Posted by HyeJung Lee, DevRel Group Supervisor and Soonson Kwon, DevRel Program Supervisor
Let’s discover highlights and achievements of huge Google Machine Studying communities over the past quarter of final yr! We’re excited and grateful about all of the actions that the communities throughout the globe do.
India-based Aakash Nain has accomplished the TF-Jax tutorial sequence with Half 9 (Autodiff in JAX) and Half 10 (Pytrees in JAX). Aakash additionally began a brand new tutorial sequence to study concerning the present strategies of constructing fashions in JAX. The primary tutorial Constructing fashions in JAX – Part1 (Stax) is launched.
On Dec twelfth, ML GDE Paolo Galeone began to resolve puzzles of the Creation of Code sequence utilizing “pure TensorFlow” (with out some other library). His resolution has been up to date in a sequence of 12 on his weblog. He defined how he designed the options, how he carried out them, and – when wanted – targeted on some TensorFlow options not broadly used. (Day 1, Day 2, Day 3, Day 4, Day 5, Day 6, Day 7, Day 8, Day 9, Day 10, Day 11, Day 12, Wrap up)
ML GDE Chansung Park (Korea) & Sayak Paul (India) printed an “Steady Adaptation for Machine Studying System to Knowledge Adjustments” article on TensorFlow weblog. They introduced a mission that implements a workflow combining batch prediction and mannequin analysis for steady analysis retraining With a purpose to seize adjustments within the knowledge.
ML GDE Elyes Manai from Tunisia wrote an article on GDE weblog about his expertise on the Google Cloud ML Engineer certification masking information to certificates and ideas.
TFUG organizer Ali Mustufa Shaikh and Rishit Dagli launched “CPPE-5: Medical Private Protecting Tools Dataset” (paper, code). This paper bought featured on Google Analysis TRC’s publication part on January 5, 2022.
TFUG New York hosted a sequence of occasions in Dec. Finish-to-Finish NLP Workshop with TensorFlow. Temporary introduction to the Kaggle competitors for Nice Barrier Reef problem by Google(Slide). TF concept for ML Initiatives with GCP.
ML GDE Elyes Manai from Tunisia wrote an article “The power to vary folks’s lives and go away one’s mark“. Are you going through difficulties rising in constrained environments? And do you suppose you are not a first-class pupil and you do not have connections within the trade? Then, take a look at Elyes’s story. He shared how Google helped him speed up his impression.
ML GDE Sayak Paul (India) and Soumik Rakshit’s Level Cloud Segmentation carried out the PointNet structure for segmenting 3D level clouds utilizing the ShapeNetCore dataset with TensorFlow 2.x. It’s a winner of #TFCommunitySpotlight too.
Annotated Analysis Papers by ML GDE Aakash Kumar Nain (India) is an effort to make papers extra accessible to a wider group. It additionally helps the net model and contains papers from Google Analysis and and so forth. This repository is fashionable sufficient to have a +2k star and a +200 fork.
Interview sequence of DevLibrary contributors : Meet the ML GDE Shai Reznik (Israel) and Doug Duhaime. And take a look at what they constructed with Google expertise and what made them passionate.
ML DevFest 2021 by GDG Cloud San Francisco. There are 5 periods that stroll you thru framing ML issues, researching ML, constructing proofs of ideas utilizing present ML APIs and fashions, constructing ML pipelines and and so forth. ML GDE Vikram Tiwari (USA) introduced Vertex, ML Ops and GCP.
Krupal Modi (India)’s weblog article and #IamaGDE video shares how he’s been main the machine studying initiatives at Haptik, a conversational AI platform, and the way the staff paired with the Indian Authorities and WhatsApp to construct a COVID-19 helpline.