Posit AI Weblog: torch 0.11.0

torch v0.11.0 is now on CRAN! This weblog submit highlights a few of the modifications included
on this launch. However you’ll be able to all the time discover the complete changelog
on the torch web site.

Improved loading of state dicts

For a very long time it has been doable to make use of torch from R to load state dicts (i.e. 
mannequin weights) skilled with PyTorch utilizing the load_state_dict() perform.
Nonetheless, it was frequent to get the error:

Error in cpp_load_state_dict(path) :  isGenericDict() INTERNAL ASSERT FAILED at

This occurred as a result of when saving the state_dict from Python, it wasn’t actually
a dictionary, however an ordered dictionary. Weights in PyTorch are serialized as Pickle recordsdata – a Python-specific format much like our RDS. To load them in C++, with no Python runtime,
LibTorch implements a pickle reader that’s in a position to learn solely a subset of the
file format, and this subset didn’t embody ordered dicts.

This launch provides help for studying the ordered dictionaries, so that you gained’t see
this error any longer.

Moreover that, studying theses recordsdata requires half of the height reminiscence utilization, and in
consequence additionally is far quicker. Listed here are the timings for studying a 3B parameter
mannequin (StableLM-3B) with v0.10.0:

system.time({
  x <- torch::load_state_dict("~/Downloads/pytorch_model-00001-of-00002.bin")
  y <- torch::load_state_dict("~/Downloads/pytorch_model-00002-of-00002.bin")
})

   consumer  system elapsed 
662.300  26.859 713.484 

and with v0.11.0

   consumer  system elapsed 
  0.022   3.016   4.016 

Which means that we went from minutes to only a few seconds.

Utilizing JIT operations

One of the frequent methods of extending LibTorch/PyTorch is by implementing JIT
operations. This permits builders to jot down customized, optimized code in C++ and
use it instantly in PyTorch, with full help for JIT tracing and scripting.
See our ‘Torch exterior the field’
weblog submit if you wish to study extra about it.

Utilizing JIT operators in R used to require bundle builders to implement C++/Rcpp
for every operator in the event that they needed to have the ability to name them from R instantly.
This launch added help for calling JIT operators with out requiring authors to
implement the wrappers.

The one seen change is that we now have a brand new image within the torch namespace, known as
jit_ops. Let’s load torchvisionlib, a torch extension that registers many alternative
JIT operations. Simply loading the bundle with library(torchvisionlib) will make
its operators out there for torch to make use of – it is because the mechanism that registers
the operators acts when the bundle DLL (or shared library) is loaded.

For example, let’s use the read_file operator that effectively reads a file
right into a uncooked (bytes) torch tensor.

library(torchvisionlib)
torch::jit_ops$picture$read_file("img.png")

torch_tensor
 137
  80
  78
  71
 ...
   0
   0
 103
... [the output was truncated (use n=-1 to disable)]
[ CPUByteType{325862} ]

We’ve made it so autocomplete works properly, such you could interactively discover the out there
operators utilizing jit_ops$ and urgent to set off RStudio’s autocomplete.

Different small enhancements

This launch additionally provides many small enhancements that make torch extra intuitive:

• Now you can specify the tensor dtype utilizing a string, eg: torch_randn(3, dtype = "float64"). (Beforehand you needed to specify the dtype utilizing a torch perform, similar to torch_float64()).

  torch_randn(3, dtype = "float64")

  torch_tensor
  -1.0919
   1.3140
   1.3559
  [ CPUDoubleType{3} ]

• Now you can use with_device() and local_device() to quickly modify the system
  on which tensors are created. Earlier than, you had to make use of system in every tensor
  creation perform name. This permits for initializing a module on a particular system:

  with_device(system="mps", {
    linear <- nn_linear(10, 1)
  })
  linear$weight$system

  torch_device(kind='mps', index=0)

• It’s now doable to quickly modify the torch seed, which makes creating
  reproducible packages simpler.

  with_torch_manual_seed(seed = 1, {
    torch_randn(1)
  })

  torch_tensor
   0.6614
  [ CPUFloatType{1} ]

Thanks to all contributors to the torch ecosystem. This work wouldn’t be doable with out
all of the useful points opened, PRs you created, and your onerous work.

If you’re new to torch and wish to study extra, we extremely advocate the lately introduced guide ‘Deep Studying and Scientific Computing with R torch’.

If you wish to begin contributing to torch, be happy to succeed in out on GitHub and see our contributing information.

The total changelog for this launch will be discovered right here.

Picture by Ian Schneider on Unsplash

Reuse

Textual content and figures are licensed below Inventive Commons Attribution CC BY 4.0. The figures which have been reused from different sources do not fall below this license and will be acknowledged by a observe of their caption: “Determine from …”.

Quotation

For attribution, please cite this work as

Falbel (2023, June 7). Posit AI Weblog: torch 0.11.0. Retrieved from https://blogs.rstudio.com/tensorflow/posts/2023-06-07-torch-0-11/

BibTeX quotation

@misc{torch-0-11-0,
  writer = {Falbel, Daniel},
  title = {Posit AI Weblog: torch 0.11.0},
  url = {https://blogs.rstudio.com/tensorflow/posts/2023-06-07-torch-0-11/},
  12 months = {2023}
}