segmondy

19 ms·

1. 511.▲
   by segmondy 1y ago

   Perform analysis? Group the articles by subject/writers. Have you detected that some writers are biased and just pushing a certain agenda? Predictions, find all articles predicting what will happen, what's the win/loss
2. 512.▲
   by segmondy 1y ago

   Well, IMO, the issue is that we are trying to merge with AI/LLM. Why must both of us understand the code base? Before it was us that just understood it, why not just have the AI understand it all? why do you need to understand it?
3. 513.▲
   by segmondy 1y ago

   caching is not a function of the model but the provider, all models can be cached. the provider serving the model decides if they are going to cache it. openrouter is not a provider but a middleman between providers, so some of their prov
4. 514.▲
   by segmondy 1y ago

   you declared a huge problem and followed up with an IF. deepseek API supports caching, stop manufacturing problems where there is none. https://api-docs.deepseek.com/guides/kv_cache
5. 515.▲
   by segmondy 1y ago

   Good way to get free drones.
6. 516.▲
   by segmondy 1y ago

   Yes you can, see this article https://www.nbcnews.com/tech/security/virginia-police-used-f...
7. 517.▲
   by segmondy 1y ago

   We are going to see a $1,000 Raspberry pi in the next decade, won't we?
8. 518.▲
   by segmondy 1y ago

   Inference is often like 200-250w without card clocked down. Then the other cards are like 20w-50w. 4 cards, 1 card is active at once. To get the full 350watt, you need to run parallel inference on the card with multiple users. So if I
9. 519.▲
   by segmondy 1y ago

   Only a matter of time till Palantir acquires them.
10. 520.▲
    by segmondy 1y ago

    The internet is worse for me with Cloudflare. I'm using a cellphone router for my internet. My guess is I don't get a dedicated IP and probably behind a NAT with other users. 85% of my request needs me to solve a cloudflare c
11. 521.▲
    by segmondy 1y ago

    ... and this is why napkin calculation is terrible. Even running a GPU at load doesn't mean you are going to use the full wattage. 4 3090 running inference on large model barely uses 350watts combined.
12. 522.▲
    by segmondy 1y ago

    I have rig of 7 3090s that I bought from crypto bros, they are lasting quite alright and have been chugging along fine for the last 2 years. GPUs are electronic devices not mechanical devices, they rarely blow up.
13. 523.▲
    by segmondy 1y ago

    Silly, if you are going to come up with a new benchmark, then add capable models, they have Opus, Gemini Pro, and then Qwen3-32B. Why not qwen3-coder-480b, qwen3-235b-instruct, deepseek-v3.1, kimi-k2, GLM-4.5, gpt-oss-120b?
14. 524.▲
    by segmondy 1y ago

    well, you believed the wrong folks Here's from my bsky feed 8 months ago after I took LLM to advent of code with local 32b model. ‪Segmond‬ @segmond.bsky.social‬ · 8mo I can confidently say that interns and junior developers can be re
15. 525.▲
    by segmondy 1y ago

    good find, there's often more than meets the eyes in these stories. folks forget that the court/case records will reveal hidden details.
16. 526.▲
    by segmondy 1y ago

    High value transfer is relative to the person making it. For some people, moving $1k-$5k daily is as normal as it is for a college kid moving $10-$50 daily. Then for some people $10k-$50k is the norm.
17. 527.▲
    by segmondy 1y ago

    When people show you themselves, believe them. Proton is no longer to be trusted. Use at your own risk.
18. 528.▲
    by segmondy 1y ago

    I was told to use ANY language in an interview. I asked them if they were sure, so I solved it with J. They were not too pleased and asked me if I could use another language, so I did prolog and we moved on to the next question. Then th
19. 529.▲
    by segmondy 1y ago

    would it have been better if I called it "shortcut" instead of cheating? all shortcuts are called cheating until people decide on it's fairness. the AI has been given a task to fix a bug, the AI figured out that looking at
20. 530.▲
    by segmondy 1y ago

    reward hacking is a thing and is also a hint of the models intelligent. We will fix this one, and the models will find a different way to reward hack in the future. "Cheating" is a sign of intelligence
21. 531.▲
    by segmondy 1y ago

    The models are getting better and better.
22. 532.▲
    by segmondy 1y ago

    I run them with under $3,000 hardware and inference is about 500-600watts with no noise.
23. 533.▲
    by segmondy 1y ago

    As it should.
24. 534.▲
    by segmondy 1y ago

    I know this is false, DeepSeekv3.1, GLM4.5, KimiK2-0905, Qwen-235B are all solid open models. Last night, I vibed rough 1300 lines of C server code in about an hour. 0 compilation error, ran without errors and got the job done. I want t
25. 535.▲
    by segmondy 1y ago

    The same way you use a cloud LLM.
26. 536.▲
    by segmondy 1y ago

    I ordered misc gadgets from aliexpress/alibaba to tinker with at least once every 2 months. I'm not buying anything else and moving from hardware experiments to software for now until the dust settles.
27. 537.▲
    by segmondy 1y ago

    And 3, people who are about to use AI to build a mental model quickly and understand the concept deeper
28. 538.▲
    by segmondy 1y ago

    How do you stop forced assisted suicide or do you think it will never happen?
29. 539.▲
    by segmondy 1y ago

    be careful what you wish for, you are giving up your freedom to movement in the name of security. you might make the argument that you can hail a cab. that's more expensive than owning your own car and with self driving cabs you wil
30. 540.▲
    by segmondy 1y ago

    if you are running a 2bit quant, you are not giving up performance but gaining 100% performance since the alternative is usually 0%. Smaller quants are for folks who won't be able to run anything at all, so you run the largest you ca