[RL] Support checkpoint-coordinate sparse NCCL weight updates (#53751)

Signed-off-by: aoshen02 <aoshen@inferact.ai>
Signed-off-by: Shuolei Wang <shuoleiwang123@gmail.com>
Co-authored-by: aoshen02 <aoshen@inferact.ai>
Co-authored-by: OpenAI Codex <codex@openai.com>
This commit is contained in:
Shuolei Wang
2026-08-27 18:19:51 +08:00
committed by GitHub
co-authored by aoshen02 OpenAI Codex
parent 50708f95ae
commit 6c4aa8bec3
5 changed files with 435 additions and 797 deletions
+1 -1
View File
@@ -33,7 +33,7 @@ engine drives on your behalf:
| ------- | --------- | -------- |
| [NCCL](nccl.md) | NCCL broadcast | Separate GPUs for training and inference |
| [IPC](ipc.md) | CUDA IPC handles | Colocated training and inference on same GPU |
| [sparse_nccl](nccl.md#sparse-nccl) | NCCL broadcast | Sparse flat-index weight patches (TP=1/PP=1) |
| [sparse_nccl](nccl.md#sparse-nccl) | NCCL broadcast | Checkpoint-coordinate sparse weight patches |
| [sharded_rdt](sharded_rdt.md) | NIXL / Ray Direct Transport (pull-based) | Very large models where each worker needs only its own slice (MoE with expert parallelism) |
## Quickstart