fix

2025-09-03 01:55:12 +00:00 · 2024-09-25 19:00:38 +08:00
parent 6fb1322db1
commit 91ed32c256
3 changed files with 3 additions and 3 deletions
--- a/colossalai/shardformer/modeling/llama.py
+++ b/colossalai/shardformer/modeling/llama.py
@@ -571,7 +571,7 @@ def get_llama_flash_attention_forward(shard_config: ShardConfig, sp_mode=None, s
                key_states,
                value_states,
                sp_group,
-                tp_group,
+                tp_group=tp_group,
                **attention_mask,
                inner_ring_size=shard_config.inner_ring_size,
            )