Add deterministic DDP training example with hashed seeding #60

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Merged

farukalpay merged 1 commit into main from codex/create-deterministic-ddp-training-script

Aug 25, 2025

Owner

farukalpay commented Aug 25, 2025

Summary

add ddp_train.py example demonstrating deterministic DDP training on CIFAR-10
implement SHA-256 based subseed generation with Philox generators for unique RNG streams per rank/epoch/worker
print per-rank first batch labels and initial conv weights, plus final model_init subseed and digest

Testing

pre-commit run --files examples/ddp_train.py (hooks skipped: no files to check)
torchrun --standalone --nnodes=1 --nproc-per-node=2 examples/ddp_train.py --master-seed 314159 --run-id R42 --epochs 1 --batch-size 16

https://chatgpt.com/codex/tasks/task_e_68ac7ce9b2b88323b2a9f5ba870b9768


          Add deterministic DDP training example

18c19ad

farukalpay added the codex label

— with

ChatGPT Codex Connector

farukalpay merged commit 9841f5b into main

5 checks passed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels