occamrazor

I’m curious, what was the rationale for forbidding it?

Apple group of companies, rather than a specific legal entity. A common distinction in financial journalism.

FYI: Lex, not Rex.

But a Rex Talionis is an interesting concept too.

On prompts only, with answers presumably from the teacher model (Gemini).

It was not trained or RLHFd on Arena replies or user preferences.

How so? Maybe in the past, but nothing announced today is open.

For Tao, spending 10x time on aproof still means spending 5x less time than an average postdoc. He is incredibly fast and productive.

On all flights I have taken the flight attendants count the passengers on board before taxiing and check the lavatories. This method wouldn't have worked even with available free seats.

Zopf means “braid” and it also denotes a medium-size bread type, made with some milk and glazed with yolk, shaped like a braid, traditionally eaten on Sunday.

I like:

- “open weights” for no training data and no restrictions on use,

- “weights available” for no training data and restrictions on use, like in this case.

It‘s more likely to be incompetence than malice: even their 73.7% is closer to 72% than to 74%.

Scholar is heavily used internally, it’s unlikely to be discontinued even if it has never brought any money to Google.

Note that the model is based on RoBERTa and has only 125m parameter. It is not competing against any of the new popular models, not even small ones like Phi or GeMMa.

Pedantic remark: HTML 4.0 is an SGML profile, but HTML 5 isn’t anymore.