tested after rebasing, fix importance sampling bug

2025-09-14 21:51:57 +00:00 · 2025-08-06 06:15:15 +00:00
parent 3746f73854
commit 32b2148670
4 changed files with 4 additions and 3 deletions
--- a/colossalai/shardformer/modeling/qwen3.py
+++ b/colossalai/shardformer/modeling/qwen3.py
@@ -273,6 +273,7 @@ class Qwen3PipelineForwards:
        hidden_states: Optional[torch.FloatTensor] = None,
        stage_index: Optional[List[int]] = None,
        shard_config: ShardConfig = None,
+        **kwargs
    ):
        r"""
        Args: