Enlarge Your Genus

Побочными эффектами с поиском работы прилетают всякие кринжовые сообщения в линкедин, буду их собирать тут

@antirez Any chance of the smallest possible GGUF?

First kinda working implementation of GLM 5.2 in DwarfStar. Will take some time to be good enough, but it is a promising start. 433 GB GGUF file. M3 Ultra 512GB.

@DJLougen well its blocking

The neon runner game is playable here Don't fail like me ! huggingface.co/spaces/DJLouge…

@Systerr It still requires babysitting, and you may need to direct it to the right decision multiple times. I would say it's at the level of the previous year's mid-year flagship models. If I had to choose between coding models (excluding commercial), I would definitely go with GLM 5.2.

@WaveCut How is glm 5.2 to you compared to opus?

a 3D game i made completely by GLM 5.2 and literally nothing else. plotva.geta.moe

@pipenetwork please next time do REAM :)

What hardware do you wish to use GLM 5.2 on? Taking requests 👇

I created a REAM version of Qwen3.6 35B, which reduces the total number of parameters to around 24B. huggingface.co/WaveCut/Qwen3.…

Evil.

@TomDavenport It also depends on how often one encounters such responses in the pretraining data pile. I believe there are many distillation datasets published targeted at Claude specifically because it showed the raw thinking chain in the past.

@WaveCut Makes sense if someone posted an example of it thinking it is CHatGPT, or humans like John Smith, etc.

People forget that the Chinese open source is probably distilling the western proprietary models anyway… glm insists it is Claude and is surprised to learn otherwise I wouldn’t be surprised if the US controls turns out to be an aggressive move to mute Chinese open source ai

@TomDavenport Well I saw that in multiple different tweets on this topic recently

@Michaelzsguo This amount of bucks gives you basically an unlimited amount of tokens of the models you can infer on Mac but served at the speed of light.

Apparently I can rent out my MacBook Pro for inference and make ~$423/month. That means the machine could pay for itself in roughly a year. Sounds too good to be true. Has anyone actually tested Darkbloom as a provider? Real demand, real payouts, or just optimistic marketplace math?

@maria_rcks But it really can't. The coding plan comes bundled with the MCP for image understanding and web search and web fetching, though

come on glm 5.2...

Disliking all the fat kitten shitposts in my feed just to teach algos not to promote fake news.

@argofowl dlaczego

codex app needs more polish

That’s what happens when you base your marketing campaign on fearmongering.

@oft @antirez Ban Mistral f.e. 😂

@antirez The EU is way more likely to do something similar, with a model way less scary

What Europe should do right now: 1. Call all the European researchers working on AI and return them back with same salary (or they can stay but switch career). 2. Fill EU places having GPUs with money, and put those people there. 3. AI partnerships with China + India.

@granular_verb @uspehink турук макто

кто

@ostrisai It would be neat to build a crowdsourced version on top of this, like there's a website with a public voting open with the sane timeouts and live gauges.

I prototyped out an idea. Want to gauge interest. It is a live reward trainer. Images are generated and you pick you favorite variant and your preference is fed straight into GRPO. It is a very involved process, but should adjust a model to your personal preference.

@sflorimm always has been tbh

The fastest way to get hacked now is to run npm install. what a time to be alive

@bridgemindai The bold assu,ptions that the API token price is fair.

You are not paying for AI. You are being subsidized. $200/month Claude Max 20x gets you up to $8,000/month in API compute. That's 40x what you pay. $200/month ChatGPT Pro 20x gets you up to $14,000/month. 70x what you pay. Anthropic and OpenAI are lighting billions on fire to win developers. This is not scalable. The rate limit "adjustments" we keep seeing are the rugpull starting in slow motion. Enjoy it while it lasts.