AI High Signal

1 curation · 626 membres · fenêtre de 48 h, demi-vie 6 h, seuil 2 relayeurs

Fenêtre 6 h 12 h 24 h 2 j 3 j 7 j

Montrer Les deux Liens 7 Tweets 12

Seuil 2 relayeurs 3 relayeurs 4 relayeurs 5 relayeurs 6 relayeurs Tout remettre

7liens

12tweets

35235posts lus

528auteurs

3176liens en base

1452 hde recul

Les liens 7

Classés par auteurs distincts, jamais par audience.

How ShipStation Global built an AI IT help desk on Amazon Bedrock AgentCore | Amazon Web Services

ShipStation Global's engineering team built Deckhand, an AI IT help desk agent for Slack on Amazon Bedrock AgentCore. It answers routine IT questions from a curated knowledge base, reads error-message screenshots, and files Jira tickets. In its first three months it resolved 97.5% of channel queries with a 96% positive feedback rating.

2 relayeurs à 2 · il y a 6 h Amazon Web Services

Qui l'a relayé

Omar Khattab @lateinteraction · 08/11 19:28 prof @MIT CSAIL | ColBERT, https://t.co/ZZaSzaRaZ7 (@DSPyOSS), GEPA, RLMs, Pedagogical RL

Nice section in this AWS post on ShipStation Global’s AI IT help desk. A dspy.RLM module holds their entire Slack history as a variable, and mines it into a knowledge base of resolved issues.

Stanford NLP Group @stanfordnlp · 08/11 21:43 Natural Language Processing/Machine Learning @chrmanning @jurafsky @percyliang @ChrisGPotts @tatsu_hashimoto @MonicaSLam @Diyi_Yang @YejinChoinka @StanfordAILab

Nice section in this AWS post on ShipStation Global’s AI IT help desk. A dspy.RLM module holds their entire Slack history as a variable, and mines it into a knowledge base of resolved issues.

https://reflection.ai/documents/beam-model-tech-report

2 relayeurs à 2 · il y a 2 h reflection.ai

Qui l'a relayé

Yao Fu @Francis_YAO_ · 01:29 Prev. Grok 4.5/4.6 pretraining scaling @xAI; Gemini 3 perception and project Astra @GoogleDeepMind

Beam’s model technical report is now available. Read the full report to learn about the methodology our team used to train Beam end-to-end in-house, here:

M Saiful Bari (MARUF) @sbmaruf · 07/11 19:19 MIT TR35 || RSI@aws, Nova, ALLaM, xCodeEval, UXLA, BLOOMZ, T0 || Scaling Maximalist

Beam’s model technical report is now available. Read the full report to learn about the methodology our team used to train Beam end-to-end in-house, here:

Utilisation, coût et rang de Inkling 2 Small | OpenCode Data

Consultez les données d'utilisation OpenCode de Inkling 2 Small, avec le volume de tokens, le rang hebdomadaire, la répartition des tokens, les coûts, le taux de cache, les sessions, les répartitions géographiques et les modèles proches.

2 relayeurs à 2 · il y a 7 h OpenCode reprise probable

Qui l'a relayé

Soumith Chintala @soumithchintala · 07/11 16:16 Building new things @thinkymachines. Also dabble in robotics at NYU. Cofounded @PyTorch. AI is delicious when it is accessible and open-source.

Inkling 2 Small on OpenCode for free for a limited time, to help us improve our models further:

Victoria X Lin @VictoriaLinML · 08/11 20:57 MTS @thinkymachines | Native Multimodal Intelligence Prev: @AIatMeta @SFResearch • PhD @uwcse

Inkling 2 Small on OpenCode for free for a limited time, to help us improve our models further:

Harness adapters: Safe RL for efficient agentic development

A Blog post by harness-hub on Hugging Face

2 relayeurs à 2 · il y a 7 h huggingface.co

Qui l'a relayé

younes @yb2698 · 07/11 08:57 :hugging_face:

This week we were blessed by: 1. @Nouamanetazi dropped MoE training handbook: 2. @yb2698 started training harness adapters: 3. @joelniklaus released the harness leaderboard: Have a lovely weekend y'all. 🤗

Leandro von Werra @lvwerra · 08/11 20:53 Head of research @huggingface

Harness Leaderboard - a Hugging Face Space by joelniklaus

Which coding-agent harness should you use for your model?

