Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GLM-5.3-Flash

z.ai

Post reply on HN
Print view View original

581–590 of 605 posts

Re: GLM-5.3-Flash

#581

Post by computerex »

saberience wrote: post #429
computerex wrote: post #373

Earlier quoted context omitted.

You heard of JEPA? LLM's have all sorts of garbage they have memorized. Reasoning in latent space instead of in text significantly reduces the number of needed parameters.

JEPA is a joke, let me know when those models do anything useful.

You should look at some literature around it. I don't have time to pull it up now but it's been shown that much smaller small million parameters JEPA model outperforms much bigger LLMS in some applications. Keep in mind JEPA is area of active research.

Re: GLM-5.3-Flash

#582

Post by cameronh90 »

oefrha wrote: post #461
tyre wrote: post #420

Earlier quoted context omitted.

Opus is a yap god. I've found it much, much better with Claude Code's output style set to `Concise` and this: https://news.ycombinator.com/item?id=49413456 We shouldn't have to resort to this, but it can be mitigated enough that it stays as my daily worker agent. Although, I mostly use Fable to farm out to Opus agents so I don't have as much exposure to what kind of blathering is going on in there.

It's crazy that sometimes I ask Opus 5 to explain what it just wrote to me, and it declares "that was word salad" (its words, not mine, without any hint from me other than "explain it").

[dead]

Re: GLM-5.3-Flash

#583

Post by tnchr »

nananana9 wrote: post #498
yanhangyhy wrote: post #471

Earlier quoted context omitted.

i tought we can just use AI to read the readme and figure it out? install things.. i always do this

We can, but a README like this is usually a very good heuristic that I don't want to install the thing.

It depends, first I ask myself if this didn't exist would I vibe code it myself, if I would, then is this good enough, might as well save my time and tokens on checking than building from scratch. Personal projects wise

Re: GLM-5.3-Flash

#584

Post by grandiego »

Aurornis wrote: post #60
mmastrac wrote: post #7

Weights on HF here: https://huggingface.co/zai-org/GLM-5.3-Flash I decided to take the plunge and get myself four sparks at a decent price (and bought the QSFP cables from AliExpress because they are literally 1/2 the price of Amazon), even knowing Apple was going to release new hardware and there's probably a spark 2 on the horizon. It looks like this is going to be a decent fit for what I need. I've been experiment…

> I would love to see an Opus-4.8-level local model but TBH I just haven't got there yet. The models I've tried so far _are_ good but they aren't able to solve tough technical challenges, regardless of harness/prompting/etc. Agree. It doesn’t even have to be local, using models in this size class through OpenRouter will reveal their limits if you work side by side with Opus level models regularly. There are a lot of…

> the difference is still huge

Same experience here. But I think the key argument for local is about being able to leverage "non castrated" models. But maybe this is not relevant at all for standard coding tasks.

Re: GLM-5.3-Flash

#585

Post by Natfan »

himata4113 wrote: post #39
epolanski wrote: post #4

I'm starting to think that this whole sanctioning China may motivate and prompt them to do more and better in every field. It's too big, bright and resourceful of a country to choose confrontation instead of collaboration.

Well the big problem with china is that they do not respect international law when it comes to technology theft. But that argument is very weak when it appears that a lot of what they do is out in the open for anyone to replicate.

if the americans didn't want IP theft, they shouldn't have taught tens of millions of people how to make their IP

Re: GLM-5.3-Flash

#586

Post by alfiedotwtf »

Balinares wrote: post #517
cdurth wrote: post #426

Earlier quoted context omitted.

I can't get 3.8 to exit thinking loops. It will just think and think and think on the most trivial topics. I wanted it to port a speed test powershell script to c#. Claude opus 5 completes it under 60 seconds. I let 3.8 churn about 6 different times for 30+ minutes and it never wrote a single line of code to disk. It wrote lots of lines in thinking. unsloth/Qwen3.8-27B-GGUF UD-Q3_K_XL DSH (pi) Any tips?

I've found it tends toward long thinking loops even for simple tasks (and any quantization seems to increase their length), but those do exit eventually, unlike with Qwen 3.6. I use the Unsloth UD_Q2_K_XL GGUF with default parameters, along with that custom template linked elsewhere in the thread, and no K/V cache quantization.

For smaller models, you'll probably find anything below Q4 will need handholding. Check Unsloth's graphs at the different quantisations VS error rates and you'll see why.

Re: GLM-5.3-Flash

#588

Post by maipen »

scotty79 wrote: post #580
gravypod wrote: post #552

Earlier quoted context omitted.

Why are these models able to reduce parameters but keep quality? I know the original intuition was scale data + params = quality but it looks like we have hit an s curve on improvements from pure scaling? Is this just because we are in a memory / data crunch? Are we learning how LLMs learn and effectively training better? Do we have a way to derive the amount of intelligence an LLM will have based on size / training…

They innovated a lot.

Hardware constrains forced this?

Re: GLM-5.3-Flash

#589

Post by HyperAI »

Jeremy1026 wrote: post #274
metadat wrote: post #168

Earlier quoted context omitted.

Qwen 3.8 27B is around Opus 4.8 level of capability on the Agentic Intelligence Index (52 vs 57). In my testing the locally hosted Qwen is good enough that looking at a given piece of work output I couldn't tell you which model was behind it. https://artificialanalysis.ai/models/qwen3-8-27b?models=gpt-...

I gave Qwen 3.8 27B and Opus 4.8 the same task in the same codebase. They both came up with the same diff. It wasn't a particularly challenging task (removing a feature flag and updating applicable specs), but it was character for character.

[flagged]

Re: GLM-5.3-Flash

#590

Post by alex7o »

KronisLV wrote: post #288
matheusmoreira wrote: post #255

Earlier quoted context omitted.

> Isn't this practically every TOS though? Not even close. Even OpenAI and Anthropic aren't bad enough that they claim literal ownership of your inputs and outputs. > HN's for example You're not paying to use HN. Getting banned here has essentially zero consequences. If Z.ai uses its absolute powers to ban you because you wrote a review about them or something, then you lose actual money. This is especially relevant…

Consider the reputation implications of them banning someone who can get their complaint about it to the front page of HN and into the YouTube drama loop. They’d get swarmed with activist cancellations. At most I suspect the A.I. providers will just come up with yellow banners like Anthropic did where naughty smut writers get put in the time out corner.

In the non China countries. But the Chinese populace would defend chinase companies unless they do something against China themselves
Post reply on HN