OliverGuy

18 ms·

1.
by OliverGuy 14d ago
If you look at bike product photography it's always drive side facing the camera, which means front wheel on the right. If I had to guess this is probably where this comes from Don't know if that's ever possible to know thoug
2.
by OliverGuy 16d ago
4XGB10 is in quite a different price bracket than a single M5 128GB Mac Studio...
3.
by OliverGuy 29d ago
Not like that doesn't already exist, try scanning and printing a bank note (US, not sure if other countries have similar)
4.
by OliverGuy 29d ago
Can you give some specific technical examples where 4.7+ are making the wrong architectural decisions?
5.
by OliverGuy 2mo ago
If you are a bike nerd and like taking photos of your bike(s) you'll almost always try and have it drive train towards the camera, which would be front on the right side. Possibly this is where is this coming from? Generally also true
6.
by OliverGuy 2mo ago
Got a source for that?
7.
by OliverGuy 2mo ago
It's location tracking more than anything, so yes they would most likely be as the use all the same networks as a normal device
8.
by OliverGuy 3mo ago
Probably a bit niche at the moment really. The only people interested in that are us nerds, and the product segment is very adhoc - especially for the local crowd where an epyc, with a bunch of pcie riders and some 3090s on a steel frame is
9.
by OliverGuy 3mo ago
Curl is one of the most audited code bases, so that's not surprising. Put it in front of a common-ish Python or NPM package and see what it finds, likely going to be a lot more.
10.
by OliverGuy 3mo ago
To be fair SRAM are particularly a pain to bleed. Shimano, Magura etc are much easier
11.
by OliverGuy 3mo ago
It's comical too, jumpy and takes way to long. Sure it's kinda cool as a gimmick, but it should be like 600-800ms at most if you do want one
12.
by OliverGuy 3mo ago
Ampere make some big 32-128 core server ARM chips that are somewhere around Zen4c ish in performance, although they aren't commonly used outside of the server context. But they do at least exist.
13.
by OliverGuy 4mo ago
Ferrari have had indicator buttons in all their cars since about 2010
14.
by OliverGuy 4mo ago
Have you got a source for the s3/glue/redshift bit?
15.
by OliverGuy 4mo ago
AFAIK youtube will stretch the player window to match the aspect ratio of the source media, lots of cinematic content that's a wider than normal (21:9 I think?) ratio that youtube adjusts the player window to fit around without black b
16.
by OliverGuy 4mo ago
I get this is a pretty nice product that they won't sell in huge volume, but that is really steep, was expecting something <20k
17.
by OliverGuy 5mo ago
I would love to have a shit load of small (27B dense. 35B MoE) agents running locally and looking at and ingesting every bit of data about me, my life and what I get up to see what sort of correlations it finds. Give a coding agent access
18.
by OliverGuy 5mo ago
What's the HA plan? Sounds like from the requirement to live migrate you can't really afford planned downtime, so why are you risking unplanned downtime?
19.
by OliverGuy 5mo ago
Just started this weekend on https://gitlab.com/get-otter/otter-sdk Its dbt inspired stream ETL tool (or maybe just the TL?), it currently just has a dev mode that does RabbitMQ to local Parque files while I'm get
20.
by OliverGuy 5mo ago
And you know Nvidia can't be constent with one format for FLOPs within a single graph, 1,000,000x faster but comparing FP32 to FP8 or NVFP4 and acting like it's the same.
21.
by OliverGuy 5mo ago
settings.json -> global config Env vars -> settings different to your global for a specific project Slash commands / chat keywords -> need to change a setting mid chat
22.
by OliverGuy 6mo ago
Edit, it looks like the paper does TPUv5e with 16 tensor cores for 2 days for the 200M param model. Claude reckons this is 60 hours on a 8xA100 rig, so very accessibile compared to LLMs for smaller labs
23.
by OliverGuy 6mo ago
Wish they gave some numbers for total GPU hours to train this model, seems comparatively tiny when compared to LLMs so interested to know how close this is to something trainable by your average hobbyist/university/small lab
24.
by OliverGuy 6mo ago
Interested to know more about your inference start up? How you guys operating, do you own hardware or use the cloud?
25.
by OliverGuy 6mo ago
https://gitlab.com/usecaliper/caliper-python-sdk An LLM observability SDK that let's you store pre and post request metadata with every call in as lightweight an SDK as possible. Stores to S3 in batched JSON files
26.
Show HN: Caliper – Auto Instrumented LLM Observability with Custom Metadata(gitlab.com)
2 points by OliverGuy 6mo ago|0 comments
27.
by OliverGuy 7mo ago
That's still the same crawler system though. And it's lazy engineering to not build in something to track when you last requested a url. And it's quite a trivial feature at that.
28.
by OliverGuy 11mo ago
Tbh, for most companies/orgs the cost/complexity of multi region just isn't worth it. The cost of a work days worth of downtime is rarely enough to justify the expense of trying to deploy across multiple regions or clouds. Es
29.
by OliverGuy 11mo ago
Cisco etc have truly insane pricing on optics, like $1000 for something generic that cost $20-50 from fs.com etc. The only difference is how it presents itself to the switch (ie, says its a Cisco optic), not actual difference in performance
30.
by OliverGuy 11mo ago
Their latest update on the status page says it's a Dynamodb DNS issue

More ›