dfdx: shape checked deep learning in rust
Ergonomics & safety focused deep learning in Rust.
Still in pre-alpha state. The next few releases are planned to be breaking releases.
Features at a glance:
- :fire: GPU accelerated tensor library with shapes up to 6d!
- Shapes with both compile and runtime sized dimensions. (e.g.
- A large library of tensor operations (including
conv2d, and much more).
- All tensor operations shape and type checked at compile time!!
- Ergonomic neural network building blocks (like
- Standard deep learning optimizers such as
RMSprop, and more.
dfdx is on crates.io! Use by adding this to your
See the documentation at docs.rs/dfdx.
- Ergonomics the whole way down (both frontend interface & internals).
- Check as much at compile time as possible (i.e. don't compile if something is not correct).
- Maximize performance.
- Minimize unsafe code
- Minimize Rc<RefCell> used in internal code
 Currently the only unsafe calls are for matrix multiplication.
 The only things that use
Arc are tensors to store their data.
Arc is used instead of
Box to reduce
allocations when tensors are cloned.
GPU acceleration with CUDA
cuda feature to start using the
Cuda device! Requires the installation of nvidia's cuda toolkit. See feature flags docs for more info.
The matrixmultiply crate is the default BLAS library. You don't need to do download/install anything for this to work!
To link to the
Intel MKL libraries (assuming you installed it already) use the
cpu-mkl-matmul feature. See feature flags docs for more info.
Check examples/ for more details.
- 👌 Simple Neural Networks API, completely shape checked at compile time.
type Mlp = ;
- 📈 Ergonomic Optimizer API
type Model = ... let mut model = dev.; let mut grads = model.alloc_grads; let mut sgd = new; let loss = ... grads = loss.backward; sgd.update;
- 💡 Const tensors can be converted to and from normal rust arrays
let t0: = dev.tensor; assert_eq!; let t1 /*: Tensor<Rank1<3>, f32, _>*/ = dev.tensor; assert_eq!; let t2: = dev.sample_normal; assert_ne!;
Fun/notable implementation details
From this flexible trait we get:
- Single & batched inputs (just have multiple impls!)
- Multiple inputs/outputs (multi-headed modules, or rnns)
- Behavior different when tape is present or not (not the .train()/.eval() behavior present in other libraries!).
Tuples represent feedforward (a.k.a sequential) modules
Since we can implement traits for tuples, which is not possible in other languages AFAIK, they provide a very nice frontend for sequentially executing modules.
// no idea why you would do this, but you could! type Model = ; let model = dev.;
type Model = let model = dev.;
How implementing Module for a 2-tuple looks:
Modules implemented for Tuples up to 6 elements, but you can arbitrarily nest them!
Rc<RefCells<T>> used - Gradient tape is not kept behind a cell!
Other implementations may store a reference to the gradient tape directly on tensors, which requires mutating tensors or using Rc/Refcells all over the place.
We've figured out an elegant way to avoid this, reducing references and dynamic borrow checks to 0!
Since all operations result in exactly 1 child, we can always move the gradient tape to the child of the last operation. Additionally, no model parameters (all tensors) will ever own the gradient tape because they will never be the result of any operation. This means we know exactly which tensor owns the gradient tape, and the tensors that have it will always be intermediate results that don't need to be maintained across gradient computation.
All of this together gives users unprecedented control/precision over what tensors are recorded on the gradient tape!
One advanced use case requires that tensors be re-used multiple times in a computation graph. This can be handled by cloning the tensor, and manually moving the gradient tape around.
Type checked backward
tl;dr: If you forget to include a call to
traced(), the program won't compile!
-let pred = module.forward(x); +let pred = module.forward(x.traced(grads)); let loss = (y - pred).square().mean(); let gradients = loss.backward();
Since we know exactly what tensors own the gradient tape, we can require the tensor passed into
.backward() to own the gradient tape!
And further, we can require it be moved into
.backward(), so it can destruct the tape and construct the gradients!
All of this can be checked at compile time 🎉
📄 Validated against pytorch
All functions & operations are tested against behavior shown by similar code in pytorch.
Dual-licensed to be compatible with the Rust project.
Licensed under the Apache License, Version 2.0 http://www.apache.org/licenses/LICENSE-2.0 or the MIT license http://opensource.org/licenses/MIT, at your option. This file may not be copied, modified, or distributed except according to those terms.