[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"summary-68a7b0ecb194f703-fix-tokenization-drift-by-matching-sft-token-patte-summary":3,"summaries-facets-categories":136,"summary-related-68a7b0ecb194f703-fix-tokenization-drift-by-matching-sft-token-patte-summary":3705},{"id":4,"title":5,"ai":6,"body":13,"categories":102,"created_at":104,"date_modified":104,"description":96,"extension":105,"faq":104,"featured":106,"kicker_label":104,"meta":107,"navigation":118,"path":119,"published_at":120,"question":104,"scraped_at":121,"seo":122,"sitemap":123,"source_id":124,"source_name":125,"source_type":126,"source_url":127,"stem":128,"tags":129,"thumbnail_url":104,"tldr":133,"tweet":104,"unknown_tags":134,"__hash__":135},"summaries\u002Fsummaries\u002F68a7b0ecb194f703-fix-tokenization-drift-by-matching-sft-token-patte-summary.md","Fix Tokenization Drift by Matching SFT Token Patterns",{"provider":7,"model":8,"input_tokens":9,"output_tokens":10,"processing_time_ms":11,"cost_usd":12},"openrouter","x-ai\u002Fgrok-4.1-fast",9688,1789,16407,0.0028096,{"type":14,"value":15,"toc":95},"minimark",[16,21,34,37,41,44,47,51,54,82,85],[17,18,20],"h2",{"id":19},"leading-spaces-and-formatting-create-entirely-new-token-sequences","Leading Spaces and Formatting Create Entirely New Token Sequences",[22,23,24,25,29,30,33],"p",{},"Tokenization drift occurs when subtle changes like adding a leading space alter token IDs and sequence lengths, pushing inputs outside the model's trained distribution. Using GPT-2 tokenizer (vocab size 50,257, same BPE as GPT-4\u002FLLaMA\u002FMistral), test pairs like \" classify\" vs \"classify\": space version gets single token ",[26,27,28],"span",{},"36509",", no-space splits to ",[26,31,32],{},"4871, 1958",". All 7 tested words (classify, answer, positive, negative, sentiment, output, label) produce different IDs—deltas range from \u003C100 (low risk, e.g., label) to >500 (high risk, e.g., classify at 31,638 delta). This changes attention computation since sequence lengths differ, making \"apple\" and \" apple\" as distinct to the model as unrelated words.",[22,35,36],{},"SFT models learn specific structures (newlines, colons, prefixes). Deviations like removing newlines drop Jaccard overlap with canonical SFT template (\"Below is a customer review. Classify the sentiment.\\n\\nReview: {review}\\n\\nSentiment:\") to 80%; no leading space on \"Review\" to 85%; colon-to-dash to 70%; rewording instruction to 50%. Lower overlap signals higher OOD risk: >80% low risk, 60-80% medium, \u003C60% high, correlating to accuracy drops.",[17,38,40],{"id":39},"jaccard-overlap-quantifies-ood-risk-from-prompt-variants","Jaccard Overlap Quantifies OOD Risk from Prompt Variants",[22,42,43],{},"Canonical SFT overlap is 100%. Variants show: no newlines 80% (medium risk), missing space 85% (low), dash instead of colon 70% (medium), reworded (\"Determine the sentiment... Answer:\") 50% (high). On sample \"The product exceeded all my expectations. Highly recommend!\", these shifts mean the model processes unfamiliar token space, leading to unpredictable outputs despite unchanged logic or data.",[22,45,46],{},"Visual deltas confirm: high-ID gaps (>500) for most words indicate severe drift. Thresholds guide safety—stay above 80% overlap to mimic training distribution, avoiding degradation without retraining.",[17,48,50],{"id":49},"apo-loop-auto-selects-high-overlap-prompts-for-stable-performance","APO Loop Auto-Selects High-Overlap Prompts for Stable Performance",[22,52,53],{},"Implement Automated Prompt Optimization on 8-sample validation set (balanced positive\u002Fnegative\u002Fneutral reviews). Test 5 candidates:",[55,56,57,61,64,67,79],"ul",{},[58,59,60],"li",{},"A (no formatting: \"Classify: {review} Answer:\");",[58,62,63],{},"B (minimal: \"Review: {review}\\nSentiment:\");",[58,65,66],{},"C (SFT-aligned: full template with newlines\u002Fcolons);",[58,68,69,70,74,75],{},"D (XML: \"",[71,72,73],"review",{},"{review}","\\n",[76,77,78],"sentiment",{},"\");",[58,80,81],{},"E (full instruction: \"You are a sentiment classifier... Output...\").",[22,83,84],{},"Simulate accuracy: base 85%, scaled by overlap factor (0.5 + 0.5*Jaccard) minus OOD penalty (e.g., 0.18 for A, 0.02 for C), clipped 40-95%, plus noise. Results: A 38%, B 50%, C 88%, D 63%, E 75%. APO picks C (\"Variant C -- SFT-aligned\") at 88% accuracy—33% better than worst, proving closest SFT match wins.",[22,86,87,88,94],{},"In production, replace simulation with real model evals on validation data. Full code: ",[89,90,91],"a",{"href":91,"rel":92},"https:\u002F\u002Fgithub.com\u002FMarktechpost\u002FAI-Agents-Projects-Tutorials\u002Fblob\u002Fmain\u002FNLP\u002FTokenization_Drift.ipynb",[93],"nofollow",". This keeps prompts in-distribution, stabilizing performance across pipeline changes.",{"title":96,"searchDepth":97,"depth":97,"links":98},"",2,[99,100,101],{"id":19,"depth":97,"text":20},{"id":39,"depth":97,"text":40},{"id":49,"depth":97,"text":50},[103],"AI & LLMs",null,"md",false,{"content_references":108,"triage":113},[109],{"type":110,"title":111,"url":91,"context":112},"other","Tokenization_Drift.ipynb","mentioned",{"relevance":114,"novelty":115,"quality":115,"actionability":115,"composite":116,"reasoning":117},5,4,4.35,"Category: AI & LLMs. The article provides a deep dive into tokenization drift, a critical issue for AI product builders, and offers actionable strategies like Jaccard token overlap to measure risk and Automated Prompt Optimization to enhance model performance. This directly addresses the audience's need for practical applications in AI integration.",true,"\u002Fsummaries\u002F68a7b0ecb194f703-fix-tokenization-drift-by-matching-sft-token-patte-summary","2026-05-03 07:06:45","2026-05-03 17:01:43",{"title":5,"description":96},{"loc":119},"68a7b0ecb194f703","MarkTechPost","article","https:\u002F\u002Fwww.marktechpost.com\u002F2026\u002F05\u002F03\u002Fwhat-is-tokenization-drift-and-how-to-fix-it\u002F","summaries\u002F68a7b0ecb194f703-fix-tokenization-drift-by-matching-sft-token-patte-summary",[130,131,132],"prompt-engineering","llm","python","Minor formatting like spaces or newlines causes tokenization drift, shifting prompts out-of-distribution and dropping accuracy. Use Jaccard token overlap (>80% safe) to measure risk; Automated Prompt Optimization (APO) selects best templates, boosting simulated accuracy from 40-50% to 83%.",[],"fLlnSu0xltWQumszVZ_1GwcHUHHo4fwQpqnVb1YmSMM",[137,140,143,145,148,151,153,155,157,159,161,163,166,168,170,172,174,176,178,180,182,184,187,190,192,194,197,199,201,204,206,208,210,212,214,216,218,220,222,224,226,228,230,232,234,236,238,240,242,244,246,248,250,252,254,256,258,260,262,264,266,268,270,272,274,276,278,280,282,284,286,288,290,292,294,296,298,300,302,304,306,308,310,312,314,316,318,320,322,324,326,328,330,332,334,336,338,340,342,344,346,348,350,352,354,356,358,360,362,364,366,368,370,372,374,376,378,380,382,384,386,388,390,392,394,396,398,400,402,404,406,408,410,412,414,416,418,420,422,424,426,428,430,432,434,436,438,440,442,444,446,448,450,452,454,456,459,461,463,465,467,469,471,473,475,477,479,481,483,485,487,489,491,493,495,497,499,501,503,505,507,509,511,513,515,517,519,521,523,525,527,529,531,533,535,537,539,541,543,545,547,549,551,553,555,557,559,561,563,565,567,569,571,573,575,577,579,581,583,585,587,589,591,593,595,597,599,601,603,605,607,609,611,613,615,617,619,621,623,625,627,629,631,633,635,637,639,641,643,645,647,649,651,653,655,657,659,661,663,665,667,669,671,673,675,677,679,681,683,685,687,689,691,693,695,697,699,701,703,705,707,709,711,713,715,717,719,721,723,725,727,729,731,733,735,737,739,741,743,745,747,749,751,753,755,757,759,761,763,765,767,769,771,773,775,777,779,781,783,785,787,789,791,793,795,797,799,801,803,805,807,809,811,813,815,817,819,821,823,825,827,829,831,833,835,837,839,841,843,845,847,849,851,853,855,857,859,861,863,865,867,869,871,873,875,877,879,881,883,885,887,889,891,893,895,897,899,901,903,905,907,909,911,913,915,917,919,921,923,925,927,929,931,933,935,937,939,941,943,945,947,949,951,953,955,957,959,961,963,965,967,969,971,973,975,977,979,981,983,985,987,989,991,993,995,997,999,1001,1003,1005,1007,1009,1011,1013,1015,1017,1019,1021,1023,1025,1027,1029,1031,1033,1035,1037,1039,1041,1043,1045,1047,1049,1051,1053,1055,1057,1059,1061,1063,1065,1067,1069,1071,1073,1075,1077,1079,1081,1083,1085,1087,1089,1091,1093,1095,1097,1099,1101,1103,1105,1107,1109,1111,1113,1115,1117,1119,1121,1123,1125,1127,1129,1131,1133,1135,1137,1139,1141,1143,1145,1147,1149,1151,1153,1155,1157,1159,1161,1163,1165,1167,1169,1171,1173,1175,1177,1179,1181,1183,1185,1187,1189,1191,1193,1195,1197,1199,1201,1203,1205,1207,1209,1211,1213,1215,1217,1219,1221,1223,1225,1227,1229,1231,1233,1235,1237,1239,1241,1243,1245,1247,1249,1251,1253,1255,1257,1259,1261,1263,1265,1267,1269,1271,1273,1275,1277,1279,1281,1283,1285,1287,1289,1291,1293,1295,1297,1299,1301,1303,1305,1307,1309,1311,1313,1315,1317,1319,1321,1323,1325,1327,1329,1331,1333,1335,1337,1339,1341,1343,1345,1347,1349,1351,1353,1355,1357,1359,1361,1363,1365,1367,1369,1371,1373,1375,1377,1379,1381,1383,1385,1387,1389,1391,1393,1395,1397,1399,1401,1403,1405,1407,1409,1411,1413,1415,1417,1419,1421,1423,1425,1427,1429,1431,1433,1435,1437,1439,1441,1443,1445,1447,1449,1451,1453,1455,1457,1459,1461,1463,1465,1467,1469,1471,1473,1475,1477,1479,1481,1483,1485,1487,1489,1491,1493,1495,1497,1499,1501,1503,1505,1507,1509,1511,1513,1515,1517,1519,1521,1523,1525,1527,1529,1531,1533,1535,1537,1539,1541,1543,1545,1547,1549,1551,1553,1555,1557,1559,1561,1563,1565,1567,1569,1571,1573,1575,1577,1579,1581,1583,1585,1587,1589,1591,1593,1595,1597,1599,1601,1603,1605,1607,1609,1611,1613,1615,1617,1619,1621,1623,1625,1627,1629,1631,1633,1635,1637,1639,1641,1643,1645,1647,1649,1651,1653,1655,1657,1659,1661,1663,1665,1667,1669,1671,1673,1675,1677,1679,1681,1683,1685,1687,1689,1691,1693,1695,1697,1699,1701,1703,1705,1707,1709,1711,1713,1715,1717,1719,1721,1723,1725,1727,1729,1731,1733,1735,1737,1739,1741,1743,1745,1747,1749,1751,1753,1755,1757,1759,1761,1763,1765,1767,1769,1771,1773,1775,1777,1779,1781,1783,1785,1787,1789,1791,1793,1795,1797,1799,1801,1803,1805,1807,1809,1811,1813,1815,1817,1819,1821,1823,1825,1827,1829,1831,1833,1835,1837,1839,1841,1843,1845,1847,1849,1851,1853,1855,1857,1859,1861,1863,1865,1867,1869,1871,1873,1875,1877,1879,1881,1883,1885,1887,1889,1891,1893,1895,1897,1899,1901,1903,1905,1907,1909,1911,1913,1915,1917,1919,1921,1923,1925,1927,1929,1931,1933,1935,1937,1939,1941,1943,1945,1947,1949,1951,1953,1955,1957,1959,1961,1963,1965,1967,1969,1971,1973,1975,1977,1979,1981,1983,1985,1987,1989,1991,1993,1995,1997,1999,2001,2003,2005,2007,2009,2011,2013,2015,2017,2019,2021,2023,2025,2027,2029,2031,2033,2035,2037,2039,2041,2043,2045,2047,2049,2051,2053,2055,2057,2059,2061,2063,2065,2067,2069,2071,2073,2075,2077,2079,2081,2083,2085,2087,2089,2091,2093,2095,2097,2099,2101,2103,2105,2107,2109,2111,2113,2115,2117,2119,2121,2123,2125,2127,2129,2131,2133,2135,2137,2139,2141,2143,2145,2147,2149,2151,2153,2155,2157,2159,2161,2163,2165,2167,2169,2171,2173,2175,2177,2179,2181,2183,2185,2187,2189,2191,2193,2195,2197,2199,2201,2203,2205,2207,2209,2211,2213,2215,2217,2219,2221,2223,2225,2227,2229,2231,2233,2235,2237,2239,2241,2243,2245,2247,2249,2251,2253,2255,2257,2259,2261,2263,2265,2267,2269,2271,2273,2275,2277,2279,2281,2283,2285,2287,2289,2291,2293,2295,2297,2299,2301,2303,2305,2307,2309,2311,2313,2315,2317,2319,2321,2323,2325,2327,2329,2331,2333,2335,2337,2339,2341,2343,2345,2347,2349,2351,2353,2355,2357,2359,2361,2363,2365,2367,2369,2371,2373,2375,2377,2379,2381,2383,2385,2387,2389,2391,2393,2395,2397,2399,2401,2403,2405,2407,2409,2411,2413,2415,2417,2419,2421,2423,2425,2427,2429,2431,2433,2435,2437,2439,2441,2443,2445,2447,2449,2451,2453,2455,2457,2459,2461,2463,2465,2467,2469,2471,2473,2475,2477,2479,2481,2483,2485,2487,2489,2491,2493,2495,2497,2499,2501,2503,2505,2507,2509,2511,2513,2515,2517,2519,2521,2523,2525,2527,2529,2531,2533,2535,2537,2539,2541,2543,2545,2547,2549,2551,2553,2555,2557,2559,2561,2563,2565,2567,2569,2571,2573,2575,2577,2579,2581,2583,2585,2587,2589,2591,2593,2595,2597,2599,2601,2603,2605,2607,2609,2611,2613,2615,2617,2619