J.L

archer, pefter, Incoming PhD @cuhk

[J.L @Jo1uck]

Could this be a direction for future model architectures?

Introducing Memory Attention (MA).

https://t.co/5bDpP0uQPs

[J.L @Jo1uck]

@liiiaff 越早越好，目前frontier labs的hc都缩的严重

[J.L @Jo1uck]

最有正反馈的一集，phd要么一次做个大，要么快速收割低垂果实。顶会贬值的速度应该会越来越快，速速实习吧

[Xiuyu Li @sheriyuo]

造 bench 对于学术界算负反馈，因为很多会 bench 很好中，会先入为主地认为你有水 paper 之嫌疑（但是只是有 bias）

对于工业界来说又算正反馈，能给你带来相当大的曝光，并且给 TB4 / TB-Science / ALE 交过 PR、供过题是当下基模算法岗位的硬通货敲门砖

被迫跟随 frontier bench 不如自己创造 frontier bench

[J.L @Jo1uck]

@Haoquan03 😂

[J.L @Jo1uck]

国模还在预告Fable5.2已经来了，没发的已经可以开启新一轮蒸馏了，无限延期。

[Claude @claudeai]

Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.

They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.

[Naved @Navedux]

my take on this. mostly webgl and custom code, with a little https://t.co/dYm6F7C6V1 shader style mixed in

[J.L @Jo1uck]

@sun_hanchi 轻奢😏

[Chetaslua @chetaslua]

Opus 5.5 made these ui to toggle light/dark mode

[J.L @Jo1uck]

@MikaStars39 羡慕了？

[J.L @Jo1uck]

@sheriyuo @StepFun_ai @liulicheng10 @MikaStars39 甜🤭

[J.L @Jo1uck]

@thsottiaux @zoedotai 🤣

[Wenhao Chai @wenhaocha1]

RSI starts with infra 🔁 mlsys is in a golden age. agents are great researchers now. with tokens, "low-hanging fruit" is everywhere.

agents optimize speculative decoding for parallel drafters (DFlash2, DSpark) without training or quality loss. their 12-bit format beats cuBLAS.

[Chetaslua @chetaslua]

Fable 5.5 its over for video and image models 🤯

THIS IS PURE CODE

one shotted this video holy shit , how this is more beautiful than ai video models

Two ways to check this yourself:

Ask: “Where did @karpathy go in 2026?” with web search and memory turned off. Only the new Fable knows the answer. Then give it the prompt.
Run the same prompt I used for the test (it’s in the screenshot) and see for yourself.

Don’t rely on Grok for basic reasoning like this. Grok doesn’t have up-to-date knowledge of niche stuff. Either join the community or run the test yourself.

Also, a knowledge cutoff doesn’t mean a model knows everything before that date. Sometimes a model doesn’t know about an event even if it happened before its cutoff. This test shows that too.

Run the same prompt on regular Opus 5.5 and the quality drops. On Fable 5.1 it’s worse.

[J.L @Jo1uck]

🎃

[Tiezhen WANG @Xianbao_QIAN]

Got a chance to visit the @deepseek_ai office. The team is cooking!

[J.L @Jo1uck]

实在太好笑了🤣

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[熊师傅 weight decay 了吗 @bigeagle_xd]

能预见到一系列 Weight-Space Programming (WSP) 的灌水出来了
SelfWSP: 把golden answer 放hint里，让模型改自己的 weight
WSP-OPD: 大模型蒸小模型，但是通过WSP的方式
WSPO: GRPO+WSP
asyncWSP: 异步更新weight
（大家继续

[J.L @Jo1uck]

@leon2mcp 这个真要去😋

[J.L @Jo1uck]

so cute🥰

[Wenhao Chai @wenhaocha1]

Gemini 4 Carbon scores 53 FCI and 54 pass rate on FrontierCS 2. It’s an impressive jump from Argon to Carbon in just a few weeks. We also work with @antigravity to improve the evaluation in long-horizon runs. model and harness are co-evolving! 🐱

[J.L @Jo1uck]

Memory Attention has also made it into the Attention Zoo! 🦁

Huge thanks to @rshia_afz for including our work!
https://t.co/0NWkMThZh3

[J.L @Jo1uck]

Memory Attention is now integrated into Flash Linear Attention (FLA)!

Huge thanks to @uniartisan and @yzhang_cs !

⚡ FLA: https://t.co/pVYTVKkMZy

📄 Paper: https://t.co/AMzl2P7RIg