A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 9, 02:30 UTC
Update get_expert_mapping to include self parameter (#33525)
Signed-off-by: Borushiki <38628261+Otsutsukii@users.noreply.github.com>
Borushiki committed
Feb 2, 2026 at 13:29 UTC
b398e5c819a6fab7d7c4a4482dfa9893935261a4
1 file changed
+1
vllm/model_executor/models/glm4_moe_lite.py
+1
@@ -617,6 +617,7 @@ class Glm4MoeLiteForCausalLM(
617
# Params for weights, fp8 weight scales, fp8 activation scales
618
# (param_name, weight_name, expert_id, shard_id)
619
return SharedFusedMoE.make_expert_params_mapping(
620
+ self,
621
ckpt_gate_proj_name="gate_proj",
622
ckpt_down_proj_name="down_proj",
623
ckpt_up_proj_name="up_proj",