Hacker News: alex7o

New comment by alex7o in "StackOverflow now gets fewer than 1000 questions a month"

alex7o — Thu, 29 Oct 2026 07:43:42 +0000

Almost feels like training an llm on that caused it to ignore certain constraints that make a problem hard and solve an easier version of you problem instead.

Also a lot of communities do their own QA on discord where they can modererate themselves.

New comment by alex7o in "The Rise and Fall of Jev"

alex7o — Fri, 23 Oct 2026 11:51:48 +0000

E every 2 weeks we have a new hype cycle so all in according to how stuff already works I don't see something special here. I use jev, but it does not solve all my problems.

New comment by alex7o in "Run full iOS 16.5 firmware on Android Devices"

alex7o — Fri, 23 Oct 2026 09:29:52 +0000

Just use the jailbroken store and you get all apps from free from there

New comment by alex7o in "WorldTUI: I trained a 360M model to pretend to be a terminal"

alex7o — Thu, 22 Oct 2026 07:31:59 +0000

I think the toolcalling should be system prompts then the agent is the userspace

New comment by alex7o in "The Other Epidemic: Trying to get off benzodiazepines"

alex7o — Wed, 21 Oct 2026 09:13:09 +0000

I mean maybe a bit of a controversial opinion but this is how I hide at my desk job to try to build my own stuff. Like I give 3-4h of focused work to my employer and 6to myself every day. Some days I need to work more for work some I can do more stuff for myself.

New comment by alex7o in "Timestral: Mistral release time-series foundation models"

alex7o — Mon, 19 Oct 2026 15:44:39 +0000

https://www.nixtla.io/docs/introduction/introduction I remember using that and was quite good so happy that there is more open competition now in the space

New comment by alex7o in "Tmux and Claude Code: Sessions, Send-Keys, and the Ceiling"

alex7o — Mon, 19 Oct 2026 14:45:47 +0000

Now we are in the state where we have a thousand multiplexers

New comment by alex7o in "The gap in shared understanding of LLM capability is widening"

alex7o — Mon, 12 Oct 2026 11:23:09 +0000

I have a friend with a job like that and he is not being replaced not because it is Impossible but because it would be too expensive atm. But they replaced his whole team with workers south east Asia.

New comment by alex7o in "FTL: A new operating system for clouds"

alex7o — Sun, 04 Oct 2026 10:56:55 +0000

Stuff like u images still do that but allow you to bundle your existing software so you don't have to rewrite everything to run on it. The cool thing is that you can freeze at a syscall/hypervisor call like accept. And have very fast response times from cold

New comment by alex7o in "Pi 1.0"

alex7o — Thu, 01 Oct 2026 21:14:12 +0000

Look I can convince my boss to pay for jev, but I won't convince him to run our prod stuff on a rented vast.ai 5090. And the pricing wouldn't be worth it. If you have ideas I would be glad to hear them

New comment by alex7o in "Pi 1.0"

alex7o — Thu, 01 Oct 2026 20:10:27 +0000

Even jev is not truly novel, but it's latency is, you can use a reranker and get the same things but not the same speed.

New comment by alex7o in "Clef: Open-source decision models, and new RL fine-tuning platform"

alex7o — Thu, 01 Oct 2026 18:30:49 +0000

Oldy enough I tired this 2h ago as I was testing jev on cf and was like oh this should be a better replacement but it takes 3s which is useless to me

New comment by alex7o in "Price Increases for 2GB Raspberry Pi 4 and Raspberry Pi 5"

alex7o — Thu, 01 Oct 2026 15:40:46 +0000

Yes but also find a phone that is not top of the line and can be flashed with a custom ROM/kernel like linage OS, or Ubuntu touch, postmakret OS, then you can customize and have your pie and eat it too. Since you won't be needing the phone to have security for you banking stuff

New comment by alex7o in "Figma restricts MCP access to whitelisted clients, excluding Pi"

alex7o — Thu, 01 Oct 2026 15:30:02 +0000

That is very old, I use a figma CLI patched to look like Claude code so I can use it for everything

New comment by alex7o in "Ollaya – Ollama for open-source, Jev-style decision models"

alex7o — Sat, 26 Sep 2026 07:23:48 +0000

If you mean that I can compare the probilities between different tasks fair. This is a thing you can not do with a reranker. This sounds cool but not the amount of hype we got cool

New comment by alex7o in "Ollaya – Ollama for open-source, Jev-style decision models"

alex7o — Sat, 26 Sep 2026 07:18:13 +0000

Yea but a re-ranker can do that too if instruct trainrd. You update the instruction on which it is ranking the the values, a.k.a it is classifying them.

New comment by alex7o in "Ollaya – Ollama for open-source, Jev-style decision models"

alex7o — Fri, 25 Sep 2026 20:12:08 +0000

Guys I have a real q, what is the difference between an instruct based re-ranker and laya/jev I just don't see it.

Edit: One is that jev/laya are tuned to have better probabilities, but a reranker can be fine tuned to do that as well. And jev/laya use RLCD?

New comment by alex7o in "Jevmem – automatic project memory for Claude Code, built on Jev"

alex7o — Fri, 25 Sep 2026 18:50:03 +0000

The people that run local jev like models do sth similar. Train a seperate nn adapter on top of qwen 4b so it has balanced probabilities whatever that means and apparently it works quite well.

Side note: conspiracy theorists say that jev is a qwen model fine tuned but who know if true

New comment by alex7o in "Show HN: AgentRun: DSL to turn agents into workflows"

alex7o — Fri, 25 Sep 2026 05:43:30 +0000

This is just a custom workflow engine, but for sth like that you could have used any of the existing ones and given it to agents. Stuff like, Temporal, Erandril have a durable workflow solution build on pg only as well, or windmill/airflow depending on how you look at it. But without the nice ui given by these tools.

New comment by alex7o in "Making Tailscale Faster"

alex7o — Thu, 24 Sep 2026 08:24:02 +0000

I think it is the opposite google allows some apps from opting out of VPN.