rockinghigh

With decisions models, you don't pay for output tokens. Also this OpenAI API is multimodal.

Their marketing language is misleading. They must still use some transformer language model backbone to encode the text input (BERT or decoder-only LLM). The biggest difference is the output, instead of auto-regressively generating tokens, they produce probabilities over a bounded set of decisions (more flexible classification).

Human reviews are going away for most PRs these days. If the LLM decides the risk level of the change is low, the PR can be merged without another human in the loop. That's at least the direction many companies are taking. Only require a human judgement when necessary.

Using tools to fix your writing style or grammar is not exactly cheating.

For small models, tokenization can reach 1-10% of total inference time.

A coding agent driven by a large LLM can delegate smaller tasks to a faster model. For example searching through the codebase for references, examples, or established patterns. They are treated as tools and don't pollute the main agent's context.

Open-weight models were lagging 4 months behind OpenAI/Anthropic at the beginning of the year. They are now just 4-6 weeks behind.

The US is also falling behind Chinese manufacturing. They had to ban Chinese cars because legacy American automakers couldn't compete.

Same here, I fine tune LLMs for specific use cases. How can I trust Anthropic models not to introduce bugs to preserve their moat?

Imagine if your IDE started injecting bugs into your project just because your code looked like it implemented a competing IDE.

Consumers don't control zoning laws or risk mitigation details.

Gemini tends to be faster and the Flash and Flash Lite models outperform ChatGPT's equivalent models.

It may be a push from advertisers who want access to this format. Google Search competes for their money against the competition.

They own tons of engagement data associated with that index from being the default search engine on most devices.

Why wouldn't getting more customers the plan? Anthropic doesn't acquire companies to have a lower market share. There is clearly a consolidation and a rush to get as much of the developer market as possible.

The MoE experts are quantized to int4, all other weights like the shared expert weights are excluded from quantization and use bf16.

They could release data to back up that claim.

Are there any protections from industrial espionage when using Anthropic, Cursor, Gemini, or OpenAI?

Their revenue was $57.4 billion last year. Just in Q4; cloud revenue $6.7 billion, cloud infrastructure $3.0 billion, cloud application $3.7 billion, Fusion Cloud ERP $1.0 billion, NetSuite cloud ERP $1.0 billion.

It's the number of attempts at answering the question.

He founded the team that worked on fasttext, llama and other similarly impactful projects.

He founded FAIR and the team in Paris that ultimately worked on the early Llama versions.

It can also be used to simplify existing code bases.

It's a lot simpler. These models are not optimized for ambiguous riddles.

How is this riddle relevant to a coding model?

They do. Pretty much all agentic models call linting, compiling and testing tools as part of their flow.

It's called problem decomposition and agentic coding systems do some of this by themselves now: generate a plan, break the tasks into subgoals, implement first subgoal, test if it works, continue.

A language model in computer science is a model that predicts the probability of a sentence or a word given a sentence. This definition predates LLMs.

How do you join two datasets using r-trees? In a business setting, having a static and constant projection is critical. As long as you agree on zoom level, joining two datasets with S2 and H3 is really easy.

I wouldn't say R-trees solve the problem better. Joining multiple spatial dataset indexed with r-trees is more complex as the nodes are dynamic and data dependent. Neighborhood search is also more complicated because parent nodes overlap.