[ConvNeXT] Fix drop_path_rate (huggingface#17280)

* Fix drop_path_rate

* Fix TF's drop path rate

Author: NielsRogge

src/transformers/models/convnext/modeling_convnext.py

@@ -209,8 +209,9 @@ class ConvNextEncoder(nn.Module):
     def __init__(self, config):
         super().__init__()
         self.stages = nn.ModuleList()
-        drop_path_rates = [x.item() for x in torch.linspace(0, config.drop_path_rate, sum(config.depths))]
-        cur = 0
+        drop_path_rates = [
+            x.tolist() for x in torch.linspace(0, config.drop_path_rate, sum(config.depths)).split(config.depths)
+        ]
         prev_chs = config.hidden_sizes[0]
         for i in range(config.num_stages):
             out_chs = config.hidden_sizes[i]
@@ -220,10 +221,9 @@ def __init__(self, config):
                 out_channels=out_chs,
                 stride=2 if i > 0 else 1,
                 depth=config.depths[i],
-                drop_path_rates=drop_path_rates[cur],
+                drop_path_rates=drop_path_rates[i],
             )
             self.stages.append(stage)
-            cur += config.depths[i]
             prev_chs = out_chs
 
     def forward(

src/transformers/models/convnext/modeling_tf_convnext.py

@@ -235,8 +235,9 @@ class TFConvNextEncoder(tf.keras.layers.Layer):
     def __init__(self, config, **kwargs):
         super().__init__(**kwargs)
         self.stages = []
-        drop_path_rates = [x for x in tf.linspace(0.0, config.drop_path_rate, sum(config.depths))]
-        cur = 0
+        drop_path_rates = tf.linspace(0.0, config.drop_path_rate, sum(config.depths))
+        drop_path_rates = tf.split(drop_path_rates, config.depths)
+        drop_path_rates = [x.numpy().tolist() for x in drop_path_rates]
         prev_chs = config.hidden_sizes[0]
         for i in range(config.num_stages):
             out_chs = config.hidden_sizes[i]
@@ -246,11 +247,10 @@ def __init__(self, config, **kwargs):
                 out_channels=out_chs,
                 stride=2 if i > 0 else 1,
                 depth=config.depths[i],
-                drop_path_rates=drop_path_rates[cur],
+                drop_path_rates=drop_path_rates[i],
                 name=f"stages.{i}",
             )
             self.stages.append(stage)
-            cur += config.depths[i]
             prev_chs = out_chs
 
     def call(self, hidden_states, output_hidden_states=False, return_dict=True):