{
  "id": "qwen3-embedding-0-6b",
  "identity": {
    "family": "qwen3-embedding",
    "display_name": "Qwen3 Embedding 0.6B",
    "hub_id": "Qwen/Qwen3-Embedding-0.6B",
    "api_name": null,
    "revision": null,
    "variants": [],
    "license": "Apache-2.0",
    "commercial_use": true,
    "commercial_use_note": "Apache-2.0 weights; no use restrictions.",
    "release_date": "2025-06",
    "paper_url": "https://arxiv.org/abs/2506.05176",
    "repo_url": "https://github.com/QwenLM/Qwen3-Embedding"
  },
  "task": {
    "category": "embedding",
    "modalities_in": [
      "text"
    ],
    "modalities_out": [
      "vector"
    ],
    "best_at": "Instruction-aware multilingual text vectors from a small model (0.6B, 1024 dims) on CPU; in the Live Nation plugin it embeds Gemini tags plus the words spoken over each shot.",
    "worst_at": "Speech-only queries with one relevant shot (Live Nation speech category F1@10 0.249); it needs the Instruct/Query prompt on the query side, so a plain query embed underperforms."
  },
  "io": {
    "input": {
      "text": {
        "max_tokens": "32768 (model limit)"
      },
      "prompts": "queries: 'Instruct: <task>\nQuery: <text>'; documents: no prefix"
    },
    "output": {
      "vector": {
        "dims": 1024,
        "distance": "cosine",
        "pooling": "last token",
        "normalized": true
      }
    }
  },
  "mixpeek_wiring": {
    "extractors": [
      {
        "extractor": "ln_shots@v1",
        "selected_by": "fixed (customer _local plugin in customers/livenation/manifest-loop)"
      }
    ],
    "feature_uris": [
      "mixpeek://ln_shots@v1/ln_text"
    ],
    "vector_indexes": [
      "ln_shots_v1_ln_text"
    ],
    "retriever_stages": [
      "feature_search"
    ],
    "composes_with": [
      "ln_shots: whisper-large-v3-turbo speech + Gemini tags -> Qwen3 text vector, beside a SigLIP 2 so400m visual vector and BM25 on search_text, fused by DBSF"
    ]
  },
  "run": {
    "status": "plugin",
    "status_reason": "Runs only inside the committed Live Nation _local plugin (ln_shots v1), loaded from /data/models as safetensors with remote code off; no Serve app or built-in extractor uses it.",
    "serve_apps": [],
    "runtime": "sentence_transformers.SentenceTransformer (Qwen3Model, last-token pooling, Normalize), device cpu",
    "weights_format": "safetensors",
    "trust_remote_code": false,
    "size_on_disk_mb": 1192,
    "params_millions": 596,
    "preprocessing": "Documents are the shot's tags and spoken sentences; queries get the Instruct/Query template with the Live Nation task.",
    "postprocessing": "Last-token pooling and L2 normalization (sentence-transformers modules), batch 16.",
    "serving_shape": "both",
    "batch_size": 16,
    "serve_batch": null,
    "scale_to_zero": {
      "min_replicas": null,
      "max_replicas": null,
      "idle_timeout_s": null,
      "memory_gb": null,
      "num_cpus": null,
      "num_gpus": null,
      "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."
