datadeft

27 ms·

451.
by datadeft 4y ago
Please no. I use it every do on mac and I do not want to use such bloatware ever again. We need open standards and much better (simpler, faster, less bloated, yes , less features) implementations for pretty much everything in the Office sui
452.
by datadeft 4y ago
> What has your personal website/blog done for you? Nothing, but it helped many people to get started with something or fix problems for themselves.
453.
by datadeft 4y ago
Yes, I am not sure at what extent is MPS a viable alternative to CUDA. You seem to write a lot about ML models. Do you have a detailed write about this subject?
454.
by datadeft 4y ago
Could you give a bit more details about this? Do you have a link?
455.
by datadeft 4y ago
>> At Farba, we are building a zero-cost CDN infrastructure where all edge servers belong to anonymous Peers. Zero-cost CDN for 2 USD/TB! Sounds like a great deal to me. On a more serious note, is price the problem with other CDN
456.
by datadeft 4y ago
Sorry but who cares what HN does? Is this a business? No. Should we follow any of the practices of HN? I do not think so. My personal website has a more scalable infrastructure than HN.
457.
by datadeft 4y ago
This is exactly why I do not want to use such companies. I do not want to be an accessory in this racist crap. Your IP does not define who you are. Hetzner could two things: - stop operating in countries they don't want business from -
458.
by datadeft 4y ago
No. It is only a risk if an attacker can use it somewhow. Would you show me a scenario how could an attacker gain knowledge of which physical server my lambda function is running AWS and break out from the container and get access to my con
459.
by datadeft 4y ago
Based on my two decade experience in the field ranging being part of the team that was building super computers to one that was moving amazon.com to aws, there are many dimensions that you need to consider for computational workloads. The
460.
by datadeft 4y ago
Yeah companies and their small minded "sysadmins" are very eager to use an IP address as the primary predictor of hostile intent, sometimes they add user-agent as well. Not sure why but this is the sort of behaviour why I cannot t
461.
by datadeft 4y ago
Because I could not test it on anything else.
462.
by datadeft 4y ago
brew install python@3.11
463.
by datadeft 4y ago
Not bad with these settings: ./main -m ./models/7B/ggml-model-q4_0.bin \ --top_p 2 --top_k 40 \ --repeat_penalty 1.176 \ --temp 0.7 -p 'async fn download_url(url: &str)' async f
464.
by datadeft 4y ago
Yes it does. The blog post has the proof. You can use the nightly builds.
465.
by datadeft 4y ago
It was a typo. I fixed the url. Thanks for pointing out.
466.
Using LLaMA with M1 Mac and Python 3.11(dev.l1x.be)
617 points by datadeft 4y ago|165 comments
467.
by datadeft 4y ago
> posts are utterly worthless without data afterwards Same goes to any subject where a multivariate analysis is almost impossible due to the lack of control over your experiments. I seriously doubt that anybody has any data that could p
468.
by datadeft 4y ago
> Electricity costs are basically irrelevant because the cards are so expensive. You mean in terms of money. I think this is exactly the problem that we have in CS, nobody really cares about CO2.
469.
by datadeft 4y ago
How is this different from k8s? Maybe I am missing the point?
470.
Using Python 3.11 with AWS Lambda(dev.l1x.be)
1 points by datadeft 4y ago|0 comments
471.
by datadeft 4y ago
TL;DR: "Game developer optimizes code for execution as opposed to readability that 'clean-code' people suggest". There are few considerations: - most code is not CPU bound so his claims that you are eroding progress beca
472.
Using Rust for ETL(dev.l1x.be)
3 points by datadeft 4y ago|0 comments
473.
by datadeft 4y ago
> poor containerization support I use UTM and Docker on a daily basis and it is extremely smooth. What exactly is missing? > an annoying development toolchain What are you talking about exactly? For example most Python and Rust builds
474.
by datadeft 4y ago
> Hyperscalers are already looking to evolve from ARM citation needed https://www.theregister.com/2023/02/08/5_percent_cloud_arm/ "5% of the cloud now runs on Arm as chip designer plans 2023 IPO&
475.
by datadeft 4y ago
Could we make the chips go slower like around 1Ghz? Maybe that is not feasible with the current software architecture to achieve great user experience.
476.
by datadeft 4y ago
I have a iPhone SE II and a Huawei P30 and I refuse to buy newer / more expensive phones. All of needs were met when iPhone 5s came out and 99% of new features are just pointless. Unfortunately we live in a world partially thanks to Go
477.
by datadeft 4y ago
I was there was an alternative to iPhone that just works, the UI is consistent and "shallow" and the defaults are sane. I keep an Android phone around for the lolz. Sometimes I can do what I would like when there is no interruptio