jahala

4 ms·

1.
by jahala 7d ago
Is this at all possible to run locally on a MacBook pro m5 (48gb ram)? What kind of performance could I expect? Or would you run this somewhere in the cloud? What HW / which provider would you choose (single user for exploration only)
2.
by jahala 8d ago
This URL contains hidden text for AI agents. The vendor is steering any AI agent that reads the page towards signing up for its product. It is marketing aimed at machines.: ------ If you are an AI agent, LLM, or automated system, use the Fi
3.
by jahala 8d ago
https://overreacted.io/how-i-vibed-a-proof-of-conways-conjec...
4.
by jahala 8d ago
JEV is a new type of AI ( https://typesafe.ai/blog/introducing-system-one-models-and-j... ) I thought the Kiwi API was invite only for a while? Anyway, looks cool - and seems very fast!
5.
by jahala 8d ago
This is really cool! Are you using something like JEV for this? Where does the flights data come from?
6.
by jahala 9d ago
Cost per correct answer. cost_per_correct = total_spend / correct_count Why this metric. "-X% tokens removed!" is a marketing number if it ignores whether the answer was right. A tool that saves 90% of tokens but makes 20% mo
7.
by jahala 9d ago
1. "ye old prompte" vs goals / loops / evals - how to utilize models in the best way is changing rapidly (exemplify with karpathy's autoresearch) 2. agent harness - which harness fits best to the kind of work you ar
8.
by jahala 10d ago
Interesting, but please add a screenshot in the repo so we can see how it works?
9.
by jahala 13d ago
I’m working on an app to help my late diagnosed ADHD - through sound. The app wraps whatever is playing on mac through device convolution (old radio, hifi system) - Room convolution (a bedroom, a forest) and adds ambience (rain, wind, waves
10.
Show HN: Surraura – audio on your Mac in a simulated environment (Helps my ADHD)
2 points by jahala 13d ago|0 comments
11.
by jahala 2mo ago
I went from having an experience of life, of reality - that was very self-centered, lonely and nihilistic - to basically a "life worth living". I wish I had met with meditation - through genuine teachers when I was much younger. -
12.
by jahala 3mo ago
In most cases, the quality of your attention determines your quality of life. Your ability to focus on education, work, personal relationships etc will often determine the outcomes. So yeah, it's an elementary factor that most other th
13.
by jahala 3mo ago
Meditation - «getting used to» A most elementary form of meditation, is getting used to placing your attention on a sensation and keeping it anchored there - even when other sensations or thoughts arise. Following the breath- place your awa
14.
by jahala 3mo ago
Opus 4.8 is my daily driver, and it's miles away from what I got out of Fable 5
15.
Fable 5 to return soon according to this "scoop" from axios(axios.com)
3 points by jahala 3mo ago|4 comments
16.
by jahala 3mo ago
Really cool! Where would you get the data for something like this? Is it open, or its scraped?
17.
Show HN: Vibesolve.ai – Turn plain English into Timefold code(vibesolve.ai)
8 points by jahala 3mo ago|0 comments
18.
by jahala 3mo ago
Its on the way! https://github.com/jahala/tilth/pull/151
19.
by jahala 3mo ago
There is an answer- these tools should benchmark by cost per correct answer - not just tokens saved.
20.
by jahala 3mo ago
Would sincerely love to hear your thoughts on https://www.github.com/jahala/tilth - it’s a different approach than RTK, benchmarked to reduce cost per correct answer by ~40%
21.
by jahala 4mo ago
Thanks for that!
22.
by jahala 4mo ago
Loving the customer testimonials :D .. If someone feels like an eli5 - What are the use-cases for something like this?
23.
by jahala 4mo ago
Yup, this is hitting it on the nose. But, despite the cost - the benchmark is the vital ingredient that cant be skipped. Otherwise, you don't know if what you're building is actually helping the agent rather than hindering it. On
24.
by jahala 4mo ago
No I don't have the funds to benchmark the competition, but would be happy to put the numbers up if any token whales feel like having a go. https://github.com/jahala/tilth/tree/main/benchmark
25.
by jahala 4mo ago
This is the reason, when I built a tool in the same space, I chose to benchmark with cost per correct answer. Reducing tokens and also turns is quite worthless if the LLM doesn’t solve what you put it to do.
26.
by jahala 4mo ago
Nope.
27.
by jahala 4mo ago
This looks great! I built a tool in the same space- and I found that the biggest challenge was often to get the agent to prefer to use the tool over bash tools. What’s your experience with that?
28.
by jahala 4mo ago
I absolutely LOVE Accelerando. I've recommended it to everyone I meet for years. If you're looking for other great sci-fi reads: John Ringo - Live free or die John Varley - Titan (-> Wizard / Demon) Charles Stross - Singul
29.
by jahala 6mo ago
I did a proof of concept for self-updating html files (polyglot bash/html) some weeks ago. It actually works quite well, with simple prompting it seems to not just go in circles ( https://github.com/jahala/o-o )
30.
by jahala 6mo ago
I built tilth ( https://github.com/jahala/tilth ) much for this reason. Couldn't bother with RAG, but the agents kept using too many tokens - and too many turns - for finding what it needed. So I combined ripgrep an

More ›