In reflecting on the essence of Computing Science, the author explores various perspectives on the nature of this discipline. He discusses the different ways computers can be viewed, such as industrial products, tools for international competition, or the center of philosophical and psychological questions. The author suggests that for computing to thrive as an academic […]
Read more »
A few years ago, I created YAIOUOM, a static analyzer for Rust that ensured the correct use of units of measure in code. This refinement type system was surprisingly easy to write but could only work on a specific version of Rust Nightly. Unfortunately, Rust lacks the ability to specify certain type equivalence, hindering advanced […]
Due to high cocoa bean prices, concerns about deforestation and child labor, the Swiss chocolate industry is seeking new ingredients. Cocoa prices have reached record highs, leading to a deficit in cocoa beans production. The European Union’s “zero deforestation” regulation and pressure from the market are driving this search. Start-ups like Nukoko and ChoViva are […]
I’ve been heavily involved with Rust since 2016 and have encountered several surprising limitations of the borrow checker while working professionally with it since 2021. I’ve shared insights on common issues like checking and returning values from a hashmap, limitations with async functions, and problems with reborrowing captures in closures. Additionally, I discussed the Send […]
The author expresses a dislike for benchmarking code in JavaScript due to the inherent complexities and challenges involved, such as the impact of the JIT compiler on accuracy. Different JavaScript engines like V8, JSC, and SpiderMonkey can significantly affect performance, leading to varying results during benchmarking. The author highlights the importance of testing code across […]
Authors attacked the state-of-the-art Go-playing AI system KataGo by training adversarial policies, achieving over a 97% win rate against it. Surprisingly, adversaries did not win by playing Go well, but by tricking KataGo into making serious blunders. This attack, transferable to other superhuman AIs, can be understood by human experts to consistently beat superhuman AIs […]
Our team has trained sparse autoencoders (SAEs) on Llama 3.3 70B and made the interpretive model available through an API, creating the most accessible and capable model with interpretability tools. By providing these tools, we hope to facilitate new research and product development. Through our exploration of the feature space of Llama 3.3-70B, we have […]
LEGO Island Decompilation Development Vlog has successfully decompiled ISLE.EXE and LEGO1.DLL, aiming for accuracy by matching recompiled instructions to the original machine code. While ongoing work is focused on improving accuracy, naming, and structure, the source code allows for modifications, improvements, and porting to other platforms. Using the CMake build system and Microsoft Visual C++ […]
AI-generated worlds by Genesis allow for the creation of realistic physics, cameras, and object behaviors through text commands. The system produces physically accurate ray-traced videos and data for robot training. This innovative prompt-based system can create complex testing environments for robots by using natural language commands instead of manual programming. Genesis has the potential to […]
Two people with identical cancers seek treatment at different hospitals, highlighting the complexities of medical treatment. Clinical guidelines aim to standardize care based on evidence-based recommendations, with the NCCN guidelines being the gold standard. However, guidelines can be difficult to navigate when trapped in dense PDF documents. The author developed a tool to extract and […]