"""GPT-2 SAE last-token programs, layer 7 -- features 02400 to 02499 (100 programs).

One function per SAE feature. Grouped 100 to a file so the set stays browsable;
the programs themselves are unchanged.
"""

def prog_L07F02400(input_tokens):
    """4 token families, 7 surface forms (train F1 of the uncapped optimum: 0.178).
    Fires at 139,931 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to organizations, events or places"""
    # --- end external annotation ---
    fire = {
        # ,  (127,479 firings)
        ',', ' ,',
        # ),  (1,970 firings)
        '),', ' ),',
        # .,  (650 firings)
        '.,',
        # ",  (379 firings)
        '",', ' ",',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return ',-family', 1

def prog_L07F02401(input_tokens):
    """2 token families, 3 surface forms (train F1 of the uncapped optimum: 0.124).
    Fires at 96,073 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to importance or significance"""
    # --- end external annotation ---
    fire = {
        # ,  (86,103 firings)
        ',',
        # ",  (394 firings)
        '",', ' ",',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return ',-family', 1

def prog_L07F02402(input_tokens):
    """31 token families, 41 surface forms (train F1 of the uncapped optimum: 0.673).
    Fires at 9,928 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to types of oils"""
    # --- end external annotation ---
    fire = {
        # oil  (3,673 firings)
        ' oil', ' Oil', 'oil', ' oils', 'Oil', 'OIL',
        # fat  (584 firings)
        ' fat', ' fats',
        # olive  (209 firings)
        ' olive', ' Olive',
        # wax  (202 firings)
        ' wax', ' Wax',
        # veget  (146 firings)
        ' vegetable',
        # spray  (136 firings)
        ' spray',
        # sauce  (116 firings)
        ' sauce',
        # venom  (115 firings)
        ' venom',
        # fry  (111 firings)
        ' fry', ' frying',
        # glos  (82 firings)
        ' gloss',
        # poison  (77 firings)
        ' poison',
        # fatty  (76 firings)
        ' fatty',
        # thc  (71 firings)
        ' THC',
        # oxide  (65 firings)
        ' oxide', 'oxide',
        # grease  (59 firings)
        ' grease',
        # acid  (56 firings)
        ' acids',
        # inflamma  (54 firings)
        ' inflammation',
        # petroleum  (52 firings)
        ' petroleum',
        # rae  (51 firings)
        ' Rae',
        # lubr  (48 firings)
        ' lubric',
        # acryl  (47 firings)
        ' acrylic',
        # silicone  (45 firings)
        ' silicone',
        # prope  (44 firings)
        ' prope',
        # towel  (38 firings)
        ' towels',
        # oily  (35 firings)
        ' oily',
        # shampoo  (32 firings)
        ' shampoo',
        # ι  (30 firings)
        'ι',
        # resin  (29 firings)
        ' resin',
        # moistur  (27 firings)
        ' moistur',
        # omega  (27 firings)
        ' omega',
        # coat  (27 firings)
        ' coated',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'oil-family', 1

def prog_L07F02403(input_tokens):
    """44 token families, 50 surface forms (train F1 of the uncapped optimum: 0.611).
    Fires at 22,505 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to political or social movements and changes"""
    # --- end external annotation ---
    fire = {
        # iza  (1,933 firings)
        'ization',
        # ation  (1,484 firings)
        'ation', 'ATION',
        # isa  (831 firings)
        'isation',
        # ifica  (655 firings)
        'ification',
        # integra  (371 firings)
        ' integration', ' Integration',
        # ila  (333 firings)
        'ILA', 'ilation',
        # transforma  (295 firings)
        ' transformation', ' Transformation',
        # occupa  (210 firings)
        ' occupation',
        # automa  (209 firings)
        ' automation',
        # ina  (202 firings)
        'ination',
        # ula  (195 firings)
        'ulation',
        # ization  (192 firings)
        'izations',
        # ica  (187 firings)
        'ication',
        # transla  (182 firings)
        ' translation',
        # legal  (176 firings)
        ' legalization',
        # ver  (170 firings)
        'version',
        # genera  (170 firings)
        ' Generation',
        # osi  (167 firings)
        'osis',
        # libera  (141 firings)
        ' Liberation', ' liberation',
        # vaccina  (124 firings)
        ' vaccination',
        # visual  (122 firings)
        ' visualization',
        # optim  (112 firings)
        ' optimization',
        # exploita  (110 firings)
        ' exploitation',
        # ign  (109 firings)
        'ignment',
        # ain  (100 firings)
        'ainment',
        # restructur  (99 firings)
        ' restructuring',
        # manipula  (97 firings)
        ' manipulation',
        # igra  (95 firings)
        'igration',
        # forma  (94 firings)
        'formation', 'FORMATION',
        # redevelop  (88 firings)
        ' redevelopment',
        # advance  (88 firings)
        ' advancement',
        # custom  (83 firings)
        ' customization',
        # rifica  (81 firings)
        'rification',
        # ular  (75 firings)
        'ularity',
        # privat  (73 firings)
        ' privatization',
        # global  (73 firings)
        ' globalization',
        # manage  (73 firings)
        'management',
        # consolida  (71 firings)
        ' consolidation',
        # annexa  (67 firings)
        ' annexation',
        # empower  (65 firings)
        ' empowerment',
        # enrich  (59 firings)
        ' enrichment',
        # ogenesi  (59 firings)
        'ogenesis',
        # deforesta  (59 firings)
        ' deforestation',
        # omic  (58 firings)
        'omics',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'iza-family', 1

