[PD] Support PD disaggregation with Prefill PP (#8846)

Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
Signed-off-by: Shangming Cai <csmthu@gmail.com>
Co-authored-by: root <huzhiyuan@xiaohongshu.com>
Co-authored-by: Ying Sheng <sqy1415@gmail.com>
Co-authored-by: Francis <38564764+ssssnow@users.noreply.github.com>
Co-authored-by: zitto <zhjc1124@gmail.com>

python/sglang/srt/models/deepseek_nextn.py

@@ -20,7 +20,7 @@ import torch
 from torch import nn
 from transformers import PretrainedConfig
 
-from sglang.srt.distributed import get_tensor_model_parallel_world_size
+from sglang.srt.distributed import get_pp_group, get_tensor_model_parallel_world_size
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
 from sglang.srt.layers.dp_attention import is_dp_attention_enabled
 from sglang.srt.layers.layernorm import RMSNorm
@@ -135,6 +135,8 @@ class DeepseekV3ForCausalLMNextN(DeepseekV3ForCausalLM):
         self.config = config
         self.tp_size = get_tensor_model_parallel_world_size()
         self.quant_config = quant_config
+        # if not set, model load will be broken in DeepseekV3ForCausalLM load_weights()
+        self.pp_group = get_pp_group()
         self.determine_num_fused_shared_experts("DeepseekV3ForCausalLMNextN")
 
         self.model = DeepseekModelNextN(