Today we're starting to open up access to Bard, our early experiment that lets you collaborate with generative AI. You can use Bard to boost your productivity, accelerate your ideas and fuel your curiosity. Learn more, including how to sign up ↓ https://t.co/4zDI5RD1fr
Today's announcement with @AnthropicAI is another great example of the ecosystem of AI startups leveraging Google Cloud's reliable and open infrastructure to build their businesses. https://t.co/VB2CjFTd88
1/ In 2021, we shared next-gen language + conversation capabilities powered by our Language Model for Dialogue Applications (LaMDA). Coming soon: Bard, a new experimental conversational #GoogleAI service powered by LaMDA.
https://t.co/cYo6iYdmQ1
the ai world timeline is almost completely detached from the other ones i'm following, which are full of turbulences (general tech industry, financial markets, macro economics and geopolitics, to name a few) but maybe that's a good thing~
@kadarakos energy problem seems much easier to solve than agi and brings much more net positive to the humanity. just feel the current hype is kind of misaligned with the reality
well put! text is inherently redundant and ambiguous. it's unlikely to be the universal interface between humans and machines imho (ppl definitely will not use text for some more effective ways of communication, like using mouses or touchscreens, etc)
Small rant about LLMs and how I see them being put, rather thoughtlessly IMO, into productivity tools. 📄
TL;DR — Most knowledge work isn't a text-generation task, and your product shouldn't ship an implementation detail of LLMs as the end-user interface
https://t.co/N1NJoKzUqA
Can multi-100B param language models be served efficiently? We think so! Today we’re announcing the PaLM inference paper and releasing code for low-latency, high-throughput inference of 8B–540B models on TPU v4.
Paper: https://t.co/ZBEZOu5usf
Code: https://t.co/mgGNsSxOnq 1/5
Excited this is out! CaP uses code generation LLMs to plan and execute robotics tasks, since code is more expressive and can represent complex logics, loops, conditions, etc. I want to highlight a few contributions of this work that goes beyond vanilla program synthesis:
A new way to work w. LMs!
Binder, an easy neuro-symbolic paradigm:
1.Parse input➡️SQL/Python bound w. GPT3 Codex API calls
2.Codex+PL interpreter execute➡️answer
No train&few-shot!➡️SOTA
🆚chain-of-thought: interpretable&robust⬆️
🆚NL2Code: coverage⬆️
https://t.co/YxymeDYv8X
📜🚨📜🚨
NN loss landscapes are full of permutation symmetries, ie. swap any 2 units in a hidden layer. What does this mean for SGD? Is this practically useful?
For the past 5 yrs these Qs have fascinated me. Today, I am ready to announce "Git Re-Basin"!
https://t.co/mRu5k3ONUm