def prog_L07F02404(input_tokens):
    """32 token families, 39 surface forms (train F1 of the uncapped optimum: 0.522).
    Fires at 14,066 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: terms related to sports franchises and their players"""
    # --- end external annotation ---
    fire = {
        # franchise  (1,809 firings)
        ' franchise', ' Franchise',
        # reboot  (315 firings)
        ' reboot', ' Reboot',
        # empire  (267 firings)
        ' empire',
        # franchis  (255 firings)
        ' franchises',
        # studio  (241 firings)
        ' Studios',
        # legion  (237 firings)
        ' Legion', ' legion', ' Legions',
        # republ  (161 firings)
        ' republic',
        # mascot  (133 firings)
        ' mascot',
        # dynasty  (127 firings)
        ' dynasty', ' Dynasty',
        # fandom  (120 firings)
        ' fandom',
        # moff  (117 firings)
        ' Moff',
        # vet  (108 firings)
        'Veter',
        # incarna  (102 firings)
        ' incarnation',
        # fiance  (102 firings)
        ' fiance',
        # apost  (101 firings)
        ' apost', ' Apost',
        # cousin  (95 firings)
        ' Cousins',
        # fanbase  (81 firings)
        ' fanbase',
        # bow  (77 firings)
        'bow',
        # dealership  (75 firings)
        ' dealership',
        # entitle  (68 firings)
        ' entitle',
        # homegrown  (68 firings)
        ' homegrown',
        # hold  (60 firings)
        ' Holdings',
        # custom  (58 firings)
        ' customization',
        # unrestrict  (58 firings)
        ' unrestricted',
        # kem  (57 firings)
        ' Kem',
        # archie  (53 firings)
        ' Archie',
        # lot  (46 firings)
        'lot',
        # gener  (45 firings)
        ' generational',
        # patriarch  (43 firings)
        ' patriarch',
        # lent  (40 firings)
        ' Lent',
        # franch  (40 firings)
        ' franch', ' Franch',
        # ranch  (37 firings)
        'ranch',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'franchise-family', 1

def prog_L07F02405(input_tokens):
    """1 token families, 1 surface forms (train F1 of the uncapped optimum: 0.155).
    Fires at 121,689 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: punctuation marks at the end of sentences"""
    # --- end external annotation ---
    fire = {
        # ,  (110,492 firings)
        ',',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return ',-family', 1

def prog_L07F02406(input_tokens):
    """32 token families, 51 surface forms (train F1 of the uncapped optimum: 0.651).
    Fires at 7,729 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: terms related to web development and coding, particularly focusing on HTML and CSS elements"""
    # --- end external annotation ---
    fire = {
        # ele  (1,472 firings)
        ' elements', ' element', ' ele', ' Elements', ' Element', ' ELE', 'Element', 'element',
        # firefight  (432 firings)
        ' firefighters', ' firefighter',
        # wheel  (238 firings)
        ' Wheeler', 'Wheel',
        # tenant  (207 firings)
        ' tenants', ' tenant',
        # attribut  (197 firings)
        ' attributes', ' Attributes', 'Attributes',
        # ori  (180 firings)
        'ori',
        # lement  (155 firings)
        'lement', 'lements',
        # attribute  (143 firings)
        ' attribute', 'Attribute', 'attribute',
        # garn  (140 firings)
        ' Garner',
        # sibl  (112 firings)
        ' siblings',
        # dom  (105 firings)
        ' DOM', 'DOM',
        # lopez  (103 firings)
        ' Lopez',
        # gallagh  (82 firings)
        ' Gallagher',
        # elector  (77 firings)
        ' electors',
        # essent  (73 firings)
        ' Essential',
        # flan  (73 firings)
        ' Flan',
        # ero  (72 firings)
        ' erosion',
        # ancestor  (64 firings)
        ' ancestor',
        # element  (60 firings)
        ' elemental', ' Elemental',
        # harrington  (51 firings)
        ' Harrington',
        # goodwin  (45 firings)
        ' Goodwin',
        # component  (45 firings)
        'Component', ' Component',
        # eaton  (44 firings)
        ' Eaton',
        # emblem  (44 firings)
        ' Emblem',
        # sutton  (41 firings)
        ' Sutton',
        # alloy  (40 firings)
        ' alloy', ' Alloy',
        # hybrid  (36 firings)
        ' Hybrid',
        # dimension  (31 firings)
        ' Dimensions',
        # ]]  (27 firings)
        ']]',
        # ik  (26 firings)
        ' Ik',
        # frame  (26 firings)
        ' Frame',
        # magnesium  (21 firings)
        ' magnesium',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ele-family', 1

def prog_L07F02407(input_tokens):
    """31 token families, 42 surface forms (train F1 of the uncapped optimum: 0.227).
    Fires at 69,350 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases relating to effects or consequences"""
    # --- end external annotation ---
    fire = {
        # mean  (1,564 firings)
        ' means', ' mean', ' Means', ' Mean',
        # mak  (1,475 firings)
        ' makes', ' Makes', 'makes',
        # look  (1,286 firings)
        ' looks',
        # doe  (881 firings)
        ' does',
        # doesn  (770 firings)
        ' doesn',
        # turn  (669 firings)
        ' turns',
        # tak  (523 firings)
        ' takes',
        # work  (489 firings)
        ' works',
        # giv  (479 firings)
        ' gives',
        # includ  (465 firings)
        ' includes', 'Includes',
        # allow  (458 firings)
        ' allows', ' Allows',
        # sound  (386 firings)
        ' sounds', ' sounded',
        # feel  (339 firings)
        ' feels',
        # provid  (307 firings)
        ' provides',
        # off  (293 firings)
        ' offers',
        # affect  (270 firings)
        ' affect', ' affects',
        # represent  (268 firings)
        ' represents',
        # requir  (265 firings)
        ' requires', 'Requires',
        # tell  (259 firings)
        ' tells',
        # help  (250 firings)
        ' helps',
        # follow  (239 firings)
        ' follows',
        # involv  (230 firings)
        ' involves',
        # bring  (225 firings)
        ' brings',
        # unlock  (217 firings)
        ' unlocks',
        # add  (210 firings)
        ' adds',
        # contain  (205 firings)
        ' contains',
        # mark  (195 firings)
        ' marks',
        # hurt  (193 firings)
        ' hurt', ' hurts',
        # appl  (178 firings)
        ' applies',
        # focus  (178 firings)
        ' focuses',
        # depend  (177 firings)
        ' depends',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'mean-family', 1

def prog_L07F02408(input_tokens):
    """11 token families, 14 surface forms (train F1 of the uncapped optimum: 0.081).
    Fires at 76,065 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to giving or providing"""
    # --- end external annotation ---
    fire = {
        # him  (782 firings)
        ' him',
        # them  (732 firings)
        ' them', 'them',
        # us  (599 firings)
        ' us',
        # children  (356 firings)
        ' children', 'children',
        # student  (345 firings)
        ' students',
        # fan  (341 firings)
        ' fans',
        # fc  (340 firings)
        ' FC',
        # everyone  (286 firings)
        ' everyone',
        # user  (281 firings)
        ' users', 'users',
        # custom  (259 firings)
        ' customers',
        # shar  (194 firings)
        ' Shares',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'him-family', 1