2 relayeurs à 2 · il y a 7 h huggingface.co reprise probable

Qui l'a relayé

younes @yb2698 · 07/11 08:57 :hugging_face:

Leandro von Werra @lvwerra · 08/11 20:53 Head of research @huggingface

The Ultimate MoE Training Handbook - a Hugging Face Space by huggingface

A hands-on guide to training Mixture-of-Experts models

2 relayeurs à 4 · il y a 3 j huggingface.co reprise probable

Qui l'a relayé

younes @yb2698 · 07/11 08:57 :hugging_face:

Leandro von Werra @lvwerra · 08/11 20:53 Head of research @huggingface

The End of AI Frontiers

3 relayeurs à 4 · il y a 24 h hunch.net

Qui l'a relayé

Zachary Nado @zacharynado · 08/11 02:27 Gemini pretraining research eng @GoogleDeepMind. Personal account!!! Past: swe intern @SpaceX, ugrad researcher in @tserre lab. All opinions my own.

The End of AI Frontiers . There's a bit of a train wreck here that's painful to express. Nevertheless, it's a significant event for me and quite a few other folks involved so I wanted to memorialize a bit.

Eric Hartford @QuixiAI · 08/11 02:18 We make AI models Dolphin and Samantha BTC 3ENBV6zdwyqieAXzZP2i3EjeZtVwEmAuo4 https://t.co/3ri2GbXrQB https://t.co/zH0F3pTjjY @dphnAI

Terry Yue Zhuo @terryyuezhuo · 08/11 03:26 futurist

Les tweets 12

Axe indépendant : repris sans regarder les liens.

Justin Lo @loresjpg · 01:37

Nice DeepSWE score! Does it survive a lowercase filename? Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement. Models clustered around 74% on DeepSWE end up 23 pp apart. Good

11 relayeurs à 11 · il y a 2 h Voir sur X

Qui l'a repris

J Julian Schrittwieser”

Hamel Husain”

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)”

Ofir Press”

Lisan al Gaib

Mat Velloso”

Ce qu'ils en ont dit

Julian Schrittwieser @Mononofu a cité · 01:37

Making novel variations of evals is very useful, public evals tend to get memorized!

Hamel Husain @HamelHusain a cité · 01:16

I see some are drawing the wrong inference from this study: they are focusing on this meta benchmark as yet another benchmark for model picking, i.e. open vs closed models, model families etc. That’s too narrow of an interpretation! You need to do your own private evals on

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 08/11 22:00

holy shit, an art form of scientific grift next METR/Epoch/AA employee right here you all need to work on your epistemic hygiene

Ofir Press @OfirPress a cité · 08/11 21:26

GPT-5.6 Sol and Luna were built by roughly the same people using roughly the same infra, I don't think we can blame the 20 point difference in Luna just on memorization. There's something else being tracked here in addition to memorization. Maybe MaxxBench questions are slightly

Mat Velloso @matvelloso a cité · 07/11 21:20

I'm both surprised and also not at all

Hey, It's Erin! @TheTruestErin · 02:41

Watch Jesse’s smile fade micrometer by micrometer as he realizes that yes actually, that is indeed the plan. Watch him slowly realize that his party will never receive votes again. You can literally see his mouth straighten out if you watch the video

2 relayeurs à 2 · il y a 53 min Voir sur X

Qui l'a repris

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)”

Ben (no treats)

Ce qu'ils en ont dit

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 02:41

I didn’t realize that libs genuinely want One Billion Americans

will brown @willcb · 03:00

2027 is gonna be about engines

2 relayeurs à 2 · il y a 35 min Voir sur X

Qui l'a repris

swyx 🔜 aie.codes”

Joanne Jang”

Ce qu'ils en ont dit

swyx 🔜 aie.codes @swyx a cité · 03:00

are you listening anon?

Joanne Jang @joannejang a cité · 08/11 18:18

vroomberg terminals

Pushmeet Kohli @pushmeet · 08/11 20:37

Across our projects @GoogleDeepMind, we are witnessing a genuine phase transition in the pace of scientific discovery. You will see a glimpse of this in our releases in the next few weeks.

4 relayeurs à 4 · il y a 7 h Voir sur X

Qui l'a repris

Lucas Beyer (bl16)”

elvis”

Chubby♨️”

Demis Hassabis

Ce qu'ils en ont dit

Lucas Beyer (bl16) @giffmana a cité · 08/11 20:37

Pre-announcing the pre-announcement

elvis @omarsar0 a cité · 08/11 17:08

Been working for this moment for over a decade. What I've witnessed behind the scenes, together with what we have been building, tells me we are about to enter a truly incredible era of AI-powered scientific discovery. It's an incredible time to be doing research. I will be

Chubby♨️ @kimmonismus a cité · 08/11 15:36

2026 is the turning point for humanity. Google DeepMind’s Pushmeet Kohli says AI is accelerating scientific discovery and that upcoming releases will offer a glimpse of the progress. He says the team also aims to help the wider scientific community use these tools responsibly.

