Update config.json

config.json

@@ -1,40 +1,13 @@
 {
   "_name_or_path": "microsoft/Florence-2-base",
-  "architectures": [
-    "Florence2ForConditionalGeneration"
-  ],
+  "architectures": ["Florence2ForConditionalGeneration"],
   "vision_config": {
     "model_type": "davit",
-    "depths": [
-      1,
-      1,
-      9,
-      1
-    ],
-    "dim_embed": [
-      128,
-      256,
-      512,
-      1024
-    ],
-    "num_heads": [
-      4,
-      8,
-      16,
-      32
-    ],
-    "patch_size": [
-      7,
-      3,
-      3,
-      3
-    ],
-    "patch_stride": [
-      4,
-      2,
-      2,
-      2
-    ],
+    "depths": [1, 1, 9, 1],
+    "dim_embed": [128, 256, 512, 1024],
+    "num_heads": [4, 8, 16, 32],
+    "patch_size": [7, 3, 3, 3],
+    "patch_stride": [4, 2, 2, 2],
     "window_size": 12
   },
   "auto_map": {
@@ -49,7 +22,6 @@
   "pad_token_id": 1,
   "projection_dim": 768,
   "text_config": {
-    "_name_or_path": "",
     "activation_dropout": 0.1,
     "activation_function": "gelu",
     "add_bias_logits": false,
@@ -82,52 +54,5 @@
     "vocab_size": 51289
   },
   "torch_dtype": "float32",
-  "transformers_version": "4.41.2",
-  "vision_config": {
-    "model_type": "davit",
-    "depths": [
-      1,
-      1,
-      9,
-      1
-    ],
-    "dim_embed": [
-      128,
-      256,
-      512,
-      1024
-    ],
-    "num_heads": [
-      4,
-      8,
-      16,
-      32
-    ],
-    "patch_size": [
-      7,
-      3,
-      3,
-      3
-    ],
-    "patch_stride": [
-      4,
-      2,
-      2,
-      2
-    ],
-    "window_size": 12,
-    "image_feature_source": [
-      "spatial_avg_pool",
-      "temporal_avg_pool"
-    ],
-    "image_pos_embed": {
-      "max_pos_embeddings": 50,
-      "type": "learned_abs_2d"
-    },
-    "visual_temporal_embedding": {
-      "max_temporal_embeddings": 100,
-      "type": "COSINE"
-    }
-  },
-  "vocab_size": 51289
+  "transformers_version": "4.41.2"
 }