Core functionality to allow multi-device workflow

README.md

@@ -22,7 +22,7 @@ As well as:
 First, [download LibTorch](https://pytorch.org/get-started/locally/). For Mac arm64, use:
 
 ```sh
-curl -L https://download.pytorch.org/libtorch/cpu/libtorch-macos-arm64-2.9.0.zip > libtorch.zip
+curl -L https://download.pytorch.org/libtorch/cpu/libtorch-macos-arm64-2.9.1.zip > libtorch.zip
 unzip -q libtorch.zip
 ```
 
@@ -42,6 +42,8 @@ gem "torch-rb"
 
 It can take 5-10 minutes to compile the extension. Windows is not currently supported.
 
+For distributed data parallel helpers, add the optional `torch-ddp` gem alongside this one.
+
 ## Getting Started
 
 A good place to start is [Deep Learning with Torch.rb: A 60 Minute Blitz](tutorials/blitz/README.md).
@@ -329,6 +331,8 @@ net.load_state_dict(Torch.load("net.pth"))
 net.eval
 ```
 
+`Torch.load` mirrors the Python API and accepts `map_location` and `weights_only` keyword arguments for compatibility with existing PyTorch checkpoints.
+
 When saving a model in Python to load in Ruby, convert parameters to tensors (due to outstanding bugs in LibTorch)
 
 ```python

bin/torchrun

@@ -0,0 +1,6 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require_relative "../lib/torch/torchrun"
+
+Torch::TorchRun.start(ARGV)