The New York Times @nytimes · 08/11 22:18

Breaking News: Nvidia opened an investigation into Jensen Huang, its CEO, after he acknowledged to the company’s board that he had been intimate with a business executive who U.S. intelligence determined to be a Chinese spy.

nytimes.com Nvidia Opens Investigation Into Jensen Huang’s Relationship With Chinese Spy Mr. Huang, the company’s chief executive, acknowledged on Sunday that he was intimate with a business executive who U.S. officials had determined was working for Chinese intelligence.

2 relayeurs à 2 · il y a 5 h Voir sur X

Qui l'a repris

john allard”

Ethan Perez

Ce qu'ils en ont dit

john allard @john__allard a cité · 08/11 22:18

having an affair with a suspected chinese spy is bad obvi, but it doesn’t materially update my opinion of jensen. i already thought he was a pos when he started groveling at trump’s feet for permission to sell h200s to china

Richard Sutton @RichardSSutton · 08/11 14:26

Brute force computation is not enough for intelligence. Clever new methods are needed. The science of AI is far from finished. IMHO

incompleteideas.net https://incompleteideas.net/IncIdeas/BitterLesson.html

13 relayeurs à 13 · il y a 13 h Voir sur X

Qui l'a repris

chansung

elie”

François Fleuret” M M Saiful Bari (MARUF)”

Denny Zhou”

Satya Mallick”

Ce qu'ils en ont dit

elie @eliebakouch a cité · 08/11 09:58

i disagree with the first point, there is nothing that says that scaling up current transformer recipe is not enough for "intelligence". you take kimi K3 or your favorite open arch + some stability fixes, you scale it to 50T parameters, you do classical training end to end, you

François Fleuret @francoisfleuret a cité · 08/11 07:44

We need a model that learns *in spite* of its inductive bias, not thanks to it. We have to stop being cross-entropy fetishists.

M Saiful Bari (MARUF) @sbmaruf a cité · 08/11 07:02

True! This is the era of the RL objective. The biggest economy on this earth (US stock market) runs on subscriptions. Subscriptions require vendor lock-in. Currently, the frontier is pushing “Variance” learning from customer data, like how Google won the search market. But I

Denny Zhou @denny_zhou a cité · 08/11 06:26

The Bitter Lesson 2.0

Satya Mallick @LearnOpenCV a cité · 08/11 05:00

That’s what I call a bitter sense of humor.

DSPy @DSPyOSS · 08/11 21:43

Nice section in this AWS post on ShipStation Global’s AI IT help desk. A dspy.RLM module holds their entire Slack history as a variable, and mines it into a knowledge base of resolved issues.

Amazon Web Services How ShipStation Global built an AI IT help desk on Amazon Bedrock AgentCore | Amazon Web Services ShipStation Global's engineering team built Deckhand, an AI IT help desk agent for Slack on Amazon Bedrock AgentCore. It answers routine IT questions from a curated knowledge base, reads error-message screenshots, and files Jira tickets. In its first three months it resolved 97.5% of channel queries with a 96% positive feedback rating.

2 relayeurs à 2 · il y a 6 h Voir sur X

Qui l'a repris

Stanford NLP Group

Omar Khattab

alphaXiv @askalphaxiv · 08/11 21:24

We mapped funding raised vs founder citations across 32 neo-labs Their founding teams have amassed 8.35M citations and have raised nearly $26B Discovery Loop leads with 1.59M founder citations, followed by SSI (915K) and AMI Labs (752K) The two are correlated, but a huge

3 relayeurs à 3 · il y a 6 h Voir sur X

Qui l'a repris

rohan anil”

Sara Hooker

Julian Togelius”

Ce qu'ils en ont dit

rohan anil @_arohan_ a cité · 08/11 21:24

X axis can be number of uncs. Never beating allegations.

Julian Togelius @togelius a cité · 08/11 04:22

How can I not retweet this …hey, investors, look here

Reflection @reflection_ai · 01:29

Beam’s model technical report is now available. Read the full report to learn about the methodology our team used to train Beam end-to-end in-house, here:

reflection.ai https://reflection.ai/documents/beam-model-tech-report

2 relayeurs à 2 · il y a 2 h Voir sur X

Qui l'a repris

Yao Fu M M Saiful Bari (MARUF)

Soumith Chintala @soumithchintala · 08/11 20:57

Inkling 2 Small on OpenCode for free for a limited time, to help us improve our models further:

OpenCode Utilisation, coût et rang de Inkling 2 Small | OpenCode Data Consultez les données d'utilisation OpenCode de Inkling 2 Small, avec le volume de tokens, le rang hebdomadaire, la répartition des tokens, les coûts, le taux de cache, les sessions, les répartitions géographiques et les modèles proches.

2 relayeurs à 2 · il y a 7 h Voir sur X

Qui l'a repris

Victoria X Lin