def prog_L07F02409(input_tokens):
    """13 token families, 41 surface forms (train F1 of the uncapped optimum: 0.095).
    Fires at 80,883 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: strong verbs indicating action or movement"""
    # --- end external annotation ---
    fire = {
        # with  (2,412 firings)
        ' with',
        # form  (832 firings)
        ' former', ' Former', ' forming', 'formed', 'former', 'forms',
        # ing  (659 firings)
        'ing',
        # ed  (578 firings)
        'ed',
        # born  (575 firings)
        'Born', ' born', 'born', ' Born',
        # tak  (479 firings)
        ' taking', 'Taking', ' Taking', 'taking', ' Takes',
        # hav  (284 firings)
        ' having', 'having',
        # mak  (278 firings)
        ' making',
        # winn  (233 firings)
        ' winning', ' winner', 'winning', 'winner', ' Winner', 'Winner', ' Winning',
        # suffer  (220 firings)
        ' suffering', ' suffered',
        # join  (211 firings)
        ' joining', ' joined', 'join', 'Joined', ' joins',
        # doing  (210 firings)
        ' doing', 'doing', ' Doing',
        # forc  (205 firings)
        ' forced', ' forcing', ' Forced',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'with-family', 1

def prog_L07F02410(input_tokens):
    """41 token families, 54 surface forms (train F1 of the uncapped optimum: 0.605).
    Fires at 14,471 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to pedestrian infrastructure and safety"""
    # --- end external annotation ---
    fire = {
        # walk  (962 firings)
        ' walking', 'walk', 'walking',
        # cyclist  (589 firings)
        ' cyclists', ' cyclist',
        # pedestrian  (586 firings)
        ' pedestrian', ' pedestrians',
        # tourist  (582 firings)
        ' tourist', ' tourists',
        # sidewalk  (381 firings)
        ' sidewalk', ' sidewalks',
        # bicycle  (372 firings)
        ' bicycle', ' Bicycle',
        # landmark  (293 firings)
        ' landmark', ' landmarks',
        # deer  (198 firings)
        ' deer',
        # decrimin  (172 firings)
        ' decriminal',
        # bicycl  (166 firings)
        ' bicycl', ' bicycles',
        # electr  (164 firings)
        ' electr',
        # vand  (151 firings)
        ' vandal',
        # ped  (149 firings)
        ' Ped', 'Ped',
        # wendy  (137 firings)
        ' Wendy',
        # pass  (136 firings)
        ' passers', ' passer',
        # pave  (134 firings)
        ' pavement',
        # vandalism  (120 firings)
        ' vandalism',
        # christine  (116 firings)
        ' Christine',
        # bystand  (114 firings)
        ' bystanders', ' bystand',
        # ging  (105 firings)
        ' ginger',
        # kind  (102 firings)
        ' kindness',
        # motorist  (91 firings)
        ' motorists',
        # ux  (88 firings)
        ' UX',
        # visual  (82 firings)
        ' visually',
        # touchscreen  (78 firings)
        ' touchscreen',
        # typ  (76 firings)
        ' typ',
        # intersection  (75 firings)
        ' intersections',
        # roadway  (70 firings)
        ' roadway',
        # neon  (65 firings)
        ' neon',
        # crow  (59 firings)
        ' crow',
        # usabil  (57 firings)
        ' usability',
        # icular  (57 firings)
        'icular',
        # signage  (54 firings)
        ' signage',
        # tresp  (53 firings)
        ' tresp',
        # rolley  (50 firings)
        'rolley',
        # veh  (47 firings)
        ' veh', ' Veh',
        # http  (45 firings)
        ' HTTPS',
        # bicy  (43 firings)
        ' bicy',
        # robat  (41 firings)
        'robat',
        # landsl  (38 firings)
        ' landsl',
        # tactile  (38 firings)
        ' tactile',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'walk-family', 1

def prog_L07F02411(input_tokens):
    """1 token families, 5 surface forms (train F1 of the uncapped optimum: 0.461).
    Fires at 30,388 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to making a decision not to do something"""
    # --- end external annotation ---
    fire = {
        # not  (25,224 firings)
        ' not', ' Not', 'not', ' NOT', 'NOT',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'not-family', 1

def prog_L07F02412(input_tokens):
    """32 token families, 81 surface forms (train F1 of the uncapped optimum: 0.378).
    Fires at 70,769 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to jars and cans"""
    # --- end external annotation ---
    fire = {
        # box  (1,946 firings)
        ' box', ' boxes', ' Box', ' boxed', 'boxes',
        # bag  (1,172 firings)
        ' bag', ' bags', ' Bag',
        # pack  (717 firings)
        ' pack', ' packs', ' Pack', 'pack', 'packs', ' Packs', 'Pack', ' PACK',
        # cas  (704 firings)
        ' cases', ' cas', 'cases', ' casing', ' Cases',
        # file  (649 firings)
        ' file', ' FILE',
        # bottle  (510 firings)
        ' bottle', ' Bottle',
        # package  (457 firings)
        ' package',
        # packag  (453 firings)
        ' packages', ' packaging', ' packaged',
        # tank  (452 firings)
        ' tank', ' tanks', ' tanker',
        # contain  (437 firings)
        ' container', ' containers', ' Container', 'container',
        # sampl  (416 firings)
        ' samples',
        # bottl  (389 firings)
        ' bottles', ' bottled',
        # cup  (384 firings)
        ' cup', ' cups',
        # barrel  (383 firings)
        ' barrels', ' barrel', ' Barrel',
        # copy  (381 firings)
        ' copy',
        # cop  (364 firings)
        ' copies',
        # fil  (330 firings)
        ' files', ' filings', 'files',
        # glas  (298 firings)
        ' glass',
        # sample  (296 firings)
        ' sample', ' Sample', 'sample', 'Sample',
        # bowl  (290 firings)
        ' bowl', ' bowls',
        # piec  (280 firings)
        ' pieces',
        # stack  (267 firings)
        ' stack', ' stacks', ' stacked', 'stack',
        # batch  (251 firings)
        ' batch', ' batches',
        # field  (242 firings)
        ' fields',
        # pill  (233 firings)
        ' pills', ' pill',
        # load  (232 firings)
        ' loaded', 'load', 'loads',
        # ship  (209 firings)
        ' shipment', ' shipments',
        # can  (193 firings)
        ' cans',
        # pile  (191 firings)
        ' pile',
        # jar  (191 firings)
        ' jar', ' jars', 'Jar',
        # tin  (181 firings)
        ' tin', ' Tin', 'tin',
        # ball  (179 firings)
        ' balls', 'balls',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'box-family', 1

