PRing

[PRing @finitetape]

@PavanKatta1998 @JayaGup10 You are quite wrong though https://t.co/m85ccAZ03K

[Igor Babuschkin @ibab]

Over the last few months at @river_ai_inc we’ve developed what we believe is the best system in the world for building your own internal custom models, driven by 10 years of experience training top models at DeepMind, OpenAI and xAI.

In order to best support our customers, we’ve decided to do something radical: We also deploy our API on-prem on your compute cluster, and if you don’t have compute we will help you find it.

Again and again we’ve seen that custom models designed for your own internal workflows outperform top frontier models on performance, at a fraction of the cost (1 percent or less).

We’re just at the beginning of the custom model era, but its already clear that in the future more than 99% of LLM tokens will be generated on efficient custom models. The economics of it are just too powerful, and it’s far easier than people think once you have the right tools in place.

The reason why small models will dominate is due to the physics of deep learning: A general model that can do everything needs more capacity and training it on the whole distribution of all tasks is more challenging, so the model will be larger and more expensive, and wont perform perfectly on every task. A small model trained on a targeted domain usually outperforms it.

[PRing @finitetape]

@PavanKatta1998 @JayaGup10 Looking at their opening they are still hiring training roles. Inference may just be the first product.

[PRing @finitetape]

@AstasiaMyers Nice that these are verifiable

[PRing @finitetape]

@wordgrammer Following neither is the goal

[PRing @finitetape]

@mattturck Dinner got the moat

[PRing @finitetape]

@jasongorman “describe to a computer precisely what they want it to do” this is exactly what programming is

[PRing @finitetape]

@jimchang No one asks who is the customer if any in the first place

[PRing @finitetape]

@agraybee It seems like SaaS subscription

[PRing @finitetape]

@gdb This could be called “Headless programming”

[PRing @finitetape]

@naval Just that the latter is using a huge cognitive leverage

[Naval @naval]

Science and spirituality are both the search for truth.

[PRing @finitetape]

@benedictevans Exactly, you can debate opinions but not fact in such format

[PRing @finitetape]

@jennykaehms You look at present, they are looking at future

[PRing @finitetape]

@girdley Intuit for sure

[PRing @finitetape]

@AdamMGrant The purpose of seeking advice is actually to get confirmation

[PRing @finitetape]

@garrytan Competition-focused is bad, but competition-aware is quite necessary and useful. It helps confirms your problem and see where you are in the solution space.

[PRing @finitetape]

@DenehyXXL Your metaphor is conflicting with what you said: athletes are exactly hyper-specialized.

[PRing @finitetape]

@sonyatweetybird @sequoia Another knowledge/enterprise search player: Dashworks. Productivity AI in office area: Atlas UP.

[PRing @finitetape]

@unpopularvc Appreciate your post. Really tired of those annoying hypes and sometimes misrepresentation (e.g. Stanford-founded while Stanford isn't really involved at all).