Upload 3 files

config.json

@@ -19,7 +19,7 @@
   "decoder_attention_heads": 8,
   "decoder_depth_residual": false,
   "decoder_ffn_dim": 256,
-  "decoder_layers": 1,
+  "decoder_layers": 3,
   "decoder_n_points": 4,
   "decoder_self_attn": false,
   "decoder_text_residual": false,
@@ -32,7 +32,7 @@
   "dropout": 0.1,
   "encoder_attention_heads": 8,
   "encoder_ffn_dim": 256,
-  "encoder_layers": 3,
+  "encoder_layers": 6,
   "encoder_n_points": 4,
   "focal_alpha": 0.25,
   "freeze_backbone": true,
@@ -69,7 +69,7 @@
   "num_depth_bins": 80,
   "num_feature_levels": 4,
   "num_queries": 1,
-  "num_text_output_layers": 1,
+  "num_text_output_layers": 6,
   "position_embedding_type": "sine",
   "pretrained_backbone_path": "pretrained-models/swin_large_patch4_window7_224/model.safetensors",
   "text_encoder_type": "pretrained-models/roberta-base",
@@ -78,8 +78,8 @@
   "two_stage": false,
   "use_dab": true,
   "use_pretrained_backbone": true,
-  "use_text_guided_adapter": true,
+  "use_text_guided_adapter": false,
   "use_timm_backbone": true,
-  "vl_encoder_type": "fusion",
+  "vl_encoder_type": "simple-bridge-tower",
   "with_box_refine": true
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eeae527ec76a06e746b913e8e4b4a984107567a03d67494cacdae822835aba3b
-size 1324513302
+oid sha256:b548993b0f92b4f4e1c18f2823f32c9b39cae8225277afa4b00a29447d940ea3
+size 1365911480