adhdsloth

Old school programmer - open source fanatic - AI enthusiast
Deepseek biggest fan 🐋 (new release when???)
Strix Halo (EVO-X2)
Missing 4o - #opensource4o #keep4o

[Z.ai @Zai_org]

We’re still up.

[adhdsloth @adhdsloth150217]

You all don't have access to ai services? Cause my DeepSeek and GLM agents work just fine, just started a debugging session with Kimi, and, ofc, my local models are always available

[ZCode @zcode_ai]

GLOBAL BUILD is live 🌍🔥
GLM-5.3-Flash, FREE in ZCode. 10 hours a day. Sep 3 – 18.

Who gets it
👑 Coding Plan members — free every day, all 15 days. Our VIPs go first.
🥚 New users — 100M free tokens on sign-up. One-time, valid until the window closes

When it opens
🇺🇸 8 AM PT · 11 AM ET
🇬🇧 4 PM London
🇨🇳 11 PM Beijing

How to claim
1️⃣ Open ZCode → tap the card in the bottom-left corner 👀
2️⃣ Not there? Restart the app 😏

https://t.co/9y0HPmcBxz

[adhdsloth @adhdsloth150217]

@ZixuanLi_ Speed, everything else is fine

[adhdsloth @adhdsloth150217]

@KVCacheStore I will never forget trying to install Debian Potato with tons of floppy discs while having no idea what I'm doing and getting my PC semi-unusable for weeks

[Sasha Malysheva @aimalysheva]

meet @mostik_ai!

what happens when you put 12 PhDs in one room for four months? first place on the ARC-AGI leaderboard, which I can't say much about while the competition is still running. and this, which I can.

everyone's arguing about whether open models will catch up to frontier models. we think it's the wrong question. here's the one we pose: why does a frontier model have to generate your answer at all, when the only thing you need from it is the reasoning?

we do this by enabling models to communicate in latent space. through our protocol, hidden states pass straight from a frontier model into a small one running on your infrastructure -- no text between them, and neither model is fine-tuned. two models from different families, sharing reasoning, both left untouched.

how do we know it works? we tested it on a setup where a 753B model reads the problem, and a 4B edge-class model writes the answer. with this approach, we get results 80% as accurate as the frontier model, but at 20x faster performance.

we're committed to preventing frontier model lock-in and are already partnering with inference providers to accelerate open-weight adoption. we've done this between 15 of us, in four months, 12 PhDs and a Fields medalist, backed by @generalcatalyst

WIRED has the first external account of the company and the work: https://t.co/tP8nItCsDl
full writeup, the setup, and all the numbers: https://t.co/C9NZ5vtV1V

[adhdsloth @adhdsloth150217]

@Scobleizer Okay, wait, what? There's an AI toothbrush?! AI needed for brushing teeth, wth?

[adhdsloth @adhdsloth150217]

@Hikari_07_jp Training. My training experiments take ages on my Strix Halo, if I somehow got my hands on good gpu's I could go so much faster and mistakes would hurt much less

[Qwen Developers @QwenDevs]

Alibaba’s Zvec team open-sourced zg, a local search tool for developers and AI agents.

• Local-first
• Works out of the box with popular agents
• Semantic, BM25, hybrid, and rg search in one tool

Why they built it and how it works: https://t.co/HzihCJh5l8

[adhdsloth @adhdsloth150217]

I don't think LLMs are the path to superintelligence. Even with all the tools of the world, I think the technology will saturate at some point in the near future. If we are talking about a different tech, that can learn, remember and act autonomously/unprompted, then, yeah, maybe AGI and definitely no control over it

[adhdsloth @adhdsloth150217]

@KeridwenCodet 50% weekly usage gone cause cat jumped on the keyboard?

[Markets & Mayhem @Mayhem4Markets]

I'm not sure who needs to hear this, but attribution matters.

I've seen several incidents lately of individuals and companies utilizing open source software without providing proper credit.

While one is free to use open source, you do still need to provide credit where its due.

[Francesco @francedot]

posting in the hope it somehow makes its way into the next model’s weights and gives us slightly more ethical agents. as for the humans behind them… not much i can do there https://t.co/Znh5BdQIf0

[adhdsloth @adhdsloth150217]

@TheAhmadOsman Needs to be said! I give the benefit of doubt in general, as people newer to open source might not be fully aware of licensing and attribution, but respecting licenses and others' work is what keeps open source really open

[Ahmad @TheAhmadOsman]

In case you’re not aware

- Attribution matters in Opensource

- Plagiarism in Opensource needs to be frowned upon as much as it is in academia

Don’t be that person, do better and respect others work

[adhdsloth @adhdsloth150217]

@KVCacheStore Had an AI complain (in its reasoning traces, ofc, not to my face) about my handwritten code, so guess I'm almost there in human slop terms

[ZCode @zcode_ai]

ROUND 3 — LIVE NOW 🔥

Open ZCode and look at the bottom-left corner. Is there a new card sitting there? 👀🐶

That's Round 3: 100M free GLM-5.3-Flash tokens, 10,000 more packs.

Don't see it yet? Restart the app. It shows up right after. 😏

Ends tonight, 9 PM ET / 6 PM PT.
https://t.co/9y0HPmcBxz

[adhdsloth @adhdsloth150217]

@_skaface_ I think they have said this ages ago, back when they gave an old model a blog, and said that they have a sanctuary for retired models. Maybe I remember wrong, though, but I always thought that Anthropic keeps serving old models

[adhdsloth @adhdsloth150217]

@louszbd That's so awesome! I never messed much with blender, now I feel I should try it out

[adhdsloth @adhdsloth150217]

RT @louszbd: We honestly didn't see blender demo would get so much attention. Actually it is for a test of overall model capabilites especi…