def prog_L07F02413(input_tokens):
    """1 token families, 1 surface forms (train F1 of the uncapped optimum: 0.047).
    Fires at 42,571 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: the word "the." """
    # --- end external annotation ---
    fire = {
        # the  (32,998 firings)
        ' the',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'the-family', 1

def prog_L07F02414(input_tokens):
    """5 token families, 8 surface forms (train F1 of the uncapped optimum: 0.034).
    Fires at 63,251 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: relevant entities (specifically instances of "its" or "the") along with descriptors that indicate membership or association"""
    # --- end external annotation ---
    fire = {
        # the  (24,185 firings)
        ' the', ' THE', 'THE',
        # two  (665 firings)
        ' two', ' TWO',
        # ]  (407 firings)
        ']',
        # form  (252 firings)
        ' former',
        # as  (169 firings)
        ' As',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'the-family', 1

def prog_L07F02415(input_tokens):
    """15 token families, 24 surface forms (train F1 of the uncapped optimum: 0.821).
    Fires at 5,800 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases containing the word "Trans" """
    # --- end external annotation ---
    fire = {
        # tran  (2,946 firings)
        ' Trans', ' trans', 'Trans', 'trans', ' TRANS', ' TRAN',
        # rawdownloadcloneembedreportprint  (204 firings)
        'rawdownloadcloneembedreportprint',
        # transform  (189 firings)
        ' Transformers', 'Transform', 'transform',
        # tele  (178 firings)
        ' Tele', 'Tele', 'tele',
        # transc  (140 firings)
        ' transc',
        # transcript  (136 firings)
        ' Transcript',
        # kind  (76 firings)
        ' Kinder',
        # ran  (74 firings)
        'rans',
        # transl  (73 firings)
        ' transl',
        # ο  (72 firings)
        'ο',
        # transf  (60 firings)
        ' transf',
        # transm  (45 firings)
        ' transm',
        # trin  (32 firings)
        ' Trin',
        # transforma  (24 firings)
        ' Transformation',
        # mex  (21 firings)
        ' Mex',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'tran-family', 1

def prog_L07F02416(input_tokens):
    """1 token families, 1 surface forms (train F1 of the uncapped optimum: 0.195).
    Fires at 164,101 of 36,786,566 positions.
    Fires at 0.45% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to reports or studies specifically mentioning different organizations or entities"""
    # --- end external annotation ---
    fire = {
        # the  (148,119 firings)
        ' the',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'the-family', 1

def prog_L07F02417(input_tokens):
    """47 token families, 55 surface forms (train F1 of the uncapped optimum: 0.658).
    Fires at 7,596 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: questions or uncertainties expressed using punctuation symbols"""
    # --- end external annotation ---
    fire = {
        # ™  (470 firings)
        '™',
        # ?  (327 firings)
        ' ?',
        # ~  (297 firings)
        ' ~',
        # !  (259 firings)
        ' !',
        # %  (187 firings)
        ' %',
        # \  (176 firings)
        ' \\',
        # ??  (166 firings)
        '??', ' ??',
        # ;  (154 firings)
        ' ;',
        # plasma  (154 firings)
        ' plasma', ' Plasma',
        # ​  (105 firings)
        ' \u200b',
        # ***  (101 firings)
        ' ***',
        # «  (78 firings)
        ' «',
        # ..  (74 firings)
        ' ..',
        # ^  (74 firings)
        ' ^',
        # orbit  (71 firings)
        ' Orbit',
        # ||  (69 firings)
        ' ||', '||',
        # eclipse  (67 firings)
        ' Eclipse',
        # 000  (67 firings)
        ' 000',
        # ???  (63 firings)
        '???', ' ???',
        # flux  (62 firings)
        ' flux',
        # nightmare  (59 firings)
        ' Nightmare',
        # 192  (58 firings)
        ' 192', '192',
        # .?  (55 firings)
        '.?',
        # ""  (54 firings)
        ' ""', '""',
        # vu  (52 firings)
        ' Vu',
        # ---  (48 firings)
        ' ---',
        # $$  (43 firings)
        ' $$',
        # ....  (41 firings)
        ' ....',
        # 196  (39 firings)
        ' 196',
        # du  (38 firings)
        ' DU',
        # 191  (37 firings)
        ' 191', '191',
        # [[  (37 firings)
        ' [[', '[[',
        # °  (37 firings)
        ' °',
        # </  (34 firings)
        ' </',
        # {{  (31 firings)
        ' {{',
        # !?  (31 firings)
        '!?',
        # --------  (30 firings)
        ' --------',
        # 197  (30 firings)
        ' 197',
        # /?  (29 firings)
        '/?',
        # ~~  (28 firings)
        '~~',
        # ®  (27 firings)
        ' ®',
        # !!  (24 firings)
        ' !!',
        # este  (24 firings)
        ' este',
        # ((  (22 firings)
        ' ((',
        # volk  (20 firings)
        ' Volks',
        # ????  (19 firings)
        '????',
        # unsure  (19 firings)
        ' Unsure',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return '™-family', 1

def prog_L07F02418(input_tokens):
    """15 token families, 24 surface forms (train F1 of the uncapped optimum: 0.649).
    Fires at 10,739 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: warnings or cautions issued by various entities or individuals"""
    # --- end external annotation ---
    fire = {
        # warn  (4,490 firings)
        ' warned', ' warning', ' warn', ' warns', ' warnings', ' Warn', 'warn', 'WARN',
        # adv  (266 firings)
        ' advised',
        # accus  (197 firings)
        ' accusing',
        # alert  (151 firings)
        ' alerted',
        # advis  (136 firings)
        ' advising', ' advises',
        # wary  (117 firings)
        ' wary',
        # flagg  (112 firings)
        ' flagged',
        # caution  (102 firings)
        ' cautioned',
        # beware  (99 firings)
        ' beware', ' Beware',
        # avert  (68 firings)
        ' avert',
        # remind  (54 firings)
        ' reminding',
        # admon  (38 firings)
        ' admon',
        # rail  (32 firings)
        ' railing',
        # old  (31 firings)
        'olded',
        # eware  (28 firings)
        'eware',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'warn-family', 1

