curse of knowledge and 1 to n mentality (minds go straight to optimization, reliability, cost, primarily respect scale) meanwhile the disruptive 0 to 1 stuff (capability, creativity, content) benefits from a totally different background classical ml folks dont have (so perhaps
MACHINE LEARNING
-
Google Maps Largest Human Brain Connectome Achievement
By
–
It's been amazing watching the progress of this @GoogleResearch collaboration w/HHMI, @MCB_Harvard, … over the last decade. Starting w/fly brains years ago, progressing through parts of mouse brain, and now the largest-ever connectome for part of a human brain! 🧠 https://t.co/RSWfpgdH3z
— Jeff Dean (@JeffDean) 10 mai 2024It's been amazing watching the progress of this @GoogleResearch collaboration w/HHMI, @MCB_Harvard
, … over the last decade. Starting w/fly brains years ago, progressing through parts of mouse brain, and now the largest-ever connectome for part of a human brain! -
Machine Learning Productivity: Balancing Exploration and System Building
By
–
machine learning productivity is a mix between curiosity-driven exploration to understand every surprising observation and a focused pursuit of the otherwise singular goal of building a working system
-
LLMs Tips and Tutorials for Sharing
By
–
If you find this useful, RT to share it with your friends. Don't forget to follow me @Saboo_Shubham_ for more such LLMs tips and tutorials.
-
Mixed Precision Neural Operators Boost Throughput ICLR2024
By
–
Check out our #ICLR2024 paper on mixed precision neural operators. We show >50% increase in throughput with almost no loss in accuracy! Poster # 58, Fri 4:30 – 6:30pm CEST https://
arxiv.org/abs/2307.15034 @crwhite_ml @tu_renbo @JeanKossaifi Boris, Gennady, @Azizzadenesheli Ours is a -
NeuralPlexer Surpasses AlphaFold3 in Protein Binding Predictions
By
–
While AlphaFold3 improves upon the previous version, it still can't handle conformational changes during binding like our NeuralPlexer does. Paper: https://
nature.com/articles/s4225
6-024-00792-z
… Summary: https://
nature.com/articles/s4159
2-024-02249-y
…
"NeuralPLexer can predict both apo and bound forms of proteins and can -
Tesla FSD 12.3 Review: One Month of Autonomous Driving Experience
By
–
I have been trying Tesla FSD 12.3 for a month and am completely blown away.
— Satya Mallick (@LearnOpenCV) 9 mai 2024
In the entire month i have not driven one continuous mile manually. I faced 3 situations where it did something obviously wrong like making a left turn from a lane adjacent to the lane marked for left… https://t.co/Ms0af5JmJxI have been trying Tesla FSD 12.3 for a month and am completely blown away. In the entire month i have not driven one continuous mile manually. I faced 3 situations where it did something obviously wrong like making a left turn from a lane adjacent to the lane marked for left
-
Attention-Driven Training-Free Efficiency Enhancement for Diffusion Models
By
–
Attention-Driven Training-Free Efficiency Enhancement of Diffusion Models Wang et al.: https://
arxiv.org/abs/2405.05252 #ArtificialIntelligence #DeepLearning #MachineLearning -
ORPO Fine-tuning Configuration for Llama3-8B on Single GPU
By
–
Here's the best config to orpo finetune llama3-8b on argilla capybara on a single 24GB home GPU! Github: https://
github.com/huggingface/au
totrain-advanced/blob/main/configs/llm_finetuning/llama3-8b-orpo.yml
…
You can run this config using AutoTrain: $ pip install -U autotrain-advanced
$ autotrain –config config.yaml/URL to raw config file -
Command R Fine-Tuning Reduces Inference Costs by 15x
By
–
Maximize efficiency and speed with Command R fine-tuning. This reduces inference costs by 15x compared to larger models and achieves faster throughput, lower latency and superior performance.