AI Risks that Could Lead to Catastrophe is a CAIS resource focused on mitigating AI risks that could cause catastrophic societal harm. It covers four key risk categories: malicious use, AI race, organizational risks, and rogue AIs. CAIS conducts related research, field-building, and advocacy projects, and operates a compute cluster to support this work.
Dlib is a C++ library whose image processing page documents routines for managing and manipulating images using array2d objects rather than an explicit image type. It lists pixel types including RGB, RGB alpha, HSI, HSV, LAB and grayscale, alongside functions for image I/O, object detection, feature extraction, filtering, scaling and rotating.
A Hacker News discussion thread about the ChemCrow preprint on augmenting large-language models with chemistry tools. Commenters raise citation problems and question how a program can make an already widely used medicine less toxic.
A reference page in the Clojure documentation covering Java interop, including class access, member access, the dot special form, implementing interfaces and extending classes, arrays, vararg methods and type hints.
This DeepSeek API documentation page lists models and per-token pricing. It describes the deepseek-flash and deepseek-v4-pro models, their features, context length, concurrency limits, and peak and off-peak token rates, and notes that the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted.
LMQL is a programming language for interacting with large language models, using types, templates, constraints and an optimizing runtime. Prompt statements have associated where clauses that specify logical constraints on output, which LMQL translates to subtoken masks enforced during generation. It works across llama.cpp, OpenAI and Transformers backends and supports nested queries.
Codeflash uses AI to automatically find the most optimized version of your Python code through benchmarking — while verifying it's correct.
Community notes
Be the first to share how this project worked for you.
Sign in with a verified account to leave a review or join the discussion.
Sign in to reviewKeep exploring
Dlib is a C++ library whose image processing page documents routines for managing and manipulating images using array2d objects rather than an explicit image type. It lists pixel types including RGB, RGB alpha, HSI, HSV, LAB and grayscale, alongside functions for image I/O, object detection, feature extraction, filtering, scaling and rotating.
A Hacker News discussion thread about the ChemCrow preprint on augmenting large-language models with chemistry tools. Commenters raise citation problems and question how a program can make an already widely used medicine less toxic.
A reference page in the Clojure documentation covering Java interop, including class access, member access, the dot special form, implementing interfaces and extending classes, arrays, vararg methods and type hints.
This DeepSeek API documentation page lists models and per-token pricing. It describes the deepseek-flash and deepseek-v4-pro models, their features, context length, concurrency limits, and peak and off-peak token rates, and notes that the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted.
LMQL is a programming language for interacting with large language models, using types, templates, constraints and an optimizing runtime. Prompt statements have associated where clauses that specify logical constraints on output, which LMQL translates to subtoken masks enforced during generation. It works across llama.cpp, OpenAI and Transformers backends and supports nested queries.
Codeflash uses AI to automatically find the most optimized version of your Python code through benchmarking — while verifying it's correct.
