[Jan 2024] ML Community - Highlights and Achievements
Let’s explore highlights and accomplishments of the vast Google Machine Learning communities over the month. We appreciate all the activities and commitment by the community members. Without further ado, here are the key highlights!
Featured Stories
ML Developer Journey
Interview with Aakash Kumar Nain, MLE at Merlyn Minds by Abheesht Sharma (India) and Anshuman Mishra (India) is a conversation with AI/ML GDE Aakash Nain. This article outlines what it takes to be an open source contributor and how it would be like to continuously learn and stay updated as an ML expert, and includes Aakash’s skillful experiences and insightful thoughts on career and self-development.
Collaboration in Africa
Imen Masmoudi, Sara El-Ateif, Armel Yara, and Louis Kouassi who all met in person at the ML Community Summit 2023 (Tashkent), started working together to introduce Keras 3.0 to their communities, and grow all together! This session will be streamed on Youtube on March 9th!
- Imen Masmoudi (Tunisia): MLAct (3rd party community) lead
- Sara El-Ateif (Morroco): AI/ML GDE, TFUG Casablanca lead
- Armel Yara (Canada): TFUG Abidjan lead
- Louis Kouassi (Ivory Coast): TFUG Bassam lead
Focus Area Highlights
Gemini
Natural Language to SQL using Google’s Gemini Pro Model by AI/ML GDE Bhavesh Bhatt (India) talks about how to leverage Gemini Pro to query a SQL database using natural language.
Hyper-Personalized Ad Campaigns using Generative AI + Quick Demo using Google Gemini Pro Vision by AI/ML GDE Jigyasa Grover (US) shares how to generate human-like personalized advertisement text in real-time using LLMs. A demo for this project is shared via Colab.
Daily Newsletter for 🤗 Daily Papers by AI/ML GDE Chansung Park (Korea) is a newsletter bot using Gemini API to generate paper summary in a friendly email format.
Bookmarks AI by Taha Bouhsine (US) is a Gemini Pro-powered Chrome extension that provides summaries of bookmarked pages.
LlamaIndex 101 GRATIS: Tutorial RAG by AI/ML GDE Carlos Alarcon explains how to use LLaMA index and Gemini Pro for RAG systems.
Getting started with Gemini and Google AI Studio by TFUG Malaysia and RAG — Talking to your data by TFUG São Paulo introduced Gemini and covered LLM trends.
Generative AI / LLM
LLM Fine-Tuning and Model Selection Using Neptune and Transformers by AI/ML GDE Pedro Gengo (Brazil) is about selecting the best model and conducting efficient fine-tuning when resources are constrained.
Getting Started with Diffusers for Text-to-Image by AI/ML GDE Ritwik Raha (India) is a tutorial on how to generate images from text descriptions using the Diffusers library from Hugging Face.
AI Architect — Design your home with Stable Diffusion by AI/ML GDE Nitin Tiwari (India) and Aashi Dutt (India) is a project demonstrating the implementation of fine-tuning SDXL model using DreamBooth and LoRA to generate room interior designs on custom data.
Keras
Navigating Keras 3 and Model Deployment Options by AI/ML GDE Marvin Ngesa (Kenya) was a workshop guiding attendees on how to construct a simple Keras 3 model and executing it with a preferred backend, PyTorch, TensorFlow, or JAX.
Machine and Deep Learning at INAF-IASF Milano by AI/ML GDE Umberto Michelucci (Switzerland) was a 5 days school on ML focusing on TensorFlow and Keras for astrophysics and researchers at the Postdoc level. More than 210 people (50 in person, 160 online) participated in the course from 15+ universities.
Introduction to ML and NLP using KerasNLP by AI/ML GDE Karthik Muthuswamy (Germany) at Keras Community Day GDG Hannover introduced how to use KerasNLP or KerasCV to train and deploy a model using FastAPI.
Kaggle
Kaggle Notebooks (Keras3-KerasCV-MixerTransformer[Training], [Inference]) by Usha Rengaraju (India) are for the HMS — Harmful Brain Activity Classification competition aiming to develop predictive models that can accurately identify and classify harmful brain activity patterns, contributing to advancements in neurology and medical diagnosis.
How to Detect AI Generated Content With TensorFlow (Kaggle notebook) by AI/ML GDE Derrick Mwiti (Kenya) explains how to detect AI-generated content in an on-going Kaggle competition.
ML Research
Tutorials on Google’s SigLIP Model by AI/ML GDE Merve Noyan (France) shows how you can utilize SigLIP for search in different modalities.
The Chronicles of RAG: The Retriever, the Chunk and the Generator by AI/ML GDE Pedro Gengo (Brazil) and AI/ML GDE Vinicius Caridá (Brazil) showcased various setups for RAG in the Brazilian context. The authors explored diverse methods to answer questions about the first Harry Potter book, and used Gemini Pro and GPT to generate answers.
Image Animation with Generative AI (Colab notebook) by AI/ML GDE Hugo Zanini (Brazil) is an implementation and summary of the Magic Animate paper.
Cloud
Fine-tuning and Deploying Open-Source LLMs by AI/ML GDE Ruqiya Bin Safi (Saudi Arabia) delved into the intricacies of fine-tuning and deploying open-source LLMs aiming to provide a comprehensive understanding of model adaptation and deployment, covering model variations, tuning techniques, and the prominent steps required to execute these processes. She also highlighted the LLMs in Model Garden in Vertex AI.
Generative AI for SMEs on the Google Cloud Platform (slides) by AI/ML GDE Svetlana Meissner (Germany) covered the basics of Gen AI, usage of the GCP for Gen AI, and an overview of Vertex AI and Dialogflow CX.
In TU5.2 — How developers are shaping the future with Google Workspace and GenAI, AI/ML GDE Allen Firstenberg (US) talked with Googlers about how Vertex AI and other Gen AI solutions can be used to extend the capabilities of Google Workspace with targeted solutions that integrate with Google Workspace Add-ons, Editor Add-ons, Chat apps, and etc.
JAX
NanoDL by Henry Ndubuaku (UK) is a JAX/Flax library designed for building transformer models, featuring a wide range of blocks and layers for custom creation from scratch.
Convolutional Neural Networks in JAX: Ultimate Guide by AI/ML GDE Derrick Mwiti (Kenya) is a quick guide on how to define and train CNN in JAX.
