Upload model

config.json

@@ -1,89 +1,15 @@
 {
-  "add_all_frames_to_correct_as_cond": false,
   "architectures": [
     "Sam2Model"
   ],
-  "backbone_stride": 16,
   "binarize_mask_from_pts_for_mem_enc": true,
   "enable_occlusion_spatial_embedding": true,
   "enable_temporal_pos_encoding_for_object_pointers": true,
   "fill_hole_area": 8,
-  "image_encoder_config": {
-    "backbone_channel_list": [
-      768,
-      384,
-      192,
-      96
-    ],
-    "backbone_feature_sizes": [
-      [
-        256,
-        256
-      ],
-      [
-        128,
-        128
-      ],
-      [
-        64,
-        64
-      ]
-    ],
-    "dim_mul": 2.0,
-    "drop_path_rate": 0.0,
-    "fpn_hidden_size": 256,
-    "fpn_interpolation_mode": "nearest",
-    "fpn_kernel_size": 1,
-    "fpn_padding": 0,
-    "fpn_stride": 1,
-    "fpn_top_down_levels": [
-      2,
-      3
-    ],
-    "fuse_type": "sum",
-    "global_attention_blocks": [
-      5,
-      7,
-      9
-    ],
-    "head_mul": 2.0,
-    "hidden_act": "gelu",
-    "hidden_size": 96,
-    "image_size": 1024,
-    "layer_norm_eps": 1e-06,
-    "model_type": "",
-    "num_channels": 3,
-    "num_feature_levels": 3,
-    "num_heads": 1,
-    "patch_kernel_size": 7,
-    "patch_padding": 3,
-    "patch_stride": 4,
-    "q_pool": 3,
-    "q_stride": [
-      2,
-      2
-    ],
-    "stages": [
-      1,
-      2,
-      7,
-      2
-    ],
-    "window_positional_embedding_background_size": [
-      7,
-      7
-    ],
-    "window_spec": [
-      8,
-      4,
-      14,
-      7
-    ]
-  },
   "image_size": 1024,
   "initializer_range": 0.02,
-  "iou_prediction_use_sigmoid": true,
   "mask_decoder_config": {
+    "attention_downsample_rate": 2,
     "dynamic_multimask_stability_delta": 0.05,
     "dynamic_multimask_stability_thresh": 0.98,
     "dynamic_multimask_via_stability": true,
@@ -92,37 +18,31 @@
     "hidden_size": 256,
     "iou_head_depth": 3,
     "iou_head_hidden_dim": 256,
-    "iou_prediction_use_sigmoid": true,
+    "mlp_dim": 2048,
     "model_type": "",
+    "num_attention_heads": 8,
+    "num_hidden_layers": 2,
     "num_multimask_outputs": 3,
-    "two_way_transformer_activation": "relu",
-    "two_way_transformer_attention_downsample_rate": 2,
-    "two_way_transformer_depth": 2,
-    "two_way_transformer_embedding_dim": 256,
-    "two_way_transformer_mlp_dim": 2048,
-    "two_way_transformer_num_heads": 8,
-    "use_multimask_token_for_object_pointer": true
+    "two_way_transformer_activation": "relu"
   },
-  "max_cond_frames_in_attn": -1,
   "max_object_pointers_in_encoder": 16,
   "memory_attention_config": {
     "apply_pe_at_cross_attn_keys": true,
     "apply_pe_at_cross_attn_queries": false,
     "apply_pe_at_self_attn": false,
+    "attention_downsample_rate": 1,
     "dim_feedforward": 2048,
     "dropout": 0.1,
     "hidden_act": "relu",
     "hidden_size": 256,
     "model_type": "",
+    "num_attention_heads": 1,
     "num_layers": 4,
-    "rope_downsample_rate": 1,
     "rope_dropout": 0.1,
-    "rope_embedding_dim": 256,
     "rope_feat_sizes": [
-      32,
-      32
+      64,
+      64
     ],
-    "rope_num_heads": 1,
     "rope_theta": 10000
   },
   "memory_encoder_config": {
@@ -143,7 +63,6 @@
     "model_type": "",
     "output_channels": 64
   },
-  "memory_temporal_stride_for_eval": 1,
   "model_type": "sam2",
   "multimask_max_pt_num": 1,
   "multimask_min_pt_num": 0,
@@ -157,6 +76,7 @@
   "prompt_encoder_config": {
     "hidden_act": "gelu",
     "hidden_size": 256,
+    "image_embedding_size": 64,
     "image_size": 1024,
     "layer_norm_eps": 1e-06,
     "mask_input_channels": 16,
@@ -168,8 +88,78 @@
   "sigmoid_bias_for_mem_enc": -10.0,
   "sigmoid_scale_for_mem_enc": 20.0,
   "torch_dtype": "float32",
-  "transformers_version": "4.53.0.dev0",
-  "use_mask_input_as_output_without_sam": true,
-  "use_multimask_token_for_object_pointer": true,
-  "use_object_pointers_in_encoder": true
+  "transformers_version": "4.54.0.dev0",
+  "vision_config": {
+    "backbone_channel_list": [
+      768,
+      384,
+      192,
+      96
+    ],
+    "backbone_feature_sizes": [
+      [
+        256,
+        256
+      ],
+      [
+        128,
+        128
+      ],
+      [
+        64,
+        64
+      ]
+    ],
+    "dim_mul": 2.0,
+    "drop_path_rate": 0.0,
+    "fpn_hidden_size": 256,
+    "fpn_interpolation_mode": "nearest",
+    "fpn_kernel_size": 1,
+    "fpn_padding": 0,
+    "fpn_stride": 1,
+    "fpn_top_down_levels": [
+      2,
+      3
+    ],
+    "fuse_type": "sum",
+    "global_attention_blocks": [
+      5,
+      7,
+      9
+    ],
+    "head_mul": 2.0,
+    "hidden_act": "gelu",
+    "hidden_size": 96,
+    "image_size": 1024,
+    "initializer_range": 0.02,
+    "layer_norm_eps": 1e-06,
+    "model_type": "sam2_vision_model",
+    "num_attention_heads": 1,
+    "num_channels": 3,
+    "num_feature_levels": 3,
+    "patch_kernel_size": 7,
+    "patch_padding": 3,
+    "patch_stride": 4,
+    "q_pool": 3,
+    "q_stride": [
+      2,
+      2
+    ],
+    "stages": [
+      1,
+      2,
+      7,
+      2
+    ],
+    "window_positional_embedding_background_size": [
+      7,
+      7
+    ],
+    "window_spec": [
+      8,
+      4,
+      14,
+      7
+    ]
+  }
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:030a362dae8f614f7aa8afc76097ecacd3a22b00baf76f8f1d92819a5eedf3f0
-size 155906184
+oid sha256:ff4f3e742dfcaff5b6073362fdd9b74a48cb673b3cb6637d9d9d5cd2b3f5bfef
+size 155906944