TGViewer
Channel Public Channel
Hugging Face

Hugging Face

@huggingface

Open in Telegram
Subscribers
302
Photos
1.3K
Videos
408
Links
2.2K

Showing posts older than #2116 · Back to latest

Older Posts 18 shown
Post #2114 18
Hugging Face (Twitter)

RT @ctmorley: Hundreds of questions were asked by my 4-year-old son as we assembled the Reachy Mini robot (2 hours), connected it with Claude Code (5 minutes), integrated real-time VLMs/web search APIs (10 minutes) and brought embodied Al to life at our kitchen table.

May the curiosity and creativity of this Generation Alpha, the first Al-native generation, be a wellspring of daily inspiration to us all.

Epic foundational work by @huggingface @pollenrobotics @claudeai
Post #2105 14
Hugging Face (Twitter)

RT @eliebakouch: Here is more detail on how this is working. There are 5 teams for this first round and only 4 teams will go to the second round. There are 3 constraints:
> from-scratch training
> somewhat open (can be used commercially)
> ambitious (omni model, large scale, etc.)

breakdown of the different costs:

> Compute:
~$110M in compute leasing (sk telecom and naver are providing the compute)

> Data:
~$7M for data accessible to each team, ~$14M for a video dataset (only naver ended up training an omni model afaik). Each team gets an additional ~$2M for their own dataset curation

total cost of this first round is ~140M USD https://twitter.com/eliebakouch/status/2006364076977336552#m
Post #2104 12
Hugging Face (Twitter)

RT @ClementDelangue: As spotted by @eliebakouch & @natolambert, Korea's AI is on fire these days thanks to the public support for open-source AI with the "Sovereign AI Foundation Model” project!

Just in the past few days, three beautiful open models were announced:

- K-Exagone from @LGUS, a 236B fine-grained MoE design (23B active) that shines in Korean and multilingual, optimized with Multi-Token Prediction (MTP), enabling self-speculative decoding that boosts inference throughput by approximately 1.5x.

- Solar Open by @upstageai: 102B / 12B Mixture-of-Experts architecture models with 102B total / 12B active parameters trained entirely from scratch: https://huggingface.co/upstage/Solar-Open-100B

- A.X K1 by @SKtelecom: a 519B total param, planned to release on 4th of January from SKT: huggingface.co/skt/A.X-K1

In 2026, all countries in the world should make it the number 1 priority to push for open-source AI as the most promising factor for...

Перейти на оригинальный пост
Post #2103 12
Hugging Face (Twitter)

RT @eliebakouch: If you are wondering what is happening: last summer, the Ministry of Science of Korea created a program with 5 companies to train sovereign AI models and release them under a permissive license, so other South Korean companies can use them and hence expand their domestic AI ecosystem

to recap we got:
- SK Telecom: A(.)X-K1 519B total, 33B active
- LG: K-EXAONE 236B total, 23B active
- NC-AI: VAETKI 112B total, 10B active
- Upstage: Solar-Open 102B total, 12B active
- Naver: HyperCLOVAX-SEED-Think 32B Dense

this is REALLY a big deal https://twitter.com/eliebakouch/status/2006359083776201059#m
Post #2100 14
Hugging Face (Twitter)

RT @eliebakouch: wowwww another large open model from Korea about to be released! this time from sk telecom with 519B total param, release planned on 4th of january 2026
Older posts →
Threads Profile ViewerView any public Threads profile without an account.Open ThreadLook →Writing with AI? Make it sound human.Metric37 rewrites AI drafts so they read naturally. Free AI detector, 1,500 words free.Try Metric37 →