From 97f69a6b8142b4446b6bd5b0139e796fda557f67 Mon Sep 17 00:00:00 2001 From: Zayan Khan Date: Sun, 26 Jul 2026 19:17:05 -0400 Subject: [PATCH 1/2] cifar: fix --batch flag in distributed training example main.py's argparse defines --batch_size (main.py line 19); the documented mlx.launch command fails with 'unrecognized arguments: --batch 256' as written. Co-Authored-By: Claude Fable 5 --- cifar/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cifar/README.md b/cifar/README.md index 2016200df..734da7ab7 100644 --- a/cifar/README.md +++ b/cifar/README.md @@ -60,5 +60,5 @@ $ cat >hostfile.json {"ssh": "host-to-ssh-to", "ips": ["ip-to-bind-to"]}, {"ssh": "host-to-ssh-to", "ips": ["ip-to-bind-to"]} ] -$ mlx.launch --verbose --hostfile hostfile.json main.py --batch 256 --epochs 5 --arch resnet20 +$ mlx.launch --verbose --hostfile hostfile.json main.py --batch_size 256 --epochs 5 --arch resnet20 ``` From 91d566f4ee48e2e39b88ea5974a09dbee591592a Mon Sep 17 00:00:00 2001 From: Zayan Khan Date: Sun, 26 Jul 2026 19:17:05 -0400 Subject: [PATCH 2/2] cvae: fix train.py reference in example output The training entry point is main.py (train.py does not exist in this example); the README's own earlier instructions already say 'python main.py'. Co-Authored-By: Claude Fable 5 --- cvae/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cvae/README.md b/cvae/README.md index 8cfa26ae0..2e1620917 100644 --- a/cvae/README.md +++ b/cvae/README.md @@ -25,7 +25,7 @@ To see the supported options, do `python main.py -h`. Training with the default options should give: ```shell -$ python train.py +$ python main.py Options: Device: GPU Seed: 0