Search results

ShardingOptimizer.save()/load() resurrects constraints removed via remove_constraints()

Status: Open.

#476 In meta-pytorch/autoparallel;

· fmassa opened on May 26, 2026

AutoParallel fails on HF transformers Llama-3 / Qwen-2.5: AssertionError in scaled_dot_product_cudnn_attention strategy

Status: Open.

#456 In meta-pytorch/autoparallel;

· apbose opened on May 12, 2026

Move AutoParallel activation checkpointing out of the compile backend

Status: Open.

#447 In meta-pytorch/autoparallel;

· sanketpurandare opened on May 5, 2026

Unifying operator.getitem with the standard code path

enhancementNew feature or requestNew feature or request

Status: Open.

#388 In meta-pytorch/autoparallel;

· fmassa opened on Mar 22, 2026

run_with_rng_state + checkpointing interactions

Status: Open.

#273 In meta-pytorch/autoparallel;

· xmfan opened on Dec 8, 2025

Compile-based Pipeline Parallelism

Status: Open.

#199 In meta-pytorch/autoparallel;

· fmassa opened on Oct 9, 2025

Llama3 doesn't work when there's only one GPU

Status: Open.

#73 In meta-pytorch/autoparallel;

· ezyang opened on Aug 4, 2025

Integrate memory tools into default examples

Status: Open.

#55 In meta-pytorch/autoparallel;

· ezyang opened on Jul 28, 2025