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 10, 12:45 UTC
[Bugfix] Remove the num_hidden_layers override for glm4_moe (#31745)
Andy Liu committed
Jan 7, 2026 at 23:45 UTC
e5173d3bac054edc1b556005d804a5acf94eaf15
1 file changed
-1
vllm/config/speculative.py
-1
@@ -199,7 +199,6 @@ class SpeculativeConfig:
199
n_predict = getattr(hf_config, "num_nextn_predict_layers", None)
200
hf_config.update(
201
{
202
- "num_hidden_layers": 0,
202
"n_predict": n_predict,
203
"architectures": ["Glm4MoeMTPModel"],
204
}