The University of Toronto made a video for a non technical audience in which I explain how deep learning works, the enormous promise of this technology and some of the potential risks. https://
youtu.be/-9cW4Gcn5WY
RESEARCH
-
University of Toronto Explains Deep Learning to Non-Technical Audiences
By
–
-
HyperReel: High-Fidelity 6-DoF Video with Ray-Conditioned Sampling
By
–
HyperReel: High-Fidelity 6-DoF Video with Ray-Conditioned Sampling
— AK (@_akhaliq) 22 juin 2023
paper page: https://t.co/TkYI5ibRny
Volumetric scene representations enable photorealistic view synthesis for static scenes and form the basis of several existing 6-DoF video techniques. However, the volume… pic.twitter.com/7zIO8MXavDHyperReel: High-Fidelity 6-DoF with Ray-Conditioned Sampling paper page: https://
huggingface.co/papers/2301.02
238
… Volumetric scene representations enable photorealistic view synthesis for static scenes and form the basis of several existing 6-DoF video techniques. However, the volume -
MosaicML Releases New 30B Open Source Language Model
By
–
The just release 30B model from MosaicML looks really great! Nice (big) size, OSS apache-2 licence and long context! take a look at the thread for more details
-
AI Ethics Panel Discussion with Experts Concludes Successfully
By
–
hi, the AI ethics panel with @alexhanna and @mer__edith hosted by @sarahfrier that just concluded was fantastic, stay tuned for a video link! pic.twitter.com/yRoLy4BOHj
— Rachel Metz (@rachelmetz) 22 juin 2023hi, the AI ethics panel with @alexhanna and @mer__edith hosted by @sarahfrier that just concluded was fantastic, stay tuned for a video link!
-
SoundStorm: Parallel Decoding for Efficient Audio Generation
By
–
Many generative audio models rely on auto-regressive decoding, which produces tokens one by one and can be slow. Read all about SoundStorm, a new method tailored for audio tokens that uses parallel decoding for efficient and high-quality audio generation ↓
-
Data Preprocessing for Large Language Models: RedPajama Analysis
By
–
This 30-minute interactive Q&A will cover several topics, including: * Why data pre-processing is important
* An analysis of RedPajama and SlimPajama
* What Cerebras did to pre-process this massive dataset
* How this impacts organizations like yours -
Cerebras Tech Talk: Dataset Processing for LLM Training
By
–
Join us for our next Cerebras Tech Talk! On Wednesday, June 28th at 11:00 AM PT, we will discuss how we cleaned and processed one of the largest public datasets so others can train higher-quality large language models. See next tweet for info. Register: https://
hubs.li/Q01VxRqb0 -
AI-Descartes: Scientific Renaissance in Artificial Intelligence
By
–
AI-Descartes: A Scientific Renaissance in the World of Artificial Intelligence https://
bit.ly/3M4MJV3 #AI #ethics -
Model Ideological Bias Measurement and Mitigation
By
–
For example, when we prompt a model to vote on key issues, it tends to align with certain opinion groups more than others. As a result, model-based ideological biases (which human facilitators and moderators may also have) must be carefully measured and considered.
-
Language Models Risks Require Deeper Scientific Study
By
–
At the same time, we also find that LMs applied in this context pose risks that require (and illuminate areas for) deeper study.