def prog_L07F02419(input_tokens):
    """41 token families, 51 surface forms (train F1 of the uncapped optimum: 0.551).
    Fires at 8,406 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: contexts or situations that are crucial for understanding people or things"""
    # --- end external annotation ---
    fire = {
        # context  (1,070 firings)
        ' context', ' contexts', ' Context', 'Context', 'context',
        # occa  (203 firings)
        ' occasion',
        # ango  (148 firings)
        'ango',
        # cue  (148 firings)
        ' cues', ' cue', ' Cue',
        # background  (131 firings)
        'Background',
        # accent  (107 firings)
        ' accent',
        # irony  (106 firings)
        ' irony',
        # trayvon  (95 firings)
        ' Trayvon',
        # cultural  (89 firings)
        ' culturally',
        # domin  (81 firings)
        ' Dominic',
        # caleb  (76 firings)
        ' Caleb',
        # vas  (75 firings)
        'vas',
        # butt  (71 firings)
        ' Butt',
        # text  (68 firings)
        'text', 'TEXT',
        # glos  (66 firings)
        ' gloss', ' Gloss',
        # ongo  (63 firings)
        'ongo',
        # circumst  (63 firings)
        ' circumstance', ' circumst',
        # flavour  (62 firings)
        ' flavour',
        # mongo  (61 firings)
        ' Mongo',
        # juxtap  (60 firings)
        ' juxtap',
        # loc  (56 firings)
        'loc', 'Loc',
        # ethn  (52 firings)
        ' ethn',
        # rapport  (51 firings)
        ' rapport',
        # fact  (50 firings)
        ' Fact',
        # contextu  (44 firings)
        ' contextual',
        # endo  (40 firings)
        'endo',
        # nuance  (37 firings)
        ' nuance',
        # dic  (37 firings)
        ' diction',
        # spat  (37 firings)
        ' spatial',
        # emphasi  (36 firings)
        'emphasis',
        # info  (36 firings)
        'Info',
        # itu  (35 firings)
        'itu',
        # compton  (34 firings)
        ' Compton',
        # notation  (31 firings)
        'notations',
        # aleb  (27 firings)
        'aleb',
        # cla  (25 firings)
        ' cla',
        # locale  (24 firings)
        ' locale',
        # span  (24 firings)
        ' Span',
        # color  (23 firings)
        'COLOR',
        # nota  (23 firings)
        'notation',
        # situ  (22 firings)
        ' situational',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'context-family', 1

def prog_L07F02420(input_tokens):
    """5 token families, 9 surface forms (train F1 of the uncapped optimum: 0.267).
    Fires at 19,459 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases expressing strong opinions or beliefs"""
    # --- end external annotation ---
    fire = {
        # say  (5,363 firings)
        ' say', ' saying', ' Say', ' SAY', ' Saying',
        # mention  (311 firings)
        ' mentioned',
        # pick  (77 firings)
        ' Picks',
        # swear  (55 firings)
        ' swear',
        # guess  (54 firings)
        ' guessed',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'say-family', 1

def prog_L07F02421(input_tokens):
    """6 token families, 30 surface forms (train F1 of the uncapped optimum: 0.246).
    Fires at 137,639 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to technology and digital content"""
    # --- end external annotation ---
    fire = {
        # video  (591 firings)
        ' Video', 'Video', 'video', ' Videos', ' VIDEO',
        # game  (508 firings)
        ' Game', 'Game', ' GAME', 'GAME',
        # play  (433 firings)
        ' Play', 'Play', 'Playing', ' Player', ' Players', ' Playing', 'Player', 'Players', ' Played', ' PLAY', ' Plays', 'PLAY',
        # fire  (393 firings)
        ' Fire', 'Fire', ' FIRE',
        # back  (383 firings)
        ' Back', 'Back',
        # art  (370 firings)
        ' Art', 'Art', 'arter', ' ART',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'video-family', 1

def prog_L07F02422(input_tokens):
    """46 token families, 52 surface forms (train F1 of the uncapped optimum: 0.744).
    Fires at 9,831 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of specific words, likely related to a particular topic or entity named "Stone" """
    # --- end external annotation ---
    fire = {
        # stone  (1,816 firings)
        ' Stone', 'stone', 'Stone',
        # glen  (340 firings)
        ' Glen',
        # rodriguez  (312 firings)
        ' Rodriguez',
        # bould  (229 firings)
        ' Boulder',
        # wat  (198 firings)
        ' Waters',
        # drake  (198 firings)
        ' Drake',
        # clay  (190 firings)
        ' Clay',
        # slate  (185 firings)
        ' Slate',
        # steele  (167 firings)
        ' Steele',
        # sheffield  (152 firings)
        ' Sheffield',
        # ston  (134 firings)
        ' Stones',
        # field  (134 firings)
        ' Fields',
        # blade  (131 firings)
        ' Blade', 'Blade',
        # anglo  (130 firings)
        ' Anglo',
        # berg  (122 firings)
        ' Berg',
        # estone  (116 firings)
        'estone',
        # payne  (102 firings)
        ' Payne',
        # hardy  (96 firings)
        ' Hardy',
        # barton  (90 firings)
        ' Barton',
        # andowski  (89 firings)
        'andowski',
        # usk  (85 firings)
        'usk',
        # spear  (82 firings)
        ' Spears', ' Spear',
        # woodward  (80 firings)
        ' Woodward',
        # bone  (80 firings)
        ' Bone', 'Bone',
        # starr  (78 firings)
        ' Starr',
        # clo  (74 firings)
        ' Clo',
        # arton  (72 firings)
        'arton',
        # wit  (70 firings)
        ' Wit',
        # stephen  (68 firings)
        ' Stephens',
        # razor  (65 firings)
        ' Razor',
        # spar  (65 firings)
        ' spar',
        # wern  (62 firings)
        ' Werner',
        # smithsonian  (61 firings)
        ' Smithsonian',
        # thiel  (57 firings)
        ' Thiel',
        # peck  (55 firings)
        ' Peck',
        # orb  (55 firings)
        ' Orb',
        # plat  (55 firings)
        ' Plat',
        # argent  (55 firings)
        ' Argent',
        # spo  (54 firings)
        ' Spo',
        # balk  (53 firings)
        ' Balk',
        # splash  (47 firings)
        ' Splash',
        # wen  (47 firings)
        ' Wen',
        # sword  (42 firings)
        ' Swords', 'Sword',
        # roche  (40 firings)
        ' Roche',
        # quartz  (34 firings)
        ' Quartz',
        # sphere  (25 firings)
        ' Sphere',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'stone-family', 1

