Foobar8568

4 ms·

1.
by Foobar8568 6d ago
Qwen can't train anymore with openai reasoning tokens? I kid I kid.
2.
by Foobar8568 7d ago
I am more curious about a 60k prompt... I haven't seen much discussion about large prompts, is it still < 500ms?
3.
by Foobar8568 8d ago
Well the Doom demo is again passing a textual structure....I am not really convinced on how it's different than any other llm that execute small context within 100ms. On a MBP M3Max with LFM 2.5B, I get about 500ms -600ms on &quot
4.
by Foobar8568 8d ago
I still don't get the point of jev....it's basically an optimized models/runner on really short context and output?
5.
by Foobar8568 8d ago
Ninfer is compatible with a nvfp4 model for the 27b. Also nowadays I prefer to use the byteshape one, I get less loops, and I am not sure if I really see a difference in speed or quality. Pure vibe agentic coding on a C++ codebase or ocaml
6.
by Foobar8568 11d ago
Considering the ability of men to aim in urinals, Cornucopia is really one of the worst idea.
7.
by Foobar8568 11d ago
I wanted so much to play with VR back in the 90s, but most 3d games give me motion sickness. I could barely play 5min of Goldeneye on N64, 2min playing Metroid and I was sick to death (headache, vomiting if I forced myself, feeling sick for
8.
by Foobar8568 12d ago
Also, I think the toe box durability test has a big flaw... At least in my experience with running shoes from Asics (Kayano and the other similar one). The toe box durability test is just from the top, not both the top and the bottom. My As
9.
by Foobar8568 15d ago
And don't forget contract management with the supplier, L1-L2-L3 support, all in one, and integrating as the supplier is useless and your contract is shit. Oh and you will do also business analysis with the business as claude is too co
10.
by Foobar8568 15d ago
From my experience building different systems, there is no moat anymore ( beside your commercial connections for B2B). If one lack understanding of a domain, he will get it by testing the system. So basically nothing has changed from before
11.
by Foobar8568 15d ago
How so?
12.
by Foobar8568 17d ago
With weak models, yeah if you get things compiled, with any Opus 4.6 or Codex, not really... It's a sad parody that people will take as reality.
13.
by Foobar8568 18d ago
Lol, and Switzerland has high working hours, but I achieve less in 3 months than 3weeks in Paris. My US colleagues, like German one,just bullshit on reality of what they produce. Real problem is more on VC, autonomy and small companies.
14.
by Foobar8568 18d ago
Not just legacy dying industry. Whole IT in France and Switzerland runs on consultancy.
15.
by Foobar8568 21d ago
Oh and how is it any different than most software engineers? How many times I heard ORM are bad only to recreate the same shit? How many times I heard ORM had bad performance and see 1+n stuff everywhere? How many times I have seen tight co
16.
by Foobar8568 22d ago
Look at the cost of a FP screen replacement, 100 €/chf for the FP4. Mine is going black randomly until I reboot the FP4 one to five times.
17.
by Foobar8568 22d ago
I have an FP4, and when it'll die, I will just buy an iphone. Security updates are slow, and after the bullshit of the android upgrade, yeah but no.
18.
by Foobar8568 23d ago
AIPocalypse.
19.
by Foobar8568 23d ago
> If they keep doing it they'll get a letter from the school and probably a police visit. It depends of the town/location and region. In our town, police had to be involved several times due to different issues with creeps targ
20.
by Foobar8568 23d ago
Where is my Gemma 5?
21.
by Foobar8568 24d ago
I post once with a fake/hidden email address to be contacted. I received thousands of DocuSign emails...
22.
by Foobar8568 24d ago
Take a look at the most viewed videos... It's gross. Basically 90% are just crap done by computers (pre AI), targeting preschoolers.
23.
by Foobar8568 26d ago
Memory used : 38GB, and I haven't even started a LLM nor podman, I always fight with memory when using LLM on my mac with 48gb. And I don't remember to have been able to have pushed to 200k context Qwen 3.6. 3.8 is running on my R
24.
by Foobar8568 1mo ago
I know companies that are using github, even using public repo, and request their teams to not use SOTA models, but are ok with local models. Just stupid policy.
25.
by Foobar8568 1mo ago
Well... Everything new is old "A vector space model for automatic indexing" 1975 - https://dl.acm.org/doi/10.1145/361219.361220
26.
by Foobar8568 1mo ago
Companies are avoid risks, and at this stage, sometimes, I feel that all cloud providers are just buying RAM that they would have bought anyway. Now it's ensure that no company will be willing to pay x digits just for cards. One of my
27.
by Foobar8568 1mo ago
Well for the last two years, chatgpt ( and I guess claude) were already better than most of my coworkers ( read IT in F500 style organizations ).
28.
by Foobar8568 1mo ago
Even for most organizations, testing AI systems is too cost prohibitive, so they YOLO in production, including public facing systems.
29.
by Foobar8568 1mo ago
And it's across the industry, thinking banks, private banks, insurance, pharamcy etc don't outsource their IT, including development... I believe US outsource even more than Europe on this matter.
30.
by Foobar8568 1mo ago
7) can't use reranker with it, asked by people for one year or two... 8) ...

More ›