Spaces:

THULab
/

CompressionBenchmark

Running

Apply uniform min128 dataset scope and compact Overall rows
86c9d33 verified 13 days ago
30.5 kB
export const FINAL160_SCOPE = {"active_algorithm_count": 85, "created_at": "2026-10-13T20:23:43.365976+08:00", "dataset_count": 137, "datasets": ["Air-Quality", "Bar-Crawl", "Bird-migration", "Comments", "ETDataset-ett", "IGlocations2", "METR-LA", "Metro-Traffic", "NYC_Yellow_Taxi_Trip", "PEMS-BAY", "RealEstate1", "RealEstate2", "Romance", "SpatialEpiBench", "Telco", "TrainsUK2", "Uberlandia", "Wins", "YaleLanguages", "ahr999-dataset", "ast_spacemobile_fleet_data", "australia_solar", "australian_electricity_demand", "auto_production_sf", "azure2019_u", "bitcoin", "bitcoin_price_timeseries", "coastal_t_s", "commodity_import", "commodity_production", "covid_deaths-monash", "cphl", "crypto", "crypto_data", "current_velocity", "deep-space-probes", "dst-index", "ecdc_covid", "eea_pm25_forecasting", "electricity_demand", "electricity_load_2011_2014", "electricity_production", "enedis-with-holidays", "epf_electricity_price", "eweld_load", "exchange_rate", "faang_engineered_time_series_features_2013_2025", "finland_traffic", "forex_factory_calendar", "fred_md", "freshretailnet_lt", "fulus_lyd_rates_2025", "geomagnetic_kp_index", "global_influenza", "global_price", "gluco_tsfm_benchmark", "goes_omni_electron_flux_forecasting", "greatplains_multisource_2000_2024", "hourly-stock-data-2023", "hourly_electricity_demand_brazil", "indian_stock_hourly_2017_2021", "job_claims", "jolts", "kaggle_web_traffic", "kdd_cup_2018", "knowable", "kraken-trading-data", "kuiper_fleet_data", "london_smart_meters", "m2oe2_timeseries", "meteoblue", "metropt_3", "multi_model_trading_data", "national_illness", "ne_china_wind", "nifty50_market_regime", "nn5_daily", "oikolab_weather", "oil_price", "omie_blockchain_study", "online_retail_2_uci", "ooni_censorship_historical", "openelectricity_nem", "orbital_chaos_nasa_ssc", "pedestrian_counts", "phenology_normal_hawaii", "polymarket_crypto_5m_15m", "port_activity", "quitobench", "rating", "reg_forecasting_v2", "rideshare", "salesforecastingwomartstore", "saugeenday-monash", "scaphandre_cpu_usage", "sg_carpark", "sg_pm25", "sg_weather", "sleep_score_fitbit", "solar_10_minutes", "solar_4_seconds", "solusdt_enhanced_180", "spx", "stock_dataset", "substorm_onsets", "sunspot", "telangana_weather", "temperature_rain", "timeseries-1m-QQQ-10y", "timeseries_1mn_sp500", "timeseries_1mn_stocks", "tourism_monthly", "tourism_monthly-zaai-ai", "tourism_residuals", "traffic", "traffic_hourly", "trxusdt_spot_1m_05_2021_to_03_2026", "uk_electricity_generation_mix_2019_2026", "uncertainty_1m", "us_births-monash", "us_labor", "us_term_structure", "vehicle_sales", "vehicle_supply", "vehicle_trips", "water_quality_darwin", "weather_data_bavaria", "wids2023_competition_weather_forecasting", "wind_4_seconds", "wind_farms_minutely", "wordle_mcm_dataset", "wui_global", "xauusd", "xauusd_1m", "xinyi_public_gym", "xrpusdt_spot_1m_05_2021_to_03_2026", "yongchun_public_gym"], "excluded_dataset_count": 23, "excluded_datasets": [{"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "africa_senegal_mauritania_crops_spi", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 30.87160525309193, "nonempty_numeric_sequence_count": 7843, "source_bindings": {"block_catalog_bytes": 18563392, "block_catalog_path": "/data/root/web_compression/datasets164/africa_senegal_mauritania_crops_spi/typed/full/blocks.jsonl", "block_catalog_sha256": "69e195e6a31c16e86bd8fb450ca8b82c7142b41c274534b6571e561b7a98163f", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/africa_senegal_mauritania_crops_spi/typed/dataset_manifest.json", "typed_manifest_sha256": "7d13ad8ef221f1996ee5f7ccc75d7c33e9dffd763021c546a25d18905e72c728"}, "valid_value_count": 