,2621,2623,2625,2627,2629,2631,2633,2635,2637,2639,2641,2643,2645,2647,2649,2651,2653,2655,2657,2659,2661,2663,2665,2667,2669,2671,2673,2675,2677,2679,2681,2683,2685,2687,2689,2691,2693,2695,2697,2699,2701,2703,2705,2707,2709,2711,2713,2715,2717,2719,2721,2723,2725,2727,2729,2731,2733,2735,2737,2739,2741,2743,2745,2747,2749,2751,2753,2755,2757,2759,2761,2763,2765,2767,2769,2771,2773,2775,2777,2779,2781,2783,2785,2787,2789,2791,2793,2795,2797,2799,2801,2803,2805,2807,2809,2811,2813,2815,2817,2819,2821,2823,2825,2827,2829,2831,2833,2835,2837,2839,2841,2843,2845,2847,2849,2851,2853,2855,2857,2859,2861,2863,2865,2867,2869,2871,2873,2875,2877,2879,2881,2883,2885,2887,2889,2891,2893,2895,2897,2899,2901,2903,2905,2907,2909,2911,2913,2915,2917,2919,2921,2923,2925,2927,2929,2931,2933,2935,2937,2939,2941,2943,2945,2947,2949,2951,2953,2955,2957,2959,2961,2963,2965,2967,2969,2971,2973,2975,2977,2979,2981,2983,2985,2987,2989,2991,2993,2995,2997,2999,3001,3003,3005,3007,3009,3011,3013,3015,3017,3019,3021,3023,3025,3027,3029,3031,3033,3035,3037,3039,3041,3043,3045,3047,3049,3051,3053,3055,3057,3059,3061,3063,3065,3067,3069,3071,3073,3075,3077,3079,3081,3083,3085,3087,3089,3091,3093,3095,3097,3099,3101,3103,3105,3107,3109,3111,3113,3115,3117,3119,3121,3123,3125,3127,3129,3131,3133,3135,3137,3139,3141,3143,3145,3147,3149,3151,3153,3155,3157,3159,3161,3163,3165,3167,3169,3171,3173,3175,3177,3179,3181,3183,3185,3187,3189,3191,3193,3195,3197,3199,3201,3203,3205,3207,3209,3211,3213,3215,3217,3219,3221,3223,3225,3227,3229,3231,3233,3235,3237,3239,3241,3243,3245,3247,3249,3251,3253,3255,3257,3259,3261,3263,3265,3267,3269,3271,3273,3275,3277,3279,3281,3283,3285,3287,3289,3291,3293,3295,3297,3299,3301,3303,3305,3307,3309,3311,3313,3315,3317,3319,3321,3323,3325,3327,3329,3331,3333,3335,3337,3339,3341,3343,3345,3347,3349,3351,3353,3355,3357,3359,3361,3363,3365,3367,3369,3371,3373,3375,3377,3379,3381,3383,3385,3387,3389,3391,3393,3395,3397,3399,3401,3403,3405,3407,3409,3411,3413,3415,3417,3419,3421,3423,3425,3427,3429,3431,3433,3435,3437,3439,3441,3443,3445,3447,3449,3451,3453,3455,3457,3459,3461,3463,3465,3467,3469,3471,3473,3475,3477,3479,3481,3483,3485,3487,3489,3491,3493,3495,3497,3499,3501,3503,3505,3507,3509,3511,3513,3515,3517,3519,3521,3523,3525,3527,3529,3531,3533,3535,3537,3539,3541,3543,3545,3547,3549,3551,3553,3555,3557,3559,3561,3563,3565,3567,3569,3571,3573,3575,3577,3579,3581,3583,3585,3587,3589,3591,3593,3595,3597,3599,3601,3603,3605,3607,3609,3611,3613,3615,3617,3619,3621,3623,3625,3627,3629,3631,3633,3635,3637,3639,3641,3643,3645,3647,3649,3651,3653,3655,3657,3659,3661,3663,3665,3667,3669,3671,3673,3675,3677,3679,3681,3683,3685,3687,3689,3691,3693,3695,3697,3699,3701,3703],{"categories":138},[139],"Developer Productivity",{"categories":141},[142],"Business & SaaS",{"categories":144},[103],{"categories":146},[147],"AI Automation",{"categories":149},[150],"Product Strategy",{"categories":152},[103],{"categories":154},[139],{"categories":156},[142],{"categories":158},[],{"categories":160},[103],{"categories":162},[],{"categories":164},[165],"AI News & Trends",{"categories":167},[147],{"categories":169},[165],{"categories":171},[147],{"categories":173},[147],{"categories":175},[103],{"categories":177},[103],{"categories":179},[165],{"categories":181},[103],{"categories":183},[],{"categories":185},[186],"Design & Frontend",{"categories":188},[189],"Data Science & Visualization",{"categories":191},[165],{"categories":193},[],{"categories":195},[196],"Software Engineering",{"categories":198},[103],{"categories":200},[147],{"categories":202},[203],"Marketing & Growth",{"categories":205},[103],{"categories":207},[147],{"categories":209},[],{"categories":211},[],{"categories":213},[186],{"categories":215},[147],{"categories":217},[139],{"categories":219},[186],{"categories":221},[103],{"categories":223},[147],{"categories":225},[165],{"categories":227},[],{"categories":229},[],{"categories":231},[147],{"categories":233},[196],{"categories":235},[],{"categories":237},[142],{"categories":239},[],{"categories":241},[],{"categories":243},[147],{"categories":245},[147],{"categories":247},[103],{"categories":249},[],{"categories":251},[196],{"categories":253},[],{"categories":255},[],{"categories":257},[],{"categories":259},[103],{"categories":261},[203],{"categories":263},[186],{"categories":265},[186],{"categories":267},[103],{"categories":269},[147],{"categories":271},[103],{"categories":273},[103],{"categories":275},[147],{"categories":277},[147],{"categories":279},[189],{"categories":281},[165],{"categories":283},[147],{"categories":285},[203],{"categories":287},[147],{"categories":289},[150],{"categories":291},[],{"categories":293},[147],{"categories":295},[],{"categories":297},[147],{"categories":299},[196],{"categories":301},[186],{"categories":303},[103],{"categories":305},[],{"categories":307},[],{"categories":309},[147],{"categories":311},[],{"categories":313},[103],{"categories":315},[],{"categories":317},[139],{"categories":319},[196],{"categories":321},[142],{"categories":323},[165],{"categories":325},[103],{"categories":327},[],{"categories":329},[103],{"categories":331},[],{"categories":333},[196],{"categories":335},[189],{"categories":337},[],{"categories":339},[103],{"categories":341},[186],{"categories":343},[],{"categories":345},[186],{"categories":347},[147],{"categories":349},[],{"categories":351},[147],{"categories":353},[165],{"categories":355},[103],{"categories":357},[],{"categories":359},[147],{"categories":361},[103],{"categories":363},[150],{"categories":365},[],{"categories":367},[103],{"categories":369},[147],{"categories":371},[147],{"categories":373},[],{"categories":375},[189],{"categories":377},[103],{"categories":379},[],{"categories":381},[139],{"categories":383},[142],{"categories":385},[103],{"categories":387},[147],{"categories":389},[196],{"categories":391},[103],{"categories":393},[],{"categories":395},[],{"categories":397},[103],{"categories":399},[],{"categories":401},[186],{"categories":403},[],{"categories":405},[103],{"categories":407},[],{"categories":409},[147],{"categories":411},[103],{"categories":413},[186],{"categories":415},[],{"categories":417},[103],{"categories":419},[103],{"categories":421},[142],{"categories":423},[147],{"categories":425},[103],{"categories":427},[186],{"categories":429},[147],{"categories":431},[],{"categories":433},[],{"categories":435},[165],{"categories":437},[],{"categories":439},[103],{"categories":441},[142,203],{"categories":443},[],{"categories":445},[103],{"categories":447},[],{"categories":449},[],{"categories":451},[103],{"categories":453},[],{"categories":455},[103],{"categories":457},[458],"DevOps & Cloud",{"categories":460},[],{"categories":462},[165],{"categories":464},[186],{"categories":466},[],{"categories":468},[165],{"categories":470},[165],{"categories":472},[103],{"categories":474},[203],{"categories":476},[],{"categories":478},[142],{"categories":480},[],{"categories":482},[103,458],{"categories":484},[103],{"categories":486},[103],{"categories":488},[147],{"categories":490},[103,196],{"categories":492},[189],{"categories":494},[103],{"categories":496},[203],{"categories":498},[147],{"categories":500},[147],{"categories":502},[],{"categories":504},[147],{"categories":506},[103,142],{"categories":508},[],{"categories":510},[186],{"categories":512},[186],{"categories":514},[],{"categories":516},[],{"categories":518},[165],{"categories":520},[],{"categories":522},[139],{"categories":524},[196],{"categories":526},[103],{"categories":528},[186],{"categories":530},[147],{"categories":532},[196],{"categories":534},[165],{"categories":536},[186],{"categories":538},[],{"categories":540},[103],{"categories":542},[103],{"categories":544},[103],{"categories":546},[165],{"categories":548},[139],{"categories":550},[103],{"categories":552},[147],{"categories":554},[458],{"categories":556},[186],{"categories":558},[147],{"categories":560},[],{"categories":562},[],{"categories":564},[186],{"categories":566},[165],{"categories":568},[189],{"categories":570},[],{"categories":572},[103],{"categories":574},[103],{"categories":576},[142],{"categories":578},[103],{"categories":580},[103],{"categories":582},[165],{"categories":584},[],{"categories":586},[147],{"categories":588},[196],{"categories":590},[],{"categories":592},[103],{"categories":594},[103],{"categories":596},[147],{"categories":598},[],{"categories":600},[],{"categories":602},[103],{"categories":604},[],{"categories":606},[142],{"categories":608},[147],{"categories":610},[],{"categories":612},[139],{"categories":614},[103],{"categories":616},[142],{"categories":618},[165],{"categories":620},[],{"categories":622},[],{"categories":624},[],{"categories":626},[165],{"categories":628},[165],{"categories":630},[],{"categories":632},[],{"categories":634},[142],{"categories":636},[],{"categories":638},[],{"categories":640},[139],{"categories":642},[],{"categories":644},[203],{"categories":646},[147],{"categories":648},[142],{"categories":650},[147],{"categories":652},[],{"categories":654},[150],{"categories":656},[186],{"categories":658},[196],{"categories":660},[103],{"categories":662},[147],{"categories":664},[142],{"categories":666},[103],{"categories":668},[],{"categories":670},[],{"categories":672},[196],{"categories":674},[189],{"categories":676},[150],{"categories":678},[147],{"categories":680},[103],{"categories":682},[],{"categories":684},[458],{"categories":686},[],{"categories":688},[147],{"categories":690},[],{"categories":692},[],{"categories":694},[103],{"categories":696},[186],{"categories":698},[203],{"categories":700},[147],{"categories":702},[],{"categories":704},[139],{"categories":706},[],{"categories":708},[165],{"categories":710},[103,458],{"categories":712},[165],{"categories":714},[103],{"categories":716},[142],{"categories":718},[103],{"categories":720},[],{"categories":722},[142],{"categories":724},[],{"categories":726},[196],{"categories":728},[186],{"categories":730},[165],{"categories":732},[189],{"categories":734},[139],{"categories":736},[103],{"categories":738},[196],{"categories":740},[],{"categories":742},[],{"categories":744},[150],{"categories":746},[],{"categories":748},[103],{"categories":750},[],{"categories":752},[186],{"categories":754},[186],{"categories":756},[186],{"categories":758},[],{"categories":760},[],{"categories":762},[165],{"categories":764},[147],{"categories":766},[103],{"categories":768},[103],{"categories":770},[103],{"categories":772},[142],{"categories":774},[103],{"categories":776},[],{"categories":778},[196],{"categories":780},[196],{"categories":782},[142],{"categories":784},[],{"categories":786},[103],{"categories":788},[103],{"categories":790},[142],{"categories":792},[165],{"categories":794},[203],{"categories":796},[147],{"categories":798},[],{"categories":800},[186],{"categories":802},[],{"categories":804},[103],{"categories":806},[],{"categories":808},[142],{"categories":810},[147],{"categories":812},[],{"categories":814},[458],{"categories":816},[189],{"categories":818},[196],{"categories":820},[203],{"categories":822},[196],{"categories":824},[147],{"categories":826},[],{"categories":828},[],{"categories":830},[147],{"categories":832},[139],{"categories":834},[147],{"categories":836},[150],{"categories":838},[142],{"categories":840},[],{"categories":842},[103],{"categories":844},[150],{"categories":846},[103],{"categories":848},[103],{"categories":850},[203],{"categories":852},[186],{"categories":854},[147],{"categories":856},[],{"categories":858},[],{"categories":860},[458],{"categories":862},[196],{"categories":864},[],{"categories":866},[147],{"categories":868},[103],{"categories":870},[186,103],{"categories":872},[139],{"categories":874},[],{"categories":876},[103],{"categories":878},[139],{"categories":880},[186],{"categories":882},[147],{"categories":884},[196],{"categories":886},[],{"categories":888},[103],{"categories":890},[],{"categories":892},[139],{"categories":894},[],{"categories":896},[147],{"categories":898},[150],{"categories":900},[103],{"categories":902},[103],{"categories":904},[186],{"categories":906},[147],{"categories":908},[458],{"categories":910},[186],{"categories":912},[147],{"categories":914},[103],{"categories":916},[103],{"categories":918},[103],{"categories":920},[165],{"categories":922},[],{"categories":924},[150],{"categories":926},[147],{"categories":928},[186],{"categories":930},[147],{"categories":932},[196],{"categories":934},[186],{"categories":936},[147],{"categories":938},[165],{"categories":940},[],{"categories":942},[103],{"categories":944},[186],{"categories":946},[103],{"categories":948},[139],{"categories":950},[165],{"categories":952},[103],{"categories":954},[203],{"categories":956},[103],{"categories":958},[103],{"categories":960},[147],{"categories":962},[147],{"categories":964},[103],{"categories":966},[147],{"categories":968},[186],{"categories":970},[103],{"categories":972},[],{"categories":974},[],{"categories":976},[196],{"categories":978},[],{"categories":980},[139],{"categories":982},[458],{"categories":984},[],{"categories":986},[139],{"categories":988},[142],{"categories":990},[203],{"categories":992},[],{"categories":994},[142],{"categories":996},[],{"categories":998},[],{"categories":1000},[],{"categories":1002},[],{"categories":1004},[],{"categories":1006},[103],{"categories":1008},[147],{"categories":1010},[458],{"categories":1012},[139],{"categories":1014},[103],{"categories":1016},[196],{"categories":1018},[150],{"categories":1020},[103],{"categories":1022},[203],{"categories":1024},[103],{"categories":1026},[103],{"categories":1028},[103],{"categories":1030},[103,139],{"categories":1032},[196],{"categories":1034},[196],{"categories":1036},[186],{"categories":1038},[103],{"categories":1040},[],{"categories":1042},[],{"categories":1044},[],{"categories":1046},[196],{"categories":1048},[189],{"categories":1050},[165],{"categories":1052},[186],{"categories":1054},[],{"categories":1056},[103],{"categories":1058},[103],{"categories":1060},[],{"categories":1062},[],{"categories":1064},[147],{"categories":1066},[103],{"categories":1068},[142],{"categories":1070},[],{"categories":1072},[139],{"categories":1074},[103],{"categories":1076},[139],{"categories":1078},[103],{"categories":1080},[196],{"categories":1082},[203],{"categories":1084},[103,186],{"categories":1086},[165],{"categories":1088},[186],{"categories":1090},[],{"categories":1092},[458],{"categories":1094},[186],{"categories":1096},[147],{"categories":1098},[],{"categories":1100},[],{"categories":1102},[],{"categories":1104},[],{"categories":1106},[196],{"categories":1108},[147],{"categories":1110},[147],{"categories":1112},[103],{"categories":1114},[103],{"categories":1116},[],{"categories":1118},[186],{"categories":1120},[],{"categories":1122},[],{"categories":1124},[147],{"categories":1126},[],{"categories":1128},[],{"categories":1130},[203],{"categories":1132},[203],{"categories":1134},[147],{"categories":1136},[],{"categories":1138},[103],{"categories":1140},[103],{"categories":1142},[196],{"categories":1144},[186],{"categories":1146},[186],{"categories":1148},[147],{"categories":1150},[139],{"categories":1152},[103],{"categories":1154},[186],{"categories":1156},[186],{"categories":1158},[147],{"categories":1160},[147],{"categories":1162},[103],{"categories":1164},[],{"categories":1166},[],{"categories":1168},[103],{"categories":1170},[147],{"categories":1172},[165],{"categories":1174},[196],{"categories":1176},[139],{"categories":1178},[103],{"categories":1180},[],{"categories":1182},[147],{"categories":1184},[147],{"categories":1186},[],{"categories":1188},[139],{"categories":1190},[103],{"categories":1192},[139],{"categories":1194},[139],{"categories":1196},[],{"categories":1198},[],{"categories":1200},[147],{"categories":1202},[147],{"categories":1204},[103],{"categories":1206},[103],{"categories":1208},[165],{"categories":1210},[189],{"categories":1212},[150],{"categories":1214},[165],{"categories":1216},[186],{"categories":1218},[],{"categories":1220},[165],{"categories":1222},[],{"categories":1224},[],{"categories":1226},[],{"categories":1228},[],{"categories":1230},[196],{"categories":1232},[189],{"categories":1234},[],{"categories":1236},[103],{"categories":1238},[103],{"categories":1240},[189],{"categories":1242},[196],{"categories":1244},[],{"categories":1246},[],{"categories":1248},[147],{"categories":1250},[165],{"categories":1252},[165],{"categories":1254},[147],{"categories":1256},[139],{"categories":1258},[103,458],{"categories":1260},[],{"categories":1262},[186],{"categories":1264},[139],{"categories":1266},[147],{"categories":1268},[186],{"categories":1270},[],{"categories":1272},[147],{"categories":1274},[147],{"categories":1276},[103],{"categories":1278},[203],{"categories":1280},[196],{"categories":1282},[186],{"categories":1284},[],{"categories":1286},[147],{"categories":1288},[103],{"categories":1290},[147],{"categories":1292},[147],{"categories":1294},[147],{"categories":1296},[203],{"categories":1298},[147],{"categories":1300},[103],{"categories":1302},[],{"categories":1304},[203],{"categories":1306},[165],{"categories":1308},[147],{"categories":1310},[],{"categories":1312},[],{"categories":1314},[103],{"categories":1316},[147],{"categories":1318},[165],{"categories":1320},[147],{"categories":1322},[],{"categories":1324},[],{"categories":1326},[],{"categories":1328},[147],{"categories":1330},[],{"categories":1332},[],{"categories":1334},[189],{"categories":1336},[103],{"categories":1338},[189],{"categories":1340},[165],{"categories":1342},[103],{"categories":1344},[103],{"categories":1346},[147],{"categories":1348},[103],{"categories":1350},[],{"categories":1352},[],{"categories":1354},[458],{"categories":1356},[],{"categories":1358},[],{"categories":1360},[139],{"categories":1362},[],{"categories":1364},[],{"categories":1366},[],{"categories":1368},[],{"categories":1370},[196],{"categories":1372},[165],{"categories":1374},[203],{"categories":1376},[142],{"categories":1378},[103],{"categories":1380},[103],{"categories":1382},[142],{"categories":1384},[],{"categories":1386},[186],{"categories":1388},[147],{"categories":1390},[142],{"categories":1392},[103],{"categories":1394},[103],{"categories":1396},[139],{"categories":1398},[],{"categories":1400},[139],{"categories":1402},[103],{"categories":1404},[203],{"categories":1406},[147],{"categories":1408},[165],{"categories":1410},[142],{"categories":1412},[103],{"categories":1414},[147],{"categories":1416},[],{"categories":1418},[103],{"categories":1420},[139],{"categories":1422},[103],{"categories":1424},[],{"categories":1426},[165],{"categories":1428},[103],{"categories":1430},[],{"categories":1432},[142],{"categories":1434},[103],{"categories":1436},[],{"categories":1438},[],{"categories":1440},[],{"categories":1442},[103],{"categories":1444},[],{"categories":1446},[458],{"categories":1448},[103],{"categories":1450},[],{"categories":1452},[103],{"categories":1454},[103],{"categories":1456},[103],{"categories":1458},[103,458],{"categories":1460},[103],{"categories":1462},[103],{"categories":1464},[186],{"categories":1466},[147],{"categories":1468},[],{"categories":1470},[147],{"categories":1472},[103],{"categories":1474},[103],{"categories":1476},[103],{"categories":1478},[139],{"categories":1480},[139],{"categories":1482},[196],{"categories":1484},[186],{"categories":1486},[147],{"categories":1488},[],{"categories":1490},[103],{"categories":1492},[165],{"categories":1494},[103],{"categories":1496},[142],{"categories":1498},[],{"categories":1500},[458],{"categories":1502},[186],{"categories":1504},[186],{"categories":1506},[147],{"categories":1508},[165],{"categories":1510},[147],{"categories":1512},[103],{"categories":1514},[],{"categories":1516},[103],{"categories":1518},[],{"categories":1520},[],{"categories":1522},[103],{"categories":1524},[103],{"categories":1526},[103],{"categories":1528},[147],{"categories":1530},[103],{"categories":1532},[],{"categories":1534},[189],{"categories":1536},[147],{"categories":1538},[],{"categories":1540},[103],{"categories":1542},[165],{"categories":1544},[],{"categories":1546},[186],{"categories":1548},[458],{"categories":1550},[165],{"categories":1552},[196],{"categories":1554},[196],{"categories":1556},[165],{"categories":1558},[165],{"categories":1560},[458],{"categories":1562},[],{"categories":1564},[165],{"categories":1566},[103],{"categories":1568},[139],{"categories":1570},[165],{"categories":1572},[],{"categories":1574},[189],{"categories":1576},[165],{"categories":1578},[196],{"categories":1580},[165],{"categories":1582},[458],{"categories":1584},[103],{"categories":1586},[103],{"categories":1588},[],{"categories":1590},[142],{"categories":1592},[],{"categories":1594},[],{"categories":1596},[103],{"categories":1598},[103],{"categories":1600},[103],{"categories":1602},[103],{"categories":1604},[],{"categories":1606},[189],{"categories":1608},[139],{"categories":1610},[],{"categories":1612},[103],{"categories":1614},[103],{"categories":1616},[458],{"categories":1618},[458],{"categories":1620},[],{"categories":1622},[147],{"categories":1624},[165],{"categories":1626},[165],{"categories":1628},[103],{"categories":1630},[147],{"categories":1632},[],{"categories":1634},[186],{"categories":1636},[103],{"categories":1638},[103],{"categories":1640},[],{"categories":1642},[],{"categories":1644},[458],{"categories":1646},[103],{"categories":1648},[196],{"categories":1650},[142],{"categories":1652},[103],{"categories":1654},[],{"categories":1656},[147],{"categories":1658},[139],{"categories":1660},[139],{"categories":1662},[],{"categories":1664},[103],{"categories":1666},[186],{"categories":1668},[147],{"categories":1670},[],{"categories":1672},[103],{"categories":1674},[103],{"categories":1676},[147],{"categories":1678},[],{"categories":1680},[147],{"categories":1682},[196],{"categories":1684},[],{"categories":1686},[103],{"categories":1688},[],{"categories":1690},[103],{"categories":1692},[],{"categories":1694},[103],{"categories":1696},[103],{"categories":1698},[],{"categories":1700},[103],{"categories":1702},[165],{"categories":1704},[103],{"categories":1706},[103],{"categories":1708},[139],{"categories":1710},[103],{"categories":1712},[165],{"categories":1714},[147],{"categories":1716},[],{"categories":1718},[103],{"categories":1720},[203],{"categories":1722},[],{"categories":1724},[],{"categories":1726},[],{"categories":1728},[139],{"categories":1730},[165],{"categories":1732},[147],{"categories":1734},[103],{"categories":1736},[186],{"categories":1738},[147],{"categories":1740},[],{"categories":1742},[147],{"categories":1744},[],{"categories":1746},[103],{"categories":1748},[147],{"categories":1750},[103],{"categories":1752},[],{"categories":1754},[103],{"categories":1756},[103],{"categories":1758},[165],{"categories":1760},[186],{"categories":1762},[147],{"categories":1764},[186],{"categories":1766},[142],{"categories":1768},[],{"categories":1770},[],{"categories":1772},[103],{"categories":1774},[139],{"categories":1776},[165],{"categories":1778},[],{"categories":1780},[],{"categories":1782},[196],{"categories":1784},[186],{"categories":1786},[],{"categories":1788},[103],{"categories":1790},[],{"categories":1792},[203],{"categories":1794},[103],{"categories":1796},[458],{"categories":1798},[196],{"categories":1800},[],{"categories":1802},[147],{"categories":1804},[103],{"categories":1806},[147],{"categories":1808},[147],{"categories":1810},[103],{"categories":1812},[],{"categories":1814},[139],{"categories":1816},[103],{"categories":1818},[142],{"categories":1820},[196],{"categories":1822},[186],{"categories":1824},[],{"categories":1826},[],{"categories":1828},[],{"categories":1830},[147],{"categories":1832},[186],{"categories":1834},[165],{"categories":1836},[103],{"categories":1838},[165],{"categories":1840},[186],{"categories":1842},[],{"categories":1844},[186],{"categories":1846},[165],{"categories":1848},[142],{"categories":1850},[103],{"categories":1852},[165],{"categories":1854},[203],{"categories":1856},[],{"categories":1858},[],{"categories":1860},[189],{"categories":1862},[103,196],{"categories":1864},[165],{"categories":1866},[103],{"categories":1868},[147],{"categories":1870},[147],{"categories":1872},[103],{"categories":1874},[],{"categories":1876},[196],{"categories":1878},[103],{"categories":1880},[189],{"categories":1882},[147],{"categories