{
  "id": "dinov2-base",
  "identity": {
    "family": "dinov2",
    "display_name": "DINOv2 Base",
    "hub_id": "facebook/dinov2-base",
    "api_name": null,
    "revision": null,
    "variants": [],
    "license": "Apache-2.0",
    "commercial_use": true,
    "commercial_use_note": "Apache-2.0 weights and code; no use restrictions.",
    "release_date": "2023-04",
    "paper_url": "https://arxiv.org/abs/2304.07193",
    "repo_url": "https://github.com/facebookresearch/dinov2"
  },
  "task": {
    "category": "embedding",
    "modalities_in": [
      "image"
    ],
    "modalities_out": [
      "vector"
    ],
    "best_at": "Image-to-image similarity on visual structure (layout, near-duplicates, the same object or scene) with no text in the loop.",
    "worst_at": "Text queries (no text tower, so text-to-image search is impossible); semantic matches across different-looking images, where SigLIP is better."
  },
  "io": {
    "input": {
      "image": {
        "resolution": "resize shortest side 256, center crop 224",
        "color": "RGB",
        "normalize": "ImageNet mean/std"
      }
    },
    "output": {
      "vector": {
        "dims": 768,
        "distance": "cosine",
        "type": "dense"
      }
    }
  },
  "mixpeek_wiring": {
    "extractors": [
      {
        "extractor": "web_scraper@v1",
        "selected_by": "generate_structure_embeddings=true (default true); image documents only"
      }
    ],
    "feature_uris": [
      "mixpeek://web_scraper@v1/facebook__dinov2_base"
    ],
    "vector_indexes": [
      "facebook__dinov2_base"
    ],
    "retriever_stages": [
      "feature_search"
    ],
    "composes_with": [
      "web_scraper: DINOv2 structure vector beside SigLIP image vectors for the same page images"
    ]
  },
  "run": {
    "status": "served",
    "status_reason": "Ray Serve app in the inference registry (image queries); web_scraper loads facebook/dinov2-base in-process (DocTypeStructureEmbedder).",
    "serve_apps": [
      "facebook__dinov2_base"
    ],
    "runtime": "transformers AutoModel + AutoImageProcessor (Dinov2Model)",
    "weights_format": "safetensors",
    "trust_remote_code": false,
    "size_on_disk_mb": 346,
    "params_millions": 86.6,
    "preprocessing": "AutoImageProcessor defaults for the checkpoint (resize, center crop 224, ImageNet normalization).",
    "postprocessing": "CLS token of last_hidden_state, L2-normalized (768 dims).",
    "serving_shape": "both",
    "batch_size": null,
    "serve_batch": null,
    "scale_to_zero": {
      "min_replicas": 0,
      "max_replicas": 4,
      "idle_timeout_s": 300,
      "memory_gb": 1,
      "num_cpus": 0.5,
      "num_gpus": {
        "expr": "hardware_config.gpu_fraction if hardware_config.gpu_concurrency > 0 else 0"
      },
      "cold_start_s": {
        "warm_node_s": null,
        "new_node_s": null,
        "source": null
      },
      "cpu_fallback": true
    }
  },
  "cost": {
    "unmeasured": true,
    "measured": null,
    "note": "No committed throughput or cost measurement for this model."
  },
  "evidence": {
    "unmeasured": true,
    "items": []
  },
  "refs": [
    "hf://facebook/dinov2-base@f9e44c814b77203eaa57a6bdbbd535f21ede1415",
    "mixpeek://system/dinov2_vit_b_14@1.0.0"
  ],
  "formats": [
    {
      "ref": "hf://facebook/dinov2-base@f9e44c814b77203eaa57a6bdbbd535f21ede1415",
      "format": "safetensors",
      "file": "model.safetensors",
      "loader": "transformers / open_clip from_pretrained; Mixpeek Hub import (api/models/huggingface.py)",
      "supported_today": true
    },
    {
      "ref": "hf://facebook/dinov2-base@f9e44c814b77203eaa57a6bdbbd535f21ede1415",
      "format": "pytorch_bin",
      "file": "pytorch_model.bin",
      "loader": "torch.load (pickle); Mixpeek refuses pickle weights (shared/security/torch_pickle.py)",
      "supported_today": false
    }
  ],
  "use_cases": [
    "visual-search",
    "web-pages"
  ],
  "recommended": {
    "value": false,
    "for": [],
    "reason": null
  },
  "demos": [],
  "seo": {
    "title": "DINOv2 Base: embedding model",
    "description": "DINOv2 Base is a image embedding model (Apache-2.0). Served on Mixpeek, scaling to zero when idle.",
    "keywords": [
      "DINOv2 Base",
      "dinov2",
      "embedding",
      "facebook/dinov2-base",
      "visual-search",
      "web-pages"
    ],
    "faqs": [
      {
        "q": "What is DINOv2 Base best at?",
        "a": "Image-to-image similarity on visual structure (layout, near-duplicates, the same object or scene) with no text in the loop."
