Introducing Small 3, our most efficient and versatile model yet! Pre-trained and instructed version, Apache 2.0, 24B, 81% MMLU, 150 tok/s. No synthetic data so great base for anything reasoning – happy building!
OPEN SOURCE
-
Mistral-Small-3-Instruct Model Released for Download
By
–
magnet:?xt=urn:btih:11f2d1ca613ccf5a5c60104db9f3babdfa2e6003&dn=Mistral-Small-3-Instruct&tr=udp%
http://
2Ftracker.opentrackr.org1337announce&tr=http%https://t.co/ua2yzvEYLu1337announce -
Claude to Add Web Search Support
By
–
Claude is working on web search support. Finally 👀
— 🚨 AI News | TestingCatalog (@testingcatalog) 30 janvier 2025
I think that adding a web search feature to 3.5 Sonnet will boost Claude's performance a lot w/o even a need to release any smarter model. https://t.co/u3G5YHMtVzClaude is working on web search support. Finally I think that adding a web search feature to 3.5 Sonnet will boost Claude's performance a lot w/o even a need to release any smarter model.
-
TinySwallow ChatUI Local: Self-Contained AI Web Application
By
–
また、完全なローカル実行を希望される方向けに、モデルの重みを含むself-contained版のウェブアプリも用意しています。
— Sakana AI (@SakanaAILabs) 30 janvier 2025
GitHub: https://t.co/G2mreMkxFS… pic.twitter.com/YzRHKFazYWまた、完全なローカル実行を希望される方向けに、モデルの重みを含むself-contained版のウェブアプリも用意しています。 GitHub: https://
github.com/SakanaAI/TinyS
wallow-ChatUI-Local
… -
TinySwallow-1.5B: Knowledge Distillation for Japanese Language Models
By
–
この度、新手法「TAID」を用いて学習された小規模日本語言語モデル「TinySwallow-1.5B」を公開しました。https://t.co/U7qpbz2BgL
— Sakana AI (@SakanaAILabs) 30 janvier 2025
私たちは、大規模言語モデル（LLM）の知識を効率的に小規模モデルへ転移させる新しい知識蒸留手法「TAID (Temporally Adaptive Interpolated… pic.twitter.com/OUCy71ho42この度、新手法「TAID」を用いて学習された小規模日本語言語モデル「TinySwallow-1.5B」を公開しました。 https://
sakana.ai/taid-jp 私たちは、大規模言語モデル（LLM）の知識を効率的に小規模モデルへ転移させる新しい知識蒸留手法「TAID (Temporally Adaptive Interpolated -
OpenAI Realtime API Solar System Demo Now Open Source
By
–
One of our most popular @OpenAI DevDay demos is now open source!
— Romain Huet (@romainhuet) 29 janvier 2025
Dive into @kagigz’s interactive, voice-navigable solar system, and learn how you can build your own speech-to-speech experiences with the Realtime API.https://t.co/5tHJDeVloY https://t.co/jUlkGzqiXbOne of our most popular @OpenAI DevDay demos is now open source! Dive into @kagigz
’s interactive, voice-navigable solar system, and learn how you can build your own speech-to-speech experiences with the Realtime API. https://
github.com/openai/openai-
realtime-solar-system
… -
Cost comparison: AI vs human brain
By
–
—
Just so we're clear: Buy dual RTX 4090s and 128 GB RAM = $5k Download DeepSeek R1 70b weights, run offline – unlimited use You can buy a 130 IQ human brain for the cost of a used Civic.
— -
Stanford OVAL Launches Open-Source AI Assistant for Knowledge Access
By
–
The @StanfordOVAL is on a mission to make knowledge accessible to all. Its new open-source public AI assistant aims to empower researchers, journalists, and historians. See how it works IRL on Feb. 14: https://
bit.ly/3WFb5K3 -
Crowdsourced Distributed Fine-Tuning: A Year-Long Advocacy
By
–
I've been arguing for something like this for over a year: crowdsourced distributed fine-tuning.
-
Groq’s Llama-3.3 Distilled from DeepSeek-R1 Open Source
By
–
Groq's Llama-3.3 distilled from DeepSeek-R1 would belong to the first line.
#OpenSourceMagic