AI Daily Brief — 01 January 2025
The new year opened with no fresh frontier-lab announcements but a quiet earthquake in US AI law: California's first wave of statutes signed by Governor Newsom…
Latest
QVQ: To See the World with Wisdom
GITHUB HUGGING FACE MODELSCOPE KAGGLE DEMO DISCORD Language and vision intertwine in the human mind, shaping how we perceive and understand the world around us. Our…
2024 Year in Review
A peaceful year of steady progress on my craft and health.
o1 tops aider’s new polyglot leaderboard
o1 scores the top result on aider's new multi-language, more challenging coding benchmark.
Building effective agents
We've worked with dozens of teams building LLM agents across industries. Consistently, the most successful implementations use simple, composable patterns rather than complex frameworks.
Is AI progress slowing down?
Making sense of recent technology trends and claims
We Looked at 78 Election Deepfakes. Political Misinformation is not an AI Problem.
Technology Isn’t the Problem—or the Solution.
nbsanity – Share Notebooks as Polished Web Pages in Seconds
nbsanity - Share Notebooks as Polished Web Pages in Seconds
SAEs trained on the same data don’t learn the same features
In this post, we show that when two TopK SAEs are trained on the same data, with the same batch order but with different random initializations,…
OpenAI o1 System Card
This report outlines the safety work carried out prior to releasing OpenAI o1 and o1-mini, including external red teaming and frontier risk evaluations according to our…
QwQ is a code architect, not an editor
QwQ is reasoning model like o1, and needs to be used as an architect with another model as editor.
Seemingly Paradoxical Rules of Writing
With regard to writing, there are many rules and also no rules at all.
Building an Audience Through Technical Writing: Strategies and Mistakes
Building an Audience Through Technical Writing: Strategies and Mistakes
Reward Hacking in Reinforcement Learning
Reward hacking occurs when a reinforcement learning (RL) agent exploits flaws or ambiguities in the reward function to achieve high rewards, without genuinely learning or completing…
QwQ: Reflect Deeply on the Boundaries of the Unknown
GITHUB HUGGING FACE MODELSCOPE DEMO DISCORD Note: This is the pronunciation of QwQ: /kwju:/ , similar to the word “quill”. What does it mean to think,…
Yi and Yi 1.5 are evolving🌳
Yi and Yi 1.5 are evolving🌳Omar Sanseviero: The (non-exhaustive) evolution of base modelsIf you want to learn more about it and how to use these models,…
Merge pull request #620 from 01-ai/Mia-xia-patch-3
Merge pull request #620 from 01-ai/Mia-xia-patch-3 Update README.md
Adds controlnet images, updates README (#21)
Adds controlnet images, updates README (#21) * Adds blur and depth images * Cosmetic changes to REEADME --------- Co-authored-by: Vikram Voleti
Merge pull request #20 from Stability-AI/bf/controlnet
Merge pull request #20 from Stability-AI/bf/controlnet ControlNet support
fixed latent encoder behavior based on control type
fixed latent encoder behavior based on control type
How to Run a Weekly Paper Club (and Build a Learning Community)
Benefits of running a weekly paper club, how to start one, and how to read and facilitate papers.
Details matter with open source models
Open source LLMs are becoming very powerful, but pay attention to how you (or your provider) are serving the model. It can affect code editing skill.
Fixes for VAE logic and 2B ControlNets, and speed up model loading by…
Fixes for VAE logic and 2B ControlNets, and speed up model loading by loading ControlNets to CUDA if available
minor changes to image saving and controlnet loading
minor changes to image saving and controlnet loading