Expand description
2026-09-25: Device weight handles for one GLM-5.3 MLP site, already sharded for this rank.
Owner: model-arch (GLM-5.3).
Invariants: none beyond the types; build.rs fills them.
Structs§
- Glm5
Next Dense MlpWeights - 2026-09-25: A BF16 SwiGLU MLP: the dense layers
0..first_k_dense_replace, and the shared expert of every routed layer. - Glm5
Next Expert PtrTable - 2026-09-25: Device pointer tables for one projection over the full expert set.
- Glm5
Next Expert Weights - 2026-09-25: One routed NVFP4 expert, owned whole by one EP rank.
- Glm5
Next MoePtr Tables - 2026-09-25: The three projections’ pointer tables for one routed site.
- Glm5
Next MoeWeights - 2026-09-25: A routed MoE site’s weights for this rank.
- Nvfp4
Proj - 2026-09-25: One NVFP4 projection: packed
e2m1pairs, per-16-elemente4m3block scales, and one globalf32scale.