Hacker News: ilaksh

New comment by ilaksh in "Coding Is Not Solved"

ilaksh — Mon, 28 Sep 2026 16:13:16 +0000

Those are embarrassing bugs, but they don't prove that LLMs of that time couldn't be effective overall, and they don't mean that the latest LLMs aren't generally very good at coding.

New comment by ilaksh in "Coding is not solved"

ilaksh — Mon, 28 Sep 2026 16:07:14 +0000

Is that the fault of AI or management for not firing them?

New comment by ilaksh in "Z80 REPL (2018)"

ilaksh — Thu, 24 Sep 2026 09:07:58 +0000

Is there screen memory or anything? How do I output a character?

New comment by ilaksh in "Feds Target AI Critics as "Foreign Agents""

ilaksh — Thu, 24 Sep 2026 05:53:46 +0000

I am very pro-AI (although I do think we need to have a culture of caution) but this is another red-alert for democracy. People should be terrified if the president is able to make you into a state enemy just because you said you are against some technology.

New comment by ilaksh in "Feds Target AI Critics as "Foreign Agents""

ilaksh — Thu, 24 Sep 2026 05:45:09 +0000

Right or anyone who says they are anti-fascist.

New comment by ilaksh in "Meta VR Glasses"

ilaksh — Thu, 24 Sep 2026 03:32:23 +0000

It's not a headset, it's glasses.

New comment by ilaksh in "Stripe's Knowledge AI Platform"

ilaksh — Wed, 23 Sep 2026 16:57:52 +0000

Maybe they have jobs to do so they offloaded a lot of the presentation work to AI, assuming reasonable people would not judge based on surface level copy style.

New comment by ilaksh in "Claude Opus 5.5"

ilaksh — Tue, 22 Sep 2026 20:28:19 +0000

With the frequency of model releases, pelicans seem to have become a part-time job for you. But unpaid :/

New comment by ilaksh in "MiMo v2.6"

ilaksh — Tue, 22 Sep 2026 04:13:29 +0000

Because they can brag about the cost effectiveness.

New comment by ilaksh in "MiMo v2.6"

ilaksh — Tue, 22 Sep 2026 04:10:32 +0000

I think we are talking about the medium term for China to pull ahead. They just released a new domestic AI chip which seems in important ways to have caught up to Nvidias previous generation. They have a LOT more manufacturing capacity. They are collaborating via open source by default. They have better materials access and much more energy capacity. They have many more people overall and more researchers.

There is a strong chance most of the researchers are pulled out of the US and Europe if WWIII really kicks off or even if there is just more global crisis and concern.

One other thing about the power needs. Within a few years, the power efficiency of AI chips is likely to improve by a factor of 20, 50 or more times by switching to true compute-in-memory architecture with new materials that have made rapid progress lately.

New comment by ilaksh in "Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM"

ilaksh — Mon, 21 Sep 2026 13:33:18 +0000

I would be okay if you shared it as a potential idea and possibly interesting early result, but the language you are actually using to characterize it is misleading or delusional.

Please get a model to the point where it seems like it has some natural language understanding and then share again with reasonable characterization.

New comment by ilaksh in "Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM"

ilaksh — Mon, 21 Sep 2026 12:39:46 +0000

You can't claim it "works" if it hasn't produced any coherent responses and is still early in your first training attempt.

New comment by ilaksh in "Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM"

ilaksh — Mon, 21 Sep 2026 12:12:48 +0000

Actually I'm mad that I wasted my time looking at it based on the claims. He implies it is trained and uses the term "AGI" and "continuous learning". He never finished a single training run or enough that he considers not "undertrained". It's not trained. And actually there is no evidence that it can actually learn anything useful.

New comment by ilaksh in "Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM"

ilaksh — Mon, 21 Sep 2026 11:52:21 +0000

If you actually scroll through the transcript he links to, you will see that something that looks like it could be training is happening, but no coherent responses are coming out at any point. At least not that I saw skimming through.

That might explain why there are no benchmarks of any kind.

New comment by ilaksh in "Gemini 3.8 Live and 3.8 Live Extended Thinking"

ilaksh — Tue, 15 Sep 2026 19:33:48 +0000

OpenAI just released the new full duplex mode to the API as gpt-live-1 or something like that. Very realistic.

New comment by ilaksh in "A 386 PC for Your RP2350"

ilaksh — Mon, 14 Sep 2026 19:51:05 +0000

yeah but for 386?

New comment by ilaksh in "A 386 PC for Your RP2350"

ilaksh — Mon, 14 Sep 2026 17:55:42 +0000

yeah but the PSRAM is going to cause a delay

New comment by ilaksh in "A 386 PC for Your RP2350"

ilaksh — Mon, 14 Sep 2026 17:23:08 +0000

I don't know if they have some kind of magically genius caching system but the chip only has 512kb of SRAM and a little cache and the rest has to be managed by flash memory which is like 30+ times slower so I dunno about the performance of that.

New comment by ilaksh in "A 386 PC for Your RP2350"

ilaksh — Mon, 14 Sep 2026 17:19:38 +0000

The problem is PC programs assume 640kb of RAM and the RP2350 has 512kb + 16kb cache SRAM so anything that goes past that has a sudden 40 cycle access delay.

So even if you account for the RP2350 being 6 times faster than the RP2350, I think the emulation would randomly start to stutter a lot in certain circumstances waiting to access that part of the memory.

New comment by ilaksh in "The Interim Computer Museum"

ilaksh — Sun, 13 Sep 2026 13:34:11 +0000

Click on the Recollections page, it has a bunch of real vintage Unix (etc.) computers you can log into.