HNHacker News
TopNewBestAskShowJobs

jychang

3,007 karma · joined October 23, 2012

submissionscomments
jychang··on Synology reverses policy banning third-party HDDs
It works fine as a NAS, but you need to run Plex on a different server.

You can't run Plex directly off the device like a DS224+ would.

jychang··on Joint Book Review: Class, by Paul Fussell
I think it’s important to read the book more as humor than as science.

He’s not trying to defend his hypothesis at a PhD thesis defense in front of a panel.

It’s more like he’s on the stage at a small TEDx talk. He’s not trying to offer proof for each statement, he’s trying to almost play stand up comedian keeping the crowd engaged.

jychang··on Launch HN: Grapevine (YC S19) – A company GPT that actually works
Onyx.app has a self hosted option. I just did the docker setup yesterday. It’s not a great home user option imo but seems like it’s functional for enterprise.
jychang··on Fire destroys S. Korean government's cloud storage system, no backups available
If the theory "north korea hacked the UPS batteries to blow" is true, though, then it makes more sense why fire suppression wasn't able to kick in on time.
jychang··on WireGuard topologies for self-hosting at home
Headscale isn't commercial
jychang··on High-resolution efficient image generation from WiFi Mapping
The image examples from the paper are absolutely insane.

Is this just extremely overfitted?

Is there a way for us to test this? Or even if the model isn't open source, I'd pay $1 to upload a capture from my wifi card on my linux box and upload it to the researchers and have them generate a picture and see if it's accurate

jychang··on Leaked Apple M5 9 core Geekbench scores
Any cpu can run at 100ghz for 1 cycle. The trick is getting it to run at 100ghz forever
jychang··on GPT-OSS Reinforcement Learning
So gpt-oss-20b is a sparse MoE model

Which means it has ~3b parameters active per token.

Qwen3-32b has 32b params active per token

jychang··on Qwen3-VL
Exactly. You're ill equipped to answer the question because you don't know. Qwen is terrible at explaining what the difference is, between the models that they serve on their API.

It's such a simple question: "For someone who does not want to run the model locally, what is the difference between these 2 models on the API?" and yet nobody can answer that question.

jychang··on Qwen3-VL
If the naming is so clear to you, then why don't you explain: for a user who wants to use Qwen3-VL through an API, which one has better performance? Qwen3-VL Plus or Qwen3-VL 235b?
jychang··on Qwen3-VL
They still suck at explaining which model they serve is which, though.

They also released today Qwen3-VL Plus [1] today alongside Qwen3-VL 235B [2] and they don't tell us which one is better. Note that Qwen3-VL-Plus is a very different model compared to Qwen-VL-Plus.

Also, qwen-plus-2025-09-11 [3] vs qwen3-235b-a22b-instruct-2507 [4]. What's the difference? Which one is better? Who knows.

You know it's bad when OpenAI has a more clear naming scheme.

[1] https://modelstudio.console.alibabacloud.com/?tab=doc#/doc/?...

[2] https://modelstudio.console.alibabacloud.com/?tab=doc#/doc/?...

[3] https://modelstudio.console.alibabacloud.com/?tab=doc#/doc/?...

[4] https://modelstudio.console.alibabacloud.com/?tab=doc#/doc/?...

jychang··on 25L Portable NV-linked Dual 3090 LLM Rig
Yeah, this page seems to be not great for beginners and also useless for people with experience.

A 2x 3090 build is okay for inference, but even with nvlink you're a bit handicapped for training. You're much better off with getting a 4090 48GB from China for $2.5k and just using that. Example: https://www.alibaba.com/trade/search?keywords=4090+48gb&pric...

Also, this phrasing is concerning:

> WARNING - these components don't fit if you try to copy this build. The bottom GPU is resting on the Arctic p12 slim fans at the bottom of the case and pushing up on the GPU. Also the top arctic p14 Max fans don't have mounting points for half of their screw holes, and are in place by being very tightly wedged against the motherboard, case, and PSU. Also, there's probably way too much pressure on the pcie cables coming off the gpus when you close the glass.

jychang··on Determination of the fifth Busy Beaver value
You guys need a blog to chase these down.

If nothing else, it’ll inspire the next generation of mathematicians.

jychang··on Notion API importer, with Databases to Bases conversion bounty
It's 1100 lines of code in a single file that nobody understands. That's instant spaghetti right there, not a valid PR.

At least have it be split into some files and structured in some way.

jychang··on Meta RayBan AR glasses shows Lumus waveguide structures in leaked video
The human body consumes merely ~2000 calories a day. Really not that much energy, about the same as a 100W light bulb.

Even if you can capture every single bit of extra energy from a tshirt, you'd end up with a tiny fraction of 100W. Certainly not enough to power a mobile device like this.

jychang··on "Your" vs. "My" in user interfaces
90% of all important work is banal. That's kind of the thing.

I'm sure a lot of engineering hours were spent on getting the door handle on your car to the exact safety/cost/functionality requirements, and at the end of the day, it's a door handle. Replace "door handle" with 99% of hardware and software that you ever see, and the same thing still applies. And yet, imagine using a car without a door handle.

Most important work isn't sexy, it's banal stuff that's boring until you remove it and realize how important it is.

jychang··on Qwen3-Next
To clarify, I should have stated: "Instead of generating tokens one at a time, you generate the second one as well WITH MTP, and then use speculative decoding on that second token (instead of having the second token be produced by a draft model like Qwen 0.6b). If the FIRST MTP token is checked and is correct, then the second token gets generated MUCH faster."
jychang··on Geedge and MESA leak: Analyzing the great firewall’s largest document leak
> My second thought is how badly Chinese communism must be doing that they need such a massive effort in order to prevent their citizens from accessing information and voicing dissent.

