Tensors is a concept in mathematics.
They have
- Type
- Rank (no of dimensions)
- Shapes
Rank0 - scalr
Rank 1/ Shape{5} - Array of size 5
Rank 2/ Shape {3} - 3X3 Matrix
In TensorFlow
- Tensors in Tensor Flow may be contained by variables
- Tensors may be mutated by operations
- Graphs contains tensors, variables and operations
It wraps only 20% of the TensorFlow classes
It includes
- Basic model creation
- Model Training
- Inline Predictions
- Predictions across the network.
Essential classes
- org.tensorflow.Tensor<T>
- org.tensorflow.Output
- This is used to reach into a running graph and help us to get the values. For example the predictions that we want
- org.tensorflow.Graph
- The database for tensors
- org.tensorflow.Session
- If the Graph is the database, then Session is the connection to Database
No comments:
Post a Comment