smsx

They have an OpenAI deal right now. https://openai.com/index/cerebras-partnership/

The 4.8 model costs the same as it's 4.7 predecessor.

Are the numbers in the H100 PCIE vs SXM table swapped for rows 3 onwards? It looks to me like the PCI is showing higher GiB/s numbers, which is counter to expectations. Or am I misunderstanding those benchmarks?

I don't think first class Linear or Notion support will be high on their list given who acquired them.

Chris Olah has never published a paper? ... https://scholar.google.com/citations?user=6dskOSUAAAAJ&hl=en...

it's being actively worked on!

The numbers are pretty incredible. Will the competition be able to match them?

Wow, incredibly fast. Can easily replace my Groq usage if they can keep up with capacity demands.

Yup, check out Rho-1 by microsoft research.

Sounded like he he was talking about a public blog post right?

The point of this post isn’t the linear transformer algorithm. They’re surveying a variety of Linear transformers and showing a general form in order to talk at large about their performance characteristics.

They didn't say they joined 11 years ago, but that they were 11 years into their career.

They're back actually - They recently raised a new round of funding and are working to release their next-gen chip.

Source: https://mythic.ai/whats-new/mythic-raises-13-million-to-brin...

Why does this sound like ChatGPT?

Actually, in the last few months there has been an absolute ton of feature work. A very sophisticated generational profile system was introduced, and the entire framework is being split into separate packages to enable more stable versioning. This is all with the goal of reducing the bus-factor from being just Henrik.

I’m just learning helix, but I believe you can move to the end of the line with gl, and i’m sure there’s an equivalent for the beginning.

This is referring to commits across all repos using those languages, not commits to the languages themselves.

I've just tested this, and it works fine. It asks for my password like normal from an ssh session, while still using touch id locally.

I would certainly be very interested in it if there was Python support (and VIM keybindings!).

Location: Toronto, ON, Canada

Remote: Yes, during COVID

Willing to relocate: Within Canada, primarily Toronto and Vancouver.

Technologies: Python, PyTorch, Tensorflow, Julia, SKLearn, Pandas, Data Visualization

Résumé/CV: https://www.dropbox.com/s/wdvxjp6bof5jzlp/SSegalHNResumeFeb2...

Email: [email protected]

----------------------------

I'm a Machine Learning Engineer that just got my MSc in Toronto, Canada, where I studied Computer Vision and ML-inspired Computational Photography.

I'm looking for a full-time ML Eng role where I can create and optimize ML pipelines and bring some new ML tools and techniques to market. I have experience as a full-stack developer before going back to school to get my Masters, and I'm also self-starter that can set and meet targets independently and under pressure.

I've performed freelance work during and after grad-school, but I'm looking to transition into a full-time position in the ML/CV space.

Once you EOL a product, you no longer have to support it. Not the same case at all as making changes to a core product and then continuing to support it.

Because Docker has a large financial incentive to get it working?

As far as I understand it, his hypergraph based theory we're reading about here is different from the cellular automata ideas in New Kind of Science?

yes, but i’m used to that already :)

By this logic, you think that something has to actually occur first before one can be morally opposed to it? I would have to first witness a genocide before I can have moral opposition to it?

Of course concepts can have moral weight. To suggest otherwise is foolish.

I had the exact same issue with a 2017 13” MBP. Switching from one of those tiny usb c hubs to a full fledged dock with external power solved it.

Unfortunately not on the newest consumer cards yet. Their CUDA competitor is ROCM which isnt available on the Navi 10 cards i.e the Radeon 5600-5700 series. I've been watching the space closely and it seems like support might land relatively soon though.

(Performance as compared to CUDA on cards that are supported is not great however)

You can very easily pair an AMD CPU and an Nvidia GPU.

I believe he's contrasting his current 16" experience vs the older butterfly switches.

Evil does support this, I use it every day. (Or rather, evil-escape is a plug-in I use that makes this setup very simple)