What if you taught transformers to reason in both latent states and tokens? This Microsoft paper adds a self-supervised objective of predicting the next latent state to the standard next-token training, where a lightweight dynamic model
AI
-
A great opportunity to learn the Agent Development Cycle with hwchase17
By
–
A great opportunity to learn the Agent Development Cycle with @hwchase17
-
If Google had an enforceable patent on transformers
By
–
Imagine if Google had an enforceable patent on transformers.
-
AI Voice Dictation 2026: Automatic Punctuation, Context and Homophones
By
–
10. Voice dictation that really works. Open any keyboard, tap the microphone and speak. The 2026 AI dictation automatically punctuates, follows context and correctly distinguishes "là", "leurs" and "ils sont". Speak a complete WhatsApp message or an email.
-
Enable Gemini to summarize, write, translate, analyze
By
–
1. Gemini, already integrated. Press and hold the home button or swipe up from a corner. A real AI opens, not the old assistant. Ask it to summarize a page you're reading, write an email, translate a sign, or analyze a photo. It was
-
Impressed by depth of model building; many fixes improve initial model
By
–
also in general i've been really impressed with the amount of DEPTH there is to building models in industry, and working on larger projects over longer timespans honestly the first model we trained was not very smart. a few clever ideas and so many little fixes/improvements
-
Initial deployment: small step, essential repeatable improvement cycle
By
–
One thing we have observed quite consistently: deploying the first version is only a small part of the work. A key part of building reliable agents is having a repeatable lifecycle to improve them over time.
-
AI agent gets phone number and voice with AgentPhoneHQ
By
–
Get your AI agent a phone number.
— Robert Scoble (@Scobleizer) 23 juin 2026
Visited @manav2modi, founder of @AgentPhoneHQ which makes a phone number and a voice to answer it.
For your agent in seconds. pic.twitter.com/MQh2GknirDGet your AI agent a phone number. Visited @manav2modi
, founder of @AgentPhoneHQ which makes a phone number and a voice to answer it. For your agent in seconds. -
Automating post-training with continual learning and OPSD-RL interpolation
By
–
on a technical level it feels like we're automating post-training we want to take a bunch of data and output a post-trained model – then iterate this process many times. continual learning by induction the ideal algorithm is an interpolation of OPSD and RL with a lot of
-
Leaving Their PhD to Co-Found Engram
By
–
Finally I share what I have devoted myself to! I left my PhD at the end of 2025 and co-founded Engram. There are a few startups in SF right now making very different bets on the right way to train AI models. Here is ours: people want models that