242126}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "car_parts", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 50.76514584891548, "nonempty_numeric_sequence_count": 5348, "source_bindings": {"block_catalog_bytes": 11551132, "block_catalog_path": "/data/root/web_compression/datasets164/car_parts/typed/full/blocks.jsonl", "block_catalog_sha256": "df9670d9af90f132228ca048bf46889c53eae59fcf4561cb727f78112f30dd33", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/car_parts/typed/dataset_manifest.json", "typed_manifest_sha256": "3deda1e663dfc0f1255b0ddac83b7859d6c6e44ab1790f64b4d311147c6842fe"}, "valid_value_count": 271492}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "dmi-aarhus-weather-data", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 48.061442006269594, "nonempty_numeric_sequence_count": 60610, "source_bindings": {"block_catalog_bytes": 145991876, "block_catalog_path": "/data/root/web_compression/datasets164/dmi-aarhus-weather-data/typed/full/blocks.jsonl", "block_catalog_sha256": "76999b090861b54c87599295d59ea84275a1657159b2fc10f3c10fcc7f5f0276", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/dmi-aarhus-weather-data/typed/dataset_manifest.json", "typed_manifest_sha256": "07816de98eda4e4567d96cfd1f0f0017b9476aad776dfb338c4dac9e0fb496d4"}, "valid_value_count": 2913004}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "freshretailnet_50k", "input_version": "corrected_full114_typed_frozen_ready", "mean_valid_values_per_sequence": 48.5, "nonempty_numeric_sequence_count": 5700000, "source_bindings": {"block_catalog": {"bytes": 13015265091, "path": "/data/user/new_ai_compress/operator_sweep/six_dataset_numeric_repair_20261012_v1/typed_attempt_v2/datasets/freshretailnet_50k/full/blocks.jsonl", "sha256": "fe0b800aed22cde1e823dd8f0861246acd71746023f0ddb095a9e2510fa7fe8c"}, "frozen_ready": {"bytes": 8026, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/fresh50k/typed_ready_v2/FROZEN_READY.json", "sha256": "1b71521fbd61e9d9ffe737d57b6ac73077b7e8e1f77661eacac45e8a4abb1dcd"}, "source_data_sha256": "6f7e162d476c2025cd1493da71939f5c20f0e7cbdb5724b4d1e3bc0b11a4ff3c", "typed_manifest": {"bytes": 82351, "path": "/data/user/new_ai_compress/operator_sweep/six_dataset_numeric_repair_20261012_v1/typed_attempt_v2/datasets/freshretailnet_50k/dataset_manifest.json", "sha256": "17b2a278f4eeacc1f27a64e1cbb348e1946e6ed173a658eca59c15d94d4ece76"}}, "valid_value_count": 276450000}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "github_top_projects", "input_version": "corrected_accepted_input", "mean_valid_values_per_sequence": 6.246782941614378, "nonempty_numeric_sequence_count": 37612, "source_bindings": {"columns_csv": {"bytes": 443, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/github_top_projects/columns.csv", "sha256": "f1ee36477b22012eab7b9567b138be0a01265377df6541bef82b6519ed087be1"}, "data_csv": {"bytes": 2855950, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/github_top_projects/data.csv", "sha256": "faa5bdd56961f095da5d6f885055ad3d05a5711814547d9be9d8449a81f43201"}, "dataset_manifest": {"bytes": 1316, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/github_top_projects/dataset_manifest.json", "sha256": "c472deb1f54446d0a7d83cb239424ece7da8e73b73737ae234ac2b68fd276d5c"}, "method": "local_exact_corrected_wide_csv_scan", "series_map_csv": {"bytes": 2136531, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/github_top_projects/series_map.csv", "sha256": "a15818feb0a413c06652f6e68ef88dd97edd19e603b726bc2ba7f102ac3232d9"}}, "valid_value_count": 234954}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "globalstar_fleet_data", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 109.25, "nonempty_numeric_sequence_count": 20, "source_bindings": {"block_catalog_bytes": 