kbumsik

6 ms·

1.
by kbumsik 2mo ago
Why own numbering instead of CVE?
2.
Ray Tracer in SQL(github.com)
58 points by kbumsik 3mo ago|22 comments
3.
Ray Tracer in SQL(github.com)
3 points by kbumsik 3mo ago|1 comments
4.
by kbumsik 3mo ago
So no multimodal support yet?
5.
Python JIT project was asked to pause development(discuss.python.org)
182 points by kbumsik 3mo ago|101 comments
6.
Python JIT project was asked to pause development(discuss.python.org)
6 points by kbumsik 3mo ago|1 comments
7.
by kbumsik 4mo ago
So the draft model's performance is directly linked to the overall speed. Thank you for the explanation! By the way, can it be slower than without speculative decoding in worst case then?
8.
by kbumsik 4mo ago
> performance often degrades under different chat templates, long-context inputs, or out-of-distribution system prompts. I heard that speculative decoding doesn't affect performance (I meant accuracy). Am I wrong about it?
9.
Kimi K2.6(huggingface.co)
11 points by kbumsik 5mo ago|1 comments
10.
by kbumsik 6mo ago
Right? Even enterprise routers, e.g. Cisco, are not produced in USA.
11.
Logitech Options+ and G Hub macOS Certificate Issue(support.logi.com)
3 points by kbumsik 8mo ago|0 comments
12.
by kbumsik 10mo ago
We switched from Celery to Temporal. Temporal is such a great piece of distributed system.
13.
by kbumsik 11mo ago
> realistic written documents? Just get a DEF 14A (Annual meeting) filing of a company from SEC EDGAR. I have seen so many mistakes when looking at the result closely. Here is a DEF 14A filing from Salseforce. You can print it to a PDF a
14.
by kbumsik 11mo ago
Right, I wouldn't use full table detection to VLM model because they tend to mistake with numbers in table...
15.
by kbumsik 11mo ago
> My impression is that OCR is basically solved at this point. Not really in practice to me. Especially they still struggle with Table format detection.
16.
by kbumsik 1y ago
Well, that has nothing to do with JS itself though.
17.
by kbumsik 1y ago
In case if you don't know, Auth.js is not a frontend-only framework. It uses a backend server to make it secure. So it basically has no difference from the alternatives you mentioned.
18.
by kbumsik 1y ago
It is really easy to tell the difference btw. You will always see "4" or "7" in the middle.
19.
by kbumsik 1y ago
How does it compare to frp, one of the most popular Open Source Cloudflare Tunnel alternative? https://github.com/fatedier/frp
20.
by kbumsik 1y ago
Artificial Analysis is the only stable source. Don't look at others like HF Leaderboard. https://artificialanalysis.ai/
21.
by kbumsik 1y ago
Because gov infra also relies on CVE?
22.
by kbumsik 1y ago
> it probably is an anomaly that this was government funded Companies can definitely fund it. But to be fair the gov, including NIST, also relies on CVE.
23.
by kbumsik 1y ago
Yup. aws-cli is not dedicated for file sync. It was strange that there is sync command when aws-cli is meant to be a wrapper for a single api calls at the beginning. Use a tool that does one thing well like rclone.
24.
by kbumsik 2y ago
It is not same as OP, but according to a similar o11y stack on top of Clickhouse, Signoz, Clickhouse based logging costs less than ELK for storage and performs better: https://github.com/SigNoz/logs-benchmark
25.
by kbumsik 2y ago
How is it different from Signoz, a complete observability stack (including Logs) built on top of Clickhouse?
26.
by kbumsik 2y ago
Right. PoE will be fantastic.
27.
by kbumsik 2y ago
As a homelab guy, the only thing I wish is multi-gig ethernet. (2.5gbe)
28.
by kbumsik 2y ago
It was released around 4 years ago, when I worked at a civil engineering tech startup. We had a lot of Cesium 3D tiles of construction sites, captured by drones. It was quite easy to place them to Unreal engine. Is was fun to place random t
29.
by kbumsik 2y ago
So does it scale to zero, and scale from zero with <1s latency like Neon DB? AWS' recent "serverless" products, Aurora Serverless v2 and MSK Serverless for example, had been very misleading with "serverless" titl
30.
by kbumsik 2y ago
Large files are good, but it may have performance issues with many (millions) small files.

More ›