js98

8 ms·

1.
by js98 4mo ago
Personally, my recent and surprisingly greatest win was to set up my old phone (samsung S21) with the addictive apps, removing them completely from my iPhone. Quite literally "cold-turkey'ed" from 4.5-ish hours/day to 2
2.
Ask HN: How much AI is in your writing?
10 points by js98 4mo ago|37 comments
3.
by js98 5mo ago
Hey all, I’ve built Nekode ( https://github.com/Jakob-98/nekode ), which is desktop-cats for monitoring your AI coding agent sessions (OpenCode/Claude Code/ Copilot CLI/VSCode Copilot). It works as a macOS
4.
by js98 7mo ago
Oh wow thats good timing. I DIY’ed a workaround for this only last week for my openclaw instance. Happy to change it to this! https://jakobs.dev/granular-notes-access-limiting-openclaw-b...
5.
by js98 8mo ago
Hey all. TL;DR: none of this is super novel or fancy, but the set of a few simple approaches (cloud VM, tailscale for connecting, strong isolation, long-running sessions, basic notifications) made using agents with all tools enabled much mo
6.
Pragmatic Notes on Running Dangerous AI Coding Agents in Cloud VMs(jakobs.dev)
4 points by js98 8mo ago|1 comments
7.
by js98 8mo ago
https://jakobs.dev Sharing learnings from engineering work!
8.
Where is Point Nemo? (the point furthest away from land)(lukatela.com)
3 points by js98 8mo ago|1 comments
9.
by js98 8mo ago
I tested ChatGPT Atlas as an alternative to writing a small scraper for collecting price distribution counts from Dubizzle (UAE Marketplace). It could extract correct numbers for a single filter, but consistently failed to repeat the same s
10.
ChatGPT Atlas breaks down on simple, scoped repetitive browsing tasks(jakobs.dev)
3 points by js98 8mo ago|1 comments
11.
by js98 11mo ago
Similar writeup I did about 1.5 years ago for processing millions of (technical) pages for RAG. Lots has stayed the same it seems https://jakobs.dev/learnings-ingesting-millions-pages-rag-az...
12.
by js98 2y ago
This website is unreadable on mobile.
13.
by js98 2y ago
Nice work! I spent a bit too much time creating a map with 5 colors… ;)
14.
by js98 2y ago
This seems just like a single argument and counter argument against the sentience (or lack thereof) of AI today. I feel like the article lacks some broader views on the nature of sentience, and is quite narrow in its approach Although, in f
15.
Learnings from ingesting 5M+ technical pages for RAG on Azure(jakobs.dev)
2 points by js98 2y ago|0 comments
16.
by js98 2y ago
Good story, but the writing is annoyingly tough to follow.
17.
Azure AI 'Hybrid' search gotcha: returning unexpected/unmatching results(jakobs.dev)
1 points by js98 2y ago|0 comments
18.
Ask HN: What is the current (Apr. 2024) gold standard of running an LLM locally?
195 points by js98 2y ago|88 comments
19.
Show HN: Inject hidden prompt in LLMs using Base64 encoding(jakobs.dev)
5 points by js98 3y ago|0 comments
20.
by js98 3y ago
Ha this is great! I built something similar to this for myself a few months back at my personal site at https://jakobs.dev/chat
21.
Show HN: GPT(-4) CLI with OS-agnostic persistence in 10 lines of code(jakobs.dev)
2 points by js98 3y ago|0 comments
22.
Show HN: Voice record, redact with GPT4, and save to Notes using Apple Shortcuts(jakobs.dev)
2 points by js98 3y ago|0 comments
23.
Evaluating RAG/LLMs in highly technical settings using synthetic QA generation(jakobs.dev)
3 points by js98 3y ago|0 comments
24.
On automating (writing of) unit tests with LLMs(jakobs.dev)
3 points by js98 3y ago|0 comments
25.
by js98 3y ago
Actually a company has been working on this for a few years now, and I believe they are currently in production. Their focus is football/soccer I believe. I was going to do a research internship at them before I dropped it for a differ
26.
by js98 3y ago
Very cool! At first the title reminded me of a project me and my colleague are working on called OpenAI-Functools [1], but your concept is quite the opposite, combining LLMs in your code rather seamlessly instead of the other way around. Qu
27.
by js98 3y ago
Author here: completely agree. I think it is not much of an achievement by itself, but it is interesting to see that it works! I will add a comment/edit to the post once I am home to clarify the relative ease of solving MNIST
28.
78% MNIST accuracy using GZIP in under 10 lines of code(jakobs.dev)
363 points by js98 3y ago|137 comments
29.
by js98 3y ago
Sounds like I should join! This is something I knew should exist, but hadn’t found it yet. Will sign up for one of the events, for now feel free to already look at https://Jakobs.dev
30.
by js98 3y ago
Easily generate function metadata (json) automatically for GPT function calling: https://github.com/Jakob-98/openai-functools My colleague and I have been working on this tool and using it in some internal projects. It

More ›