Add projection dim to text and vision model configs for CLIPVisionModelWithProjection and CLIPTextModelWithProjection support

config.json

@@ -61,6 +61,7 @@
     "pad_token_id": 1,
     "prefix": null,
     "problem_type": null,
+    "projection_dim": 1280,
     "pruned_heads": {},
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
@@ -88,7 +89,8 @@
     "hidden_size": 1280,
     "intermediate_size": 5120,
     "num_attention_heads": 20,
-    "num_hidden_layers": 32
+    "num_hidden_layers": 32,
+    "projection_dim": 1280
   },
   "torch_dtype": "float32",
   "transformers_version": null,
@@ -148,6 +150,7 @@
     "patch_size": 14,
     "prefix": null,
     "problem_type": null,
+    "projection_dim": 1280,
     "pruned_heads": {},
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
@@ -175,6 +178,7 @@
     "intermediate_size": 8192,
     "num_attention_heads": 16,
     "num_hidden_layers": 48,
-    "patch_size": 14
+    "patch_size": 14,
+    "projection_dim": 1280,
   }
 }