The Hitchhiker's Guide to Tensorflow: from Tensors to running models in production
summary
Covers hands-on TensorFlow usage from tensors and sessions to shipping models: builds logistic-regression and CNN examples with placeholders, variables and optimizers; loads pretrained Inception/ResNet/VGG, extracts 2048-d feature vector for image similarity (beats SIFT in a hackathon), indexes with Annoy nearest-neighbor index, and retrains only the final layer (transfer learning, one-shot works). Shows graph and metrics in TensorBoard, exports frozen graph, serves with TensorFlow Serving on Kubernetes and shrinks mobile model sizes (author reports up to ~28× reduction).
AI-written summary · about
chapters (15)
- 0:00Introduction to TensorFlow
- 7:11Logistic regression basics
- 11:27TensorFlow architecture overview
- 13:54Tensors and lazy evaluation
- 16:26Placeholders and variables
- 18:12Loss functions and activations
- 20:17Optimizers and training
- 22:21Pre-trained models and resources
- 23:27Inception model and feature extraction
- 26:16Image similarity with feature descriptors
- 29:00Keras high-level API
- 31:51Fine-tuning and transfer learning
- 33:37Advanced model combinations
- 34:38Model deployment options
- 35:46Optimizing models for mobile
related talks
52mThe Hitchhikers Guide to Tensorflow
01h 18mPutting Tensorflow Models into Production
41mTensorFlow ever higher
01h 18mPutting Tensor Flow Models Into Production
55mTensorflow: Learning Machines
43mDeep learning beyond the learning
31mTensorflow for Janitors
37mSimplifying Training Deep & Serving Learning Models...using Tensorflow