Update tiny models for DeepseekOcr2Model

config.json

@@ -0,0 +1,144 @@
+{
+  "architectures": [
+    "DeepseekOcr2Model"
+  ],
+  "dtype": "float32",
+  "image_token_id": 1,
+  "model_type": "deepseek_ocr2",
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 0,
+    "eos_token_id": 1,
+    "head_dim": 32,
+    "hidden_act": "silu",
+    "hidden_size": 128,
+    "initializer_range": 0.02,
+    "intermediate_size": 256,
+    "max_position_embeddings": 512,
+    "mlp_bias": false,
+    "mlp_layer_types": [
+      "dense",
+      "sparse"
+    ],
+    "model_type": "deepseek_ocr2_text",
+    "moe_intermediate_size": 64,
+    "n_group": null,
+    "n_routed_experts": 8,
+    "n_shared_experts": 1,
+    "num_attention_heads": 4,
+    "num_experts_per_tok": 2,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 4,
+    "pad_token_id": 2,
+    "pretraining_tp": 1,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "routed_scaling_factor": 1.0,
+    "tie_word_embeddings": false,
+    "topk_group": null,
+    "topk_method": "greedy",
+    "use_cache": true,
+    "vocab_size": 128827
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.16.0.dev0",
+  "vision_config": {
+    "encoder_config": {
+      "_name_or_path": "",
+      "architectures": null,
+      "attention_dropout": 0.0,
+      "bos_token_id": null,
+      "chunk_size_feed_forward": 0,
+      "dtype": null,
+      "eos_token_id": null,
+      "hidden_act": "silu",
+      "hidden_size": 64,
+      "id2label": {
+        "0": "LABEL_0",
+        "1": "LABEL_1"
+      },
+      "initializer_range": 0.02,
+      "intermediate_size": 128,
+      "is_encoder_decoder": false,
+      "label2id": {
+        "LABEL_0": 0,
+        "LABEL_1": 1
+      },
+      "layer_types": [
+        "full_attention",
+        "full_attention"
+      ],
+      "max_position_embeddings": 512,
+      "max_window_layers": 28,
+      "model_type": "deepseek_ocr2_encoder",
+      "num_attention_heads": 4,
+      "num_hidden_layers": 2,
+      "num_key_value_heads": 4,
+      "output_attentions": false,
+      "output_hidden_states": false,
+      "pad_token_id": null,
+      "problem_type": null,
+      "return_dict": true,
+      "rms_norm_eps": 1.0,
+      "rope_parameters": {
+        "rope_theta": 10000.0,
+        "rope_type": "default"
+      },
+      "sliding_window": null,
+      "tie_word_embeddings": false,
+      "use_cache": true,
+      "use_sliding_window": false,
+      "vocab_size": 151936
+    },
+    "model_type": "deepseek_ocr2_vision",
+    "sam_config": {
+      "_name_or_path": "",
+      "architectures": null,
+      "attention_dropout": 0.0,
+      "chunk_size_feed_forward": 0,
+      "downsample_channels": [
+        32,
+        64
+      ],
+      "dtype": null,
+      "global_attn_indexes": [
+        1
+      ],
+      "hidden_act": "gelu",
+      "hidden_size": 32,
+      "id2label": {
+        "0": "LABEL_0",
+        "1": "LABEL_1"
+      },
+      "image_size": 16,
+      "initializer_range": 1e-10,
+      "is_encoder_decoder": false,
+      "label2id": {
+        "LABEL_0": 0,
+        "LABEL_1": 1
+      },
+      "layer_norm_eps": 1e-06,
+      "mlp_dim": 128,
+      "mlp_ratio": 4.0,
+      "model_type": "deepseek_ocr2_sam_vision_model",
+      "num_attention_heads": 4,
+      "num_channels": 3,
+      "num_hidden_layers": 2,
+      "output_attentions": false,
+      "output_channels": 16,
+      "output_hidden_states": false,
+      "patch_size": 2,
+      "problem_type": null,
+      "qkv_bias": true,
+      "return_dict": true,
+      "scale": 16,
+      "use_abs_pos": true,
+      "use_rel_pos": true,
+      "window_size": 4
+    }
+  }
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5748d71f20abe053b5d0c85152dba9334b641172d565a89df9e762e8c48f0603
+size 68463800

preprocessor_config.json

@@ -0,0 +1,32 @@
+{
+  "background_color": [
+    127,
+    127,
+    127
+  ],
+  "crop_to_patches": true,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "DeepseekOcr2ImageProcessor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "max_patches": 6,
+  "min_patches": 2,
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 1024,
+    "width": 1024
+  },
+  "tile_size": 768
+}

tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<｜begin▁of▁sentence｜>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<｜end▁of▁sentence｜>",
+  "is_local": true,
+  "legacy": true,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<｜▁pad▁｜>",
+  "processor_class": "DeepseekOcr2Processor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "use_default_system_prompt": false
+}