Support Sharding Overlap

paddlenlp/trainer/trainer.py

@@ -1892,7 +1892,6 @@
                     optimizer._set_broadcast_overlap(True, model)
 
                 self.optimizer = optimizer
-
         # pure tesnor parallel mode, no pipeline_parallel, no sharding.
         if (
             not in_pipeline_parallel_mode
@@ -1908,6 +1907,21 @@
                 self.optimizer = mix_precision_utils.MixPrecisionOptimizer(self.optimizer)
             self.optimizer = fleet.distributed_optimizer(self.optimizer)
 
+        # stage1 has v1 and v2 version
+        if in_sharding_parallel_mode and ShardingOption.SHARD_OP in self.args.sharding:
+            if "split_param" in self.args.sharding_parallel_config:
+                if (
+                    hasattr(self.optimizer, "_set_all_gather_overlap_forward")
+                    and "enable_stage1_allgather_overlap" in self.args.sharding_parallel_config
+                ):
+                    self.optimizer._set_all_gather_overlap_forward(True, model)
+            else:
+                if (
+                    hasattr(self.optimizer, "_set_broadcast_overlap")
+                    and "enable_stage1_broadcast_overlap" in self.args.sharding_parallel_config
+                ):
+                    self.optimizer._set_broadcast_overlap(True, model)
+
         return model
 
     def _prepare_input(self, data: Union[paddle.Tensor, Any]) -> Union[paddle.Tensor, Any]:

paddlenlp/transformers/llama/modeling.py

@@ -115,11 +115,15 @@
         )
 
 
+def get_use_casual_mask():
+    """Get the value of the 'USE_CASUAL_MASK' environment variable."""
+    return os.getenv("USE_CASUAL_MASK", "False")
+
+
 def build_alibi_tensor(
     bool_attention_mask: Tensor, num_heads: int, dtype: paddle.dtype, tensor_parallel_degree=1
 ) -> Tensor:
-    attention_mask = bool_attention_mask.astype("float32")
-    batch_size, seq_length = attention_mask.shape[0], attention_mask.shape[-1]
+    batch_size, seq_length = bool_attention_mask.shape[0], bool_attention_mask.shape[-1]
     slopes = paddle.to_tensor(_get_interleave(num_heads), dtype="float32")
     alibi = slopes.unsqueeze(axis=[1, 2]) * paddle.arange(seq_length, dtype="float32").unsqueeze(axis=[0, 1]).expand(
         [num_heads, -1, -1]
@@ -307,7 +311,7 @@
 
 def _make_causal_mask(input_ids_shape, past_key_values_length):
     """
-    Make causal mask used for self-attention
+    Make casual mask used for self-attention
     """
     batch_size, target_length = input_ids_shape  # target_length: seq_len
 
@@ -1533,12 +1537,22 @@
         if position_ids is None:
             position_ids = paddle.arange(seq_length, dtype="int64").expand((batch_size, seq_length))
 
-        attention_mask = self._prepare_decoder_attention_mask(
-            attention_mask, (batch_size, seq_length), cache_length, inputs_embeds.dtype
-        )  # [bs, 1, seq_len, seq_len]
+        use_casual_mask = get_use_casual_mask()
+
+        if use_casual_mask:
+            attention_mask = None
+        else:
+            attention_mask = self._prepare_decoder_attention_mask(
+                attention_mask, (batch_size, seq_length), cache_length, inputs_embeds.dtype
+            )  # [bs, 1, seq_len, seq_len]
+
         is_casual = False
+
         if self.config.use_flash_attention and get_env_device() != "gcu":
-            is_casual = is_casual_mask(attention_mask)
+            if use_casual_mask:
+                is_casual = True
+            else:
+                is_casual = is_casual_mask(attention_mask)
             if get_env_device() != "npu":
                 if is_casual and alibi is None:
                     attention_mask = None