Tweak init for convnext block using maxxvit/coatnext.

timm/models/maxxvit.py

@@ -259,8 +259,6 @@ def _rw_max_cfg(
     # - mbconv expansion calculated from input instead of output chs
     # - mbconv shortcut and final 1x1 conv did not have a bias
     # - mbconv uses silu in timm, not gelu
-    # - avg pool with kernel_size=2 favoured downsampling (instead of maxpool for coat)
-    # - default to avg pool for mbconv downsample instead of 1x1 or dw conv
     # - expansion in attention block done via output proj, not input proj
     return dict(
         conv_cfg=MaxxVitConvCfg(
@@ -411,12 +409,6 @@ model_cfgs = dict(
             rel_pos_dim=384,  # was supposed to be 512, woops
         ),
     ),
-    coatnext_nano_rw_224=MaxxVitCfg(
-        embed_dim=(64, 128, 256, 512),
-        depths=(3, 4, 6, 3),
-        stem_width=(32, 64),
-        **_next_cfg(),
-    ),
     coatnet_nano_cc_224=MaxxVitCfg(
         embed_dim=(64, 128, 256, 512),
         depths=(3, 4, 6, 3),
@@ -424,6 +416,13 @@ model_cfgs = dict(
         block_type=('C', 'C', ('C', 'T'), ('C', 'T')),
         **_rw_coat_cfg(),
     ),
+    coatnext_nano_rw_224=MaxxVitCfg(
+        embed_dim=(64, 128, 256, 512),
+        depths=(3, 4, 6, 3),
+        stem_width=(32, 64),
+        weight_init='normal',
+        **_next_cfg(),
+    ),
 
     # Trying to be like the CoAtNet paper configs
     coatnet_0_224=MaxxVitCfg(
@@ -498,6 +497,7 @@ model_cfgs = dict(
         depths=(1, 2, 3, 1),
         block_type=('M',) * 4,
         stem_width=(32, 64),
+        weight_init='normal',
         **_next_cfg(window_size=8),
     ),