Add latent MoE support (#30203)

Signed-off-by: Shahar Mor <smor@nvidia.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
This commit is contained in:
shaharmor98
2025-12-08 17:35:01 +00:00
committed by GitHub
co-authored by Tyler Michael Smith
parent 398a596ed2
commit fcd5306f65
2 changed files with 58 additions and 6 deletions
@@ -189,6 +189,7 @@ class NemotronHConfig(PretrainedConfig):
n_shared_experts=1,
moe_intermediate_size=7688,
moe_shared_expert_intermediate_size=7688,
moe_latent_size=None,
num_experts_per_tok=2,
routed_scaling_factor=1.0,
n_group=1,
@@ -254,6 +255,7 @@ class NemotronHConfig(PretrainedConfig):
self.n_shared_experts = n_shared_experts
self.moe_intermediate_size = moe_intermediate_size
self.moe_shared_expert_intermediate_size = moe_shared_expert_intermediate_size # noqa: E501
self.moe_latent_size = moe_latent_size
self.num_experts_per_tok = num_experts_per_tok
self.routed_scaling_factor = routed_scaling_factor
self.n_group = n_group