[Minor] Fix styles for overlap mode (#2068)

python/sglang/srt/managers/scheduler.py

@@ -1002,7 +1002,7 @@ class Scheduler:
             if req.is_retracted:
                 continue
 
-            if self.server_args.enable_overlap_schedule and (req.finished()):
+            if self.enable_overlap and req.finished():
                 self.token_to_kv_pool.free(batch.out_cache_loc[i : i + 1])
                 continue
 
@@ -1319,7 +1319,7 @@ def run_scheduler_process(
     try:
         scheduler = Scheduler(server_args, port_args, gpu_id, tp_rank, dp_rank)
         pipe_writer.send("ready")
-        if server_args.enable_overlap_schedule:
+        if scheduler.enable_overlap:
             scheduler.event_loop_overlap()
         else:
             scheduler.event_loop_normal()

python/sglang/srt/managers/tp_worker_overlap_thread.py

@@ -26,7 +26,6 @@ import torch
 from sglang.srt.managers.io_struct import UpdateWeightReqInput
 from sglang.srt.managers.schedule_batch import ModelWorkerBatch
 from sglang.srt.managers.tp_worker import TpModelWorker
-from sglang.srt.model_executor.forward_batch_info import ForwardBatch
 from sglang.srt.server_args import ServerArgs
 
 logger = logging.getLogger(__name__)
@@ -176,16 +175,8 @@ class TpModelWorkerClient:
         ) % self.future_token_ids_limit
         return None, future_next_token_ids
 
-    def forward_batch_embedding(self, model_worker_batch: ModelWorkerBatch):
-        forward_batch = ForwardBatch.init_new(model_worker_batch, self.model_runner)
-        logits_output = self.model_runner.forward(forward_batch)
-        embeddings = logits_output.embeddings
-        return embeddings
-
     def update_weights(self, recv_req: UpdateWeightReqInput):
-        success, message = self.model_runner.update_weights(
-            recv_req.model_path, recv_req.load_format
-        )
+        success, message = self.worker.update_weights(recv_req)
         return success, message
 
     def __delete__(self):

python/sglang/srt/model_executor/model_runner.py

@@ -276,10 +276,6 @@ class ModelRunner:
             else None
         )
         self.dtype = self.vllm_model_config.dtype
-        if self.sliding_window_size:
-            assert (
-                self.server_args.attention_backend == "flashinfer"
-            ), "Only flashinfer supports window attention."
 
         logger.info(
             f"Load weight end. "

test/srt/test_triton_attention_backend.py

@@ -1,4 +1,8 @@
-import subprocess
+"""
+Usage:
+python3 -m unittest test_triton_attention_backend.TestTritonAttnBackend.test_mmlu
+"""
+
 import unittest
 from types import SimpleNamespace