Update tiny models for Sam2Model

config.json

@@ -0,0 +1,146 @@
+{
+  "architectures": [
+    "Sam2Model"
+  ],
+  "dtype": "float32",
+  "image_size": 128,
+  "initializer_range": 0.02,
+  "mask_decoder_config": {
+    "attention_downsample_rate": 2,
+    "dynamic_multimask_stability_delta": 0.05,
+    "dynamic_multimask_stability_thresh": 0.98,
+    "dynamic_multimask_via_stability": true,
+    "hidden_act": "relu",
+    "hidden_size": 32,
+    "iou_head_depth": 3,
+    "iou_head_hidden_dim": 32,
+    "mlp_dim": 64,
+    "model_type": "",
+    "num_attention_heads": 4,
+    "num_hidden_layers": 2,
+    "num_multimask_outputs": 3
+  },
+  "mask_downsampler_embed_dim": 32,
+  "memory_attention_feed_forward_hidden_size": 32,
+  "memory_attention_hidden_size": 12,
+  "memory_attention_num_layers": 1,
+  "memory_encoder_hidden_size": 32,
+  "memory_fuser_embed_dim": 32,
+  "model_type": "sam2",
+  "prompt_encoder_config": {
+    "hidden_act": "gelu",
+    "hidden_size": 32,
+    "image_size": 128,
+    "layer_norm_eps": 1e-06,
+    "mask_input_channels": 8,
+    "model_type": "",
+    "num_point_embeddings": 4,
+    "patch_size": 16,
+    "scale": 1
+  },
+  "transformers_version": "5.16.0.dev0",
+  "vision_config": {
+    "backbone_channel_list": [
+      96,
+      48,
+      24,
+      12
+    ],
+    "backbone_config": {
+      "_name_or_path": "",
+      "architectures": null,
+      "blocks_per_stage": [
+        1,
+        2,
+        7,
+        2
+      ],
+      "chunk_size_feed_forward": 0,
+      "dtype": null,
+      "embed_dim_per_stage": [
+        12,
+        24,
+        48,
+        96
+      ],
+      "global_attention_blocks": [
+        5,
+        7,
+        9
+      ],
+      "hidden_act": "gelu",
+      "hidden_size": 12,
+      "id2label": {
+        "0": "LABEL_0",
+        "1": "LABEL_1"
+      },
+      "image_size": 128,
+      "initializer_range": 0.02,
+      "is_encoder_decoder": false,
+      "label2id": {
+        "LABEL_0": 0,
+        "LABEL_1": 1
+      },
+      "layer_norm_eps": 1e-06,
+      "mlp_ratio": 4.0,
+      "model_type": "sam2_hiera_det_model",
+      "num_attention_heads": 1,
+      "num_attention_heads_per_stage": [
+        1,
+        2,
+        4,
+        8
+      ],
+      "num_channels": 3,
+      "num_query_pool_stages": 3,
+      "output_attentions": false,
+      "output_hidden_states": false,
+      "patch_kernel_size": 7,
+      "patch_padding": 3,
+      "patch_stride": 4,
+      "problem_type": null,
+      "query_stride": [
+        2,
+        2
+      ],
+      "return_dict": true,
+      "window_positional_embedding_background_size": [
+        7,
+        7
+      ],
+      "window_size_per_stage": [
+        8,
+        4,
+        14,
+        7
+      ]
+    },
+    "backbone_feature_sizes": [
+      [
+        32,
+        32
+      ],
+      [
+        16,
+        16
+      ],
+      [
+        8,
+        8
+      ]
+    ],
+    "fpn_hidden_size": 32,
+    "fpn_kernel_size": 1,
+    "fpn_padding": 0,
+    "fpn_stride": 1,
+    "fpn_top_down_levels": [
+      2,
+      3
+    ],
+    "hidden_act": "gelu",
+    "initializer_range": 0.02,
+    "layer_norm_eps": 1e-06,
+    "model_type": "sam2_vision_model",
+    "num_feature_levels": 3
+  }
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2207f6d6aae27fed968c96cb80d4efabb196b8d5e89c713d2b0c2b08a2d51622
+size 1950516

preprocessor_config.json

@@ -0,0 +1,31 @@
+{
+  "crop_size": {
+    "height": 128,
+    "width": 128
+  },
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "image_processor_type": "Sam2ImageProcessor",
+  "image_std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "mask_size": {
+    "height": 256,
+    "width": 256
+  },
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 128,
+    "width": 128
+  }
+}

video_preprocessor_config.json

@@ -0,0 +1,28 @@
+{
+  "crop_size": {
+    "height": 128,
+    "width": 128
+  },
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "image_std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
+  "size": {
+    "height": 128,
+    "width": 128
+  },
+  "video_processor_type": "Sam2VideoVideoProcessor"
+}