diff --git a/python/sglang/srt/models/glm4_moe_lite.py b/python/sglang/srt/models/glm4_moe_lite.py index 03ff3d201..5b08f1f45 100644 --- a/python/sglang/srt/models/glm4_moe_lite.py +++ b/python/sglang/srt/models/glm4_moe_lite.py @@ -403,6 +403,8 @@ class Glm4MoeLiteDecoderLayer(DeepseekV2DecoderLayer): config.hidden_size, eps=config.rms_norm_eps ) + self._gfx95_quant_format = self._detect_gfx95_quant_format() + self.layer_communicator = LayerCommunicator( layer_scatter_modes=self.layer_scatter_modes, input_layernorm=self.input_layernorm,