@jyhsia5174 Profile picture

夏睿陽

@jyhsia5174

Joined December 2016
Similar User
Chen Zhu photo

@chenzhucs

ABee photo

@abuoma_agajelu

Siddharth Bhatia photo

@siddharthb_

Jing Conan Wang photo

@jingconanwang

LP Gear photo

@LP_Gear

Oluaye photo

@oluayemayana

Sean Gerrish photo

@seannyg

I am photo

@super_bizi

KING 👑 BOITEL photo

@KingBoitel

Michal Ptaszynski | プタシンスキ ミハウ photo

@mich_ptaszynski

Olawale Arowolo photo

@whalesarowolo

夏睿陽 Reposted

New research from FAIR: Better & Faster Large Language Models via Multi-token Prediction Research paper ➡️ go.fb.me/wty7gj We show that replacing next token prediction tasks with multiple token prediction can result in substantially better code generation performance…

Tweet Image 1

夏睿陽 Reposted

The simulation is sending us a message 👀

Tweet Image 1

夏睿陽 Reposted

This is a second highlight clip from my upcoming Supervised FSD v12.4.1 Unprotected Left Turn video. Enjoy. @Tesla_AI


夏睿陽 Reposted

Microsoft presents The Era of 1-bit LLMs All Large Language Models are in 1.58 Bits Recent research, such as BitNet, is paving the way for a new era of 1-bit Large Language Models (LLMs). In this work, we introduce a 1-bit LLM variant, namely BitNet b1.58, in which every single…

Tweet Image 1

夏睿陽 Reposted

Nice new article: medium.com/towards-data-s… GPT-2, from scratch, in MLX Companion code: github.com/pranavjad/mlx-…

Tweet Image 1

夏睿陽 Reposted

Hanging out with the big kids: PyTorch 8 years old TF 9 years old Jax 6 years old MLX 6 months old

Tweet Image 1

夏睿陽 Reposted

Google announces Leave No Context Behind Efficient Infinite Context Transformers with Infini-attention This work introduces an efficient method to scale Transformer-based Large Language Models (LLMs) to infinitely long inputs with bounded memory and computation. A key

Tweet Image 1

夏睿陽 Reposted

Introducing Dream Machine - a next generation video model for creating high quality, realistic shots from text instructions and images using AI. It’s available to everyone today! Try for free here lumalabs.ai/dream-machine #LumaDreamMachine


夏睿陽 Reposted

🚀Introducing Mirage, a superoptimizer that automatically discovers highly-optimized GPU implementations for LLMs (and beyond). For certain attention operators, the fastest programs found by Mirage is 2x faster than existing expert-designed implementations such as FlashAttention…


夏睿陽 Reposted

Here’s the problem with “agreeing” to share your data: nobody actually reads the terms & conditions cc.com/video/e48ddg/s…


夏睿陽 Reposted

Here it comes. Supervised FSD v12.4.1 2024.15.5 is downloading to my car as I type this. Unfortunately I won't be home until after the shareholder's meeting in Austin to drive it.

Tweet Image 1

夏睿陽 Reposted

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt…


夏睿陽 Reposted

Yeah, seeing it everywhere. Btw, 12.4 goes to internal release this weekend and limited external beta next week. Roughly 5X to 10X improvement in miles per intervention vs 12.3. 12.5 will be out in late June. Will also see a major improvement in mpi and is single stack – no…


夏睿陽 Reposted

Sure. Just like the availability of calculators and computers has stopped people from learning math, science, and engineering. Oh wait.... ssti.org/blog/useful-st…

Tweet Image 1

夏睿陽 Reposted

BREAKING — Elon confirms FSD 12.4 will remove steering wheel nags as soon as next week!!!! let’s fucking goooooooooooooooooooo!!!!

Yes



夏睿陽 Reposted

profound

FSD V12 "imagines" turn signals from other vehicles' behavior



夏睿陽 Reposted

Tesla Full Self-Driving (Supervised) 12.3.6 is incredibly smooth and humanlike Here's my first drive, with zero interventions Watch in 4K: youtu.be/UVILrFldKVc


夏睿陽 Reposted

# on shortification of "learning" There are a lot of videos on YouTube/TikTok etc. that give the appearance of education, but if you look closely they are really just entertainment. This is very convenient for everyone involved : the people watching enjoy thinking they are…


夏睿陽 Reposted

Today we're releasing three new papers on large language models. This work offers a foundation for our future language research, especially in areas that will have a bearing on how models are evaluated and deployed: dpmd.ai/language-model… 1/

Tweet Image 1

Loading...

Something went wrong.


Something went wrong.