Upload HieraForImageClassification

config.json

@@ -1,13 +1,10 @@
 {
-  "_out_features": [
-    "stage4"
-  ],
-  "_out_indices": [
-    4
-  ],
   "architectures": [
     "HieraForImageClassification"
   ],
+  "decoder_depth": null,
+  "decoder_hidden_size": null,
+  "decoder_num_heads": null,
   "depths": [
     1,
     2,
@@ -1021,12 +1018,11 @@
     "998": "ear, spike, capitulum",
     "999": "toilet tissue, toilet paper, bathroom tissue"
   },
-  "initial_num_heads": 1,
-  "initializer_range": 0.02,
-  "input_size": [
+  "image_size": [
     224,
     224
   ],
+  "initializer_range": 0.02,
   "label2id": {
     "Afghan hound, Afghan": 160,
     "African chameleon, Chamaeleo chamaeleon": 47,
@@ -2030,6 +2026,7 @@
   },
   "layer_norm_eps": 1e-06,
   "layer_norm_init": 1.0,
+  "mask_ratio": 0.6,
   "masked_unit_attention": [
     true,
     true,
@@ -2042,17 +2039,30 @@
   ],
   "mlp_ratio": 4.0,
   "model_type": "hiera",
+  "norm_pix_loss": true,
   "num_channels": 3,
-  "num_head_multiplier": 2.0,
+  "num_heads": [
+    1,
+    2,
+    4,
+    8
+  ],
+  "num_layers": 4,
   "num_query_pool": 3,
-  "patch_kernel": [
-    7,
-    7
+  "out_features": [
+    "stage4"
+  ],
+  "out_indices": [
+    4
   ],
   "patch_padding": [
     3,
     3
   ],
+  "patch_size": [
+    7,
+    7
+  ],
   "patch_stride": [
     4,
     4
@@ -2061,7 +2071,6 @@
     2,
     2
   ],
-  "sep_pos_embed": false,
   "stage_names": [
     "stem",
     "stage1",
@@ -2070,5 +2079,6 @@
     "stage4"
   ],
   "torch_dtype": "float32",
-  "transformers_version": "4.40.0.dev0"
+  "transformers_version": "4.41.0.dev0",
+  "use_separate_position_embedding": false
 }