Update config

config.json

@@ -9,10 +9,10 @@
   "d_latents": 1280,
   "d_model": 768,
   "hidden_act": "gelu",
-  "hidden_dropout_prob": 0.1,
   "image_size": 56,
   "initializer_range": 0.02,
   "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 2048,
   "model_type": "perceiver",
   "num_blocks": 1,
   "num_cross_attention_heads": 8,
@@ -20,11 +20,15 @@
   "num_latents": 256,
   "num_self_attends_per_block": 26,
   "num_self_attention_heads": 8,
-  "position_embedding_init_scale": 0.02,
+  "output_shape": [
+    1,
+    16,
+    224,
+    224
+  ],
   "qk_channels": 256,
   "samples_per_patch": 16,
   "self_attention_widening_factor": 1,
-  "seq_len": 2048,
   "torch_dtype": "float32",
   "train_size": [
     368,