# Step-3.7-Flash is supported in community Docker on DGX Spark!

**URL:** <https://forums.developer.nvidia.com/t/step-3-7-flash-is-supported-in-community-docker-on-dgx-spark/371652>\
**Category:** DGX Spark / GB10\
**Created:** [May 29, 2026, 1:33am UTC](https://forums.developer.nvidia.com/t/step-3-7-flash-is-supported-in-community-docker-on-dgx-spark/371652 "2026-05-29T01:33:04Z")\
**Posts on this page:** 1\
**Showing post:** 49

<div class="post-metadata">

**Author:** ![Mike\_MK](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@Mike\_MK](https://forums.developer.nvidia.com/u/Mike_MK)\
**Post date:** [June 2, 2026, 8:06pm UTC](https://forums.developer.nvidia.com/t/step-3-7-flash-is-supported-in-community-docker-on-dgx-spark/371652/49 "2026-06-02T20:06:56Z")

</div>

Hit a bug trying to run Step 3.7 Flash NVFP4 with MTP on a 2-node Spark cluster (TP=2).

The MTP draft loader crashes when it tries to copy the full 4096-dim vocab weight into the TP-sharded 2048-dim slot. Basically the loader doesn’t know the embedding was split across ranks.

```auto
File ".../models/step3p5_mtp.py", line 273, in load_weights
RuntimeError: The size of tensor a (2048) must match the size of tensor b (4096) at non-singleton dimension 1

```

Everything else loads fine — NVFP4 MoE on CUTLASS, FlashInfer attention, NCCL multi-rail RoCE all good. No-MTP serving works great SMH on the same setup (14 tok/s SMH, correct answers). It’s just the MTP weight loading that breaks at TP=2.

StepFun tested at TP=8 so they probably never saw this. For those of us on 2 Sparks, TP=2 is the only option and MTP is dead until this gets patched.

Anyone run into this or know a workaround?

---

_[View the full topic](https://forums.developer.nvidia.com/t/step-3-7-flash-is-supported-in-community-docker-on-dgx-spark/371652)._