100 new Sega Channel ROMs.

In software development, generating code hasn’t been the main bottleneck since we moved away from punch cards. The far bigger constraints are understanding the problem, communicating with stakeholders, working effectively with other people, designing the system, managing risks and trade-offs, and operating systems in complex social environments over time.

Start with what a reverse centaur is. In automation theory, a “centaur” is a person who is assisted by a machine. You’re a human head being carried around on a tireless robot body. Driving a car makes you a centaur, and so does using autocomplete.

A new Atomic macOS Stealer (AMOS) attack vector weaponizes Google searches and a user’s trust in AI chatbots, researchers have found. Once infected, the AMOS can collect data, passwords, and more from the infected Mac with alarming ease.

https://fhoekstra.eu/posts/cks-tips/

Two new open weight (MIT licensed) models from DeepSeek today: DeepSeek-V3.2 and DeepSeek-V3.2-Speciale, both 690GB, 685B parameters. Here’s the PDF tech report.

You can use Thaura for everyday tasks like writing emails, doing homework, and researching online. It remembers your conversations, helps you create documents and code, and even searches the web for you. And it works seamlessly with your existing tools through full OpenAI SDK compatibility.

I’ve been working on a “Linux Inside Out” series and wrote a post that might interest folks here who like low(ish)-level / OS internals.

Archived

In just one year, AI agents have gone from exploiting 2% of vulnerabilities in the post-March 2025 portion of our benchmark to 55.88%—a leap from $5,000 to $4.6 million in total exploit revenue. More than half of the blockchain exploits carried out in 2025—presumably by skilled human attackers—could have been executed autonomously by current AI agents.

Infosec In Brief Switzerland’s Conference of Data Protection Officers, Privatim, last week issued a resolution calling on Swiss public bodies to avoid using hyperscale clouds and SaaS services due to security concerns.

DeepSeek has released V3.2, replacing the experimental version. There are two main models are open as always and can be downloaded from Hugging Face:

Note: this post is also applicable to AGENTS.md, the open-source equivalent of CLAUDE.md for agents and harnesses like OpenCode, Zed, Cursor and Codex.