verl-project/verl

[FR] Documentation about Resolving OOM

Geschlossen

#1.014 geöffnet am 10.04.2025

 (6 Kommentare) (7 Reaktionen) (0 zugewiesene Personen)Python (4.440 Forks)auto 404
call for contributiondocumentationgood first issue

Repository-Metriken

Stars
 (23.087 Sterne)
PR-Merge-Metriken
 (Durchschn. Merge 3T 6h) (125 gemergte PRs in 30 T)

Beschreibung

Motivation

There are many issues related to OOM, e.g. #328 . We might need a clear guide about how to resolve OOM.

Plan

A non-exclusive enumeration about related configurations:

  1. Rollout:gpu_memory_utilization
  2. Other Inference:
    1. Liger Kernel
    2. *_max_len_per_gpu / micro_batch_size_per_gpu
  3. Training:
    1. Liger Kernel
    2. Ulysses Sequence Parallelism
    3. gradient checkpointing
    4. offload

TODO

  • Complete the list of related configurations
  • Benchmark the effect & overhead of each configuration

Contributor Guide