44300, "block_catalog_path": "/data/root/web_compression/datasets164/globalstar_fleet_data/typed/full/blocks.jsonl", "block_catalog_sha256": "c8378f97eaaafb2678dc119ed5fc922077f6e96324059dcc25515c01137ebafd", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/globalstar_fleet_data/typed/dataset_manifest.json", "typed_manifest_sha256": "e13103d004e8503a68292af02b6b14704065d33318e8dbef15a623567fc616b0"}, "valid_value_count": 2185}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "hospital-monash", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 84.0, "nonempty_numeric_sequence_count": 1534, "source_bindings": {"block_catalog_bytes": 3369626, "block_catalog_path": "/data/root/web_compression/datasets164/hospital-monash/typed/full/blocks.jsonl", "block_catalog_sha256": "39e9d05162ca54709299ffe7bab36e3e8d0e6ef961a04dfa1d187475f92856b7", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/hospital-monash/typed/dataset_manifest.json", "typed_manifest_sha256": "9ccb7c12def299f710a0b5418f3fb31cea3caf22b651dccd532531e8519a1e56"}, "valid_value_count": 128856}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "housing_inventory", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 114.0, "nonempty_numeric_sequence_count": 4, "source_bindings": {"block_catalog_bytes": 8721, "block_catalog_path": "/data/root/web_compression/datasets164/housing_inventory/typed/full/blocks.jsonl", "block_catalog_sha256": "4dface5860d4a08cc0d4d3eb928358059c871a959ac2d887c445835c7dd1487b", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/housing_inventory/typed/dataset_manifest.json", "typed_manifest_sha256": "798085150daf04292d158ce9220ec446580bf2f7bef0b9c77822f738f2964d0b"}, "valid_value_count": 456}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "kaggle_web_traffic_weekly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 114.0, "nonempty_numeric_sequence_count": 290126, "source_bindings": {"block_catalog_bytes": 669107871, "block_catalog_path": "/data/root/web_compression/datasets164/kaggle_web_traffic_weekly/typed/full/blocks.jsonl", "block_catalog_sha256": "08113aae0d3d59c8a5816345c6a18f93592503590c9c7e3af03e37c687f0d38f", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/kaggle_web_traffic_weekly/typed/dataset_manifest.json", "typed_manifest_sha256": "a3f024f36d98062db4d52b6b11e403ca998ee5c1088dc84f0fddadc2cc978b94"}, "valid_value_count": 33074364}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "magseven_ml_ready_high_frequency_anomaly", "input_version": "corrected_accepted_input", "mean_valid_values_per_sequence": 33.0, "nonempty_numeric_sequence_count": 6, "source_bindings": {"columns_csv": {"bytes": 1295, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/mag7_float64_v2/magseven_ml_ready_high_frequency_anomaly/columns.csv", "sha256": "6382075fcd1109b3189b0ecb0852da6fa2d21d26bf50b4f4a22a8399b85da817"}, "data_csv": {"bytes": 2122, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/mag7_float64_v2/magseven_ml_ready_high_frequency_anomaly/data.csv", "sha256": "1bd2b20c2ad74cb89bfdec5ce4fbc8d5f99b83d423a7bc7612ebb526f1f16b32"}, "dataset_manifest": {"bytes": 1632, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/mag7_float64_v2/magseven_ml_ready_high_frequency_anomaly/dataset_manifest.json", "sha256": "4d57e891319e836e65c95cc414b37b2da1def54323562f14d35a00065753ac26"}, "method": "local_exact_corrected_wide_csv_scan", "series_map_csv": {"bytes": 322, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/mag7_float64_v2/magseven_ml_ready_high_frequency_anomaly/series_map.csv", "sha256": "4d2b0feeed3c0ea0dfbeca93e6f5bc21fb33d99652c358e533ed1e7dfeca0426"}}, "valid_value_count": 198}, {"classification": "exclude_no_admitted_scientific_numeric_payload", "dataset_id": "medium_posts", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": null, "nonempty_numeric_sequence_count": 0, "source_bindings": {"block_catalog_bytes": 25939, "block_catalog_path": "/data/root/web_compression/datasets164/medium_posts/typed/full/blocks.jsonl", "block_catalog_sha256": "a0ed3e9b2b42de7240a3f0b640f95de1179deb52f08e03fb2865c272d2cbf370", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/medium_posts/typed/dataset_manifest.json", "typed_manifest_sha256": "9ceb1dbbcc5c8708c0f63f8d7e913b56a4fe30d2aa6f70255dc334e968c90a4e"}, "valid_value_count": 0}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "niederschlag_station_4091_bavaria_germany", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 60.0, "nonempty_numeric_sequence_count": 1, "source_bindings": {"block_catalog_bytes": 2376, "block_catalog_path": "/data/root/web_compression/datasets164/niederschlag_station_4091_bavaria_germany/typed/full/blocks.jsonl", "block_catalog_sha256": "616aef8f0b72a70aad8635da48e828698389db5560c446e0dd382a1da8f34be6", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/niederschlag_station_4091_bavaria_germany/typed/dataset_manifest.json", "typed_manifest_sha256": "dc85be141a65df6aceac18b72dcd960f5f81e1118a5432e00604da6e7eb27131"}, "valid_value_count": 60}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "nn5_weekly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 113.0, "nonempty_numeric_sequence_count": 222, "source_bindings": {"block_catalog_bytes": 479726, "block_catalog_path": "/data/root/web_compression/datasets164/nn5_weekly/typed/full/blocks.jsonl", "block_catalog_sha256": "7e80c31ba5c88bedcfc2c43695f13ce116bc2829df89d2a418c12ea2943c7b8b", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/nn5_weekly/typed/dataset_manifest.json", "typed_manifest_sha256": "7fbd87675bfd087093114168c8f972f6217790061438aa4db98cf0357a4d98ca"}, "valid_value_count": 25086}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "solar_weekly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 52.0, "nonempty_numeric_sequence_count": 274, "source_bindings": {"block_catalog_bytes": 595087, "block_catalog_path": "/data/root/web_compression/datasets164/solar_weekly/typed/full/blocks.jsonl", "block_catalog_sha256": "24dd43bad6668d50137e196e8e7e9749522f43b399566fab2b013b31df195e13", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/solar_weekly/typed/dataset_manifest.json", "typed_manifest_sha256": "cac9980160d9b29b606ba108e756240a6d512ebd9d8b115e4df68240f1e9e87a"}, "valid_value_count": 14248}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "timeseries-finance-ETF", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 57.49729988052569, "nonempty_numeric_sequence_count": 83700, "source_bindings": {"block_catalog_bytes": 188084769, "block_catalog_path": "/data/root/web_compression/datasets164/timeseries-finance-ETF/typed/full/blocks.jsonl", "block_catalog_sha256": "dd19a10edf0d6f4f5825caaf137b0f7c48453d2eb5e6f9825c87fecfaf04cd83", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/timeseries-finance-ETF/typed/dataset_manifest.json", "typed_manifest_sha256": "e61a1c59855ddb5847b6ec292d1c9c7bec6163bc2470ea976f17bf9d3df437bc"}, "valid_value_count": 4812524}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "timeseries_finance_etf_missing", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 57.49729988052569, "nonempty_numeric_sequence_count": 83700, "source_bindings": {"block_catalog_bytes": 193441569, "block_catalog_path": "/data/root/web_compression/datasets164/timeseries_finance_etf_missing/typed/full/blocks.jsonl", "block_catalog_sha256": "bccf15a09e4a505d1a3658eac93fa5bf5e93c181c3c88dc06f63528c89c8c198", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/timeseries_finance_etf_missing/typed/dataset_manifest.json", "typed_manifest_sha256": "02a97e0106ab3fc55ee0b384a6f5594c8097d5c61c531dc5d2258a5c74667503"}, "valid_value_count": 4812524}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "tourism_quarterly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 99.63466042154566, "nonempty_numeric_sequence_count": 854, "source_bindings": {"block_catalog_bytes": 1896652, "block_catalog_path": "/data/root/web_compression/datasets164/tourism_quarterly/typed/full/blocks.jsonl", "block_catalog_sha256": "c6a22e4277933d7fc9cb19909dda84f1da108e192f011d32143fabb0f40a1495", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/tourism_quarterly/typed/dataset_manifest.json", "typed_manifest_sha256": "7546af1d7f49ee5480d791cb942351a4a50ac6ac644dd770a1374afd6aa42cbf"}, "valid_value_count": 85088}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "tourism_yearly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 24.627413127413128, "nonempty_numeric_sequence_count": 1036, "source_bindings": {"block_catalog_bytes": 2271148, "block_catalog_path": "/data/root/web_compression/datasets164/tourism_yearly/typed/full/blocks.jsonl", "block_catalog_sha256": "c20cdff15822ba6809d02fcfb1e0d3cffc9bb5d6eb3b05c618cf1f7f3a7a76a4", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/tourism_yearly/typed/dataset_manifest.json", "typed_manifest_sha256": "8a9f17a8b9437b4c186ddb3a8355c76f330ed2c451385cf00d4364fd9199d55d"}, "valid_value_count": 25514}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "traffic_weekly", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 104.0, "nonempty_numeric_sequence_count": 1724, "source_bindings": {"block_catalog_bytes": 3790280, "block_catalog_path": "/data/root/web_compression/datasets164/traffic_weekly/typed/full/blocks.jsonl", "block_catalog_sha256": "4aa6a797240415e2b47b9ced3ecb64d31d643c5f72bb2d97fc92500b02f63e54", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/traffic_weekly/typed/dataset_manifest.json", "typed_manifest_sha256": "9ba985cbc9ca1ead89e8c0ee0a533bd8517c596d0a872b467b3990f73a281440"}, "valid_value_count": 179296}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "traqq_weekly_activity", "input_version": "published_correct_source_semantics_unchanged_typed_input", "mean_valid_values_per_sequence": 4.895522388059701, "nonempty_numeric_sequence_count": 335, "source_bindings": {"block_catalog_bytes": 740698, "block_catalog_path": "/data/root/web_compression/datasets164/traqq_weekly_activity/typed/full/blocks.jsonl", "block_catalog_sha256": "2e35d39cbe95790c6473a49f272792e2828f623533117b12f91d9b1b5c223b61", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/traqq_weekly_activity/typed/dataset_manifest.json", "typed_manifest_sha256": "7e9b85464124cdecd8b19b0a7c6fe59632b6856e8dd3ea340add0fb9794ba1cd"}, "valid_value_count": 1640}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "trending_stocks_yahoo_finance", "input_version": "corrected_accepted_input", "mean_valid_values_per_sequence": 5.7372982158028885, "nonempty_numeric_sequence_count": 17655, "source_bindings": {"columns_csv": {"bytes": 1227, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/trending_stocks_yahoo_finance/columns.csv", "sha256": "df6236a75d3eb6f67efe9dcda7d1186b06d5d915cd7c70ea65d06750c1ca2489"}, "data_csv": {"bytes": 1073819, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/trending_stocks_yahoo_finance/data.csv", "sha256": "fd8de8b6166d6875f5bb961d423e34bb438460625c884a5c2ddc6784efea4637"}, "dataset_manifest": {"bytes": 1507, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/trending_stocks_yahoo_finance/dataset_manifest.json", "sha256": "79349da2c5a892dc988104b7ea56ee2bbc5a7128919c2caee7228107d1fc0eb6"}, "method": "local_exact_corrected_wide_csv_scan", "series_map_csv": {"bytes": 252602, "path": "/Users/xiaojinzhao/Documents/GitHub/web_compression/analysis_outputs/six_dataset_numeric_repair_20261012_v1/yahoo_github/output_v2/trending_stocks_yahoo_finance/series_map.csv", "sha256": "e29ecee1ceaf947b33be5c1d143eb12e17c540f47cea39a88a2fba63423f4ee4"}}, "valid_value_count": 101292}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "us_stock_financial_ratios", "input_version": "published_correct_source_semantics_unchanged_typed_input", "mean_valid_values_per_sequence": 9.896346929627969, "nonempty_numeric_sequence_count": 8924, "source_bindings": {"block_catalog_bytes": 20865599, "block_catalog_path": "/data/root/web_compression/datasets164/us_stock_financial_ratios/typed/full/blocks.jsonl", "block_catalog_sha256": "04e12b341f09d4aae337077377cd4c17e7922a8a259a5465699385f101add238", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/us_stock_financial_ratios/typed/dataset_manifest.json", "typed_manifest_sha256": "8d2625899f1ce97ae2a1d9c4c2b82132948a31bd7803bca2818d250483812c04"}, "valid_value_count": 88315}, {"classification": "exclude_mean_valid_values_per_sequence_below_128", "dataset_id": "zillow_viewer", "input_version": "published_baseline_typed_input", "mean_valid_values_per_sequence": 32.28351815443223, "nonempty_numeric_sequence_count": 244651, "source_bindings": {"block_catalog_bytes": 640656958, "block_catalog_path": "/data/root/web_compression/datasets164/zillow_viewer/typed/full/blocks.jsonl", "block_catalog_sha256": "0e78b5cb919cbf544e49d916fe8d72ccc1b6b61ab93c635f115d149111ea33c4", "host": "192", "mirror_acceptance": "dataset164-final-acceptance-v2", "typed_manifest_path": "/data/root/web_compression/datasets164/zillow_viewer/typed/dataset_manifest.json", "typed_manifest_sha256": "1f60337db1ce69b7562ce2d20908e7158455f36481d4bb8cc3ffc130be997494"}, "valid_value_count": 7898195}], "expected_active_pair_count": 11645, "input_version_policy": "current corrected input where available; published accepted baseline otherwise", "schema": "compressionbenchmark-min128-dataset-scope-v1", "scope_id": "min128_20261013_v1", "selection_rule": "retain admitted numeric payload datasets with mean valid values per complete scientific sequence >=128", "sequence_length_audit_sha256": "9c340b8f314984881230f0bf3122f45fde64b885ce3e26baf1ec3df1f2540df4", "source_final160_scope_sha256": "92a995e4278971077fe60dabd913e26e1afdccaeca1ec6181ae1de3f50929889", "source_min32_sequence_length_audit_sha256": "cdce9b3f988f48639164e850ac16a345bbb23066cbd60f2c8989c21adfcd9365", "status": "frozen_user_authorized_min128_test_scope"};
export const FINAL160_INPUT_STATUS = {"schema": "compressionbenchmark-current-input-status-v2", "status": "known_incomplete_input_versions_quarantined_pending_replacement", "scope_sha256": "b56f9bd610e0bf63ab50d719ce70cf92ddc1b79e6b30fd2c6e2f707944b2ba72", "format_audit_sha256": "2d3fb152f4f040a044a49fb23a0a34eb217ad0a54668f1fa0c65281d72064ba8", "scope_id": "min128_20261013_v1", "pending_datasets": [{"dataset_id": "freshretailnet_lt", "classification": "proven_old_input_incomplete_new_input_required", "reason": "The old export retained 66/114 table-qualified numeric columns and dropped 48 trailing train columns under the 1 GB cap. All 7,939,549 source rows were retained, but the value-column universe is incomplete.", "rerun_requirement": "All 85 active families on the new full114 input after independent input acceptance."}, {"dataset_id": "kaggle_web_traffic", "classification": "proven_old_input_incomplete_new_input_required", "reason": "The old export retained only 33,860,956 of 116,476,879 eligible source rows because of the 1 GB cap (82,615,923 rows lost). The one scientific value is FLOAT target; physical Time is the temporal coordinate and TAG fields (including start_timestamp) are metadata, not value payload.", "rerun_requirement": "All 85 active families on a full-row target-only input after independent input acceptance; preserve all 7,018,638 source target NaNs as canonical NaN bits."}], "policy": "Apply the user-authorized mean numerical-sequence length >=128 rule to every dataset with admitted numeric payload. The timestamp-only medium_posts input has no admitted scientific numeric payload and is separately ineligible, not assigned a zero mean. Keep archived evidence. Exclude superseded incomplete inputs across all active language/type views until new exact accepted input/result bindings are published.", "raw_science_payload_modified": false, "sequence_length_audit_sha256": "9c340b8f314984881230f0bf3122f45fde64b885ce3e26baf1ec3df1f2540df4"};
export const ACTIVE_CPP_FAMILIES = ["2DIFF", "ALP", "ANS", "Arithmetic", "BOS", "BSC", "BUFF", "Blosc", "Blosc2", "BriefLZ", "Brotli", "Bzip2", "Bzip3", "CORAD", "CSC", "Camel", "Chimp", "Chimp128", "Crush", "DBP", "Density", "Dict.", "Elf", "Elf+", "FPC", "FPCmp-R", "FPCmp-S", "FPZIP", "FSE", "FastLZ", "FastLZMA2", "FastPFor", "Gipfeli", "Gorilla", "Huff0", "Huffman", "Kanzi", "LECO", "LZ4", "LZ4 Fast", "LZ4HC", "LZAV", "LZF", "LZFSE", "LZG", "LZHAM", "LZJB", "LZMAT", "LZO1", "LZO2A", "LZRW", "LZSSE", "LZVN", "Libdeflate", "Lizard", "Lzlib", "NRV", "PFPC", "PPMD8", "Pithy", "Pongo", "QuickLZ", "REGER", "RLBE", "RLE", "SLZ", "SPDP", "SVB", "Shrinker", "Simple8b", "Snappy", "Sprintz", "TAMP", "Tornado", "WFLZ", "XPACK", "XZ", "YALZ77", "Yappy", "ZXC", "Zlib", "Zlib-ng", "Zling", "Zstd", "gzip"];
const EXPECTED_SCOPE_SHA256="b56f9bd610e0bf63ab50d719ce70cf92ddc1b79e6b30fd2c6e2f707944b2ba72";

export function applyFinal160Scope(snapshot,scope=FINAL160_SCOPE,status=FINAL160_INPUT_STATUS) {
  if(scope?.schema!=='compressionbenchmark-min128-dataset-scope-v1'||scope?.status!=='frozen_user_authorized_min128_test_scope'||scope.scope_id!==FINAL160_SCOPE.scope_id)throw new Error('Invalid numerical-sequence scope contract.');
  if(!Array.isArray(scope.datasets)||scope.datasets.length!==FINAL160_SCOPE.dataset_count||new Set(scope.datasets).size!==scope.datasets.length||scope.datasets.some((id,index)=>id!==FINAL160_SCOPE.datasets[index]))throw new Error('Dataset identities differ from the frozen scope.');
  if(status?.schema!=='compressionbenchmark-current-input-status-v2'||status.scope_id!==scope.scope_id||status.scope_sha256!==EXPECTED_SCOPE_SHA256||!Array.isArray(status.pending_datasets))throw new Error('Invalid current-input gate.');
  const ids=new Set(scope.datasets),blocked=new Set(status.pending_datasets.map(item=>item.dataset_id)),families=new Set(ACTIVE_CPP_FAMILIES);
  if(blocked.size!==status.pending_datasets.length||[...blocked].some(id=>!ids.has(id)))throw new Error('Invalid pending input identity.');
  const known=new Set((snapshot.datasets||[]).map(item=>item.key??item.dataset));
  if([...ids].some(id=>!known.has(id)))throw new Error('A selected identity is absent from the immutable source snapshot.');
  if(!Array.isArray(snapshot.dataset_results))throw new Error('Expected the receipt-bound decoded result view.');
  const selected=snapshot.dataset_results.filter(row=>ids.has(row.dataset)&&!blocked.has(row.dataset)&&(row.comparison_profile!=='cpp'||families.has(row.family)));
  return {...snapshot,datasets:snapshot.datasets.filter(item=>ids.has(item.key??item.dataset)),dataset_results:selected,
    active_dataset_scope:{scope_id:status.scope_id,dataset_count:ids.size,pending_dataset_ids:[...blocked].sort(),historical_archive_dataset_count:known.size,retained_result_rows:selected.length}};
}
export function filterFinal160Catalog(catalog) {
  const ids=new Set(FINAL160_SCOPE.datasets);
  return {...catalog,datasets:(catalog.datasets||[]).filter(item=>ids.has(item.key??item.dataset))};
}
export function final160CoverageForFamily(ledger,family) {
  const ids=new Set(FINAL160_SCOPE.datasets),pending=new Set(FINAL160_INPUT_STATUS.pending_datasets.map(item=>item.dataset_id));
  return ledger.filter(item=>item.canonical_family===family&&ids.has(item.dataset)).map(item=>({dataset:item.dataset,status:pending.has(item.dataset)?'replacement_input_pending':item.status,scopes:pending.has(item.dataset)?[]:(item.selected_scopes||[]),reason:pending.has(item.dataset)?'Old incomplete input excluded; corrected-input experiment pending.':item.reason}));
}