Add model from convert_rt_detr_original_pytorch_checkpoint_to_pytorch.py

config.json

@@ -1,5 +1,9 @@
 {
+  "_name_or_path": "rafaelpadilla/porting_rt_detr",
   "act_encoder": "silu",
+  "architectures": [
+    "RTDetrModel"
+  ],
   "backbone_config": {
     "backbone": "resnet50d",
     "freeze_batch_norm_2d": true,
@@ -41,6 +45,14 @@
   "id2label": {
     "0": "person",
     "1": "bicycle",
+    "2": "car",
+    "3": "motorcycle",
+    "4": "airplane",
+    "5": "bus",
+    "6": "train",
+    "7": "truck",
+    "8": "boat",
+    "9": "traffic light",
     "10": "fire hydrant",
     "11": "stop sign",
     "12": "parking meter",
@@ -51,7 +63,6 @@
     "17": "horse",
     "18": "sheep",
     "19": "cow",
-    "2": "car",
     "20": "elephant",
     "21": "bear",
     "22": "zebra",
@@ -62,7 +73,6 @@
     "27": "tie",
     "28": "suitcase",
     "29": "frisbee",
-    "3": "motorcycle",
     "30": "skis",
     "31": "snowboard",
     "32": "sports ball",
@@ -73,7 +83,6 @@
     "37": "surfboard",
     "38": "tennis racket",
     "39": "bottle",
-    "4": "airplane",
     "40": "wine glass",
     "41": "cup",
     "42": "fork",
@@ -84,7 +93,6 @@
     "47": "apple",
     "48": "sandwich",
     "49": "orange",
-    "5": "bus",
     "50": "broccoli",
     "51": "carrot",
     "52": "hot dog",
@@ -95,7 +103,6 @@
     "57": "couch",
     "58": "potted plant",
     "59": "bed",
-    "6": "train",
     "60": "dining table",
     "61": "toilet",
     "62": "tv",
@@ -106,7 +113,6 @@
     "67": "cell phone",
     "68": "microwave",
     "69": "oven",
-    "7": "truck",
     "70": "toaster",
     "71": "sink",
     "72": "refrigerator",
@@ -116,9 +122,7 @@
     "76": "scissors",
     "77": "teddy bear",
     "78": "hair drier",
-    "79": "toothbrush",
-    "8": "boat",
-    "9": "traffic light"
+    "79": "toothbrush"
   },
   "image_size": [
     640,
@@ -217,6 +221,7 @@
   "matcher_giou_cost": 2.0,
   "model_type": "rt_detr",
   "normalize_before": false,
+  "num_classes": 80,
   "num_decoder_layers": 6,
   "num_decoder_points": 4,
   "num_denoising": 100,
@@ -224,8 +229,12 @@
   "num_levels": 3,
   "num_queries": 300,
   "pe_temperature": 10000,
+  "torch_dtype": "float32",
   "transformers_version": "4.36.0.dev0",
   "use_aux_loss": true,
+  "use_encoder_idx": [
+    2
+  ],
   "use_focal_loss": true,
   "weight_loss_bbox": 5.0,
   "weight_loss_giou": 2.0,