For months, we've worked on building @huggingface
's new moonshot: agentic systems. So today we're very proud to announce the release of ๐๐๐๐๐๐๐๐๐๐! It's the simplest library we could make to let people build powerful agents. The main logic for agents fits in ~1000
@aymericroucher
-
Hugging Face Releases smolagents Library for AI Agent Development
By
–
-
Definition of AI Agents Involves Tool Use and Multistep Memory
By
–
@minimaxir i'd say agents are tool use + multistep (multistep means a memory that logs past steps ans errors properly)
-
Hugging Face Releases Picotron for Efficient LLM Training Parallelization
By
–
> Hugging Face releases Picotron, a microscopic lib that solves LLM training 4D parallelization Llama-3.1-405B took 39 million GPU-hours to train, which represents 4.5 thousand years. If they had needed all this time, we would have GPU stories from the time of Pharaoh
-
New LLM Research Shows Tokenizers Can Be Removed
By
–
Potential paradigm shift in LLMs: new paper by @AIatMeta shows that we can get rid of tokenizers! Current LLMs process text by first splitting it into tokens. They use a module named "tokenizer", that -spl-it-s- th-e- te-xt- in-to- arbitrary tokens depending on a fixed
-
Google Releases Gemini 2.0 AI Model with Agentic Capabilities
By
–
๐๐ผ๐ผ๐ด๐น๐ฒ ๐ฟ๐ฒ๐น๐ฒ๐ฎ๐๐ฒ๐ ๐๐ฒ๐บ๐ถ๐ป๐ถ ๐ฎ.๐ฌ, ๐๐๐ฎ๐ฟ๐๐ถ๐ป๐ด ๐๐ถ๐๐ต ๐ฎ ๐๐น๐ฎ๐๐ต ๐บ๐ผ๐ฑ๐ฒ๐น ๐๐ต๐ฎ๐ ๐๐๐ฒ๐ฎ๐บ๐ฟ๐ผ๐น๐น๐ ๐๐ฃ๐ง-๐ฐ๐ผ ๐ฎ๐ป๐ฑ ๐๐น๐ฎ๐๐ฑ๐ฒ-๐ฏ.๐ฒ ๐ฆ๐ผ๐ป๐ป๐ฒ๐! And they start a huge effort on agentic capabilities. Performance improvements:
โฃ Gemini -
Speculation on Anthropic’s Opus-3.5 and AI Scaling Laws
By
–
Are scaling laws still alive? New blog post suggests that @AnthropicAI might have an extremely strong Opus-3.5 already available, but is not releasing it to keep their edge over the competition. – Since the release of Opus-3.5 has been delayed indefinitely, there have been
-
New Multilingual AI Benchmark and Model Releases in Open Source AI
By
–
Last week was crazy in OS AI, with important models and datasets releases every day. Here are the most important ones I've pinned: Cohere relased GLobal-MMLU, a multilingual version of MMLU, to evaluate AI models' world knowledge in many languages! Meta released
-
ShowUI: Small End-to-End AI Agent Navigates Any UI
By
–
๐ฆ๐ต๐ผ๐๐จ๐: ๐ฎ ๐๐บ๐ฎ๐น๐น ๐ฒ๐ป๐ฑ-๐๐ผ-๐ฒ๐ป๐ฑ ๐ฎ๐ด๐ฒ๐ป๐ ๐๐ต๐ฎ๐ ๐ฐ๐ฎ๐ป ๐ป๐ฎ๐๐ถ๐ด๐ฎ๐๐ฒ ๐ฎ๐ป๐ ๐จ๐ ๐ฎ๐ป๐ฑ ๐ผ๐๐๐ฝ๐ฒ๐ฟ๐ณ๐ผ๐ฟ๐บ๐ ๐บ๐๐ฐ๐ต ๐ฏ๐ถ๐ด๐ด๐ฒ๐ฟ ๐๐๐๐๐ฒ๐บ๐! A team from NUS and Microsoft just released an agent that can act on any UI (Desktop, Android, Web)
-
Adobe’s Code-Generating Agent Tops GAIA Leaderboard
By
–
๐๐ฑ๐ผ๐ฏ๐ฒ'๐ ๐ฐ๐ผ๐ฑ๐ฒ-๐ด๐ฒ๐ป๐ฒ๐ฟ๐ฎ๐๐ถ๐ป๐ด ๐ฎ๐ด๐ฒ๐ป๐ ๐ฟ๐ฒ๐ฎ๐ฐ๐ต๐ฒ๐ ๐๐ต๐ฒ ๐๐ผ๐ฝ ๐ผ๐ณ ๐๐๐๐ ๐น๐ฒ๐ฎ๐ฑ๐ฒ๐ฟ๐ฏ๐ผ๐ฎ๐ฟ๐ฑ – and they cite "Roucher, 2024" in the paper! Reminder:ย Broadly defined, an "Agent" is a system where a LLM is augmented with the ability to run
-
Release of Qwen-QwQ-32B Preview Model on Hugging Face
By
–
https://
huggingface.co/chat/models/Qw
en/QwQ-32B-Preview
โฆ