def prog_L07F02423(input_tokens):
    """4 token families, 10 surface forms (train F1 of the uncapped optimum: 0.054).
    Fires at 45,595 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: recognition awards and accolades for individuals in various fields"""
    # --- end external annotation ---
    fire = {
        # of  (18,932 firings)
        ' of', 'of', ' Of', ' OF',
        # play  (463 firings)
        ' Player', ' player',
        # the  (295 firings)
        'the', ' THE',
        # choice  (185 firings)
        ' choice', ' Choice',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'of-family', 1

def prog_L07F02424(input_tokens):
    """45 token families, 59 surface forms (train F1 of the uncapped optimum: 0.733).
    Fires at 8,154 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: the word "divergent" or variations of the word"""
    # --- end external annotation ---
    fire = {
        # div  (520 firings)
        ' divers', ' diver', ' diving', ' Diver',
        # iver  (369 firings)
        'iver', 'ivers', 'IVER',
        # dire  (199 firings)
        ' dire',
        # redirect  (193 firings)
        ' redirect',
        # dolphin  (191 firings)
        ' dolphin', ' dolphins',
        # sharp  (189 firings)
        ' sharply',
        # divert  (175 firings)
        ' diverted', ' divert',
        # dil  (169 firings)
        ' Dil',
        # conv  (168 firings)
        ' conver', ' Conver',
        # drastical  (135 firings)
        ' drastically',
        # emer  (132 firings)
        ' Emer', 'Emer',
        # divul  (96 firings)
        ' divul',
        # lad  (96 firings)
        ' Lad',
        # iren  (93 firings)
        'iren',
        # degen  (89 firings)
        ' degener',
        # destabil  (87 firings)
        ' destabil',
        # dove  (87 firings)
        ' Dove', ' dove',
        # plung  (86 firings)
        ' plung',
        # stri  (86 firings)
        ' stri',
        # liv  (84 firings)
        ' Liver',
        # rin  (84 firings)
        ' Rin',
        # sev  (84 firings)
        ' sever',
        # curt  (78 firings)
        ' Curt',
        # cutt  (78 firings)
        ' cutter', ' Cutter',
        # blur  (74 firings)
        ' blur',
        # lev  (71 firings)
        ' Lever',
        # dispar  (71 firings)
        ' dispar',
        # dele  (71 firings)
        ' Dele', ' dele',
        # extingu  (70 firings)
        ' extingu',
        # disqu  (64 firings)
        ' disqual',
        # gou  (64 firings)
        ' gou', ' Gou',
        # clair  (63 firings)
        ' Clair',
        # ingu  (63 firings)
        'ingu',
        # scaven  (62 firings)
        ' scaven',
        # comb  (62 firings)
        ' Comb',
        # ernand  (61 firings)
        'ernand',
        # conve  (59 firings)
        ' conve',
        # revol  (58 firings)
        ' revol',
        # ibal  (58 firings)
        'ibal',
        # sear  (57 firings)
        ' Sear', ' sear',
        # diver  (53 firings)
        ' diversion',
        # converg  (50 firings)
        ' convergence',
        # converge  (48 firings)
        ' converge',
        # dred  (48 firings)
        ' dred',
        # lib  (45 firings)
        ' liber',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'div-family', 1

def prog_L07F02425(input_tokens):
    """2 token families, 2 surface forms (train F1 of the uncapped optimum: 0.039).
    Fires at 22,954 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to being the first to do something"""
    # --- end external annotation ---
    fire = {
        # to  (13,631 firings)
        ' to',
        # ever  (176 firings)
        ' ever',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'to-family', 1

def prog_L07F02426(input_tokens):
    """20 token families, 34 surface forms (train F1 of the uncapped optimum: 0.330).
    Fires at 27,724 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: keywords related to the physical location or direction within an object"""
    # --- end external annotation ---
    fire = {
        # top  (3,162 firings)
        ' top', 'top', ' tops', 'tops', ' TOP',
        # side  (2,294 firings)
        ' side', 'side', 'Side',
        # bottom  (966 firings)
        ' bottom', 'bottom', ' Bottom', 'Bottom',
        # base  (492 firings)
        ' base',
        # corn  (401 firings)
        ' corner', ' corners',
        # edge  (381 firings)
        ' edge',
        # sid  (354 firings)
        ' sides',
        # upp  (307 firings)
        ' upper', 'upper',
        # surface  (266 firings)
        ' surface',
        # tip  (169 firings)
        ' tip', 'tip', 'tips',
        # roof  (159 firings)
        ' roof', ' roofs',
        # edg  (150 firings)
        ' edges',
        # out  (137 firings)
        ' outer',
        # wall  (125 firings)
        ' walls',
        # floor  (118 firings)
        ' floors',
        # root  (90 firings)
        ' root',
        # lid  (80 firings)
        ' lid',
        # atop  (76 firings)
        ' atop',
        # upside  (71 firings)
        ' upside',
        # topp  (71 firings)
        ' topped',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'top-family', 1

def prog_L07F02427(input_tokens):
    """38 token families, 60 surface forms (train F1 of the uncapped optimum: 0.692).
    Fires at 21,202 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to aviation"""
    # --- end external annotation ---
    fire = {
        # flight  (1,601 firings)
        ' flight', ' Flight', 'flight', 'Flight',
        # fly  (1,484 firings)
        ' flying', ' Flying', 'Flying', 'flying',
        # pilot  (987 firings)
        ' pilot', ' pilots', ' Pilot',
        # airport  (865 firings)
        ' Airport', ' airports',
        # aircraft  (694 firings)
        ' aircraft', ' Aircraft',
        # avia  (558 firings)
        ' aviation', ' Aviation', 'avia',
        # transporta  (542 firings)
        ' Transportation',
        # soar  (478 firings)
        ' soaring', ' soared', ' soar',
        # flew  (390 firings)
        ' flew',
        # jet  (332 firings)
        ' jet',
        # wing  (285 firings)
        ' wings', 'Wing', 'wings',
        # ski  (270 firings)
        ' skies',
        # aerospace  (260 firings)
        ' aerospace', ' Aerospace',
        # ufo  (233 firings)
        ' UFO', ' UFOs',
        # airplane  (229 firings)
        ' airplane',
        # flown  (210 firings)
        ' flown',
        # aer  (210 firings)
        ' aer', ' Aerial',
        # runway  (184 firings)
        ' runway',
        # av  (180 firings)
        ' av',
        # pil  (176 firings)
        ' Pil',
        # airlin  (151 firings)
        ' airlines', ' airliner',
        # squadron  (144 firings)
        ' Squadron', ' squadron',
        # airspace  (141 firings)
        ' airspace',
        # faa  (135 firings)
        ' FAA',
        # altitude  (131 firings)
        ' altitude',
        # cockpit  (124 firings)
        ' cockpit',
        # wreckage  (120 firings)
        ' wreckage',
        # anson  (113 firings)
        'anson',
        # airbu  (107 firings)
        ' Airbus',
        # plane  (105 firings)
        'plane',
        # fu  (101 firings)
        ' fu', 'FU',
        # airplan  (99 firings)
        ' airplanes',
        # fut  (95 firings)
        ' fut',
        # cir  (85 firings)
        ' Cir',
        # parachute  (85 firings)
        ' parachute',
        # autop  (82 firings)
        ' autop',
        # raf  (81 firings)
        ' RAF',
        # aut  (74 firings)
        'autical',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'flight-family', 1

