Keras 3 + JAX makes it possible to efficiently distribute large models via both data parallelism and model parallelism (variable partitioning) across arbitrarily many devices. It doesn't require modifying your model code or your training code at all — all you need is to
MACHINE LEARNING
-
Long Context Window Progress: From 65k to 1M Tokens in One Year
By
–
quite incredible to see the goalposts of long context move in the past 1 year. May 2023: asking @jefrankle and @abhi_venigalla about their 65k+ "Llongboi" model May 2024: @markatgradient casually extending Llama 3 to >1m tokens with ~perfect NIAH and mainstream ai engineers
-
NY Tech Week: Comprehensive Event Guide and Community Gathering
By
–
NY Tech Week is going to be wild. Just look at this list of events: https://
docs.google.com/spreadsheets/d
/13LeRMAgMdtqDpCKgI-Msv2h6DwpO15V7jy_yok1DWf0/edit
… Who’s coming? -
Salesforce Databricks Integration for Custom AI Models
By
–
Introducing Salesforce Bring Your Own Model for Databricks — a powerful new integration! Build, train, and deploy custom #AI models in Databricks and effortlessly integrate them into Salesforce to deliver intelligent and personalized customer experiences:
-
Fine-Tuning Custom Embedding Models with AutoTrain Guide
By
–
NEW BLOG: How to Fine-Tune Custom Embedding Models Using AutoTrain
Learn:
– what should be the data format
– how to map columns properly
– example datasets
– custom configs
– train locally
– train on hugging face spaces -
RSS 2024 Workshop on Embodied Voices: Call for Submissions
By
–
Last call 1st Workshop on Embodied Voices @ RSS 2024 @RoboticsSciSys
! https://
rosielab.github.io/wev/
Deadline: June 1st for 2-4 page submissions.
Excited to bring together communities across ML, speech, robotics, interaction. -
AI Identifies New Potential Treatments for Parkinson’s Disease
By
–
AI Identifies New Potential Treatments For Parkinson’s Disease https://
forbes.com/sites/victoria
forster/2024/04/18/ai-identifies-new-potential-treatments-for-parkinsons-disease/?utm_source=dlvr.it&utm_medium=twitter
… #Innovation #ArtificialIntelligence #MachineLearning -
AMD Performance and Determinism in ML Training Optimization
By
–
super nice that it just runs on AMD!
have to look into why you seem to be getting less noisy charts, this is on current todo list under getting full determinism.
and yes possibly not exactly comparable to look at electricity costs alone. but i'm not sure there are XTX boxes on -
Daily Content Sharing on Python, Data Science, and Machine Learning
By
–
That's a wrap! If you are interested in any of these below topics: – Python – Data Science – Machine Learning – Data Analysis – LLMs – MLOps Find me → @Sumanth_077 I'm sharing daily content over here.
-
Predict Method Implementation with Binary Classification Threshold
By
–
Finally the predict method: It uses the trained model to make predictions based on input features. This method applies a threshold of 0.5 to make binary predictions, returning 1 for positive class predictions and 0 for negative class predictions.