Reinventing Machine Learning with Transformers and Hugging Face
summary
Hugging Face demonstrates building and deploying Transformer-based apps using the Transformers library, the Hub, AutoTrain and Diffusers: the Hub hosts 83,000 models and 13,000 datasets with >1M downloads/day; AutoTrain fine-tunes a model to 91.45% accuracy on Food-101; generating ~200 synthetic Swedish‑meatball images with Stable Diffusion and training for ~30 minutes produces a recognizer that scores ~29.8% on meatballs but deploys as a managed inference endpoint for scalable, private production use.
AI-written summary · about
chapters (14)
- 0:00Welcome and intro
- 2:20Deep learning challenges
- 6:33Bridging ML and software
- 9:45Rise of transformers
- 13:38Hugging Face platform
- 18:47Workflow demo
- 24:24Navigating the hub
- 28:42AutoTrain demo
- 32:23Generating training data
- 37:10Fine-tuning models
- 44:12Evaluating models
- 48:51Deploying with endpoints
- 52:36Getting started resources
- 55:27Q&A session
related talks
33mMachine Learning 2.0 with Hugging Face
19mBuilding Machine Learning Applications using Hugging Face
14mKeynote: Building AI with Open Source & Hugging Face- 14mKeynote: Building AI with Open Source & Hugging Face
58mAn Introduction to Transfer Learning and Hugging Face
34mTransformers End to End Experience training...
34mExploring Hugging Face: A Hub for Machine Learning Models
32mAn Introduction to Transfer Learning in NLP and HuggingFace