":1884},[203],{"categories":1886},[458],{"categories":1888},[],{"categories":1890},[139],{"categories":1892},[147],{"categories":1894},[147],{"categories":1896},[196],{"categories":1898},[103],{"categories":1900},[103],{"categories":1902},[],{"categories":1904},[],{"categories":1906},[],{"categories":1908},[458],{"categories":1910},[165],{"categories":1912},[103],{"categories":1914},[103],{"categories":1916},[103],{"categories":1918},[],{"categories":1920},[189],{"categories":1922},[142],{"categories":1924},[],{"categories":1926},[147],{"categories":1928},[458],{"categories":1930},[],{"categories":1932},[186],{"categories":1934},[186],{"categories":1936},[],{"categories":1938},[196],{"categories":1940},[186],{"categories":1942},[103],{"categories":1944},[],{"categories":1946},[165],{"categories":1948},[103],{"categories":1950},[186],{"categories":1952},[147],{"categories":1954},[165],{"categories":1956},[],{"categories":1958},[147],{"categories":1960},[186],{"categories":1962},[103],{"categories":1964},[],{"categories":1966},[103],{"categories":1968},[103],{"categories":1970},[458],{"categories":1972},[165],{"categories":1974},[189],{"categories":1976},[189],{"categories":1978},[],{"categories":1980},[],{"categories":1982},[],{"categories":1984},[147],{"categories":1986},[196],{"categories":1988},[196],{"categories":1990},[],{"categories":1992},[],{"categories":1994},[103],{"categories":1996},[],{"categories":1998},[147],{"categories":2000},[103],{"categories":2002},[],{"categories":2004},[103],{"categories":2006},[142],{"categories":2008},[103],{"categories":2010},[203],{"categories":2012},[147],{"categories":2014},[103],{"categories":2016},[196],{"categories":2018},[165],{"categories":2020},[147],{"categories":2022},[],{"categories":2024},[165],{"categories":2026},[147],{"categories":2028},[147],{"categories":2030},[],{"categories":2032},[142],{"categories":2034},[147],{"categories":2036},[],{"categories":2038},[103],{"categories":2040},[139],{"categories":2042},[165],{"categories":2044},[458],{"categories":2046},[147],{"categories":2048},[147],{"categories":2050},[139],{"categories":2052},[103],{"categories":2054},[],{"categories":2056},[],{"categories":2058},[186],{"categories":2060},[103,142],{"categories":2062},[],{"categories":2064},[139],{"categories":2066},[189],{"categories":2068},[103],{"categories":2070},[196],{"categories":2072},[103],{"categories":2074},[147],{"categories":2076},[103],{"categories":2078},[103],{"categories":2080},[165],{"categories":2082},[147],{"categories":2084},[],{"categories":2086},[],{"categories":2088},[147],{"categories":2090},[103],{"categories":2092},[458],{"categories":2094},[],{"categories":2096},[103],{"categories":2098},[147],{"categories":2100},[],{"categories":2102},[103],{"categories":2104},[203],{"categories":2106},[189],{"categories":2108},[147],{"categories":2110},[103],{"categories":2112},[458],{"categories":2114},[],{"categories":2116},[103],{"categories":2118},[203],{"categories":2120},[186],{"categories":2122},[103],{"categories":2124},[],{"categories":2126},[203],{"categories":2128},[165],{"categories":2130},[103],{"categories":2132},[103],{"categories":2134},[139],{"categories":2136},[],{"categories":2138},[],{"categories":2140},[186],{"categories":2142},[103],{"categories":2144},[189],{"categories":2146},[203],{"categories":2148},[203],{"categories":2150},[165],{"categories":2152},[],{"categories":2154},[],{"categories":2156},[103],{"categories":2158},[],{"categories":2160},[103,196],{"categories":2162},[165],{"categories":2164},[147],{"categories":2166},[196],{"categories":2168},[103],{"categories":2170},[139],{"categories":2172},[],{"categories":2174},[],{"categories":2176},[139],{"categories":2178},[203],{"categories":2180},[103],{"categories":2182},[],{"categories":2184},[186,103],{"categories":2186},[458],{"categories":2188},[139],{"categories":2190},[],{"categories":2192},[142],{"categories":2194},[142],{"categories":2196},[103],{"categories":2198},[196],{"categories":2200},[147],{"categories":2202},[165],{"categories":2204},[203],{"categories":2206},[186],{"categories":2208},[103],{"categories":2210},[103],{"categories":2212},[103],{"categories":2214},[139],{"categories":2216},[103],{"categories":2218},[147],{"categories":2220},[165],{"categories":2222},[],{"categories":2224},[],{"categories":2226},[189],{"categories":2228},[196],{"categories":2230},[103],{"categories":2232},[186],{"categories":2234},[189],{"categories":2236},[103],{"categories":2238},[103],{"categories":2240},[147],{"categories":2242},[147],{"categories":2244},[103,142],{"categories":2246},[],{"categories":2248},[186],{"categories":2250},[],{"categories":2252},[103],{"categories":2254},[165],{"categories":2256},[139],{"categories":2258},[139],{"categories":2260},[147],{"categories":2262},[103],{"categories":2264},[142],{"categories":2266},[196],{"categories":2268},[203],{"categories":2270},[],{"categories":2272},[165],{"categories":2274},[103],{"categories":2276},[103],{"categories":2278},[165],{"categories":2280},[196],{"categories":2282},[103],{"categories":2284},[147],{"categories":2286},[165],{"categories":2288},[103],{"categories":2290},[186],{"categories":2292},[103],{"categories":2294},[103],{"categories":2296},[458],{"categories":2298},[150],{"categories":2300},[147],{"categories":2302},[103],{"categories":2304},[165],{"categories":2306},[147],{"categories":2308},[203],{"categories":2310},[103],{"categories":2312},[],{"categories":2314},[103],{"categories":2316},[],{"categories":2318},[],{"categories":2320},[],{"categories":2322},[142],{"categories":2324},[103],{"categories":2326},[147],{"categories":2328},[165],{"categories":2330},[165],{"categories":2332},[165],{"categories":2334},[165],{"categories":2336},[],{"categories":2338},[139],{"categories":2340},[147],{"categories":2342},[165],{"categories":2344},[139],{"categories":2346},[147],{"categories":2348},[103],{"categories":2350},[103,147],{"categories":2352},[147],{"categories":2354},[458],{"categories":2356},[165],{"categories":2358},[165],{"categories":2360},[147],{"categories":2362},[103],{"categories":2364},[],{"categories":2366},[165],{"categories":2368},[203],{"categories":2370},[139],{"categories":2372},[103],{"categories":2374},[103],{"categories":2376},[],{"categories":2378},[196],{"categories":2380},[],{"categories":2382},[139],{"categories":2384},[147],{"categories":2386},[165],{"categories":2388},[103],{"categories":2390},[165],{"categories":2392},[139],{"categories":2394},[165],{"categories":2396},[165],{"categories":2398},[],{"categories":2400},[142],{"categories":2402},[147],{"categories":2404},[165],{"categories":2406},[165],{"categories":2408},[165],{"categories":2410},[165],{"categories":2412},[165],{"categories":2414},[165],{"categories":2416},[165],{"categories":2418},[165],{"categories":2420},[165],{"categories":2422},[165],{"categories":2424},[189],{"categories":2426},[139],{"categories":2428},[103],{"categories":2430},[103],{"categories":2432},[],{"categories":2434},[103,139],{"categories":2436},[],{"categories":2438},[147],{"categories":2440},[165],{"categories":2442},[147],{"categories":2444},[103],{"categories":2446},[103],{"categories":2448},[103],{"categories":2450},[103],{"categories":2452},[103],{"categories":2454},[147],{"categories":2456},[142],{"categories":2458},[186],{"categories":2460},[165],{"categories":2462},[103],{"categories":2464},[],{"categories":2466},[],{"categories":2468},[147],{"categories":2470},[186],{"categories":2472},[103],{"categories":2474},[],{"categories":2476},[],{"categories":2478},[203],{"categories":2480},[103],{"categories":2482},[],{"categories":2484},[],{"categories":2486},[139],{"categories":2488},[142],{"categories":2490},[103],{"categories":2492},[142],{"categories":2494},[186],{"categories":2496},[],{"categories":2498},[165],{"categories":2500},[],{"categories":2502},[186],{"categories":2504},[103],{"categories":2506},[203],{"categories":2508},[],{"categories":2510},[203],{"categories":2512},[],{"categories":2514},[],{"categories":2516},[147],{"categories":2518},[],{"categories":2520},[142],{"categories":2522},[139],{"categories":2524},[186],{"categories":2526},[196],{"categories":2528},[],{"categories":2530},[],{"categories":2532},[103],{"categories":2534},[139],{"categories":2536},[203],{"categories":2538},[],{"categories":2540},[147],{"categories":2542},[147],{"categories":2544},[165],{"categories":2546},[103],{"categories":2548},[147],{"categories":2550},[103],{"categories":2552},[147],{"categories":2554},[103],{"categories":2556},[150],{"categories":2558},[165],{"categories":2560},[],{"categories":2562},[203],{"categories":2564},[196],{"categories":2566},[147],{"categories":2568},[],{"categories":2570},[103],{"categories":2572},[147],{"categories":2574},[142],{"categories":2576},[139],{"categories":2578},[103],{"categories":2580},[186],{"categories":2582},[196],{"categories":2584},[196],{"categories":2586},[103],{"categories":2588},[189],{"categories":2590},[103],{"categories":2592},[147],{"categories":2594},[142],{"categories":2596},[147],{"categories":2598},[103],{"categories":2600},[103],{"categories":2602},[147],{"categories":2604},[165],{"categories":2606},[],{"categories":2608},[139],{"categories":2610},[103],{"categories":2612},[147],{"categories":2614},[103],{"categories":2616},[103],{"categories":2618},[],{"categories":2620},[186],{"categories":2622},[142],{"categories":2624},[165],{"categories":2626},[103],{"categories":2628},[103],{"categories":2630},[186],{"categories":2632},[203],{"categories":2634},[189],{"categories":2636},[103],{"categories":2638},[165],{"categories":2640},[103],{"categories":2642},[147],{"categories":2644},[458],{"categories":2646},[103],{"categories":2648},[147],{"categories":2650},[189],{"categories":2652},[],{"categories":2654},[147],{"categories":2656},[196],{"categories":2658},[186],{"categories":2660},[103],{"categories":2662},[139],{"categories":2664},[142],{"categories":2666},[196],{"categories":2668},[],{"categories":2670},[147],{"categories":2672},[103],{"categories":2674},[],{"categories":2676},[165],{"categories":2678},[],{"categories":2680},[165],{"categories":2682},[103],{"categories":2684},[147],{"categories":2686},[147],{"categories":2688},[147],{"categories":2690},[],{"categories":2692},[],{"categories":2694},[103],{"categories":2696},[103],{"categories":2698},[],{"categories":2700},[186],{"categories":2702},[147],{"categories":2704},[203],{"categories":2706},[139],{"categories":2708},[],{"categories":2710},[],{"categories":2712},[165],{"categories":2714},[196],{"categories":2716},[103],{"categories":2718},[103],{"categories":2720},[103],{"categories":2722},[196],{"categories":2724},[165],{"categories":2726},[186],{"categories":2728},[103],{"categories":2730},[103],{"categories":2732},[103],{"categories":2734},[165],{"categories":2736},[103],{"categories":2738},[165],{"categories":2740},[147],{"categories":2742},[147],{"categories":2744},[196],{"categories":2746},[147],{"categories":2748},[103],{"categories":2750},[196],{"categories":2752},[186],{"categories":2754},[],{"categories":2756},[147],{"categories":2758},[],{"categories":2760},[],{"categories":2762},[142],{"categories":2764},[103],{"categories":2766},[147],{"categories":2768},[139],{"categories":2770},[147],{"categories":2772},[203],{"categories":2774},[],{"categories":2776},[147],{"categories":2778},[],{"categories":2780},[139],{"categories":2782},[147],{"categories":2784},[],{"categories":2786},[147],{"categories":2788},[103],{"categories":2790},[165],{"categories":2792},[103],{"categories":2794},[147],{"categories":2796},[165],{"categories":2798},[147],{"categories":2800},[196],{"categories":2802},[186],{"categories":2804},[139],{"categories":2806},[],{"categories":2808},[147],{"categories":2810},[186],{"categories":2812},[165],{"categories":2814},[103],{"categories":2816},[186],{"categories":2818},[139],{"categories":2820},[],{"categories":2822},[147],{"categories":2824},[147],{"categories":2826},[103],{"categories":2828},[],{"categories":2830},[147],{"categories":2832},[150],{"categories":2834},[165],{"categories":2836},[147],{"categories":2838},[142],{"categories":2840},[],{"categories":2842},[103],{"categories":2844},[150],{"categories":2846},[103],{"categories":2848},[147],{"categories":2850},[165],{"categories":2852},[139],{"categories":2854},[458],{"categories":2856},[103],{"categories":2858},[103],{"categories":2860},[103],{"categories":2862},[165],{"categories":2864},[142],{"categories":2866},[103],{"categories":2868},[186],{"categories":2870},[165],{"categories":2872},[458],{"categories":2874},[103],{"categories":2876},[],{"categories":2878},[],{"categories":2880},[458],{"categories":2882},[189],{"categories":2884},[147],{"categories":2886},[147],{"categories":2888},[165],{"categories":2890},[103],{"categories":2892},[139],{"categories":2894},[186],{"categories":2896},[147],{"categories":2898},[103],{"categories":2900},[203],{"categories":2902},[103],{"categories":2904},[147],{"categories":2906},[],{"categories":2908},[103],{"categories":2910},[103],{"categories":2912},[165],{"categories":2914},[139],{"categories":2916},[],{"categories":2918},[103],{"categories":2920},[103],{"categories":2922},[196],{"categories":2924},[186],{"categories":2926},[103,147],{"categories":2928},[203,142],{"categories":2930},[103],{"categories":2932},[],{"categories":2934},[147],{"categories":2936},[],{"categories":2938},[196],{"categories":2940},[103],{"categories":2942},[165],{"categories":2944},[],{"categories":2946},[147],{"categories":2948},[],{"categories":2950},[147],{"categories":2952},[139],{"categories":2954},[147],{"categories":2956},[103],{"categories":2958},[458],{"categories":2960},[203],{"categories":2962},[142],{"categories":2964},[142],{"categories":2966},[139],{"categories":2968},[139],{"categories":2970},[103],{"categories":2972},[147],{"categories":2974},[103],{"categories":2976},[103],{"categories":2978},[139],{"categories":2980},[103],{"categories":2982},[203],{"categories":2984},[165],{"categories":2986},[103],{"categories":2988},[147],{"categories":2990},[103],{"categories":2992},[],{"categories":2994},[196],{"categories":2996},[],{"categories":2998},[147],{"categories":3000},[139],{"categories":3002},[],{"categories":3004},[458],{"categories":3006},[103],{"categories":3008},[],{"categories":3010},[165],{"categories":3012},[147],{"categories":3014},[196],{"categories":3016},[103],{"categories":3018},[147],{"categories":3020},[196],{"categories":3022},[147],{"categories":3024},[165],{"categories":3026},[139],{"categories":3028},[165],{"categories":3030},[196],{"categories":3032},[103],{"categories":3034},[186],{"categories":3036},[103],{"categories":3038},[103],{"categories":3040},[103],{"categories":3042},[103],{"categories":3044},[147],{"categories":3046},[103],{"categories":3048},[147],{"categories":3050},[103],{"categories":3052},[139],{"categories":3054},[103],{"categories":3056},[147],{"categories":3058},[186],{"categories":3060},[139],{"categories":3062},[147],{"categories":3064},[186],{"categories":3066},[],{"categories":3068},[103],{"categories":3070},[103],{"categories":3072},[196],{"categories":3074},[],{"categories":3076},[147],{"categories":3078},[203],{"categories":3080},[103],{"categories":3082},[165],{"categories":3084},[203],{"categories":3086},[147],{"categories":3088},[142],{"categories":3090},[142],{"categories":3092},[103],{"categories":3094},[139],{"categories":3096},[],{"categories":3098},[103],{"categories":3100},[],{"categories":3102},[139],{"categories":3104},[103],{"categories":3106},[147],{"categories":3108},[147],{"categories":3110},[],{"categories":3112},[196],{"categories":3114},[196],{"categories":3116},[203],{"categories":3118},[186],{"categories":3120},[],{"categories":3122},[103],{"categories":3124},[139],{"categories":3126},[103],{"categories":3128},[196],{"categories":3130},[139],{"categories":3132},[165],{"categories":3134},[165],{"categories":3136},[],{"categories":3138},[165],{"categories":3140},[147],{"categories":3142},[186],{"categories":3144},[189],{"categories":3146},[103],{"categories":3148},[],{"categories":3150},[165],{"categories":3152},[196],{"categories":3154},[142],{"categories":3156},[103],{"categories":3158},[139],{"categories":3160},[458],{"categories":3162},[139],{"categories":3164},[],{"categories":3166},[],{"categories":3168},[165],{"categories":3170},[],{"categories":3172},[147],{"categories":3174},[147],{"categories":3176},[147],{"categories":3178},[],{"categories":3180},[103],{"categories":3182},[],{"categories":3184},[165],{"categories":3186},[139],{"categories":3188},[186],{"categories":3190},[103],{"categories":3192},[165],{"categories":3194},[165],{"categories":3196},[],{"categories":3198},[165],{"categories":3200},[139],{"categories":3202},[103],{"categories":3204},[],{"categories":3206},[147],{"categories":3208},[147],{"categories":3210},[139],{"categories":3212},[],{"categories":3214},[],{"categories":3216},[],{"categories":3218},[186],{"categories":3220},[147],{"categories":3222},[103],{"categories":3224},[],{"categories":3226},[],{"categories":3228},[],{"categories":3230},[186],{"categories":3232},[],{"categories":3234},[139],{"categories":3236},[],{"categories":3238},[],{"categories":3240},[186],{"categories":3242},[103],{"categories":3244},[165],{"categories":3246},[],{"categories":3248},[203],{"categories":3250},[165],{"categories":3252},[203],{"categories":3254},[103],{"categories":3256},[],{"categories":3258},[],{"categories":3260},[147],{"categories":3262},[],{"categories":3264},[],{"categories":3266},[147],{"categories":3268},[103],{"categories":3270},[],{"categories":3272},[147],{"categories":3274},[165],{"categories":3276},[203],{"categories":3278},[189],{"categories":3280},[147],{"categories":3282},[147],{"categories":3284},[],{"categories":3286},[],{"categories":3288},[],{"categories":3290},[165],{"categories":3292},[],{"categories":3294},[],{"categories":3296},[186],{"categories":3298},[139],{"categories":3300},[],{"categories":3302},[142],{"categories":3304},[203],{"categories":3306},[103],{"categories":3308},[196],{"categories":3310},[139],{"categories":3312},[189],{"categories":3314},[142],{"categories":3316},[196],{"categories":3318},[],{"categories":3320},[],{"categories":3322},[147],{"categories":3324},[139],{"categories":3326},[186],{"categories":3328},[139],{"categories":3330},[147],{"categories":3332},[458],{"categories":3334},[147],{"categories":3336},[],{"categories":3338},[103],{"categories":3340},[165],{"categories":3342},[196],{"categories":3344},[],{"categories":3346},[186],{"categories":3348},[165],{"categories":3350},[139],{"categories":3352},[147],{"categories":3354},[103],{"categories":3356},[142],{"categories":3358},[147,458],{"categories":3360},[147],{"categories":3362},[196],{"categories":3364},[103],{"categories":3366},[189],{"categories":3368},[203],{"categories":3370},[147],{"categories":3372},[],{"categories":3374},[147],{"categories":3376},[103],{"categories":3378},[142],{"categories":3380},[],{"categories":3382},[],{"categories":3384},[103],{"categories":3386},[189],{"categories":3388},[103],{"categories":3390},[],{"categories":3392},[165],{"categories":3394},[],{"categories":3396},[165],{"categories":3398},[196],{"categories":3400},[147],{"categories":3402},[103],{"categories":3404},[203],{"categories":3406},[196],{"categories":3408},[],{"categories":3410},[165],{"categories":3412},[103],{"categories":3414},[],{"categories":3416},[103],{"categories":3418},[147],{"categories":3420},[103],{"categories":3422},[147],{"categories":3424},[103],{"categories":3426},[103],{"categories":3428},[103],{"categories":3430},[103],{"categories":3432},[142],{"categories":3434},[],{"categories":3436},[150],{"categories":3438},[165],{"categories":3440},[103],{"categories":3442},[],{"categories":3444},[196],{"categories":3446},[103],{"categories":3448},[103],{"categories":3450},[147],{"categories":3452},[165],{"categories":3454},[103],{"categories":3456},[103],{"categories":3458},[142],{"categories":3460},[147],{"categories":3462},[186],{"categories":3464},[],{"categories":3466},[189],{"categories":3468},[103],{"categories":3470},[],{"categories":3472},[165],{"categories":3474},[203],{"categories":3476},[],{"categories":3478},[],{"categories":3480},[165],{"categories":3482},[165],{"categories":3484},[203],{"categories":3486},[139],{"categories":3488},[147],{"categories":3490},[147],{"categories":3492},[103],{"categories":3494},[142],{"categories":3496},[],{"categories":3498},[],{"categories":3500},[165],{"categories":3502},[189],{"categories":3504},[196],{"categories":3506},[147],{"categories":3508},[186],{"categories":3510},[189],{"categories":3512},[189],{"categories":3514},[],{"categories":3516},[165],{"categories":3518},[103],{"categories":3520},[103],{"categories":3522},[196],{"categories":3524},[],{"categories":3526},[165],{"categories":3528},[165],{"categories":3530},[165],{"categories":3532},[],{"categories":3534},[147],{"categories":3536},[103],{"categories":3538},[],{"categories":3540},[139],{"categories":3542},[142],{"categories":3544},[],{"categories":3546},[103],{"categories":3548},[103],{"categories":3550},[],{"categories":3552},[196],{"categories":3554},[],{"categories":3556},[],{"categories":3558},[],{"categories":3560},[],{"categories":3562},[103],{"categories":3564},[165],{"categories":3566},[],{"categories":3568},[],{"categories":3570},[103],{"categories":3572},[103],{"categories":3574},[103],{"categories":3576},[189],{"categories":3578},[103],{"categories":3580},[189],{"categories":3582},[],{"categories":3584},[189],{"categories":3586},[189],{"categories":3588},[458],{"categories":3590},[147],{"categories":3592},[196],{"categories":3594},[],{"categories":3596},[],{"categories":3598},[189],{"categories":3600},[196],{"categories":3602},[196],{"categories":3604},[196],{"categories":3606},[],{"categories":3608},[139],{"categories":3610},[196],{"categories":3612},[196],{"categories":3614},[139],{"categories":3616},[196],{"categories":3618},[142],{"categories":3620},[196],{"categories":3622},[196],{"categories":3624},[196],{"categories":3626},[189],{"categories":3628},[165],{"categories":3630},[165],{"categories":3632},[103],{"categories":3634},[196],{"categories":3636},[189],{"categories":3638},[458],{"categories":3640},[189],{"categories":3642},[189],{"categories":3644},[189],{"categories":3646},[],{"categories":3648},[142],{"categories":3650},[],{"categories":3652},[458],{"categories":3654},[196],{"categories":3656},[196],{"categories":3658},[196],{"categories":3660},[147],{"categories":3662},[165,142],{"categories":3664},[189],{"categories":3666},[],{"categories":3668},[],{"categories":3670},[189],{"categories":3672},[],{"categories":3674},[189],{"categories":3676},[165],{"categories":3678},[147],{"categories":3680},[],{"categories":3682},[196],{"categories":3684},[103],{"categories":3686},[186],{"categories":3688},[],{"categories":3690},[103],{"categories":3692},[],{"categories":3694},[165],{"categories":3696},[139],{"categories":3698},[189],{"categories":3700},[],{"categories":3702},[196],{"categories":3704},[165],[3706,3838,4008,4458],{"id":3707,"title":3708,"ai":3709,"body":3714,"categories":3816,"created_at":104,"date_modified":104,"description":96,"extension":105,"faq":104,"featured":106,"kicker_label":104,"meta":3817,"navigation":118,"path":3826,"published_at":3827,"question":104,"scraped_at":3828,"seo":3829,"sitemap":3830,"source_id":3831,"source_name":125,"source_type":126,"source_url":3832,"stem":3833,"tags":3834,"thumbnail_url":104,"tldr":3835,"tweet":104,"unknown_tags":3836,"__hash__":3837},"summaries\u002Fsummaries\u002Fb7634f3fd3506434-5-prompt-techniques-for-reliable-llm-outputs-summary.md","5 Prompt Techniques for Reliable LLM Outputs",{"provider":7,"model":8,"input_tokens":3710,"output_tokens":3711,"processing_time_ms":3712,"cost_usd":3713},8938,1700,29836,0.00261495,{"type":14,"value":3715,"toc":3811},[3716,3720,3723,3726,3730,3741,3753,3800,3804,3807],[17,3717,3719],{"id":3718},"condition-model-responses-with-personas-and-constraints","Condition Model Responses with Personas and Constraints",[22,3721,3722],{},"Assign domain-specific roles in the system prompt to filter the model's knowledge and shift framing toward expert priorities. For a web app storing session tokens in localStorage, a generic assistant notes XSS risks and tradeoffs, but a 'senior application security researcher specializing in web authentication vulnerabilities' frames it as an attack surface: attackers steal tokens via XSS to hijack sessions, referencing OWASP guidelines and recommending HttpOnly cookies.",[22,3724,3725],{},"Combine roles with negative prompting to eliminate RLHF-induced noise like hedging, analogies, filler phrases ('great question'), and redundant summaries. Prompt a 'senior backend engineer writing internal documentation' with rules: no marketing language, resolve 'it depends' immediately, limit analogies to one sentence if needed, and stop after making the point. For explaining database indexes, this cuts verbose baseline (with headers, analogies, conclusion) to concise facts: indexes speed queries on WHERE\u002FJOIN\u002FORDER BY clauses via B-trees, use on high-cardinality filtered columns, avoid on low-cardinality or write-heavy tables.",[17,3727,3729],{"id":3728},"enforce-parseable-structures-with-json-and-arq","Enforce Parseable Structures with JSON and ARQ",[22,3731,3732,3733,3736,3737,3740],{},"Define exact JSON schemas in prompts to constrain outputs for code consumption, eliminating inconsistent free-form text. For product review parsing, specify schema with 'overall_sentiment' (positive\u002Fnegative\u002Fmixed), 'rating' (1-5 integer), 'pros'\u002F'cons' arrays, 'recommended_for'\u002F'not_recommended_for' strings. System prompt: 'You MUST return only a valid JSON object. No preamble, no explanation.' Baseline mixes pros\u002Fcons in narrative; JSON yields {'overall_sentiment': 'mixed', 'rating': 3, 'pros': ",[26,3734,3735],{},"'Stunning display', 'Comfortable keyboard'",", 'cons': ",[26,3738,3739],{},"'Poor battery life (6-hour workday)', 'Aggressive fan noise'",", 'recommended_for': 'Light work users', 'not_recommended_for': 'Heavy software runners'}. Parse directly with json.loads() for storage\u002Fquerying.",[22,3742,3743,3744,3748,3749,3752],{},"Attentive Reasoning Queries (ARQ) impose ordered, domain-specific checklists to cover all angles, surpassing unstructured chain-of-thought. For code review of unsafe SQL (",[3745,3746,3747],"code",{},"f\"SELECT * FROM users WHERE id = {user_id}\"","), list Q1-Security (SQL injection via unsanitized user_id), Q2-Error handling (unhandled db.execute() exception crashes), Q3-Performance (SELECT * fetches unnecessary columns, scales poorly), Q4-Correctness (result",[26,3750,3751],{},"0"," assumes single row, fails multi-row), Q5-Fix (parameterized query, SELECT specific columns, fetchone(), error handling). Baseline drifts; ARQ delivers systematic analysis and fixed code:",[3754,3755,3758],"pre",{"className":3756,"code":3757,"language":132,"meta":96,"style":96},"language-python shiki shiki-themes github-light github-dark","def get_user(user_id):\n    try:\n        query = \"SELECT id, username, email FROM users WHERE id = %s\"\n        result = db.execute(query, (user_id,))\n        return dict(result.fetchone()) if result.fetchone() else None\n    except Exception:\n        return None\n",[3745,3759,3760,3767,3772,3778,3783,3788,3794],{"__ignoreMap":96},[26,3761,3764],{"class":3762,"line":3763},"line",1,[26,3765,3766],{},"def get_user(user_id):\n",[26,3768,3769],{"class":3762,"line":97},[26,3770,3771],{},"    try:\n",[26,3773,3775],{"class":3762,"line":3774},3,[26,3776,3777],{},"        query = \"SELECT id, username, email FROM users WHERE id = %s\"\n",[26,3779,3780],{"class":3762,"line":115},[26,3781,3782],{},"        result = db.execute(query, (user_id,))\n",[26,3784,3785],{"class":3762,"line":114},[26,3786,3787],{},"        return dict(result.fetchone()) if result.fetchone() else None\n",[26,3789,3791],{"class":3762,"line":3790},6,[26,3792,3793],{},"    except Exception:\n",[26,3795,3797],{"class":3762,"line":3796},7,[26,3798,3799],{},"        return None\n",[17,3801,3803],{"id":3802},"generate-multiple-hypotheses-to-reveal-uncertainty","Generate Multiple Hypotheses to Reveal Uncertainty",[22,3805,3806],{},"Verbalized sampling prompts for 3+ ranked hypotheses with confidence (0.0-1.0), failure modes, validation info, and agent action, countering single confident outputs. For support ticket ('can't log in, password reset email missing'), baseline picks one issue; verbalized lists: 1. Email Delivery (0.85: no email arrives; confirm spam\u002FDNS), 2. Account State (0.70: new account locked; check flags), 3. Authentication (0.40: bad creds; verify recent login). Recommends: Ask for email provider and check spam. This aids prioritization without ensemble sampling.",[3808,3809,3810],"style",{},"html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":96,"searchDepth":97,"depth":97,"links":3812},[3813,3814,3815],{"id":3718,"depth":97,"text":3719},{"id":3728,"depth":97,"text":3729},{"id":3802,"depth":97,"text":3803},[],{"content_references":3818,"triage":3823},[3819],{"type":110,"title":3820,"url":3821,"context":3822},"Prompt_Techniques.ipynb","https:\u002F\u002Fgithub.com\u002FMarktechpost\u002FAI-Agents-Projects-Tutorials\u002Fblob\u002Fmain\u002FLLM%20Projects\u002FPrompt_Techniques.ipynb","recommended",{"relevance":114,"novelty":115,"quality":115,"actionability":114,"composite":3824,"reasoning":3825},4.55,"Category: AI & LLMs. The article provides practical techniques for prompt engineering that directly address the audience's need for actionable strategies in building AI-powered products. It details specific methods like using JSON schemas and role-specific personas, which can be immediately applied to improve LLM outputs.","\u002Fsummaries\u002Fb7634f3fd3506434-5-prompt-techniques-for-reliable-llm-outputs-summary","2026-05-03 21:41:48","2026-05-04 16:13:42",{"title":3708,"description":96},{"loc":3826},"b7634f3fd3506434","https:\u002F\u002Fwww.marktechpost.com\u002F2026\u002F05\u002F03\u002Fa-developers-guide-to-systematic-prompting-mastering-negative-constraints-structured-json-outputs-and-multi-hypothesis-verbalized-sampling\u002F","summaries\u002Fb7634f3fd3506434-5-prompt-techniques-for-reliable-llm-outputs-summary",[131,130,132],"Role-specific personas, negative constraints, JSON schemas, ARQ checklists, and verbalized sampling make LLM prompts produce consistent, structured results without fine-tuning or model changes.",[],"yBqYLAeToNXiC0rI1ewcoqNB5fac2Z-m-wU2j3V-ZVo",{"id":3839,"title":3840,"ai":3841,"body":3846,"categories":3967,"created_at":104,"date_modified":104,"description":96,"extension":105,"faq":104,"featured":106,"kicker_label":104,"meta":3968,"navigation":118,"path":3994,"published_at":3995,"question":104,"scraped_at":3996,"seo":3997,"sitemap":3998,"source_id":3999,"source_name":4000,"source_type":126,"source_url":4001,"stem":4002,"tags":4003,"thumbnail_url":104,"tldr":4005,"tweet":104,"unknown_tags":4006,"__hash__":4007},"summaries\u002Fsummaries\u002F67334f5912c7ef54-agent-harness-9-components-beyond-frameworks-summary.md","Agent Harness: 9 Components Beyond Frameworks",{"provider":7,"model":8,"input_tokens":3842,"output_tokens":3843,"processing_time_ms":3844,"cost_usd":3845},7919,2183,28813,0.00216845,{"type":14,"value":3847,"toc":3962},[3848,3852,3855,3858,3862,3865,3923,3926,3930,3933,3936,3939,3950,3953,3956,3959],[17,3849,3851],{"id":3850},"harness-delivers-ready-agents-frameworks-require-wiring","Harness Delivers Ready Agents, Frameworks Require Wiring",[22,3853,3854],{},"Turn one-shot LLMs into agents by wrapping them in a fixed harness architecture: a while loop that lets the model act (via tools), observe results, and iterate until solving the goal or hitting an iteration cap. This contrasts with frameworks like LangChain, LangGraph, AutoGen, and CrewAI, which provide abstractions (chains, memory, retrievers) for humans to assemble agents. Harnesses ship pre-wired for immediate use—you input a goal, it handles the rest. Examples include coding tools like Cursor and Claude Code, which evolved similar architectures for repo-wide code editing, starting from concrete problems rather than general abstractions.",[22,3856,3857],{},"Trade-off: Frameworks offer flexibility for custom agents but demand architecture work; harnesses prioritize out-of-box reliability, assuming the fixed loop + registry covers 80% of needs.",[17,3859,3861],{"id":3860},"_9-components-for-production-harnesses","9 Components for Production Harnesses",[22,3863,3864],{},"Build robust agents with these interconnected parts, drawn from tools like Claude Code (200k tokens budget, now 1M for Opus):",[3866,3867,3868,3875,3881,3887,3893,3899,3905,3911,3917],"ol",{},[58,3869,3870,3874],{},[3871,3872,3873],"strong",{},"While Loop Engine",": Core iteration—model reads system prompt, calls tools, feeds results back, repeats until text-only response or max iterations (prevents infinite loops).",[58,3876,3877,3880],{},[3871,3878,3879],{},"Context Management & Compaction",": Tree-like context grows with messages\u002Ftools; at 80-90% of limit (e.g., half of 1M tokens), keep recent messages verbatim, summarize older ones. Poor compaction loses critical history, causing failures.",[58,3882,3883,3886],{},[3871,3884,3885],{},"Tools vs Skills + Registry",": Tools are primitives (read file, run bash); skills encode team knowledge via Markdown files (e.g., git commit process). Registry maps names to handlers, permissions, descriptions—model sees lightweight descriptors to decide calls.",[58,3888,3889,3892],{},[3871,3890,3891],{},"Subagent Management",": For parallel\u002Fbig tasks, spawn isolated subagents with restricted tools, focused prompts, own sessions—span, restrict, collect outputs.",[58,3894,3895,3898],{},[3871,3896,3897],{},"Built-in Skills",": Ship essentials like file read\u002Fwrite\u002Fedit\u002Fsearch, bash execution, code navigation, git commits, PRs, tests. Use stdlib only for primitives to avoid deps.",[58,3900,3901,3904],{},[3871,3902,3903],{},"Session Persistence\u002FMemory",": Append-only JSON\u002FMarkdown logs every event (messages, tools, compactions) to disk for crash-proof resumption—replay rebuilds state exactly.",[58,3906,3907,3910],{},[3871,3908,3909],{},"Dynamic System Prompt Assembly",": Pipeline scans directories for files like CLAUDE.md or AGENTS.md, injects after static prefix (order preserves caching). Enables contextual instructions without hardcoding.",[58,3912,3913,3916],{},[3871,3914,3915],{},"Lifecycle Hooks",": Pre-tool: allow\u002Fdeny\u002Fmodify calls (JSON exit codes). Post-tool: audit results, log. Enables extensibility without core changes, key for enterprise.",[58,3918,3919,3922],{},[3871,3920,3921],{},"Permissions\u002FSafety",": Tools declare min perms (read-only, workspace, full). Harness enforces at dispatch; dynamic classification for bash (ls=read, rm=full); interactive user approvals for risky actions.",[22,3924,3925],{},"These make harnesses safe and durable—e.g., Anthropic separates session mgmt from core for scalability.",[17,3927,3929],{"id":3928},"python-reference-implementation-template","Python Reference Implementation Template",[22,3931,3932],{},"Core engine: While loop assembles dynamic prompt, compacts context if oversized (summarize old), handles tool\u002Fsubagent calls, caps iterations. Tools\u002Fskills as dataclasses (name, perms, handler, desc) in dict registry—descriptors for model, skills load MD on invoke.",[22,3934,3935],{},"Subagents: Archetypes (explore\u002Fgeneral\u002Fverify) with perm\u002Ftool restrictions, focused prompts.",[22,3937,3938],{},"Built-ins: Stdlib file read\u002Fbash.",[22,3940,3941,3942,3945,3946,3949],{},"Memory: ",[3745,3943,3944],{},"append(event)"," writes JSON lines (flush for durability); ",[3745,3947,3948],{},"replay()"," reconstructs.",[22,3951,3952],{},"Prompts: Static + dynamic dir scan (static first).",[22,3954,3955],{},"Hooks: Pre\u002Fpost functions on tool events.",[22,3957,3958],{},"Permissions: Check declared + dynamic parse (safe=read like grep; dangerous=full like sudo); user approve.",[22,3960,3961],{},"This ~100-line skeleton supports all 9 components—extend by registering tools\u002Fskills, no framework