{ "run_id": "2026-10-07-quick-card-prompts-f0c765", "suite": "embeddings", "tier": "quick", "suite_hash": "d499aab0d7091e3ba5b211e1cbaedf79ed4c9c86a7214156946d2ee4d21299ec", "subject": "embeddinggemma-2/hf-914f7f8/q4-onnx-community/onnxruntime-cuda-ort1.29.0-cuda12", "mode": "native", "profile": "card-prompts", "status": "complete", "label": null, "items": { "planned": 66939, "attempted": 2092, "succeeded": 2092, "failed": 0, "failed_by_status": {} }, "metrics": { "mteb": "2.22.5", "tasks": [ "NanoArguAnaRetrieval", "NanoClimateFeverRetrieval", "NanoDBPediaRetrieval", "NanoFEVERRetrieval", "NanoFiQA2018Retrieval", "NanoHotpotQARetrieval", "NanoMSMARCORetrieval", "NanoNFCorpusRetrieval", "NanoNQRetrieval", "NanoQuoraRetrieval", "NanoSCIDOCSRetrieval", "NanoSciFactRetrieval", "NanoTouche2020Retrieval", "STSBenchmark", "Banking77Classification" ], "by_dim": { "128": { "tasks": { "NanoArguAnaRetrieval": { "main_score": 0.59492, "revision": "63ac9f026e8f6b85dc0622f3d630ed7a1758e9ec" }, "NanoClimateFeverRetrieval": { "main_score": 0.26293, "revision": "ccb5552bc17e87afe0f9479fb4346898213a71db" }, "NanoDBPediaRetrieval": { "main_score": 0.55466, "revision": "2d088f80d7a640eac0be5c66408d68c3f782baa1" }, "NanoFEVERRetrieval": { "main_score": 0.85972, "revision": "a6490dfeed0f13c2a052e39c6ddaeb6086afd4a8" }, "NanoFiQA2018Retrieval": { "main_score": 0.33581, "revision": "ede84da4793cd2d4e61bba0dd08fc58de90f3681" }, "NanoHotpotQARetrieval": { "main_score": 0.79345, "revision": "a38a2615f29adba8b47a42905031aa9c137d8fae" }, "NanoMSMARCORetrieval": { "main_score": 0.51561, "revision": "5a5aa0c3df65d6883d85c8f5cdc88e679633595e" }, "NanoNFCorpusRetrieval": { "main_score": 0.28923, "revision": "2eec9d63140fb9af7c75ad0fdbf6350aaf20e94e" }, "NanoNQRetrieval": { "main_score": 0.70662, "revision": "e3973b405feb9e94d07fd971d690d24d7f45264a" }, "NanoQuoraRetrieval": { "main_score": 0.92198, "revision": "3ee153e5710da83a7f35df0cb1db4a0ad036072a" }, "NanoSCIDOCSRetrieval": { "main_score": 0.30697, "revision": "cf39b8741a6f2e984d0b8f4796b2dd5bb65390f9" }, "NanoSciFactRetrieval": { "main_score": 0.74449, "revision": "8c5f7bbe90bb865f46c67710221afa7d9096251b" }, "NanoTouche2020Retrieval": { "main_score": 0.398, "revision": "ff81c0caff8b6febe1495eae9f2ca3eddad074b8" }, "STSBenchmark": { "main_score": 0.868305, "revision": "b0fddb56ed78048fa8b90373c8a3cfc37b684831" }, "Banking77Classification": { "main_score": 0.903442, "revision": "0fd18e25b25c072e09e0d92ab615fda904d66300" } }, "mean_main_score": 0.603742, "fidelity": { "n": 66939, "mean_cosine": 0.987869, "p1_cosine": 0.963051, "min_cosine": 0.910954, "top10_overlap": 0.628096, "queries": 649, "documents": 20000, "verdict": "degraded" } }, "256": { "tasks": { "NanoArguAnaRetrieval": { "main_score": 0.65265, "revision": "63ac9f026e8f6b85dc0622f3d630ed7a1758e9ec" }, "NanoClimateFeverRetrieval": { "main_score": 0.26806, "revision": "ccb5552bc17e87afe0f9479fb4346898213a71db" }, "NanoDBPediaRetrieval": { "main_score": 