(9/12) Scaling Expert Language Models with Unsupervised Domain Discovery
Authors: @ssgrn, @margs_li
, @ml_perception
, @WeijiaShi2
, @timalthoff
, @nlpnoah
, and @LukeZettlemoyer
INNOVATION
-
Scaling Expert Language Models with Unsupervised Domain Discovery
By
–
-
Simfluence: Modeling Training Example Influence Through Simulation
By
–
(7/12) Simfluence: Modeling the Influence of Individual Training Examples by Simulating Training Runs
Authors: @kelvin_guu
, @albertwebson
, Ellie Pavlick, @iislucas
, @iftenney
, @tolgab0
, @GoogleResearch
, @Brown_NLP -
Larger Language Models Perform In-Context Learning Differently
By
–
(4/12) Larger Language Models Do In-Context Learning Differently
Authors: @JerryWeiAI
, @_jasonwei
, @YiTayML
, @dustinvtran
, @albertwebson
, @yifenglou
, @xinyun_chen_
, @Hanxiao_6
, @dhuangcn
, @denny_zhou
, @tengyuma -
Efficient LLM Inference: High-Throughput Generation on Single GPU
By
–
(5/12) High-Throughput Generative Inference of LLMs with a Single GPU
Authors: @ying11231
, @lm_zheng
, @Hades317
, @zhuohan123
, Max Ryabinin, @realDanFu
, Zhiqiang Xie, Beidi Chen, Clark Barrett, Joseph E. Gonzalez, @percyliang
, Christopher Ré @snorkelai
, Ion Stoica, Ce Zhang -
PaLM-E: Embodied Multimodal Language Model for Robotics
By
–
(2/12) PaLM-E: An Embodied Multimodal Language Model
Authors: @DannyDriess
, @xf1280
, Mehdi S. M. Sajjadi, @coreylynch
, @achowdhery
, @brian_ichter
, @ayzwah
, @JonathanTompson
, @QuanVng
, @TianheYu
, @wenlong_huang
, @YevgenChebotar
, @psermanet
, @duck et. al. -
MLPerf Results Show AI Performance Gains from Leading Tech Companies
By
–
The latest round of MLPerf results are in. It continues to be fascinating how new techniques are brought to bear by Neural Magic, cTuning, Neuchips and others. https://
zdnet.com/article/nvidia
-dell-qualcomm-speed-up-ai-results-in-latest-benchmark-tests/
… $NVDA @neuralmagic $DELL $QCOM $HPE @MLCommons #MLPerf #AI #artificialintelligence -
Language Models Training: Autoregressive vs Diffusion Approaches
By
–
Common Q: Can you train language model w diffusion?
Favorite A: read this post (the whole blog is excellent) (Roughly speaking state of the art generative AI is either trained autoregressively or with diffusion. The underlying neural net usually a Transformer.) -
AI and Generative AI Transform Healthcare and Democratize Value Creation
By
–
Thank you @ritters90
, Amicia, Kyle and Jillian for a lively and engaging discussion. #AI is transformative to any industry and particularly to Healthcare #Creation of value (of any kind) is democratised with #generativeai Cost of creati…
https://
lnkd.in/gXWB3qmV -
Stanford CS330: Deep Multi-Task and Meta-Learning Course 2022
By
–
Stanford CS330: Deep Multi-Task & Meta-Learning – 2022 This course covers topics related to multi-task and meta-learning such as self-supervised pre-training, transfer learning, lifelong learning, etc. New lectures just dropped. https://
youtube.com/playlist?list=
PLoROMvodv4rNjRoawgt72BBNwL2V7doGI
… -
Community Adoption of Baby AGI Growing Among Developers
By
–
Love seeing people use Baby AGI https://
x.com/mathis_global/
/mathis_global/status/1643204719555526656
…