You like pytorch? You like micrograd? You love tinygrad! ❤️
Go to file
George Hotz 10b1306525 binops 2021-06-10 16:52:37 -07:00
.github/workflows minor cleanups, yolo work 2021-01-03 08:14:16 -08:00
ane add an aneccompile example in Objective-C (#240) 2021-01-31 08:31:16 -08:00
docs FPGA Based Accelerator for Tinygrad (#258) 2021-06-07 17:45:09 -07:00
examples minor optimizations & cleaning (#257) 2021-06-02 09:57:15 -07:00
extra binops 2021-06-10 16:52:37 -07:00
fpga FPGA Based Accelerator for Tinygrad (#258) 2021-06-07 17:45:09 -07:00
test FPGA Based Accelerator for Tinygrad (#258) 2021-06-07 17:45:09 -07:00
tinygrad FPGA Based Accelerator for Tinygrad (#258) 2021-06-07 17:45:09 -07:00
.gitignore [WIP] YOLO working on tinygrad! (#245) 2021-04-25 18:06:52 -07:00
LICENSE readme 2020-10-18 11:27:37 -07:00
README.md Revert "Update README.md (#259)" (#260) 2021-06-04 14:41:41 -07:00
push_pypi.sh push pypi 2020-10-27 08:13:15 -07:00
requirements.txt fix pyopencl (#125) 2020-11-19 19:03:04 -08:00
setup.py All devices are equal! (#196) 2020-12-15 23:44:08 -08:00

README.md


Unit Tests

For something in between a pytorch and a karpathy/micrograd

This may not be the best deep learning framework, but it is a deep learning framework.

Due to its extreme simplicity, it aims to be the easiest framework to add new accelerators to, with support for both inference and training. Support the simple basic ops, and you get SOTA vision extra/efficientnet.py and language extra/transformer.py models. We are working on support for the Apple Neural Engine.

Eventually, we will build custom hardware for tinygrad, and it will be blindingly fast. Now, it is slow.

Installation

pip3 install git+https://github.com/geohot/tinygrad.git --upgrade

Example

from tinygrad.tensor import Tensor

x = Tensor.eye(3)
y = Tensor([[2.0,0,-2.0]])
z = y.matmul(x).sum()
z.backward()

print(x.grad)  # dz/dx
print(y.grad)  # dz/dy

Same example in torch

import torch

x = torch.eye(3, requires_grad=True)
y = torch.tensor([[2.0,0,-2.0]], requires_grad=True)
z = y.matmul(x).sum()
z.backward()

print(x.grad)  # dz/dx
print(y.grad)  # dz/dy

Neural networks?

It turns out, a decent autograd tensor library is 90% of what you need for neural networks. Add an optimizer (SGD, RMSprop, and Adam implemented) from tinygrad.optim, write some boilerplate minibatching code, and you have all you need.

Neural network example (from test/test_mnist.py)

from tinygrad.tensor import Tensor
import tinygrad.optim as optim

class TinyBobNet:
  def __init__(self):
    self.l1 = Tensor.uniform(784, 128)
    self.l2 = Tensor.uniform(128, 10)

  def forward(self, x):
    return x.dot(self.l1).relu().dot(self.l2).logsoftmax()

model = TinyBobNet()
optim = optim.SGD([model.l1, model.l2], lr=0.001)

# ... and complete like pytorch, with (x,y) data

out = model.forward(x)
loss = out.mul(y).mean()
optim.zero_grad()
loss.backward()
optim.step()

GPU and Accelerator Support

tinygrad supports GPUs through PyOpenCL.

from tinygrad.tensor import Tensor
(Tensor.ones(4,4).gpu() + Tensor.ones(4,4).gpu()).cpu()

ANE Support?!

If all you want to do is ReLU, you are in luck! You can do very fast ReLU (at least 30 MEGAReLUs/sec confirmed)

Requires your Python to be signed with ane/lib/sign_python.sh to add the com.apple.ane.iokit-user-access entitlement, which also requires amfi_get_out_of_my_way=0x1 in your boot-args. Build the library with ane/lib/build.sh

from tinygrad.tensor import Tensor

a = Tensor([-2,-1,0,1,2]).ane()
b = a.relu()
print(b.cpu())

Warning: do not rely on the ANE port. It segfaults sometimes. So if you were doing something important with tinygrad and wanted to use the ANE, you might have a bad time.

Adding an accelerator

You need to support 14 first class ops:

Relu, Log, Exp                  # unary ops
Sum, Max                        # reduce ops (with axis argument)
Add, Sub, Mul, Pow              # binary ops (with broadcasting)
Reshape, Transpose, Slice       # movement ops
Matmul, Conv2D                  # processing ops

While more ops may be added, I think this base is stable.

ImageNet inference

Despite being tiny, tinygrad supports the full EfficientNet. Pass in a picture to discover what it is.

ipython3 examples/efficientnet.py https://upload.wikimedia.org/wikipedia/commons/4/41/Chicken.jpg

Or, if you have a webcam and cv2 installed

ipython3 examples/efficientnet.py webcam

PROTIP: Set "GPU=1" environment variable if you want this to go faster.

PROPROTIP: Set "DEBUG=1" environment variable if you want to see why it's slow.

tinygrad also supports GANs

See examples/mnist_gan.py

The promise of small

tinygrad will always be below 1000 lines. If it isn't, we will revert commits until tinygrad becomes smaller.

Running tests

python3 -m pytest

TODO

  • Train an EfficientNet on ImageNet
  • Add a language model. BERT?
  • Add a detection model. EfficientDet?
  • Reduce code
  • Increase speed
  • Add features