0.61188, "revision": "2d088f80d7a640eac0be5c66408d68c3f782baa1" }, "NanoFEVERRetrieval": { "main_score": 0.89034, "revision": "a6490dfeed0f13c2a052e39c6ddaeb6086afd4a8" }, "NanoFiQA2018Retrieval": { "main_score": 0.39671, "revision": "ede84da4793cd2d4e61bba0dd08fc58de90f3681" }, "NanoHotpotQARetrieval": { "main_score": 0.87519, "revision": "a38a2615f29adba8b47a42905031aa9c137d8fae" }, "NanoMSMARCORetrieval": { "main_score": 0.55003, "revision": "5a5aa0c3df65d6883d85c8f5cdc88e679633595e" }, "NanoNFCorpusRetrieval": { "main_score": 0.34203, "revision": "2eec9d63140fb9af7c75ad0fdbf6350aaf20e94e" }, "NanoNQRetrieval": { "main_score": 0.71216, "revision": "e3973b405feb9e94d07fd971d690d24d7f45264a" }, "NanoQuoraRetrieval": { "main_score": 0.94904, "revision": "3ee153e5710da83a7f35df0cb1db4a0ad036072a" }, "NanoSCIDOCSRetrieval": { "main_score": 0.35772, "revision": "cf39b8741a6f2e984d0b8f4796b2dd5bb65390f9" }, "NanoSciFactRetrieval": { "main_score": 0.85166, "revision": "8c5f7bbe90bb865f46c67710221afa7d9096251b" }, "NanoTouche2020Retrieval": { "main_score": 0.4339, "revision": "ff81c0caff8b6febe1495eae9f2ca3eddad074b8" }, "STSBenchmark": { "main_score": 0.872416, "revision": "b0fddb56ed78048fa8b90373c8a3cfc37b684831" }, "Banking77Classification": { "main_score": 0.915714, "revision": "0fd18e25b25c072e09e0d92ab615fda904d66300" } }, "mean_main_score": 0.6453, "fidelity": { "n": 66939, "mean_cosine": 0.983247, "p1_cosine": 0.948761, "min_cosine": 0.877816, "top10_overlap": 0.691919, "queries": 649, "documents": 20000, "verdict": "degraded" } }, "512": { "tasks": { "NanoArguAnaRetrieval": { "main_score": 0.64781, "revision": "63ac9f026e8f6b85dc0622f3d630ed7a1758e9ec" }, "NanoClimateFeverRetrieval": { "main_score": 0.28356, "revision": "ccb5552bc17e87afe0f9479fb4346898213a71db" }, "NanoDBPediaRetrieval": { "main_score": 0.62332, "revision": "2d088f80d7a640eac0be5c66408d68c3f782baa1" }, "NanoFEVERRetrieval": { "main_score": 0.92358, "revision": "a6490dfeed0f13c2a052e39c6ddaeb6086afd4a8" }, "NanoFiQA2018Retrieval": { "main_score": 0.44085, "revision": "ede84da4793cd2d4e61bba0dd08fc58de90f3681" }, "NanoHotpotQARetrieval": { "main_score": 0.89098, "revision": "a38a2615f29adba8b47a42905031aa9c137d8fae" }, "NanoMSMARCORetrieval": { "main_score": 0.56377, "revision": "5a5aa0c3df65d6883d85c8f5cdc88e679633595e" }, "NanoNFCorpusRetrieval": { "main_score": 0.36326, "revision": "2eec9d63140fb9af7c75ad0fdbf6350aaf20e94e" }, "NanoNQRetrieval": { "main_score": 0.75959, "revision": "e3973b405feb9e94d07fd971d690d24d7f45264a" }, "NanoQuoraRetrieval": { "main_score": 0.95324, "revision": "3ee153e5710da83a7f35df0cb1db4a0ad036072a" }, "NanoSCIDOCSRetrieval": { "main_score": 0.38096, "revision": "cf39b8741a6f2e984d0b8f4796b2dd5bb65390f9" }, "NanoSciFactRetrieval": { "main_score": 0.86082, "revision": "8c5f7bbe90bb865f46c67710221afa7d9096251b" }, "NanoTouche2020Retrieval": { "main_score": 0.44287, "revision": "ff81c0caff8b6febe1495eae9f2ca3eddad074b8" }, "STSBenchmark": { "main_score": 0.872945, "revision": "b0fddb56ed78048fa8b90373c8a3cfc37b684831" }, "Banking77Classification": { "main_score": 0.916169, "revision": "0fd18e25b25c072e09e0d92ab615fda904d66300" } }, "mean_main_score": 0.661582, "fidelity": { "n": 66939, "mean_cosine": 0.981154, "p1_cosine": 0.94427, "min_cosine": 0.872797, "top10_overlap": 0.717418, "queries": 649, "documents": 20000, "verdict": "degraded" } }, "768": { "tasks": { "NanoArguAnaRetrieval": { "main_score": 0.65006, "revision": "63ac9f026e8f6b85dc0622f3d630ed7a1758e9ec" }, "NanoClimateFeverRetrieval": { "main_score": 0.27281, "revision": "ccb5552bc17e87afe0f9479fb4346898213a71db" }, "NanoDBPediaRetrieval": { "main_score": 0.64397, "revision": "2d088f80d7a640eac0be5c66408d68c3f782baa1" }, "NanoFEVERRetrieval": { "main_score": 0.90466, "revision": "a6490dfeed0f13c2a052e39c6ddaeb6086afd4a8" }, "NanoFiQA2018Retrieval": { "main_score": 0.44449, "revision": "ede84da4793cd2d4e61bba0dd08fc58de90f3681" }, "NanoHotpotQARetrieval": { "main_score": 0.88786, "revision": "a38a2615f29adba8b47a42905031aa9c137d8fae" }, "NanoMSMARCORetrieval": { "main_score": 0.56462, "revision": "5a5aa0c3df65d6883d85c8f5cdc88e679633595e" }, "NanoNFCorpusRetrieval": { "main_score": 0.38265, "revision": "2eec9d63140fb9af7c75ad0fdbf6350aaf20e94e" }, "NanoNQRetrieval": { "main_score": 0.76702, "revision": "e3973b405feb9e94d07fd971d690d24d7f45264a" }, "NanoQuoraRetrieval": { "main_score": 0.95268, "revision": "3ee153e5710da83a7f35df0cb1db4a0ad036072a" }, "NanoSCIDOCSRetrieval": { "main_score": 0.3667, "revision": "cf39b8741a6f2e984d0b8f4796b2dd5bb65390f9" }, "NanoSciFactRetrieval": { "main_score": 0.8677, "revision": "8c5f7bbe90bb865f46c67710221afa7d9096251b" }, "NanoTouche2020Retrieval": { "main_score": 0.43419, "revision": "ff81c0caff8b6febe1495eae9f2ca3eddad074b8" }, "STSBenchmark": { "main_score": 0.872399, "revision": "b0fddb56ed78048fa8b90373c8a3cfc37b684831" }, "Banking77Classification": { "main_score": 0.916071, "revision": "0fd18e25b25c072e09e0d92ab615fda904d66300" } }, "mean_main_score": 0.661859, "fidelity": { "n": 66939, "mean_cosine": 0.980472, "p1_cosine": 0.94351, "min_cosine": 0.87278, "top10_overlap": 0.722079, "queries": 649, "documents": 20000, "verdict": "degraded" } } } }, "throughput": { "texts": 66939, "tokens": 11589994, "batches": 2092, "texts_per_s": 41.484294, "tokens_per_s": 7182.699498, "engine_tokens_per_s": 12530.227078, "batch_latency_ms": { "p50": 540.833625, "p95": 1628.01675 }, "note": "batches of 32 texts in descending length order, one request at a time; wall time includes HTTP and base64" }, "notes": [ "dimensions below 768 are the same vectors truncated and re-normalized (Matryoshka); fidelity compares each dimension with the reference at the same dimension" ], "latency_ms": { "p50": 540.833625, "p95": 1628.01675, "p99": 4037.500917, "note": "successful requests, nearest rank, measured from America/Chicago" }, "usage": { "input_tokens": 11589994, "output_tokens": 0, "cost_usd": 0 } }