Training in progress, epoch 1

config.json

@@ -10,22 +10,22 @@
   "backbone": null,
   "backbone_config": {
     "depths": [
-      2,
-      2,
-      2,
-      2
+      3,
+      4,
+      6,
+      3
     ],
     "downsample_in_bottleneck": false,
     "downsample_in_first_stage": false,
     "embedding_size": 64,
     "hidden_act": "relu",
     "hidden_sizes": [
-      64,
-      128,
       256,
-      512
+      512,
+      1024,
+      2048
     ],
-    "layer_type": "basic",
+    "layer_type": "bottleneck",
     "model_type": "rt_detr_resnet",
     "num_channels": 3,
     "out_features": [
@@ -58,7 +58,7 @@
     256,
     256
   ],
-  "decoder_layers": 3,
+  "decoder_layers": 6,
   "decoder_n_points": 4,
   "disable_custom_kernels": true,
   "dropout": 0.0,
@@ -70,9 +70,9 @@
   "encoder_ffn_dim": 1024,
   "encoder_hidden_dim": 256,
   "encoder_in_channels": [
-    128,
-    256,
-    512
+    512,
+    1024,
+    2048
   ],
   "encoder_layers": 1,
   "eos_coefficient": 0.0001,
@@ -85,7 +85,7 @@
   "focal_loss_alpha": 0.75,
   "focal_loss_gamma": 2.0,
   "freeze_backbone_batch_norms": true,
-  "hidden_expansion": 0.5,
+  "hidden_expansion": 1.0,
   "id2label": {
     "0": "Ampulla of Vater",
     "1": "Angiectasia",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:30dc2910330587ce86620405fb1f683a87fe3766c6e4ca3e4ef5f8ecddd722ef
-size 80549768
+oid sha256:78afef3a8d6aa757553b33327117e1e374fb7b5c0b1c0700bf4013e1da311a5d
+size 171608676

runs/Jun15_10-48-53_itri-kiosk-pt/events.out.tfevents.1749955733.itri-kiosk-pt.3608161.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc343e49d462ca42503e75aa5ef5d4f866a5a6e485df9a48ca18b032931de586
+size 9713

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:96d0a419f781e287ca99f08ff0a4fdee0450dd687e8563e3be86598a241653fe
+oid sha256:24565bec045b2bce7359f63c2b061d6ac0dd3ed06fb5af3330bc3684ae50f655
 size 5713