Skip to content

[NV] (PR[0]): enable stage base DeepSeek-V4-Pro on shared FS for multinode#2087

Draft
qiching wants to merge 1 commit into
mainfrom
albecheng/dsv4-gb200-multinode-fs
Draft

[NV] (PR[0]): enable stage base DeepSeek-V4-Pro on shared FS for multinode#2087
qiching wants to merge 1 commit into
mainfrom
albecheng/dsv4-gb200-multinode-fs

Conversation

@qiching

@qiching qiching commented Jul 6, 2026

Copy link
Copy Markdown
Collaborator

Two changes needed to run dsv4 fp4 multinode sweeps on the watchtower (Oracle) gb200 cluster:

  • Point the dynamo-vllm dsv4 MODEL_PATH at the base DeepSeek-V4-Pro checkpoint instead of the -NVFP4 re-quant. The pinned v0.20.1 container's deepseek_v4 loader doesn't define the NVFP4 export's extra quant params (e.g. ffn.experts.w13_input_scale) and KeyErrors at load; the base checkpoint loads and matches the recipe's served identity.
  • Add dsv4/dynamo-vllm to uses_watchtower_shared_fs() so the srt-slurm workspace/outputs land on a compute-visible shared FS (the runner home is not cross-mounted to compute nodes, otherwise the batch step dies).

Two changes needed to run dsv4 fp4 multinode sweeps on the watchtower
(Oracle) gb200 cluster:

- Point the dynamo-vllm dsv4 MODEL_PATH at the base DeepSeek-V4-Pro
  checkpoint instead of the -NVFP4 re-quant. The pinned v0.20.1
  container's deepseek_v4 loader doesn't define the NVFP4 export's extra
  quant params (e.g. ffn.experts.w13_input_scale) and KeyErrors at load;
  the base checkpoint loads and matches the recipe's served identity.
- Add dsv4/dynamo-vllm to uses_watchtower_shared_fs() so the srt-slurm
  workspace/outputs land on a compute-visible shared FS (the runner home
  is not cross-mounted to compute nodes, otherwise the batch step dies).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: No status

Development

Successfully merging this pull request may close these issues.

1 participant