AI - AI Dynamics

→ View original post on X — @aibreakfast

Watermarking AI text to detect ChatGPT-generated content

By

@aibreakfast

–

25 January 2023 22h54

"Watermarking" AI-generated text. Researchers from University of Maryland propose a way to discretely embed (and detect) special tokens in order to accurately determine whether or not text was generated from something like ChatGPT. ↓

→ View original post on X — @packym

Watt Carbon Raises $4.5M Seed Led by True Ventures

By

@packym

–

25 January 2023 22h20

We're doubling-down on @wattcarbon in its $4.5M seed led by @trueventures
. Carbon accounting –> offsets marketplace –> VPP is a killer combo and the right way to build this, but really… Would you bet against a guy who 12ft's his own paywalled Axios raise announcement?

→ View original post on X — @dominodatalab

Domino MLOps Platform Expands Through TDSYNNEX Partnership

By

@dominodatalab

–

25 January 2023 22h00

Domino has partnered with @TDSYNNEX to bring our #MLOps platform to thousands of channel partners and businesses, increasing the value of #ML and #AI investments and accelerating time-to-market. Learn more in this article by @RickWhiting1
:

Emergence as Framework for Understanding Language Model Scaling

By

–

25 January 2023 20h36

Overall, emergence is nice framework for viewing language models
– Emergent abilities cannot be predicted via scaling plots for small models
– Keep scaling, and we might see more emergent abilities 🙂 See our TMLR 2022 piece (w/ survey certification):

U-shaped Scaling: Model Performance Recovery in Larger Language Models

By

–

25 January 2023 20h36

Another newer example of emergence that I’m excited about is U-shaped scaling, where even if model performance goes down from small → medium models, it can go back up for large models. See our paper: https://
x.com/_jasonwei/stat
us/1588605909781319680
… Inverse scaling benchmark:

Instruction Tuning as an Emergent Ability in Large Language Models

By

–

25 January 2023 20h36

One of the most interesting emergent abilities IMO is instruction tuning. Anthropic and Flan-LaMDA suggest that zero-shot performance can improve from RLHF and NLP benchmark instruction tuning (although text-davinci usually loses to code-davinci). https://
arxiv.org/abs/2204.05862

Emergent Abilities in Language Models Through Scaling

By

–

25 January 2023 20h36

Yesterday I gave a lecture at @Stanford
's CS25 class on Transformers! The lecture was on how “emergent abilities” are unlocked by scaling up language models. Emergence is one of the most exciting phenomena in large LMs… Slides:

Emergent Abilities in Large Language Models Over Past Year

By

–

25 January 2023 20h36

Throughout the past year, there have been hundreds of emergent abilities, which can only be observed in large-enough language models. I previously made a list of them (more than 100):https://t.co/QLfwNUeIBx
— Jason Wei (@_jasonwei) 25 janvier 2023

Throughout the past year, there have been hundreds of emergent abilities, which can only be observed in large-enough language models. I previously made a list of them (more than 100):

→ View original post on X — @aiatmeta

Data2vec 2.0: 16x Faster Self-Supervised Learning Across Modalities

By

@aiatmeta

–

25 January 2023 20h15

Data2vec 2.0 can train self-supervised speech, vision & text models up to 16x faster than the most popular existing algorithm for images — achieving the same accuracy. Read more & get the open source code https://
bit.ly/3H9mPf7