campers

  The legal definition of genocide is precise and includes an element that is often hard to prove: "intent."
  
  Determining whether a situation constitutes genocide is factually and legally complex. It should only be made following a careful and detailed examination of the facts against relevant legislation.
  
  This examination is carried out to establish State responsibility or individual criminal responsibility for the crime of genocide. It must be done by a competent international or national court with jurisdiction to try such cases, after an investigation that meets appropriate due process standards.

As for whats happening, the over represented proportion of male combatant age deaths, the well below average ratio of civilian to combatant deaths in complex urban warfare, the war being in response to an atrocious attack on civilians, stopping for polio vaccines, population numbers remaining stable etc puts a question on the "Only Reasonable Inference" standard.

They do have a priority tier at double the cost, but haven't seen any benchmarks on how much faster that actually is.

The flex tier was an underrated feature in GPT5, batch pricing with a regular API call. GPT5.1 using flex priority is an amazing price/intelligence tradeoff for non-latency sensitive applications, without needing to extra plumbing of most batch APIs

Here's a couple of interesting paragraphs from it.

  Instead of the transistor, the basic element in superconducting logic is the Josephson-junction.
  
  For logic, a Josephson-junction loop without a persistent current indicates a logical 0, while a loop with one single flux quantum’s worth of current represents a logical 1. For memory, two Josephson junction loops are connected together. An SFQ’s worth of persistent current in the left loop is a memory 0, and a current in the right loop is a memory 1.
  
  In classical CMOS-based technology, it is very challenging to stack computational chips on top of each other because of the large amount of power, and therefore heat, that is dissipated within the chips. In superconducting technology, the little power that is dissipated is easily removed by the liquid helium. Logic chips can be directly stacked using advanced 3D integration technologies resulting in shorter and faster connections between the chips, and a smaller footprint.
  
   It is also straightforward to stack multiple boards of 3D superconducting chips on top of each other, leaving only a small space between them. We modeled a stack of 100 such boards, all operating within the same cooling environment and contained in a 20- by 20- by 12-centimeter volume, roughly the size of a shoebox. We calculated that this stack can perform 20 exaflops (in BF16 number format), 20 times the capacity of the largest supercomputer today. What’s more, the system promises to consume only 500 kilowatts of total power. This translates to energy efficiency one hundred times as high as the most efficient supercomputer today.

I built my own TypeScript AI platform https://typedai.dev with an extensive feature list where I've kept iterating on what I find the most ergonomic way to develop, using standard constructs as much as possible. I've coded enough Java streams, RxJS chains, and JavaScript callbacks and Promise chains to know what kind of code I like to read and debug.

I was having a peek at xstate but after I came across https://docs.dbos.dev/ here recently I'm pretty sure that's that path I'll go down for durable execution to keep building everything with a simple programming model.

https://cloud.google.com/blog/products/compute/updates-to-ai...

From https://cloud.google.com/tpu/pricing and https://cloud.google.com/vertex-ai/pricing#prediction-prices (search for ct5lp-hightpu-8t on the page) the cost for that appears to be $11.04/hr which is just under $100k for a year. Or half that on a 3-year commit.

That seems like a better deal than millions for a few CS-3 nodes.

And they've just announced the v6 TPU:

  Compared to TPU v5e, Trillium delivers: 
  Over 4x improvement in training performance 
  Up to 3x increase in inference throughput 
  A 67% increase in energy efficiency
  An impressive 4.7x increase in peak compute performance per chip 
  Double the High Bandwidth Memory (HBM) capacity 
  Double the Interchip Interconnect (ICI) bandwidth 

  The first implementation of inference on the Wafer Scale Engine and utilized only a fraction of its peak bandwidth, compute, and IO capacity. Today’s release is the culmination of numerous software, hardware, and ML improvements we made to our stack to greatly improve the utilization and real-world performance of Cerebras Inference.
 
  We’ve re-written or optimized the most critical kernels such as MatMul, reduce/broadcast, element wise ops, and activations. Wafer IO has been streamlined to run asynchronously from compute. This release also implements speculative decoding, a widely used technique that uses a small model and large model in tandem to generate answers faster.

"The scaling curves indicate that Diff Transformer requires only about 65% of model size or training tokens needed by Transformer to achieve comparable language modeling performance."

"Diff Transformer retains high performance even at reduced bit-widths, ranging from 16 bits to 6 bits. In comparison, Transformer’s accuracy significantly drops with 6-bit quantization. The 4-bit Diff Transformer achieves comparable accuracy as the 6-bit Transformer, and outperforms the 4-bit Transformer by about 25% in accuracy."

I'd added seperate design/implementation agents before that was added to Aider https://aider.chat/2024/09/26/architect.html

The other different is I have a file selection agent and a code review agent, which often has some good fixes/improvements.

I use both, I'll use Aider if its something I feel it will right the first time or I want control over the files in the context, otherwise I'll use the agent in Sophia.

I'll use Continue when a chat is all I want to generate some code/script to copy paste in. When I need to prepare a bigger input I'll use the CLI tool in Sophia (sophia.dev) to generate the response.

I use Aider sometimes, less so lately, although it has caught up with some features in Sophia (which builds on top of Aider), being able to compile, and lint, and separating design from the implementation LLM call. With Aider you have to manually add/drop files from the context, which is good for having precise control over which files are included.

I use the code agent in Sophia to build itself a fair bit. It has its own file selection agent, and also a review agent which helps a lot with fixing issues on the initial generated changes.

I think there's another one but I can't remember the name of it.

Also a bit further out is https://spectrum.ieee.org/superconducting-computer

"Instead of the transistor, the basic element in superconducting logic is the Josephson-junction."

> This article collects evidence from psychopharmacology, scripture, and archeology to explore several preparations for consumption described in the Old Testament: Manna, Showbread, the Holy Ointment, and the Tabernacle Incense. The Ointment and the Incense are herbal preparations used by the priestly caste to facilitate a direct experience of the Israelite God. A wide variety of psychoactive components are found in these preparations, including GABA-receptor agonists and modulators, opioid receptor agonists, and other agents. They are normally broken down by the body’s enzymes, and therefore orally inactive, but the Holy Ointment also contains inhibitors specific to the enzymes in question. The preparations indicate that the ancient Israelites had a profound understanding of synergism, and the way they are consumed and the taboos around them are highly suggestive of their use as psychoactive agents.

This^ Don't feel bad about asking for credits when you accidentally make a costly mistake.