Well, OpenAI and other companies training AI models have shown that the architecture of the model matters less than the quality of data fed into it. Same applies for humans.

I understand that the Great Firewall is mostly about censoring dissent, but it's also to keep Chinese citizens away from junk food media sources. The type of videos you see on Douyin vs Tiktok is a great example of the difference.

Yes, the videos on Douyin are politically censored, but they're also a lot less brainrot than Tiktok videos. The Tiktok algo is optimized for ad impressions and profit, whereas the Douyin algo is more tuned to some nebulous concept of Confucian social harmony, for better or worse.

A more nuanced take is that I don't think it's useful to measure Chinese govt behavior just mapped to "amount of suppressing political dissent". I actually think the level of censorship is above the level required for that. It's more useful to recognize that "suppressing political dissent" is actually a subset of Confucian "promote social harmony"- which is not strongly valued in the USA but is at least important enough to be paid lip service in China- and I suspect a big chunk of educated members of government may truly believe in that ideal. It explains behaviors like "why the Douyin algo is so different from Tiktok" and other overreaches of the Chinese govt, because it's not solely about suppressing dissent.

jychang··on Qwen3-Next
Strong disagree.

Llama 4's release in 2025 is (deservedly) panned, but Llama 3.1 405b does not deserve that slander.

https://artificialanalysis.ai/#frontier-language-model-intel...

Do not compare 2024 models to the current cutting edge. At the time, Llama 3.1 405b was the very first open source (open weights) model to come close to the closed source cutting edge. It was very very close in performance to GPT-4o and Claude 3.5 Sonnet.

In essence, it was Deepseek R1 before Deepseek R1.

jychang··on Qwen3-Next
Because if you generate token n+1 with all 48 layers of Qwen3-Next and 80 billion params, and also generate token n+2 with the 1 MTP layer at 2bil params... that n+2 token can be much lower quality than the n+1 token but mostly correct.

Let's say you have a model that generates the string "The 44th president of the United States is ___ ___". Your model will generate "Barack" as the n+1 token, and the MTP layer probably does a good enough job to generate "Obama" as the n+2 token (even though that MTP layer is a mere <2bil parameters in size). Then you just check if "Obama" is correct via the same speculative decoding process, which is a lot faster than if you had to start over from layer 1-48 and generate "Obama" the regular way.

jychang··on Qwen3-Next
Speculative decoding! It makes inference a LOT faster.

Instead of generating tokens one at a time, you generate the second one as well, and then use speculative decoding on that second token (instead of having it be produced by a draft model like Qwen 0.6b). If the token is checked and is correct, then the 2nd token gets generated MUCH faster.

If it's wrong, you have to generate it again the normal way (a lot slower than just checking it). Usually, it's correct, so inference is a lot faster.

jychang··on Claude’s memory architecture is the opposite of ChatGPT’s
Obviously yes, AI makes revenue.
jychang··on Qwen3-Next
Coolest part of Qwen3-Next, in my opinion, (after the linear attention parts) is that they do MTP without adding another un-embedding matrix.

Deepseek R1 also has a MTP layer (layer 61) https://huggingface.co/deepseek-ai/DeepSeek-R1/blob/main/mod...

But Deepseek R1 adds embed_tokens and shared_head.head tensors, which are [129280, 7168] or about 2GB in size at FP8.

Qwen3-Next doesn't have that: https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct/blob...

So it saves a few GB in active parameters for MTP, which is a Big Deal. This is one of the changes that helps significantly speeds up inference.

jychang··on Experimenting with Local LLMs on macOS
None of that uses the ANE.
jychang··on Experimenting with Local LLMs on macOS
To be fair, Apple HATES Nvidia after the 8400M and 9400M debacle. They probably saw replacing Nvidia as a bigger benefit than replacing Intel.
jychang··on Anthropic agrees to pay $1.5B to settle lawsuit with book authors
Lol. I opened that link and was like "hmmm, that UI looks familiar".

I'm pretty sure that's just a frontend for Uptime Kuma https://github.com/louislam/uptime-kuma

jychang··on I have two Amazon Echos that I never use, but they apparently burn GBs a day
Meh, your smartphone is already the ultimate spying device that comes with microphones and triangulates your location from 3 cell towers. The government doesn’t need more spyware than that.
jychang··on Finding thousands of exposed Ollama instances using Shodan
Yeah, I don't think most people who even run ollama would care. "Oh no, someone found my exposed instance, which means my computer in my bedroom is burning electricity for the past few hours. Oh well, I lost a few pennies in electricity." Shuts down Ollama on the computer.

Seriously, this is extremely mild as far as issues go. There's basically no incentive to fix this problem, because I bet even the people who lost a few pennies of electricity would still prefer the convenience of ollama not having auth.

Plus, that's the worst case scenario, in real life even if some black hat found an exposed ollama service, they have no interest in generating tokens for <insert random LLM here at 4 bit quant> at a slow speed of <50tok/sec.

jychang··on We already live in social credit, we just don't call it that
> is legally permitted to execute me.

Plenty of cops have shot/executed innocent people, and were declared not guilty due to qualified immunity.

Plus, a phone in China can't execute you either. It'd be the police goons who do that.

jychang··on Cloudflare Radar: AI Insights
Yes, I agree here.

Everyone should remember, limitations of technology is not meant to define society. Instead, we build edge cases into technology to better match society’s general expectations.

A website owner saying “yes normal humans, no bad bots, EXCEPT good bots” is totally fine.

← PreviousPage 8 of 28Next →