layoric

10 ms·

151.
by layoric 2y ago
For laptop, I personally don't know, only did a brief test on my backup system and seemed fine. However, Linux + Nvidia desktop has been good to me so far through many driver upgrades all through the standard 22.04 repositories (PopOS!
152.
by layoric 2y ago
Hardly a dumpster fire, moving to Linux from Windows is by far the best thing I’ve done for my work and sanity using a computer. For macOS users might be less of a reason, but aside from video editing workflow, has been a smooth process, an
153.
by layoric 2y ago
This looks like a useful tool but.. please review your generated docs for utility. """ Be ambitious about what you want your script to do Remember, LLMs in GenAIScript can do things that no other software has been able to do.
154.
by layoric 2y ago
I'd say you are right that solar is at your stage 4.. for countries with particular geographical features. If you have a large land mass to distribute solar utility sites over, the power curve becomes predictable since weather that imp
155.
by layoric 2y ago
Agreed. Sonnet 3.5 is still by far the most useful model I've found. o1-mini is priced similar and no where near as useful even if programming which it is suppose to excel. I recently tried o1-mini using `aider` and it would randomly s
156.
by layoric 2y ago
Yeah but tokenising all of it regardless of if it is ever used would be a lot more intensive yes?
157.
by layoric 2y ago
What’s the power consumption like for running this 24/7? Is it event based to reduce the need to process data when pc is idle? It’s an interesting idea for sure, but seems like a lot for an alternative of providing specific context.
158.
by layoric 2y ago
MI250s definitely aren’t a common card to rent so only can find Runpod at $2.10 per hour each. This results in a training cost of $4838 + fine tuning of $3225. However this doesn’t include the 11TB of storage or time taken to get the setup
159.
by layoric 2y ago
Agreed. Learn how to use EXPLAIN and interpret using whatever tools you prefer. Also monitor your queries. Something I did in a previous start up was install PgHero, and man did that tool help me optimise and prioritise performance.
160.
by layoric 2y ago
The statement about just throwing 200k tokens to get best answer for smaller datasets goes against my experience. I commonly find as my prompt gets larger, the less consistent the output becomes, and the poorer following instructions become
161.
by layoric 2y ago
Exactly my experience as well. I moved away from Windows (10) earlier this year, and login times (waiting for the user desktop to come up after putting in credentials) took ~10 seconds. Sometimes it was faster after a fresh install, but it
162.
by layoric 2y ago
I don't think it was this tutorial, but something similar I did around 2006 after completing a year of games programming study. Generating noise as height maps, mapping height ranges for texture blending and creating these terrains had
163.
by layoric 2y ago
Not as fast as the L40S, but Runpod.io has the A40 48gb for $0.28/hr spot price, so if its mainly VRAM you need, this is a lot cheaper option. Vast.ai has it for the same price as well.
164.
by layoric 2y ago
IMO if they released a mid level performer workstation card with 48-64GB of VRAM for a reasonable price (my understanding is standard GDDR6 is not overly expensive), say under $1k usd, they would not only fly off the shelves, but they would
165.
by layoric 2y ago
For those interested, this would be at a cost of: - ~$1691/hour on demand - ~$1065/hour reserved - ~$521/hour spot Not including any related data transfer costs.
166.
by layoric 2y ago
Tried, "Real time quote data is not available". Big brain move, can't crash the stock price if the nasdaq is offline..
167.
by layoric 2y ago
Nice write up, took me back to running LANded NT in the same neck of the woods as the author up in Darwin :) Small world!
168.
by layoric 2y ago
+1 to recently making the switch and to PopOS. There are still small experience pains coming from 2 decades of almost exclusively using Windows, but gaming has been pretty smooth, especially for recent titles. Valve has done an exceptional
169.
by layoric 2y ago
I think there is a close limit considering most of these gains are coming from the reduced memory bandwidth consumption that comes with the smaller data types. This would line up with Nvidia’s crazy graph from yesterday where data types wer
170.
by layoric 2y ago
Given the usage of LLMs to rank answers to questions, I think testing could shift as well. Having an LLM (operating with good faith prompt, eg trying to genuinely discover a level of understanding from a student) have a back and forth with
171.
by layoric 2y ago
I dunno, I still think the 2011-v3 platform that these Xeons can run in is still a great setup for a homelab. A bit power hungry but if you can build a dual core workstation with 36 cores, and 256GB ram for <$1000, that is a solid server
172.
by layoric 3y ago
I was first employee and only software engineer for 12 months at a startup working on increasing large scale solar power penetration in electricity networks globally. Worked there for 4 years solving all sorts of interesting problems workin
173.
by layoric 3y ago
> The main reason to go to Azure is because your bosses' boss was convinced to buy it at a golf course. This is why SharePoint got installed in so many gov/large enterprise in the past, and I have to assume why Azure has any ki
174.
by layoric 3y ago
What kind of performance do you see with an RX 6650?
175.
by layoric 3y ago
> Plus, we get to keep know how and avoid lock-in, which are extremely important things in the long term. So much this. The "keep know how" has been so greatly avoided over the past 10 years, I hope people with these skills sta
176.
by layoric 3y ago
Another Kagi user here, yes, the customization of results is way better than any other search engine I've used. Eg, personalization can be manually set to lower or raise weight of results from specific domains. This has become extremel
177.
by layoric 3y ago
Distracting is right! I watched your CrystalConf video and was happy to see the familiar Postgres + Redis combo :). I remember worrying about running out of legs with Redis (being single threaded), but with a combo of pipelining and changin
178.
by layoric 3y ago
"This didn’t exactly come as a surprise to us, as for the entirety of Kagi’s life so far we have actually used the cheapest, single-core database available to us on GCP!" Outages suck, but I love the fact that they are building su
179.
by layoric 3y ago
Hopefully it will never take hold, but I can see forcing individual episode purchases, or 'premium show' subscriptions (within an existing subscription) being a thing sadly.. Just have to look at twitch and subscribing to individu
180.
by layoric 3y ago
> The Big Bang Theory 420,400,000 That is just depressing IMO.. I don't understand the appeal to that show..

More ›