greyskull

7 ms·

1.
by greyskull 3mo ago
> mainstream What does "mainstream" refer to when we're talking about software development and LLMs? As opposed to "engineers".
2.
by greyskull 3mo ago
It's been a great multiplier for me in similar ways. The "dreamiest" thing has been that it has freed up time that I would normally have spent doing sprint work, to work on things that just don't make the cut until it&#x
3.
by greyskull 3mo ago
This really resonates with me, and I'm only a decade and change into my career. I use claude a lot day to day. I try to use it sensibly, making me more productive and produce better work. I'm also trying not to lose understanding
4.
by greyskull 4mo ago
Interesting. I'd have guessed there would be meaningful opex benefits to serving smaller models.
5.
by greyskull 4mo ago
> task focused small models This is tangential: and forgive my ignorance here, but is there an inherent reason why there aren't smaller, focused models from the frontier model providers? I'm thinking something like a software-s
6.
by greyskull 5mo ago
Thanks, I'll have to continue experimenting. I just ran this model Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL and it works, but if gemini is to be believed this is saturating too much VRAM to use for chat context. How did you land on that model?
7.
by greyskull 5mo ago
Thank you for all this, I'll give it a shot. Out of curiosity, are there any resources that sort of spell this out already? i.e., not requiring a comment like this to navigate. > nothing you can run locally, on that machine anyways,
8.
by greyskull 5mo ago
It's certainly part of the problem. Thanks, I'll give that a shot.
9.
by greyskull 5mo ago
Thanks! These things you're mentioning like "You may be able to offload some layers to GPU...", "You can keep the KV cache on GPU..." configured as part of the llama.cpp? I wouldn't know what to prompt with or
10.
by greyskull 5mo ago
I've been using Claude Code regularly at work for several months, and I successfully used it for a small personal project (a website) not long ago. Last weekend, I explored self-hosting for the first time. Does anyone have a similar e
11.
by greyskull 7mo ago
Missing "OpenAI sidesteps" from the beginning of the title article title
12.
by greyskull 2y ago
I agree.
13.
by greyskull 2y ago
Didn't get far enough along to understand the motivations and considered alternatives.
14.
by greyskull 2y ago
Didn't get far enough along to understand the motivations and considered alternatives.
15.
by greyskull 2y ago
OpenNext does model [0] incremental static regeneration, but beyond that I actually don't know, or at least don't recall. OpenNext doesn't do per-route lambdas like vercel does, so it's not like you get any behavior diff
16.
by greyskull 2y ago
1) I don't think it's related to Next, per se, but there may be behavior I didn't build the expertise to comment on. I also know that there were major inefficiencies in the application, so, for example, our P90 latency was (i
17.
by greyskull 2y ago
It offers packaging for deploying to a serverless environment (e.g. Lambda) analogous to how Vercel does it. The last question is salient, and it's possible for OpenNext to break and have to catch up to changes in Next.js, though I bel
18.
by greyskull 2y ago
OpenNext is just for packaging the Next.js build artifacts. The infrastructure is defined by projects that deploy those artifacts, examples here: https://opennext.js.org/aws/get_started Some of them are, for example, T
19.
by greyskull 2y ago
The biggest cost for us on Vercel (several hundred dollars a month) was Image Optimization, and that was because the app was being majorly inefficient with images, in part due to some default behavior in Next.js that we found unfriendly [0]
20.
by greyskull 2y ago
In the company I just left, I actually went through the process two or so months ago of migrating their Vercel deployment to AWS. I evaluated several options that are listed on the website and on GitHub, and we landed on using OpenNext via
21.
by greyskull 2y ago
Congratulations! I see that the book is incomplete. I didn't know that early access for books was a thing, very neat. It might be pertinent to note in your post that it's still being written, with an estimated release window of Sp
22.
by greyskull 2y ago
Might be pertinent to suffix this with (2023), though I see there are still recent replies
23.
by greyskull 3y ago
It's opening in Q2. Source: I'm one of the subjects of the article :^)
24.
by greyskull 3y ago
Off-topic, but what's with the strange font in the image at the top of the article? I see it's based off of the "csa" logo. I can't explain why but it's almost uncomfortable to read it with those connected glyp
25.
by greyskull 3y ago
> I would appreciate "automated rumble strips" that warns me gently if I veer out of my lane. Our Bronco Sport has this (steering wheel vibration) in addition to lane assist. I don't like the latter, so I appreciate that I
26.
by greyskull 4y ago
I have a very general question. What do you folks use these VPNs for besides, say, privacy/security or circumventing content restrictions?
27.
by greyskull 4y ago
I've never used Comcast's/Xfinity's own hardware. Always bought my own modem and router, which has generally worked out well, outside of when I'm (rarely) having quirky service issues and can't easily prove tha
28.
by greyskull 4y ago
I always wondered why reddit decided to make that inextricable link. Any insight on why post titles aren't editable? I thought maybe it's a database optimization, but it could also just be an unfortunate design decision.
29.
by greyskull 5y ago
The Honda S2000 had a throttle cable until... the 2006 model year I believe. I don't know how prevalent this was in the early 2000s.
30.
by greyskull 5y ago
Many/most teams internally using AWS use CloudFormation; an AWS service I was a part of was almost entirely built on top of other AWS services, and the standard development mechanism is to use CFN to maintain your infra. You only do dr

More ›