[docker] cherry pick qwen3.5 bugfix

docker/patch/latest/sglang.patch

@@ -2516,6 +2516,26 @@ index 2cf813bce..1250c49e4 100644
  def _canonicalize_weights(config, weights_in: Iterable[Tuple[str, torch.Tensor]]):
      weights_out_dict = dict(weights_in)
 
+diff --git a/python/sglang/srt/models/qwen3_5.py b/python/sglang/srt/models/qwen3_5.py
+index f01225487..145ee410c 100644
+--- a/python/sglang/srt/models/qwen3_5.py
++++ b/python/sglang/srt/models/qwen3_5.py
+@@ -372,6 +372,7 @@ class Qwen3_5LinearDecoderLayer(nn.Module):
+             input_layernorm=self.input_layernorm,
+             post_attention_layernorm=self.post_attention_layernorm,
+             allow_reduce_scatter=True,
++            is_last_layer=(layer_id == config.num_hidden_layers - 1),
+         )
+
+     def forward(
+@@ -549,6 +550,7 @@ class Qwen3_5AttentionDecoderLayer(nn.Module):
+             input_layernorm=self.input_layernorm,
+             post_attention_layernorm=self.post_attention_layernorm,
+             allow_reduce_scatter=True,
++            is_last_layer=(layer_id == config.num_hidden_layers - 1),
+         )
+
+         self.alt_stream = alt_stream
 diff --git a/python/sglang/srt/models/qwen3_vl.py b/python/sglang/srt/models/qwen3_vl.py
 index d641826e3..3abc39ef3 100644
 --- a/python/sglang/srt/models/qwen3_vl.py

docker/version.txt

@@ -1 +1 @@
-nightly-dev-20260307a
+nightly-dev-20260309a

examples/strands_sglang/generate_with_strands.py

@@ -24,6 +24,7 @@
 MAX_TOOL_ITERS = 5
 MAX_TOOL_CALLS = None  # No limit
 
+
 @tool
 def execute_python_code(code: str) -> str:
     """Execute Python code and return the output."""

slime/utils/distributed_utils.py

@@ -2,8 +2,8 @@
 from typing import Any
 
 import torch
-from packaging.version import parse as V
 import torch.distributed as dist
+from packaging.version import parse as V
 from torch.distributed.distributed_c10d import (
     Backend,
     PrefixStore,
@@ -14,7 +14,6 @@
     rendezvous,
 )
 
-
 GLOO_GROUP = None
 
 

train.py

@@ -2,7 +2,7 @@
 
 from slime.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models
 from slime.utils.arguments import parse_args
-from slime.utils.logging_utils import configure_logger, init_tracking, finish_tracking
+from slime.utils.logging_utils import configure_logger, finish_tracking, init_tracking
 from slime.utils.misc import should_run_periodic_action
 
 

train_async.py

@@ -2,7 +2,7 @@
 
 from slime.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models
 from slime.utils.arguments import parse_args
-from slime.utils.logging_utils import configure_logger, init_tracking, finish_tracking
+from slime.utils.logging_utils import configure_logger, finish_tracking, init_tracking
 from slime.utils.misc import should_run_periodic_action