  },
  "evidence": {
    "unmeasured": false,
    "items": [
      {
        "dataset": "Live Nation demo, 10 clips, 263 shots, 56 queries",
        "metric": "macro F1@10, variant 'text' (Qwen3 text vector alone, feature_search)",
        "k": 10,
        "score": 0.414,
        "baseline": "the demo's native pipeline",
        "baseline_score": 0.412,
        "held_out": false,
        "date": "2026-11-04",
        "source": "customers/livenation/manifest-loop/README.md"
      },
      {
        "dataset": "Live Nation held-out clips, 28 queries the driver never saw",
        "metric": "macro F1@10, SYSTEM number: tri_dbsf_p200 baseline = SigLIP 2 so400m visual + Qwen3 text + BM25, DBSF fusion",
        "k": 10,
        "score": 0.406,
        "baseline": null,
        "baseline_score": null,
        "held_out": true,
        "date": "2026-11-04",
        "source": "customers/livenation/manifest-loop/DRIVER-RUN.md"
      }
    ]
  },
  "refs": [
    "hf://Qwen/Qwen3-Embedding-0.6B@97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3"
  ],
  "formats": [
    {
      "ref": "hf://Qwen/Qwen3-Embedding-0.6B@97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3",
      "format": "safetensors",
      "file": "model.safetensors",
      "loader": "transformers / open_clip from_pretrained; Mixpeek Hub import (api/models/huggingface.py)",
      "supported_today": true
    }
  ],
  "use_cases": [
    "text-search",
    "multilingual-search",
    "transcripts"
  ],
  "recommended": {
    "value": false,
    "for": [],
    "reason": null
  },
  "demos": [],
  "seo": {
    "title": "Qwen3 Embedding 0.6B: embedding model",
    "description": "Qwen3 Embedding 0.6B is a text embedding model (Apache-2.0). Can run on Mixpeek as a custom plugin.",
    "keywords": [
      "Qwen3 Embedding 0.6B",
      "qwen3-embedding",
      "embedding",
      "Qwen/Qwen3-Embedding-0.6B",
      "text-search",
      "multilingual-search",
      "transcripts"
    ],
    "faqs": [
      {
        "q": "What is Qwen3 Embedding 0.6B best at?",
        "a": "Instruction-aware multilingual text vectors from a small model (0.6B, 1024 dims) on CPU; in the Live Nation plugin it embeds Gemini tags plus the words spoken over each shot."
      },
      {
        "q": "What are the limits of Qwen3 Embedding 0.6B?",
        "a": "Speech-only queries with one relevant shot (Live Nation speech category F1@10 0.249); it needs the Instruct/Query prompt on the query side, so a plain query embed underperforms."
      },
      {
        "q": "Can Qwen3 Embedding 0.6B be used commercially?",
        "a": "Yes. Apache-2.0 weights; no use restrictions."
      },
      {
        "q": "How does Qwen3 Embedding 0.6B run on Mixpeek?",
        "a": "Runs only inside the committed Live Nation _local plugin (ln_shots v1), loaded from /data/models as safetensors with remote code off; no Serve app or built-in extractor uses it."
      },
      {
        "q": "What does Qwen3 Embedding 0.6B output?",
        "a": "A 1024-dimensional dense vector, compared by cosine similarity."
      }
    ]
  },
  "last_verified": "2026-11-05",
  "sources": [
    "https://huggingface.co/Qwen/Qwen3-Embedding-0.6B",
    "https://huggingface.co/api/models/Qwen/Qwen3-Embedding-0.6B",
    "https://huggingface.co/Qwen/Qwen3-Embedding-0.6B/resolve/main/1_Pooling/config.json",
    "https://arxiv.org/abs/2506.05176",
    "customers/livenation/manifest-loop/README.md",
    "customers/livenation/manifest-loop/DRIVER-RUN.md",
    "customers/livenation/manifest-loop/plugin/ln_shots/v1/models.py",
    "customers/livenation/manifest-loop/plugin/ln_shots/v1/definition.py"
  ],
  "registry": null,
  "templates": [],
  "usage": {
    "namespaces": null,
    "collections": null,
    "uncounted": true,
    "templates": 0,
    "counted_at": null,
    "source": null
  },
  "facets": {
    "status": "plugin",
    "category": "embedding",
    "family": "qwen3-embedding",
    "license": "Apache-2.0",
    "commercial_use": true,
    "modalities_in": [
      "text"
    ],
    "modalities_out": [
      "vector"
    ],
    "use_cases": [
      "text-search",
      "multilingual-search",
      "transcripts"
    ],
    "recommended": false,
    "release_date": "2025-06",
    "params_millions": 596,
    "size_on_disk_mb": 1192,
    "vector_dims": 1024,
    "vector_distance": "cosine",
    "weights_format": "safetensors",
    "formats": [
      "safetensors"
    ],
    "loadable_today": true,
    "trust_remote_code": false,
    "cpu_fallback": true,
    "num_gpus": null,
    "memory_gb": null,
    "min_replicas": null,
    "serving_shape": "both",
    "served": false,
    "ref_schemes": [
      "hf"
    ],
    "has_evidence": true,
    "cost_measured": false,
    "template_count": 0,
    "usage_namespaces": null,
    "usage_collections": null,
    "usage_uncounted": true,
    "has_demo": false,
    "rank_score": 2
  },
  "rank_score": 2
}