TODO List for 0.1 release

[pavanky]

Base Classes

• [x] nn::Module
• [x] autograd::Variable
• [x] Solver
• [x] nn::Loss

Autograd

• [x] Broadcasting (sum, tile)
• [x] Math (exp, pow, abs, sqrt, log)
• [x] Binary (add, subtract, multiply, divide, min, max, logical operators)
• [x] matrix multiplication
• [ ] Convolutions and strided Convolutions
• [ ] Indexing and assignment

Neural Network

• [x] Linear
• [ ] Convolve
• [ ] Pooling (Min, Max, Average)
• [x] Activations: Sigmoid, Tanh, ReLU and related
• [ ] Recurrent neural networks (RNN, LSTM, GRU)
• [ ] Various Losses
• [x] Containers (Sequential)
• [x] Initializers

Solvers / Optimizers

• [x] SGD
• [x] ADAM

Examples

[unbornchikken]

@pavanky "The "forward" and "backward" methods help the composite nodes interface with other nodes in the larger network."

Oh, I get it! This was not obvious, I've seen many ML libraries out there that had this fw/bw only traversal hard coded in fundamentals. Now, I started to get excited to see where your design leads to. I cannot wait to put my hands on something that support this meta-network building approach. I started to imagine something like Peltarion Synapse but with performance level of x100.

[UniqueFool]

Also see: https://github.com/arrayfire/arrayfire/issues/1441

[pavanky]

Updated the issue to reflect the new architecture.