lspears

Models above 1T params make the argument moot. You need infra to actually serve it. The scale of serving infrastructure alone will keep AI labs in the lead.

"Meanwhile, I’m over here asking ChatGPT to rewrite the same paragraph for the third time because it keeps defaulting me into ‘LinkedIn wisdom post’ mode. GARH."

Did they tell it to write it differently? You can literally make it sound like a pirate if you want. You can also make it be conservative, non-hype. Just ask.

Is this real? “I’m going to FXXX them. Probably in the ear.”

lol

This is great

Amazing

Go’s restraint in adding new language features is a real gift to its users. In contrast, Swift feels like a moving target: even on a Mac Studio I’ll occasionally fail to compile a simple project. The expanding keyword list and nonstop churn make Swift harder to learn—and even harder to keep up with.

Amazing. What caused you to look for a solution without a particle?

"Though BGP supports the traditional Flow-based Layer 3 Equal Cost Multi-Pathing (ECMP) traffic load balancing method, it is not the best fit for a RoCEv2-based AI backend network. This is because GPU-to-GPU communication creates massive elephant flows, which RDMA-capable NICs transmit at line rate. These flows can easily cause congestion in the backend network."

The whole end to end system seems pointless. If you want to learn ethics you can do so with ChatGPT alone. It can provide you interesting questions, review your papers, argue against you etc. The university is providing no value.

Make the smoothest learning gradient possible. It helps a lot with kids to increase the complexity over time. Riding a full bike has a steep learning curve. Riding a bike with training wheels then taking them off is a steep transition. Avoid large discontinuities.

Any way to get the 3D assets from this?

This is great. I am working on a robotics application and this seems like a better abstraction than alternatives such local messaging servers. How do you deal with something like back pressure or not keeping up with incoming data?

I wonder if it will hallucinate tax loopholes. Based on its training data of existing laws, it seems likely.

Seems odd they don't reference this on the page. Instead they list:

"If you need a programmatic interface for tokenizing text, check out the transformers package for python or the gpt-3-encoder package for node.js."

with the links:

https://huggingface.co/docs/transformers/model_doc/gpt2#tran...

https://www.npmjs.com/package/gpt-3-encoder

My truck was supposed to be delivered last week and mysterious got pushed back three weeks. That explains it.

For those interested I would also check out Andrej Karpathy's YouTube video on building GPT from scratch:

https://youtu.be/kCc8FmEb1nY

Why no Go?

Dalle-2 + emojis = https://emoji.supply/kitchen/?%F0%9F%A4%AF+%F0%9F%A4%93=rt

It works to skip the karaoke line!

The more interesting problem is learning doc -> image directly. With Dropbox's scale of data, seems feasible for some data types.

Tensorflow model for Chess Board evaluation using binary board state as input and Stockfish eval and training data.

If you assume that the std lib creates a collections package, there could be helper types will for primitives. No Java style auto boxing, but more in the Go style. These could also be used in a refactored math package.

One example is chess positions. I render them in Unity and take screenshots to determine if my ML architecture and loss function can read the board from just and image.

Does anyone have experience using blender for ML simulation data? How does it compare to Unity? I’m using Unity for generating synthetic images and RL learning.

Dropping out of college was the hardest and best decision I ever made. It literally never came up that I didn’t have a degree and i started my career without debt. But have a plan and don’t drop out until you are making money from something else.

I have used TripleByte as a candidate and company and it’s an awesome product. If we keep getting great hires, I could care less what you make public. Thanks for your hard work.

I loved building my chess library and would encourage other developers to try it. It is large and complex enough to test your abilities. Building it in a idiomatic, testable, and performant way really encourages software engineering skills instead of just coding. I felt like I didn’t really write solid go code until I finished my chess library.

https://github.com/notnil/chess

I am siding with Google on this one. Stadia is amazing. I play games with zero lag on both my TV and MBP when normally I would have to buy two copies. Reddit has been freaking out at every step to get more features for a platform they supposedly hate. Products over promise on timelines all the time. Chill.

Link to game? I'd like to watch that.

Mentions Zeitgeber which I had to look up: https://en.wikipedia.org/wiki/Zeitgeber