Jérémy Cochoy @jeremycochoy

CEO at Redstone Solutions, Dr in Mathematics, Applied machine learning. On Twitter to post silly jokes. cochoy.fr Funchal, Portugal Joined August 2013

Tweets

927
Followers

71
Following

102
Likes

5K

Anthropic @AnthropicAI

2 weeks ago

We’ve developed Claude for Chrome, where Claude works directly in your browser and takes actions on your behalf. We’re releasing it at first as a research preview to 1,000 users, so we can gather real-world insights on how it’s used.

454 976 9K 1.4M 2K

Download Video

Yawar Siddiqui @yawarnihal

4 weeks ago

Check out our #ICCV2025 paper VertexRegen! Instead of the typical incomplete meshes you get with autoreg mesh gen, VertexRegen generates progressively more detailed meshes as the generated sequence gets longer. vertexregen.github.io Great work by xzhang.dev

3 37 339 19K 158

Download Video

Anthropic @AnthropicAI

a month ago

New Anthropic research: Persona vectors. Language models sometimes go haywire and slip into weird and unsettling personas. Why? In a new paper, we find “persona vectors"—neural activity patterns controlling traits like evil, sycophancy, or hallucination.

232 939 6K 1.4M 4K

Download Image

Jérémy Cochoy @jeremycochoy

4 months ago

arxiv.org/abs/2503.24322

0 0 0 10 0

Lex Fridman @lexfridman

11 months ago

I'm doing a podcast with Dario Amodei, CEO of Anthropic (creator of Claude) soon, all about AI. Let me know if you have questions/topic suggestions. Also, I'll stop by SF for a bit. Let me know if you have suggestions of who I should talk to.

870 275 7K 570K 496

Andrea Tagliasacchi 🇨🇦 @taiyasaki

a year ago

📢📢📢𝐕𝐨𝐥𝐮𝐦𝐞𝐭𝐫𝐢𝐜 𝐑𝐞𝐧𝐝𝐞𝐫𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐁𝐚𝐤𝐞𝐝 𝐐𝐮𝐚𝐝𝐫𝐚𝐭𝐮𝐫𝐞 𝐅𝐢𝐞𝐥𝐝𝐬 at #ECCV2024 quadraturefields.github.io Project led by Gopal Sharma (@beakywings) TL;DR: extends MobileNerf to represent 𝐭𝐫𝐚𝐧𝐬𝐩𝐚𝐫𝐞𝐧𝐭 scenes... with meshes?🤯...🧵

2 23 180 18K 89

Download Video

Akshay 🚀 @akshay_pachaar

11 months ago

Microsoft just changed the game! 🔥 They've open-sourced bitnet.cpp: a blazing-fast 1-bit LLM inference framework that runs directly on CPUs. Why is this a game-changer❓ You can now run 100B parameter models on local devices with up to 6x speed improvements and 82% less…

165 1K 7K 787K 8K

Download Video

Phil @phill__1

11 months ago

Wow nvidia just published a 72B model with is ~on par with llama 3.1 405B in math and coding evals and also has vision 🤯

33 136 966 77K 410

Download Image

BlinkDL @BlinkDL_AI

12 months ago

RWKV-7 "Goose" preview, with dynamic state evolution (using structured matrix) 🪿 Now the loss curve looks scalable, after fixing a hidden bug😀

5 22 165 26K 32

Download Image

RWKV @RWKV_AI

a year ago

The conclusive EagleX is here Based on the RWKV-v5 architecture, bringing into opensource 7B space, the best SOTA - Multi-lingual model - English perplexity model - Attention-free transformer today (10-100x+ lower inference) With comparable English performance to Mistral

3 31 151 14K 50

Download Image

Jérémy Cochoy @jeremycochoy

a year ago

SalesForce released "Unified Training of Universal Time Series Forecasting Transformers" & LOTSA, a 1B point dataset, plus Model weights & eval code for reproducibility. Their decoder transformer forecasts with a mixture distribution. A must read: arxiv.org/pdf/2402.02592…

0 0 0 70 0

Paul Graham @paulg

2 years ago

Even if AI were going to make human programmers obsolete, the best way to figure out how to adapt to its rise would be to learn to code, because software is what it's made of.

181 417 4K 326K 537

BlinkDL @BlinkDL_AI

2 years ago

RWKV6🐦 vs Mamba🐍. RWKV6 will be the strongest multilingual model (data = only 1.1T tokens), which can occupy some capacities for English, but worth it🙂 [email protected](Pile) is great at 1.4B, less so at 2.8B. Will be interesting to see the results when provided more training data.

3 19 137 37K 31

Download Image

Bob West @cervisiarius

2 years ago

In our new preprint, we ask: Do multilingual LLMs trained mostly on English use English as an “internal language”? - A key question for understanding how LLMs function. “Do Llamas Work in English? On the Latent Language of Multilingual Transformers” arxiv.org/abs/2402.10588

12 116 644 115K 407

Download Image

Shital Shah @sytelus

2 years ago

@yentinglin56 This video by @srush_nlp is pretty awesome! youtu.be/dKJEpOtVgXc?si…

1 17 198 15K 291

Jérémy Cochoy @jeremycochoy

2 years ago

This recent paper from ETH Zurich address the question of simplifying attention blocks. Under the right hypothesis, value, projection matrix and residual connexion can be removed. You can also get a little extra boost by centering the attention matrix. arxiv.org/pdf/2311.01906…

0 0 0 30 0

Михайло Подоляк @Podolyak_M

3 years ago

Russia has already lost — mobilization only delays its defeat. After Ukraine liberates its territories and RF’s defeat will be official, everything possible must be done to prevent recurrence of ru-aggression. It's simple: tribunal, sentences to war authors and 🇺🇦 as NATO member.

344 933 7K 0 13

Elon Musk @elonmusk

3 years ago

@sama What’s the average cost per chat?

2K 1K 47K 0 288

Bari Weiss @bariweiss

3 years ago

1. A new #TwitterFiles investigation reveals that teams of Twitter employees build blacklists, prevent disfavored tweets from trending, and actively limit the visibility of entire accounts or even trending topics—all in secret, without informing users.