ConvBERT fix torch <> tf weights conversion (#10314)

* convbert conversion test

* fin

* fin

* fin

* clean up tf<->pt conversion

* remove from_pt

Co-authored-by: patrickvonplaten <patrick.v.platen@gmail.com>

src/transformers/models/convbert/convert_convbert_original_tf1_checkpoint_to_pytorch_and_tf2.py

@@ -16,7 +16,7 @@
 
 import argparse
 
-from transformers import ConvBertConfig, ConvBertModel, load_tf_weights_in_convbert
+from transformers import ConvBertConfig, ConvBertModel, TFConvBertModel, load_tf_weights_in_convbert
 from transformers.utils import logging
 
 
@@ -30,6 +30,9 @@ def convert_orig_tf1_checkpoint_to_pytorch(tf_checkpoint_path, convbert_config_f
     model = load_tf_weights_in_convbert(model, conf, tf_checkpoint_path)
     model.save_pretrained(pytorch_dump_path)
 
+    tf_model = TFConvBertModel.from_pretrained(pytorch_dump_path, from_pt=True)
+    tf_model.save_pretrained(pytorch_dump_path)
+
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()

src/transformers/models/convbert/modeling_tf_convbert.py

@@ -343,7 +343,7 @@ class GroupedLinearLayer(tf.keras.layers.Layer):
     def build(self, input_shape):
         self.kernel = self.add_weight(
             "kernel",
-            shape=[self.num_groups, self.group_in_dim, self.group_out_dim],
+            shape=[self.group_out_dim, self.group_in_dim, self.num_groups],
             initializer=self.kernel_initializer,
             trainable=True,
         )
@@ -355,7 +355,7 @@ class GroupedLinearLayer(tf.keras.layers.Layer):
     def call(self, hidden_states):
         batch_size = shape_list(hidden_states)[0]
         x = tf.transpose(tf.reshape(hidden_states, [-1, self.num_groups, self.group_in_dim]), [1, 0, 2])
-        x = tf.matmul(x, self.kernel)
+        x = tf.matmul(x, tf.transpose(self.kernel, [2, 1, 0]))
         x = tf.transpose(x, [1, 0, 2])
         x = tf.reshape(x, [batch_size, -1, self.output_size])
         x = tf.nn.bias_add(value=x, bias=self.bias)