Training in progress, epoch 1

config.json

@@ -1,29 +1,68 @@
 {
-  "_name_or_path": "SenseTime/deformable-detr",
+  "_name_or_path": "jozhang97/deta-swin-large-o365",
   "activation_dropout": 0.0,
   "activation_function": "relu",
   "architectures": [
-    "DeformableDetrForObjectDetection"
+    "DetaForObjectDetection"
   ],
+  "assign_first_stage": true,
   "attention_dropout": 0.0,
   "auxiliary_loss": false,
-  "backbone": "resnet50",
-  "backbone_config": null,
+  "backbone_config": {
+    "attention_probs_dropout_prob": 0.0,
+    "depths": [
+      2,
+      2,
+      18,
+      2
+    ],
+    "drop_path_rate": 0.1,
+    "embed_dim": 192,
+    "encoder_stride": 32,
+    "hidden_act": "gelu",
+    "hidden_dropout_prob": 0.0,
+    "hidden_size": 1536,
+    "image_size": 224,
+    "initializer_range": 0.02,
+    "layer_norm_eps": 1e-05,
+    "mlp_ratio": 4.0,
+    "model_type": "swin",
+    "num_heads": [
+      6,
+      12,
+      24,
+      48
+    ],
+    "num_layers": 4,
+    "out_features": [
+      "stage2",
+      "stage3",
+      "stage4"
+    ],
+    "out_indices": [
+      2,
+      3,
+      4
+    ],
+    "patch_size": 4,
+    "path_norm": true,
+    "qkv_bias": true,
+    "use_absolute_embeddings": false,
+    "window_size": 12
+  },
   "bbox_cost": 5,
   "bbox_loss_coefficient": 5,
   "class_cost": 1,
   "d_model": 256,
   "decoder_attention_heads": 8,
-  "decoder_ffn_dim": 1024,
+  "decoder_ffn_dim": 2048,
   "decoder_layerdrop": 0.0,
   "decoder_layers": 6,
   "decoder_n_points": 4,
   "dice_loss_coefficient": 1,
-  "dilation": false,
-  "disable_custom_kernels": false,
   "dropout": 0.1,
   "encoder_attention_heads": 8,
-  "encoder_ffn_dim": 1024,
+  "encoder_ffn_dim": 2048,
   "encoder_layerdrop": 0.0,
   "encoder_layers": 6,
   "encoder_n_points": 4,
@@ -69,17 +108,14 @@
     "tetrapack": 13
   },
   "mask_loss_coefficient": 1,
-  "max_position_embeddings": 1024,
-  "model_type": "deformable_detr",
-  "num_channels": 3,
-  "num_feature_levels": 4,
-  "num_queries": 300,
+  "max_position_embeddings": 2048,
+  "model_type": "deta",
+  "num_feature_levels": 5,
+  "num_queries": 900,
   "position_embedding_type": "sine",
   "torch_dtype": "float32",
   "transformers_version": "4.32.1",
-  "two_stage": false,
+  "two_stage": true,
   "two_stage_num_proposals": 300,
-  "use_pretrained_backbone": true,
-  "use_timm_backbone": true,
-  "with_box_refine": false
+  "with_box_refine": true
 }

preprocessor_config.json

@@ -9,7 +9,7 @@
     0.456,
     0.406
   ],
-  "image_processor_type": "DeformableDetrImageProcessor",
+  "image_processor_type": "DetaImageProcessor",
   "image_std": [
     0.229,
     0.224,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69381d284cf3e2b05cfc00474d70b5908b828985f1d7f65dd088895cebabf818
-size 160818833
+oid sha256:16d32a857be405e5d0ec2431472be5ae8fd8278bb09a57a523fc9342f9b502a3
+size 878823725

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e089b15377bd4ec5d4b4ebca9ba6ab6345c4d39f32cd696b4b4d6217a6c93d5f
-size 4091
+oid sha256:15a8f17d6f9900b4198920113e77c1faed4705d5ce3d6dd0a4928158d4d9f1fc
+size 4219