[LoRA] Support 2D and 3D MoE LoRA adapter at the same time (#42242)

Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Signed-off-by: Jee Jee Li <jeejeelee@inferact.ai>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
This commit is contained in:
Jee Jee Li
2026-05-18 15:22:26 +08:00
committed by GitHub
co-authored by gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Roger Wang
parent e3aeee5ff8
commit 7d5b033782
16 changed files with 391 additions and 9 deletions
+6
View File
@@ -584,6 +584,7 @@ class EngineArgs:
lora_target_modules: list[str] | None = LoRAConfig.target_modules
enable_tower_connector_lora: bool = LoRAConfig.enable_tower_connector_lora
specialize_active_lora: bool = LoRAConfig.specialize_active_lora
enable_mixed_moe_lora_format: bool = LoRAConfig.enable_mixed_moe_lora_format
ray_workers_use_nsight: bool = ParallelConfig.ray_workers_use_nsight
num_gpu_blocks_override: int | None = CacheConfig.num_gpu_blocks_override
@@ -1272,6 +1273,10 @@ class EngineArgs:
lora_group.add_argument(
"--specialize-active-lora", **lora_kwargs["specialize_active_lora"]
)
lora_group.add_argument(
"--enable-mixed-moe-lora-format",
**lora_kwargs["enable_mixed_moe_lora_format"],
)
# Observability arguments
observability_kwargs = get_kwargs(ObservabilityConfig)
@@ -2009,6 +2014,7 @@ class EngineArgs:
target_modules=self.lora_target_modules,
enable_tower_connector_lora=self.enable_tower_connector_lora,
specialize_active_lora=self.specialize_active_lora,
enable_mixed_moe_lora_format=self.enable_mixed_moe_lora_format,
max_cpu_loras=self.max_cpu_loras
if self.max_cpu_loras and self.max_cpu_loras > 0
else None,