def prog_L07F02428(input_tokens):
    """8 token families, 14 surface forms (train F1 of the uncapped optimum: 0.109).
    Fires at 127,037 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to various concepts and ideas"""
    # --- end external annotation ---
    fire = {
        # evid  (797 firings)
        ' evidence', 'evidence',
        # reason  (738 firings)
        ' reason', ' reasons', 'reasonable',
        # problem  (552 firings)
        ' problem', ' problems',
        # thing  (458 firings)
        ' things',
        # way  (425 firings)
        ' ways', ' WAY',
        # option  (354 firings)
        ' options', ' option',
        # kill  (341 firings)
        ' killed',
        # chang  (336 firings)
        ' changes',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'evid-family', 1

def prog_L07F02429(input_tokens):
    """16 token families, 22 surface forms (train F1 of the uncapped optimum: 0.375).
    Fires at 19,914 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: thoughts, beliefs, and opinions indicated by phrases like "in my mind", "no doubt in my mind", and "in our minds" """
    # --- end external annotation ---
    fire = {
        # mind  (2,346 firings)
        ' mind', ' minds',
        # heart  (1,268 firings)
        ' heart', ' hearts',
        # brain  (778 firings)
        ' brain', ' brains',
        # atten  (673 firings)
        ' attention',
        # eye  (563 firings)
        ' eyes',
        # thought  (329 firings)
        ' thoughts',
        # ment  (293 firings)
        ' mental',
        # soul  (290 firings)
        ' soul', ' souls',
        # imagina  (285 firings)
        ' imagination',
        # mouth  (263 firings)
        ' mouth', ' mouths',
        # inn  (132 firings)
        ' inner',
        # mental  (129 firings)
        ' mentally',
        # stomach  (119 firings)
        ' stomach',
        # inbox  (116 firings)
        ' inbox',
        # conscious  (111 firings)
        ' consciousness', ' consciously',
        # ear  (98 firings)
        ' ears',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'mind-family', 1

def prog_L07F02430(input_tokens):
    """2 token families, 6 surface forms (train F1 of the uncapped optimum: 0.958).
    Fires at 42,375 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to personal interactions with a female individual"""
    # --- end external annotation ---
    fire = {
        # her  (38,368 firings)
        ' her', ' Her', 'Her', ' hers', ' HER',
        # herself  (967 firings)
        ' herself',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'her-family', 1

def prog_L07F02431(input_tokens):
    """25 token families, 34 surface forms (train F1 of the uncapped optimum: 0.666).
    Fires at 13,539 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to self-identity and self-description"""
    # --- end external annotation ---
    fire = {
        # self  (4,717 firings)
        ' self', 'self', ' Self', 'Self', ' selfies',
        # autonom  (238 firings)
        ' autonomous',
        # peer  (174 firings)
        ' peer',
        # ego  (162 firings)
        ' ego',
        # selfie  (158 firings)
        ' selfie',
        # ulk  (138 firings)
        'ulk',
        # personal  (115 firings)
        ' Personally', 'Personally',
        # selfish  (110 firings)
        ' selfish',
        # own  (108 firings)
        'Own', ' Owner',
        # elf  (107 firings)
        'elf', 'ELF',
        # rowd  (95 firings)
        'rowd',
        # bull  (91 firings)
        'bull',
        # crowd  (87 firings)
        ' Crowd',
        # mak  (83 firings)
        ' Maker',
        # obsess  (82 firings)
        ' obsessive',
        # false  (71 firings)
        ' False', 'False',
        # selv  (63 firings)
        ' selves', 'selves',
        # tractor  (61 firings)
        ' tractor',
        # narcis  (54 firings)
        ' narciss',
        # kung  (53 firings)
        ' Kung',
        # oneself  (45 firings)
        ' oneself',
        # aunt  (45 firings)
        'aunt',
        # altru  (42 firings)
        ' altru',
        # van  (40 firings)
        ' vanity',
        # aut  (39 firings)
        'AUT',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'self-family', 1

def prog_L07F02432(input_tokens):
    """5 token families, 6 surface forms (train F1 of the uncapped optimum: 0.065).
    Fires at 14,932 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to states or conditions"""
    # --- end external annotation ---
    fire = {
        # of  (341 firings)
        'of', ' OF',
        # run  (131 firings)
        'run',
        # um  (105 firings)
        'UM',
        # emergency  (82 firings)
        ' emergency',
        # review  (70 firings)
        ' Review',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'of-family', 1

