44za12

7 ms·

1.
by 44za12 1mo ago
Built rightsize exactly for this. https://nehmeailabs.com/right-size
2.
by 44za12 2mo ago
Others have already pointed out the absurdity, but just to put it across. Why not just ask first agent to update AGENTS.md adding context and next steps, and ask the next agent to continue from there?
3.
by 44za12 2mo ago
+1 came here to say this, I opened the link expecting some technical breakthrough. Misleading click bait title.
4.
by 44za12 3mo ago
Location: UAE Remote: Preferred Want to relocate: No Philosophy: Brutally Efficient More: https://aazar.me
5.
Stop generating what you already have(aazar.me)
2 points by 44za12 3mo ago|0 comments
6.
Horcrux – Distributed, Zero-Trust Secret Manager(github.com)
3 points by 44za12 3mo ago|2 comments
7.
Lattice Deduction Transformers(arxiv.org)
4 points by 44za12 4mo ago|0 comments
8.
MiniMax M3(xcancel.com)
5 points by 44za12 4mo ago|0 comments
9.
δ-mem: Efficient Online Memory for Large Language Models(arxiv.org)
240 points by 44za12 4mo ago|60 comments
10.
Beyond Semantic Similarity(arxiv.org)
68 points by 44za12 4mo ago|15 comments
11.
by 44za12 5mo ago
I read it as an article in defence of boring tech with a fancier/clickbaity title. Here’s the more honest one i wrote a while back: https://aazar.me/posts/in-defense-of-boring-technology
12.
by 44za12 7mo ago
Specialised models easily beat SOTA, case in point: https://nehmeailabs.com/flashcheck
13.
by 44za12 7mo ago
All of us use the same keyboards more or less, maybe us randomly typing a large number is not as random as we would like to think. Just like how “asdf”, “xcyb” are common strings because these keys are together, there has to be some pattern
14.
In Defense of Boring Technology(aazar.me)
1 points by 44za12 7mo ago|0 comments
15.
Show HN: RightSize CLI, Find the cheapest LLM that works for your prompt(github.com)
3 points by 44za12 8mo ago|0 comments
16.
by 44za12 8mo ago
Yes, I included a 'Model Selection Cheat Sheet' in the README (scroll down a bit). I map them by task type: Tiny (<3B): Gemma 3 1B (could try 4B as well), Phi-4-mini (Good for classification). Small (8B-17B): Qwen 3 8B, Llama 4
17.
Show HN: LLM Sanity Checks – A practical guide to not over-engineering AI(github.com)
1 points by 44za12 8mo ago|0 comments
18.
by 44za12 8mo ago
This is the way. I actually mapped out the decision tree for this exact process and more here: https://github.com/NehmeAILabs/llm-sanity-checks
19.
by 44za12 8mo ago
For simple extraction tasks, a delimiter-separated string uses 11 tokens vs 35 for JSON. Output tokens are the latency bottleneck.
20.
Stop using JSON for LLM structured output(nehmeailabs.com)
2 points by 44za12 8mo ago|1 comments
21.
FlashCheck-270M: Open weights for fact verification (Apache 2.0, WASM Demo)(huggingface.co)
2 points by 44za12 9mo ago|0 comments
22.
by 44za12 1y ago
Love the minimalism.
23.
by 44za12 1y ago
Shameless plug. I’ve been using a cli tool i had created for over 2 years now, it just works. I had more ideas but never got to incorporate those. https://github.com/44za12/horcrux
24.
by 44za12 1y ago
Have been using remove.bg for this for years now.
25.
by 44za12 1y ago
Like a sempahore?
26.
by 44za12 1y ago
I’ve had great luck with all gemma 3 variants, on certain tasks it the 27B quantized version has worked as well as 2.5 flash. Can’t wait to get my hands dirty with this one.
27.
by 44za12 1y ago
Can you benchmark Kimi K2 and GLM 4.5 as well? Would be interesting to see where they land.
28.
by 44za12 1y ago
That was quick, vibe coded, I presume?
29.
by 44za12 1y ago
The ability to submit a story using a curl would have been fun.
30.
by 44za12 1y ago
Tried that it’s taking exactly as much time as my program.

More ›