deps.",{"title":96,"searchDepth":97,"depth":97,"links":3963},[3964,3965,3966],{"id":3850,"depth":97,"text":3851},{"id":3860,"depth":97,"text":3861},{"id":3928,"depth":97,"text":3929},[103],{"content_references":3969,"triage":3992},[3970,3973,3975,3977,3979,3981,3983,3986,3989],{"type":3971,"title":3972,"context":112},"tool","LangChain",{"type":3971,"title":3974,"context":112},"LangGraph",{"type":3971,"title":3976,"context":112},"AutoGen",{"type":3971,"title":3978,"context":112},"CrewAI",{"type":3971,"title":3980,"context":112},"Cursor",{"type":3971,"title":3982,"context":112},"Claude Code",{"type":110,"title":3984,"url":3985,"context":3822},"Google AI Essentials","https:\u002F\u002Fimp.i384100.net\u002F1GW56D",{"type":110,"title":3987,"url":3988,"context":3822},"Prompt Engineering for ChatGPT","https:\u002F\u002Fimp.i384100.net\u002FgRWb9g",{"type":3971,"title":3990,"url":3991,"context":112},"localGPT","https:\u002F\u002Fbit.ly\u002FlocalGPT",{"relevance":114,"novelty":115,"quality":115,"actionability":115,"composite":116,"reasoning":3993},"Category: AI & LLMs. The article provides a detailed exploration of a specific architecture for building AI agents, addressing a core topic of interest for developers looking to implement AI features. It presents new insights into the trade-offs between harnesses and frameworks, which is valuable for those seeking practical applications in production.","\u002Fsummaries\u002F67334f5912c7ef54-agent-harness-9-components-beyond-frameworks-summary","2026-04-30 15:46:30","2026-05-03 16:54:07",{"title":3840,"description":96},{"loc":3994},"17f7ef60774eebb6","Prompt Engineering","https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=nWzXyjXCoCE","summaries\u002F67334f5912c7ef54-agent-harness-9-components-beyond-frameworks-summary",[4004,131,132,130],"agents","A harness is a fixed while-loop architecture that turns one-shot LLMs into iterative agents with tools, context control, subagents, memory, and safety—pre-wired unlike LangChain-style frameworks you assemble.",[],"JqrqQ2-5oaoD8BONn4-AoROBBGp86DciPogdq9xJo4o",{"id":4009,"title":4010,"ai":4011,"body":4016,"categories":4427,"created_at":104,"date_modified":104,"description":96,"extension":105,"faq":104,"featured":106,"kicker_label":104,"meta":4428,"navigation":118,"path":4444,"published_at":4445,"question":104,"scraped_at":4446,"seo":4447,"sitemap":4448,"source_id":4449,"source_name":4450,"source_type":126,"source_url":4451,"stem":4452,"tags":4453,"thumbnail_url":104,"tldr":4455,"tweet":104,"unknown_tags":4456,"__hash__":4457},"summaries\u002Fsummaries\u002Fad972853080121bc-build-graphrag-for-complex-queries-across-articles-summary.md","Build GraphRAG for Complex Queries Across Articles",{"provider":7,"model":8,"input_tokens":4012,"output_tokens":4013,"processing_time_ms":4014,"cost_usd":4015},8492,2634,24122,0.00299275,{"type":14,"value":4017,"toc":4420},[4018,4022,4025,4044,4047,4058,4061,4069,4075,4081,4085,4088,4094,4113,4118,4133,4136,4156,4159,4164,4168,4175,4181,4195,4202,4259,4262,4267,4275,4278,4282,4287,4298,4304,4312,4315,4350,4356,4359,4370,4373,4378,4382,4408,4413,4418],[17,4019,4021],{"id":4020},"standard-rag-fails-on-interconnected-datagraphrag-fixes-it","Standard RAG Fails on Interconnected Data—GraphRAG Fixes It",[22,4023,4024],{},"Standard RAG chunks documents, embeds them, and retrieves similar vectors for LLM context. It works for simple fact lookups in small datasets but degrades with scale: accuracy drops 12% at 100k pages due to embedding overlap. Worse, chunks are isolated—no cross-document reasoning for questions like \"main themes across articles\" or \"company connections in disputes.\"",[22,4026,4027,4028,4031,4032,4035,4036,4039,4040,4043],{},"GraphRAG layers a knowledge graph on top. LLMs extract ",[3871,4029,4030],{},"entities"," (e.g., organizations like OpenAI, events like lawsuits) and ",[3871,4033,4034],{},"relationships"," (e.g., \"defendant in\", \"trained on\"). These form nodes and edges in a property graph, capturing structure. Microsoft's approach adds ",[3871,4037,4038],{},"community detection"," (hierarchical Leiden algorithm groups related entities) and ",[3871,4041,4042],{},"community summaries"," (LLM-generated briefs per cluster).",[22,4045,4046],{},"Use GraphRAG for:",[55,4048,4049,4052,4055],{},[58,4050,4051],{},"Hundreds\u002Fthousands of interconnected docs (law, policy, research).",[58,4053,4054],{},"Global queries: patterns, trends, summaries.",[58,4056,4057],{},"Traceable answers.",[22,4059,4060],{},"Stick to standard RAG for:",[55,4062,4063,4066],{},[58,4064,4065],{},"Single-doc facts.",[58,4067,4068],{},"Speed\u002Fcost priority on small, non-relational data.",[22,4070,4071,4074],{},[3871,4072,4073],{},"Before\u002Fafter",": Standard RAG might retrieve unrelated chunks on \"AI copyright connections\"; GraphRAG traces paths like OpenAI → defendant in → NYT lawsuit → filed against → artists.",[4076,4077,4078],"blockquote",{},[22,4079,4080],{},"\"Standard rack has two more fundamental blind spots. The number one is each chunk is treated as an isolated fragment... no ability to reason across documents.\"",[17,4082,4084],{"id":4083},"scrape-real-world-data-without-browser-hassles","Scrape Real-World Data Without Browser Hassles",[22,4086,4087],{},"Start with live data: Use SerpApi's Google Search API for structured JSON results—no Selenium needed. Free tier covers testing.",[22,4089,4090,4093],{},[3871,4091,4092],{},"Collection steps",":",[3866,4095,4096,4099,4110],{},[58,4097,4098],{},"Define queries (e.g., \"AI intellectual property\", \"copyright generative AI\").",[58,4100,4101,4102,4105,4106,4109],{},"Call ",[3745,4103,4104],{},"GoogleSearchResults"," with params: ",[3745,4107,4108],{},"engine=\"google\", gl=\"us\", hl=\"en\", num=10",".",[58,4111,4112],{},"Dedupe URLs across queries → Pandas DataFrame.",[22,4114,4115,4093],{},[3871,4116,4117],{},"Enrich with full text",[55,4119,4120,4123,4130],{},[58,4121,4122],{},"Articles: Trafilatura extracts clean body text (strips nav\u002Fads\u002Ffooters).",[58,4124,4125,4126,4129],{},"YouTube: Regex video ID from URL → ",[3745,4127,4128],{},"youtube_transcript_api"," for captions.",[58,4131,4132],{},"Filter successes (paywalls\u002Fcaptionless fail) → Save as CSV.",[22,4134,4135],{},"Code snippet for search:",[3754,4137,4139],{"className":3756,"code":4138,"language":132,"meta":96,"style":96},"import serpapi\nresults = GoogleSearchResults({'q': query, 'api_key': SERPAPI_KEY})\nraw = results.get_dict()\n",[3745,4140,4141,4146,4151],{"__ignoreMap":96},[26,4142,4143],{"class":3762,"line":3763},[26,4144,4145],{},"import serpapi\n",[26,4147,4148],{"class":3762,"line":97},[26,4149,4150],{},"results = GoogleSearchResults({'q': query, 'api_key': SERPAPI_KEY})\n",[26,4152,4153],{"class":3762,"line":3774},[26,4154,4155],{},"raw = results.get_dict()\n",[22,4157,4158],{},"For 20 articles on AI copyright, this yields ~10-20 usable full-text docs. Scales to any topic—swap queries.",[4076,4160,4161],{},[22,4162,4163],{},"\"SER API is what we're using to script Google News results... real time structured clean search results... no browser automation needed.\"",[17,4165,4167],{"id":4166},"ontology-driven-extraction-ensures-reliable-graphs","Ontology-Driven Extraction Ensures Reliable Graphs",[22,4169,4170,4171,4174],{},"Define ",[3871,4172,4173],{},"ontology"," first: List entity types (e.g., ORGANIZATION, PERSON, LAWSUIT) and relations (e.g., FILED_AGAINST, REGULATES, TRAINED_ON). Domain-specific—tailor to AI copyright (e.g., defendant_in).",[22,4176,4177,4180],{},[3871,4178,4179],{},"Extraction prompt"," (via LlamaIndex):",[55,4182,4183,4186,4189,4192],{},[58,4184,4185],{},"Input: Article text.",[58,4187,4188],{},"Output: Up to 20 entity-relation triplets per article.",[58,4190,4191],{},"Per entity: name, type, description.",[58,4193,4194],{},"Per relation: source, target, type, description.",[22,4196,4197,4198,4201],{},"Use ",[3871,4199,4200],{},"Pydantic models"," for structured output:",[3754,4203,4205],{"className":3756,"code":4204,"language":132,"meta":96,"style":96},"from pydantic import BaseModel\nclass ExtractedEntity(BaseModel):\n    name: str\n    type: str\n    description: str\nclass ExtractedRelationship(BaseModel):\n    source: str\n    target: str\n    relation: str\n    description: str\n",[3745,4206,4207,4212,4217,4222,4227,4232,4237,4242,4248,4254],{"__ignoreMap":96},[26,4208,4209],{"class":3762,"line":3763},[26,4210,4211],{},"from pydantic import BaseModel\n",[26,4213,4214],{"class":3762,"line":97},[26,4215,4216],{},"class ExtractedEntity(BaseModel):\n",[26,4218,4219],{"class":3762,"line":3774},[26,4220,4221],{},"    name: str\n",[26,4223,4224],{"class":3762,"line":115},[26,4225,4226],{},"    type: str\n",[26,4228,4229],{"class":3762,"line":114},[26,4230,4231],{},"    description: str\n",[26,4233,4234],{"class":3762,"line":3790},[26,4235,4236],{},"class ExtractedRelationship(BaseModel):\n",[26,4238,4239],{"class":3762,"line":3796},[26,4240,4241],{},"    source: str\n",[26,4243,4245],{"class":3762,"line":4244},8,[26,4246,4247],{},"    target: str\n",[26,4249,4251],{"class":3762,"line":4250},9,[26,4252,4253],{},"    relation: str\n",[26,4255,4257],{"class":3762,"line":4256},10,[26,4258,4231],{},[22,4260,4261],{},"Pass as OpenAI function-calling schema—auto-validates\u002Frejects bad outputs. No regex parsing.",[22,4263,4264,4093],{},[3871,4265,4266],{},"GraphRAGExtractor class",[3866,4268,4269,4272],{},[58,4270,4271],{},"LLM extracts → Pydantic → LlamaIndex EntityNode\u002FRelation objects.",[58,4273,4274],{},"Process 50 articles in parallel (GPT-4o-mini for cost).",[22,4276,4277],{},"Common mistake: Skipping ontology → hallucinated\u002Finconsistent entities. Fix: Explicit lists in prompt.\nQuality check: Descriptions enrich context for later summaries.",[17,4279,4281],{"id":4280},"graph-construction-communities-and-local-global-querying","Graph Construction, Communities, and Local-Global Querying",[22,4283,4284,4093],{},[3871,4285,4286],{},"GraphRAGStore",[55,4288,4289,4292,4295],{},[58,4290,4291],{},"Insert extracted nodes\u002Fedges.",[58,4293,4294],{},"Run Leiden community detection → Clusters (e.g., \"OpenAI lawsuits\").",[58,4296,4297],{},"LLM summarizes each: Collect entities\u002Frelations → GPT-4o-mini brief.",[22,4299,4300,4303],{},[3871,4301,4302],{},"Query engine"," (two-step):",[3866,4305,4306,4309],{},[58,4307,4308],{},"GPT-4o-mini filters relevant communities (skip irrelevant to save tokens).",[58,4310,4311],{},"GPT-4o synthesizes: Per-community answers → Global response.",[22,4313,4314],{},"Code flow:",[3754,4316,4318],{"className":3756,"code":4317,"language":132,"meta":96,"style":96},"from llama_index.core.graph_stores import SimpleGraphStore\n# ...extraction...\ngraph_store = SimpleGraphStore()\n# Insert, detect_communities(), summarize_communities()\nquery_engine = GraphRAGQueryEngine(...)\nresponse = query_engine.query(\"Central companies in AI copyright disputes?\")\n",[3745,4319,4320,4325,4330,4335,4340,4345],{"__ignoreMap":96},[26,4321,4322],{"class":3762,"line":3763},[26,4323,4324],{},"from llama_index.core.graph_stores import SimpleGraphStore\n",[26,4326,4327],{"class":3762,"line":97},[26,4328,4329],{},"# ...extraction...\n",[26,4331,4332],{"class":3762,"line":3774},[26,4333,4334],{},"graph_store = SimpleGraphStore()\n",[26,4336,4337],{"class":3762,"line":115},[26,4338,4339],{},"# Insert, detect_communities(), summarize_communities()\n",[26,4341,4342],{"class":3762,"line":114},[26,4343,4344],{},"query_engine = GraphRAGQueryEngine(...)\n",[26,4346,4347],{"class":3762,"line":3790},[26,4348,4349],{},"response = query_engine.query(\"Central companies in AI copyright disputes?\")\n",[22,4351,4352,4355],{},[3871,4353,4354],{},"Visualization",": D3.js for interactive graph (nodes=entities, edges=relations, clusters colored).",[22,4357,4358],{},"Production tips:",[55,4360,4361,4364,4367],{},[58,4362,4363],{},"GPT-4o-mini for extraction\u002Fsummaries (volume), GPT-4o for queries (reasoning).",[58,4365,4366],{},"No chunking if articles short.",[58,4368,4369],{},"Parallel workers speed indexing.",[22,4371,4372],{},"Example query: \"Connections in AI copyright?\" → Traces OpenAI, NYT, artists via graph traversal.",[4076,4374,4375],{},[22,4376,4377],{},"\"Using a knowledge graph has been shown to improve LM response accuracy... sensemaking: understand connections, patterns and themes.