def prog_L07F02433(input_tokens):
    """47 token families, 56 surface forms (train F1 of the uncapped optimum: 0.686).
    Fires at 8,287 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: proper nouns with 'yr' followed by a number"""
    # --- end external annotation ---
    fire = {
        # yr  (500 firings)
        'yr', 'YR', ' yr',
        # pir  (473 firings)
        'pir', ' Pir', 'Pir', ' pir',
        # syracuse  (166 firings)
        ' Syracuse',
        # yll  (158 firings)
        'yll',
        # lur  (158 firings)
        ' lur', ' Lur',
        # tyr  (157 firings)
        ' Tyr', ' tyr',
        # lor  (156 firings)
        'lor',
        # dr  (154 firings)
        ' dr',
        # grim  (149 firings)
        ' grim',
        # yp  (145 firings)
        'yp',
        # oner  (138 firings)
        'oner',
        # vet  (126 firings)
        'Veter',
        # thr  (122 firings)
        ' Thr',
        # byrne  (106 firings)
        ' Byrne',
        # wy  (102 firings)
        'wy',
        # shr  (96 firings)
        ' shr',
        # rr  (95 firings)
        'rr', 'RR',
        # moon  (89 firings)
        ' moons',
        # r  (86 firings)
        'ér',
        # fury  (79 firings)
        ' Fury',
        # resur  (78 firings)
        ' resur',
        # kyr  (76 firings)
        ' Kyr',
        # upt  (73 firings)
        ' upt',
        # kier  (72 firings)
        ' Kier',
        # kry  (69 firings)
        ' Kry',
        # myr  (68 firings)
        ' Myr',
        # byrd  (67 firings)
        ' Byrd',
        # tir  (65 firings)
        ' Tir',
        # cyr  (64 firings)
        ' Cyr',
        # grizz  (63 firings)
        ' grizz',
        # ocr  (62 firings)
        'ocr',
        # garg  (61 firings)
        ' garg',
        # disgr  (60 firings)
        ' disgr',
        # razor  (58 firings)
        ' Razor',
        # erick  (54 firings)
        'erick',
        # frig  (53 firings)
        ' frig',
        # yrs  (52 firings)
        'yrs',
        # ear  (51 firings)
        'EAR',
        # ureen  (50 firings)
        'ureen',
        # wolv  (50 firings)
        ' Wolver',
        # ownt  (45 firings)
        'ownt',
        # eur  (44 firings)
        ' Eur',
        # ori  (38 firings)
        'orious',
        # yawn  (37 firings)
        ' yawn',
        # lair  (32 firings)
        ' Lair', 'Lair',
        # rhod  (30 firings)
        ' Rhod',
        # iky  (28 firings)
        'iky',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'yr-family', 1

def prog_L07F02434(input_tokens):
    """7 token families, 8 surface forms (train F1 of the uncapped optimum: 0.094).
    Fires at 74,696 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: names of news agencies or individuals accompanied by specific symbols or initials"""
    # --- end external annotation ---
    fire = {
        # \n  (65,924 firings)
        '\n',
        # )  (4,144 firings)
        ')',
        # by  (668 firings)
        'By',
        # photo  (503 firings)
        ' Photo',
        # buy  (492 firings)
        'Buy',
        # story  (263 firings)
        ' Story',
        # share  (259 firings)
        ' Share', 'SHARE',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return '\\n-family', 1

def prog_L07F02435(input_tokens):
    """7 token families, 12 surface forms (train F1 of the uncapped optimum: 0.563).
    Fires at 18,913 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: questions and topics related to various discussions and debates"""
    # --- end external annotation ---
    fire = {
        # ques  (4,727 firings)
        ' question', 'Question', ' Question', 'question',
        # question  (3,415 firings)
        ' questions', ' Questions', 'Questions',
        # doubt  (162 firings)
        ' doubts',
        # dilemma  (104 firings)
        ' dilemma',
        # myster  (97 firings)
        ' mysteries',
        # unch  (70 firings)
        ' unch',
        # undrum  (50 firings)
        'undrum',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ques-family', 1

def prog_L07F02436(input_tokens):
    """15 token families, 62 surface forms (train F1 of the uncapped optimum: 0.235).
    Fires at 368,420 of 36,786,566 positions.
    Fires at 1.00% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: sequences of steps or instructions"""
    # --- end external annotation ---
    fire = {
        # how  (5,982 firings)
        ' how', 'how', ' HOW', 'HOW',
        # way  (3,260 firings)
        ' way', ' ways', ' Ways',
        # simple  (2,360 firings)
        ' simple', ' Simple', 'Simple', 'simple',
        # follow  (1,994 firings)
        ' following', ' follow', ' followed', ' follows', 'follow',
        # learn  (1,836 firings)
        ' learn', ' learned', ' learning', ' Learn', ' Learning', 'Learn', ' Learned', ' learns', 'learn', 'Learning', 'learning',
        # easy  (1,723 firings)
        ' easy', ' Easy', 'Easy', 'easy',
        # proces  (1,460 firings)
        ' process', ' Process',
        # same  (1,297 firings)
        ' same',
        # code  (1,250 firings)
        ' code', ' Code', 'Code', ' CODE',
        # plan  (1,250 firings)
        ' plan', 'plan', ' PLAN',
        # step  (1,234 firings)
        ' steps', ' step', 'step', 'Step', ' Step', ' Steps', 'steps', ' STEP',
        # different  (1,194 firings)
        ' different', ' differently',
        # bas  (1,072 firings)
        ' basic', ' Basic', 'Basic', 'basic',
        # rul  (1,051 firings)
        ' rules', ' Rules', 'Rules', 'rules',
        # similar  (943 firings)
        ' similar', ' similarly', 'similar',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'how-family', 1

def prog_L07F02437(input_tokens):
    """5 token families, 6 surface forms (train F1 of the uncapped optimum: 0.075).
    Fires at 65,944 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of regulatory bodies or organizations, particularly related to food, drugs, and administration"""
    # --- end external annotation ---
    fire = {
        # and  (17,886 firings)
        ' and',
        # or  (2,649 firings)
        ' or',
        # &  (336 firings)
        ' &',
        # food  (231 firings)
        ' food', ' Food',
        # drug  (199 firings)
        ' Drug',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'and-family', 1

def prog_L07F02438(input_tokens):
    """47 token families, 57 surface forms (train F1 of the uncapped optimum: 0.672).
    Fires at 12,987 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: proper nouns related to the name "Kan" followed by a number"""
    # --- end external annotation ---
    fire = {
        # an  (926 firings)
        'an',
        # kan  (524 firings)
        ' Kan', 'kan', ' kan',
        # kh  (439 firings)
        'kh', ' KH', ' kh',
        # dash  (285 firings)
        ' Dash', 'Dash',
        # kathleen  (232 firings)
        ' Kathleen',
        # charl  (204 firings)
        ' Charl', 'Charl',
        # agan  (192 firings)
        'agan',
        # karn  (178 firings)
        ' Karn',
        # beng  (176 firings)
        ' Bengal',
        # oka  (165 firings)
        'oka',
        # aku  (162 firings)
        'aku',
        # ott  (152 firings)
        'Ott', ' Ott',
        # alison  (138 firings)
        ' Alison',
        # ottoman  (134 firings)
        ' Ottoman',
        # san  (134 firings)
        ' san',
        # wagn  (126 firings)
        ' Wagner',
        # flan  (123 firings)
        ' Flan',
        # kab  (120 firings)
        ' Kab',
        # oct  (120 firings)
        ' oct',
        # oku  (119 firings)
        'oku',
        # mu  (116 firings)
        'MU', ' MU',
        # haj  (113 firings)
        ' