Your curated collection of saved posts and media

Showing 32 posts ยท last 14 days ยท by score
E
eddybuild
@eddybuild
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”59891306

wrist cam changes everything https://t.co/I2G3l3LW0w

@physical_int โ€ข Wed Dec 17 01:04

We discovered an emergent property of VLAs like ฯ€0/ฯ€0.5/ฯ€0.6: as we scale up pre-training, the model learns to align human videos and robot data! This gives us a simple way to leverage human videos. Once ฯ€0.5 knows how to control robots, it can naturally learn from human video.

๐Ÿ–ผ๏ธ Media
๐Ÿ”Scobleizer retweeted
E
Eddy Xu
@eddybuild
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”59891306

wrist cam changes everything https://t.co/I2G3l3LW0w

โค๏ธ82
likes
๐Ÿ”6
retweets
๐Ÿ–ผ๏ธ Media
S
StanfordHAI
@StanfordHAI
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”60723048

๐Ÿ“ข New issue brief: Have Chinese AI models pulled ahead of their global counterparts? Our latest brief analyzes Chinaโ€™s diverse open-weight model ecosystem and examines the policy implications of their widespread global diffusion. https://t.co/zkdBKf7N5F https://t.co/MNugnSzaFb

Media 1Media 2
๐Ÿ–ผ๏ธ Media
M
MiniMax__AI
@MiniMax__AI
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”06171197

MiniMax (Hailuo) Video Team Has Open Sourced VTP (Visual Tokenizer Pre-training)! VTP is a scalable pre-training framework for visual tokenizers, built for next-gen generative models. It challenges the conventional belief in Latent Diffusion Models that scaling the stage-1 visual tokenizer (model size, compute, data) doesnโ€™t improve stage-2 generation. By combining representation learning + compressionโ€“reconstruction, we show the first scaling curve connecting visual tokenizers with Diffusion Transformers! ๐ŸงAnd the best part: no extra compute for the generator, but better generations simply by scaling up tokenizer training for better representation space! ๐ŸŽ„๐ŸŽ…Merry pre-Christmas to the community!

Media 1
๐Ÿ–ผ๏ธ Media
R
rodeoman
@rodeoman
๐Ÿ“…
Dec 16, 2025
105d ago
๐Ÿ†”26771464

itโ€™s nice that there are still people out there who remember what the internet is really for https://t.co/q4wcfy63yK

Media 1
๐Ÿ–ผ๏ธ Media
๐Ÿ”youwouldntpost retweeted
R
campbell g
@rodeoman
๐Ÿ“…
Dec 16, 2025
105d ago
๐Ÿ†”26771464

itโ€™s nice that there are still people out there who remember what the internet is really for https://t.co/q4wcfy63yK

Media 1
โค๏ธ7,277
likes
๐Ÿ”570
retweets
๐Ÿ–ผ๏ธ Media
Y
youwouldntpost
@youwouldntpost
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”47195590

https://t.co/nJEVS0p9wP

Media 1Media 2
๐Ÿ–ผ๏ธ Media
Y
youwouldntpost
@youwouldntpost
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”77180860

judge moved us one step closer to California suspending Tesla sales on a day Musk spent generating AI titty slop https://t.co/73f8g0Qf3W

Media 1Media 2
+1 more
๐Ÿ–ผ๏ธ Media
O
ohabryka
@ohabryka
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”64311514

Lightcone Infrastructure's 2026 fundraiser is live! We build beautiful things for truth-seeking and world-saving. We run LessWrong, Lighthaven, Inkhaven, designed AI-2027, and so many more things. All for the price of less than one OpenAI staff engineer ($2M/yr). More in ๐Ÿงต. https://t.co/Dqp9HxvoaM

Media 1
๐Ÿ–ผ๏ธ Media
B
briannekimmel
@briannekimmel
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”31070975

@grinich @WorkOS ๐Ÿคฏ https://t.co/0SoDQefjcZ

Media 1
๐Ÿ–ผ๏ธ Media
E
emollick
@emollick
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”57900914

Official rule for all AI labs: no more demoing your product with either telling the AI to โ€œbook a trip for meโ€ or creating AI photos/videos of your companyโ€™s CEO in crazy situations. Sorry, those are the rules now. https://t.co/PSnPM2P3u9

Media 1
๐Ÿ–ผ๏ธ Media
L
LuxAlgo
@LuxAlgo
๐Ÿ“…
Dec 15, 2025
105d ago
๐Ÿ†”45217289

We just added @grok as a model in our LuxAlgo AI and chat history. More agentic strategies, better backtesting, deeper automation, and more are coming soon. We're still building. https://t.co/iYa4PK1IpH

Media 1
๐Ÿ–ผ๏ธ Media
S
Speechmatics
@Speechmatics
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”11791265

NEW: German, Danish and Norwegian Medical Models, trained on 2 billion words of clinical data. ๐Ÿš€ European healthcare needs voice AI that works with non-native speakers, overlapping dialogue, regional accents, and medical terminology spoken under pressure. Generic speech recognition can't handle it. Ours can. ๐Ÿ‘‡

๐Ÿ–ผ๏ธ Media
9
9to5mac
@9to5mac
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”89469051

Apple Music is coming to ChatGPT, OpenAI announces https://t.co/Huxq7X1d5b by @mvcmendes

Media 1
๐Ÿ–ผ๏ธ Media
S
Scobleizer
@Scobleizer
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”70324585

Here are my lists. Unfortunately Apple doesnโ€™t let me share most of them: https://t.co/PMgOlNfd2Z

Media 1
๐Ÿ–ผ๏ธ Media
๐Ÿ”huggingface retweeted
_
AK
@_akhaliq
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”17844296

Meta just released sam-audio https://t.co/ThdN1ZcCgB

Media 1
โค๏ธ216
likes
๐Ÿ”26
retweets
๐Ÿ–ผ๏ธ Media
M
MetaNewsroom
@MetaNewsroom
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”07580155

Introducing SAM Audio: the first unified AI model that allows you to isolate and edit sound from complex audio mixtures. This could mean isolating the guitar in a video of your band, filtering out traffic noises, or removing the sound of a dog barking in your podcast, all with text, visual, and time span prompts.ย  https://t.co/RiuUACGv80

Media 2
๐Ÿ–ผ๏ธ Media
G
gregpr07
@gregpr07
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”41549119

Today, we're releasing BU-30B-A3B-Preview (OSS)๐Ÿง™ For the first time, you can run hundreds of browser tasks reliably on $1 of compute๐Ÿ‘€ > 30B parameters > 3B active > SoTA quality at real-time speed Welcome to the future of web agents. Try it locally or on our cloud todayโ˜๏ธ HF link โ†“

Media 1
๐Ÿ–ผ๏ธ Media
B
browser_use
@browser_use
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”12545166

We open sourced BU-30B-A3B-Preview. The best model for web agents ๐Ÿง™ Here is how it performs compared to other models ๐Ÿ‘€ https://t.co/2oAc8kY5pO

@gregpr07 โ€ข Tue Dec 16 22:04

Today, we're releasing BU-30B-A3B-Preview (OSS)๐Ÿง™ For the first time, you can run hundreds of browser tasks reliably on $1 of compute๐Ÿ‘€ > 30B parameters > 3B active > SoTA quality at real-time speed Welcome to the future of web agents. Try it locally or on our cloud todayโ˜๏ธ HF

Media 1
๐Ÿ–ผ๏ธ Media
E
eliebakouch
@eliebakouch
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”92264140

wow, this looks like a very solid open model by Xiaomi, competing with K2/DSV3.2 on benchmarks with fewer parameters. it's MIT licensed, with a very good tech report and base/thinking versions available it's using the same sliding window attention arch as gpt-oss (sink with SWA size = 128) but with much fewer global attention layers, multi-token prediction for speculative decoding support, and a new post-training distillation method. really seems like a beast at inference with day 0 @sgl_project support! really exciting

@XiaomiMiMo โ€ข Tue Dec 16 14:01

โšก Faster than Fast. Designed for Agentic AI. Introducing Xiaomi MiMo-V2-Flash โ€” our new open-source MoE model: 309B total params, 15B active. Blazing speed meets frontier performance. ๐Ÿ”ฅ Highlights: ๐Ÿ—๏ธ Hybrid Attention: 5:1 interleaved 128-window SWA + Global | 256K context ๐Ÿ“ˆ Pe

Media 1
๐Ÿ–ผ๏ธ Media
๐Ÿ”huggingface retweeted
_
AK
@_akhaliq
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”66598799

Microsoft just released TRELLIS 2 https://t.co/5IEzsmEfa8

โค๏ธ853
likes
๐Ÿ”106
retweets
๐Ÿ–ผ๏ธ Media
T
Tsuki_ManusAI
@Tsuki_ManusAI
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”25134562

ใƒžใ‚ธใงใ“ใ‚Œใƒคใƒใ„โ€ฆ๏ผ ่ชฐใงใ‚‚็ฐกๅ˜ใซใƒขใƒใ‚คใƒซใ‚ชใƒผใƒ€ใƒผใ‚ขใƒ—ใƒช๐Ÿ”ใŒไฝœใ‚Œใกใ‚ƒใ†ๆ™‚ไปฃใŒๆฅใŸใ€‚ Manusใ‚’ไฝฟใˆใฐใ€ใ‚ณใƒผใƒ‡ใ‚ฃใƒณใ‚ฐไธ€ๅˆ‡ใชใ— ใ‚ชใƒผใƒ€ใƒผใฎ่ฟฝๅŠ ใ‚„ใ‚ซใ‚นใ‚ฟใƒžใ‚คใ‚บใ€ๆ–™้‡‘่จˆ็ฎ—ใ‚‚๐Ÿ’ฐ ็Ÿฅ่ญ˜ใ‚ผใƒญใงใ‚‚ใ€็›ดๆ„Ÿ็š„ใชๆ“ไฝœใ ใ‘ใงๆœฌๆ ผ็š„ใชใ‚ขใƒ—ใƒชใŒๅฎŒๆˆใ™ใ‚‹ใชใ‚“ใฆไฟกใ˜ใ‚‰ใ‚Œใ‚‹๏ผŸ https://t.co/m8TtIP3c5Y

๐Ÿ–ผ๏ธ Media
C
ChatgptLunatics
@ChatgptLunatics
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”27364383

https://t.co/yOluGJEzI2

Media 1
๐Ÿ–ผ๏ธ Media
๐Ÿ”youwouldntpost retweeted
C
AI being dumb
@ChatgptLunatics
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”27364383

https://t.co/yOluGJEzI2

Media 1
โค๏ธ67
likes
๐Ÿ”6
retweets
๐Ÿ–ผ๏ธ Media
C
code
@code
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”70744819

In the fast-changing world of AI, you're still in control. The latest @code release introduces new security and governance features, like the Private Marketplace, enterprise policy updates, tool auto-approval management, and more. ๐Ÿงต https://t.co/F8TzfMB72L

Media 1
๐Ÿ–ผ๏ธ Media
C
code
@code
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”06713088

๐Ÿ”’ Control extension access with Private Marketplace. Organizations can now curate which extensions are available to developers. Restrict installations to approved extensions only, ensuring compliance and security across your teams. https://t.co/PhG9mEmoS1 https://t.co/45TbjEur24

Media 2
๐Ÿ–ผ๏ธ Media
C
code
@code
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”07396229

๐Ÿ”— Fine-grained URL approval gives you more control over external requests. When the fetch tool contacts a URL, you now see a two-step approval flow: 1๏ธโƒฃ Approve the domain to prevent sensitive data from reaching untrusted sites 2๏ธโƒฃ Review fetched content before it enters chat or follow-up tool calls

Media 1
๐Ÿ–ผ๏ธ Media
C
code
@code
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”10925441

โš ๏ธ When enterprise policy disables agent mode, the Agents picker now clearly explains why. No more confusion for developers. They see exactly what policy restrictions apply. https://t.co/afENZ4KlJr

Media 1
๐Ÿ–ผ๏ธ Media
C
code
@code
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”55946180

๐Ÿ“– Read the full release notes: https://t.co/wpQdDKcK9L Happy coding! ๐Ÿ’™

Media 1
๐Ÿ–ผ๏ธ Media
T
tejalpatwardhan
@tejalpatwardhan
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”00175683

we're open-sourcing a new frontier science eval in biology, chemistry, and physics. there are 2 tracks: olympiad level and advanced research level. as models become saturated on GPQA, this is a nice unsaturated alternative with clean test-time compute scaling. kudos to @MilesKWang for driving this release!

Media 1
๐Ÿ–ผ๏ธ Media
O
OfirPress
@OfirPress
๐Ÿ“…
Dec 17, 2025
104d ago
๐Ÿ†”34288846

Training on tough benchmarks leads to better results on other benchmarks as well. Lots more to investigate here but this is the exact reason we work on these types of benchmarks. From the @XiaomiMiMo paper released today. https://t.co/nCq5GDJXyF

Media 1
๐Ÿ–ผ๏ธ Media
E
exec_sum
@exec_sum
๐Ÿ“…
Dec 16, 2025
104d ago
๐Ÿ†”49485132

BREAKING: Databricks is raising over $4 billion in Series L funding at a $134 billion valuation. https://t.co/Qmx1iPZap4

Media 1
๐Ÿ–ผ๏ธ Media