\"",[17,4379,4381],{"id":4380},"key-takeaways","Key Takeaways",[55,4383,4384,4387,4390,4393,4396,4399,4402,4405],{},[58,4385,4386],{},"Switch to GraphRAG for cross-document reasoning; standard RAG for isolated facts.",[58,4388,4389],{},"Always define domain ontology first—prevents extraction drift.",[58,4391,4392],{},"SerpApi + Trafilatura = reliable scraping pipeline; dedupe and filter aggressively.",[58,4394,4395],{},"Pydantic + function calling = bulletproof structured extraction.",[58,4397,4398],{},"Community summaries enable efficient local-global querying—filter first, synthesize second.",[58,4400,4401],{},"Use cheaper models for indexing, premium for queries to optimize costs.",[58,4403,4404],{},"Visualize with D3.js to debug\u002Ftrace graph quality.",[58,4406,4407],{},"Test on real data like AI copyright: Start with GitHub repo, adapt ontology.",[4076,4409,4410],{},[22,4411,4412],{},"\"The ontology... tells the LLM exactly what types of entities and relationships it's allowed to extract.\"",[4076,4414,4415],{},[22,4416,4417],{},"\"At query time, these summaries are queried instead of the raw graph, which makes it particularly effective and fast for big picture questions.\"",[3808,4419,3810],{},{"title":96,"searchDepth":97,"depth":97,"links":4421},[4422,4423,4424,4425,4426],{"id":4020,"depth":97,"text":4021},{"id":4083,"depth":97,"text":4084},{"id":4166,"depth":97,"text":4167},{"id":4280,"depth":97,"text":4281},{"id":4380,"depth":97,"text":4381},[103],{"content_references":4429,"triage":4442},[4430,4436,4439],{"type":4431,"title":4432,"author":4433,"url":4434,"context":4435},"paper","GraphRAG paper","Microsoft Research","https:\u002F\u002Farxiv.org\u002Fpdf\u002F2404.16130","cited",{"type":3971,"title":4437,"url":4438,"context":3822},"SerpApi","https:\u002F\u002Fserpapi.link\u002Fthu-vu",{"type":110,"title":4440,"url":4441,"context":3822},"graphRAG Git repo","https:\u002F\u002Fgithub.com\u002Fthu-vu92\u002FgraphRAG",{"relevance":114,"novelty":115,"quality":115,"actionability":115,"composite":116,"reasoning":4443},"Category: AI & LLMs. The article provides a detailed explanation of how GraphRAG improves upon standard RAG for complex queries, addressing a specific pain point of interconnected data analysis. It includes actionable steps for implementing the solution, such as using SerpApi for data collection.","\u002Fsummaries\u002Fad972853080121bc-build-graphrag-for-complex-queries-across-articles-summary","2026-04-14 08:04:31","2026-04-19 01:20:43",{"title":4010,"description":96},{"loc":4444},"ad972853080121bc","AI Summaries (evaluation playlist)","https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=JTVx6i6MzVw","summaries\u002Fad972853080121bc-build-graphrag-for-complex-queries-across-articles-summary",[131,130,132,4454],"ai-automation","GraphRAG builds knowledge graphs from scraped articles to enable reasoning over interconnected data, outperforming standard RAG on global questions like themes and relationships in AI copyright disputes.",[4454],"341Gx8g9rvMLH63Xt_hzigIvq8ab6Il3BJfkLjVzfZ0",{"id":4459,"title":4460,"ai":4461,"body":4466,"categories":4712,"created_at":104,"date_modified":104,"description":96,"extension":105,"faq":104,"featured":106,"kicker_label":104,"meta":4713,"navigation":118,"path":4717,"published_at":104,"question":104,"scraped_at":4718,"seo":4719,"sitemap":4720,"source_id":4721,"source_name":4722,"source_type":126,"source_url":4723,"stem":4724,"tags":4725,"thumbnail_url":104,"tldr":4726,"tweet":104,"unknown_tags":4727,"__hash__":4728},"summaries\u002Fsummaries\u002Fc62e1f5b7f154135-three-multi-llm-patterns-chain-parallel-route-summary.md","Three Multi-LLM Patterns: Chain, Parallel, Route",{"provider":7,"model":8,"input_tokens":4462,"output_tokens":4463,"processing_time_ms":4464,"cost_usd":4465},4991,1548,11897,0.0017497,{"type":14,"value":4467,"toc":4707},[4468,4472,4475,4478,4508,4511,4589,4592,4596,4599,4602,4627,4630,4634,4637,4640,4685,4688,4691,4702,4705],[17,4469,4471],{"id":4470},"sequential-chaining-builds-complex-outputs-step-by-step","Sequential Chaining Builds Complex Outputs Step-by-Step",[22,4473,4474],{},"Chain multiple LLM calls where each step refines the previous output, ideal for tasks needing progressive transformation like data extraction and formatting. This trades latency for precision since calls run one after another.",[22,4476,4477],{},"Implement with a simple loop:",[3754,4479,4481],{"className":3756,"code":4480,"language":132,"meta":96,"style":96},"def chain(input: str, prompts: list[str]) -> str:\n    result = input\n    for i, prompt in enumerate(prompts, 1):\n        result = llm_call(f\"{prompt}\\nInput: {result}\")\n    return result\n",[3745,4482,4483,4488,4493,4498,4503],{"__ignoreMap":96},[26,4484,4485],{"class":3762,"line":3763},[26,4486,4487],{},"def chain(input: str, prompts: list[str]) -> str:\n",[26,4489,4490],{"class":3762,"line":97},[26,4491,4492],{},"    result = input\n",[26,4494,4495],{"class":3762,"line":3774},[26,4496,4497],{},"    for i, prompt in enumerate(prompts, 1):\n",[26,4499,4500],{"class":3762,"line":115},[26,4501,4502],{},"        result = llm_call(f\"{prompt}\\nInput: {result}\")\n",[26,4504,4505],{"class":3762,"line":114},[26,4506,4507],{},"    return result\n",[22,4509,4510],{},"For a Q3 performance report, four chained prompts extract metrics (e.g., \"92 points customer satisfaction\"), convert to percentages (\"92%: customer satisfaction\"), sort descending, then format as a markdown table:",[4512,4513,4514,4529],"table",{},[4515,4516,4517],"thead",{},[4518,4519,4520,4525],"tr",{},[4521,4522,4524],"th",{"align":4523},"left","Metric",[4521,4526,4528],{"align":4527},"center","Value",[4530,4531,4532,4541,4549,4557,4565,4573,4581],"tbody",{},[4518,4533,4534,4538],{},[4535,4536,4537],"td",{"align":4523},"Customer Satisfaction",[4535,4539,4540],{"align":4527},"92%",[4518,4542,4543,4546],{},[4535,4544,4545],{"align":4523},"Employee Satisfaction",[4535,4547,4548],{"align":4527},"87%",[4518,4550,4551,4554],{},[4535,4552,4553],{"align":4523},"Product Adoption",[4535,4555,4556],{"align":4527},"78%",[4518,4558,4559,4562],{},[4535,4560,4561],{"align":4523},"Operating Margin",[4535,4563,4564],{"align":4527},"34%",[4518,4566,4567,4570],{},[4535,4568,4569],{"align":4523},"Revenue Growth",[4535,4571,4572],{"align":4527},"45%",[4518,4574,4575,4578],{},[4535,4576,4577],{"align":4523},"Market Share",[4535,4579,4580],{"align":4527},"23%",[4518,4582,4583,4586],{},[4535,4584,4585],{"align":4523},"Customer Churn",[4535,4587,4588],{"align":4527},"5%",[22,4590,4591],{},"This breaks down intricate formatting that a single prompt might hallucinate or mishandle.",[17,4593,4595],{"id":4594},"parallel-execution-speeds-up-multi-stakeholder-analysis","Parallel Execution Speeds Up Multi-Stakeholder Analysis",[22,4597,4598],{},"Run identical prompts on multiple inputs concurrently using ThreadPoolExecutor (default 3 workers), cutting total latency for independent tasks like impact analysis across groups.",[22,4600,4601],{},"Code:",[3754,4603,4605],{"className":3756,"code":4604,"language":132,"meta":96,"style":96},"def parallel(prompt: str, inputs: list[str], n_workers: int = 3) -> list[str]:\n    with ThreadPoolExecutor(max_workers=n_workers) as executor:\n        futures = [executor.submit(llm_call, f\"{prompt}\\nInput: {x}\") for x in inputs]\n        return [f.result() for f in futures]\n",[3745,4606,4607,4612,4617,4622],{"__ignoreMap":96},[26,4608,4609],{"class":3762,"line":3763},[26,4610,4611],{},"def parallel(prompt: str, inputs: list[str], n_workers: int = 3) -> list[str]:\n",[26,4613,4614],{"class":3762,"line":97},[26,4615,4616],{},"    with ThreadPoolExecutor(max_workers=n_workers) as executor:\n",[26,4618,4619],{"class":3762,"line":3774},[26,4620,4621],{},"        futures = [executor.submit(llm_call, f\"{prompt}\\nInput: {x}\") for x in inputs]\n",[26,4623,4624],{"class":3762,"line":115},[26,4625,4626],{},"        return [f.result() for f in futures]\n",[22,4628,4629],{},"Example analyzes market changes for customers (price-sensitive, tech-wanting), employees (job security), investors (growth-focused), and suppliers (capacity issues). Each gets tailored impacts and actions in parallel, e.g., for customers: highlight pricing strategies and eco-features. Without parallelism, this serializes to 4x longer; concurrency delivers all results near-simultaneously at higher API cost.",[17,4631,4633],{"id":4632},"routing-directs-inputs-to-specialized-experts","Routing Directs Inputs to Specialized Experts",[22,4635,4636],{},"Classify input content first, then route to a tailored prompt, improving relevance for varied tasks like support tickets. Adds upfront classification latency but leverages specialist personas for better outputs.",[22,4638,4639],{},"Router uses chain-of-thought in XML:",[3754,4641,4643],{"className":3756,"code":4642,"language":132,"meta":96,"style":96},"def route(input: str, routes: dict[str, str]) -> str:\n    selector_prompt = f\"\"\"\n    Analyze... select from {routes.keys()}\n    \u003Creasoning>Explanation\u003C\u002Freasoning>\n    \u003Cselection>Team\u003C\u002Fselection>\n    Input: {input}\"\"\"\n    route_key = extract_xml(llm_call(selector_prompt), \"selection\").strip().lower()\n    return llm_call(f\"{routes[route_key]}\\nInput: {input}\")\n",[3745,4644,4645,4650,4655,4660,4665,4670,4675,4680],{"__ignoreMap":96},[26,4646,4647],{"class":3762,"line":3763},[26,4648,4649],{},"def route(input: str, routes: dict[str, str]) -> str:\n",[26,4651,4652],{"class":3762,"line":97},[26,4653,4654],{},"    selector_prompt = f\"\"\"\n",[26,4656,4657],{"class":3762,"line":3774},[26,4658,4659],{},"    Analyze... select from {routes.keys()}\n",[26,4661,4662],{"class":3762,"line":115},[26,4663,4664],{},"    \u003Creasoning>Explanation\u003C\u002Freasoning>\n",[26,4666,4667],{"class":3762,"line":114},[26,4668,4669],{},"    \u003Cselection>Team\u003C\u002Fselection>\n",[26,4671,4672],{"class":3762,"line":3790},[26,4673,4674],{},"    Input: {input}\"\"\"\n",[26,4676,4677],{"class":3762,"line":3796},[26,4678,4679],{},"    route_key = extract_xml(llm_call(selector_prompt), \"selection\").strip().lower()\n",[26,4681,4682],{"class":3762,"line":4244},[26,4683,4684],{},"    return llm_call(f\"{routes[route_key]}\\nInput: {input}\")\n",[22,4686,4687],{},"Routes: billing (acknowledge charges, steps), technical (numbered fixes), account (security-first), product (feature education).",[22,4689,4690],{},"Ticket examples:",[55,4692,4693,4696,4699],{},[58,4694,4695],{},"Login fail → account: Verifies security, recovery steps.",[58,4697,4698],{},"Unexpected charge → billing: Explains discrepancy, adjustment timeline.",[58,4700,4701],{},"Data export → product: Step-by-step guide, docs links.",[22,4703,4704],{},"Routing reasoning cites keywords (\"invalid password\" → security urgency) and intent, avoiding generic responses.",[3808,4706,3810],{},{"title":96,"searchDepth":97,"depth":97,"links":4708},[4709,4710,4711],{"id":4470,"depth":97,"text":4471},{"id":4594,"depth":97,"text":4595},{"id":4632,"depth":97,"text":4633},[103],{"content_references":4714,"triage":4715},[],{"relevance":114,"novelty":115,"quality":115,"actionability":114,"composite":3824,"reasoning":4716},"Category: AI & LLMs. The article provides practical patterns for using multiple LLMs, addressing specific pain points like latency and accuracy in AI feature development. It includes actionable code examples for chaining and parallel execution, making it immediately applicable for developers building AI-powered products.","\u002Fsummaries\u002Fc62e1f5b7f154135-three-multi-llm-patterns-chain-parallel-route-summary","2026-04-15 15:32:57",{"title":4460,"description":96},{"loc":4717},"c62e1f5b7f154135","__oneoff__","https:\u002F\u002Fplatform.claude.com\u002Fcookbook\u002Fpatterns-agents-basic-workflows","summaries\u002Fc62e1f5b7f154135-three-multi-llm-patterns-chain-parallel-route-summary",[131,130,132,4454],"Chain LLMs sequentially for step-by-step refinement, run parallel calls for concurrent multi-input tasks, and route inputs to specialized prompts via classification—trading latency or cost for better accuracy.",[4454],"lsJK-8dk5NDMRZSnFkDUDlNMTadplVPwYxG0L-RtIpo"]