      },
      {
        "q": "What are the limits of DINOv2 Base?",
        "a": "Text queries (no text tower, so text-to-image search is impossible); semantic matches across different-looking images, where SigLIP is better."
      },
      {
        "q": "Can DINOv2 Base be used commercially?",
        "a": "Yes. Apache-2.0 weights and code; no use restrictions."
      },
      {
        "q": "How does DINOv2 Base run on Mixpeek?",
        "a": "Ray Serve app in the inference registry (image queries); web_scraper loads facebook/dinov2-base in-process (DocTypeStructureEmbedder)."
      }
    ]
  },
  "last_verified": "2026-11-05",
  "sources": [
    "https://huggingface.co/facebook/dinov2-base",
    "https://huggingface.co/api/models/facebook/dinov2-base",
    "https://arxiv.org/abs/2304.07193",
    "server/engine/inference/facebook/dinov2/services.py",
    "server/engine/extractors/web_scraper/v1/definition.py",
    "server/engine/extractors/web_scraper/v1/processors/doctype_embedders.py"
  ],
  "registry": {
    "facebook__dinov2_base": {
      "deployment": "facebook__dinov2_base",
      "service_id": "facebook/dinov2-base",
      "routes": "engine/inference/facebook/dinov2/routes.py",
      "route_prefix": "/facebook_dinov2_base/v1",
      "enabled": true,
      "description": "Facebook DINOv2 Base self-supervised vision model. Generates 768-dimensional embeddings for fine-grained visual structure comparison. Captures layout, spatial arrangement, and visual patterns rather than semantic content. Use alongside SigLIP for complementary visual analysis in drift detection.",
      "capabilities": {
        "category": "embedding",
        "modality_input": [
          "image"
        ],
        "modality_output": "vector",
        "supports_batch": true,
        "reranking_mode": null
      },
      "vector_config": {
        "dimensions": 768,
        "distance": "cosine",
        "vector_type": "dense"
      },
      "serve": {
        "min_replicas": 0,
        "max_replicas": 4,
        "idle_timeout_s": 300,
        "target_ongoing_requests": 5,
        "num_cpus": 0.5,
        "num_gpus": {
          "expr": "hardware_config.gpu_fraction if hardware_config.gpu_concurrency > 0 else 0"
        },
        "memory_gb": 1,
        "pinnable": false,
        "placement": "worker",
        "enabled": true
      },
      "interactive_twin": false,
      "interactive_serve": null,
      "schemas": {
        "input": "shared.inference.facebook.dinov2.models.InferenceRequest",
        "output": "shared.inference.facebook.dinov2.models.InferenceResponse",
        "config": "shared.inference.facebook.dinov2.models.InferenceConfigs"
      }
    }
  },
  "templates": [],
  "usage": {
    "namespaces": null,
    "collections": null,
    "uncounted": true,
    "templates": 0,
    "counted_at": null,
    "source": null
  },
  "facets": {
    "status": "served",
    "category": "embedding",
    "family": "dinov2",
    "license": "Apache-2.0",
    "commercial_use": true,
    "modalities_in": [
      "image"
    ],
    "modalities_out": [
      "vector"
    ],
    "use_cases": [
      "visual-search",
      "web-pages"
    ],
    "recommended": false,
    "release_date": "2023-04",
    "params_millions": 86.6,
    "size_on_disk_mb": 346,
    "vector_dims": 768,
    "vector_distance": "cosine",
    "weights_format": "safetensors",
    "formats": [
      "pytorch_bin",
      "safetensors"
    ],
    "loadable_today": true,
    "trust_remote_code": false,
    "cpu_fallback": true,
    "num_gpus": null,
    "memory_gb": 1,
    "min_replicas": 0,
    "serving_shape": "both",
    "served": true,
    "ref_schemes": [
      "hf",
      "mixpeek"
    ],
    "has_evidence": false,
    "cost_measured": false,
    "template_count": 0,
    "usage_namespaces": null,
    "usage_collections": null,
    "usage_uncounted": true,
    "has_demo": false,
    "rank_score": 3
  },
  "rank_score": 3
}