[Feature] block sparse attention (#3668)

* 支持稀疏attn

* fix bug

* code style

* fix moba attn get kv shape

* 修复a100编译

* codestyle

* code style

* code style

* code style

* fix conflict

* 增加单侧

* code style

* 增加eblite 加载时间

* fix bug

* for ci

* for ci

* for ci

* for ci

* 支持mlp block size 128

* 增加小算子单测

* fix 单测 mlp

* 将环境变量加入到config里面

* fix rollout config

* 修复显存

* add test server

* add test server

* fix mlp  最后一层使用full attn

fastdeploy/rl/rollout_config.py

@@ -59,6 +59,7 @@ class RolloutModelConfig:
         graph_optimization_config: str = None,
         early_stop_config: str = None,
         local_rank: int = 0,
+        moba_attention_config: str = None,
     ):
         # Required parameters
         self.model = model_name_or_path
@@ -103,6 +104,7 @@ class RolloutModelConfig:
         self.local_rank = local_rank
         self.early_stop_config = early_stop_config
         self.ips = None
+        self.moba_attention_config = moba_attention_config
 
     def __str__(self):
         return "\n".join(f"{k}: {v}" for k, v in self.__dict__.items())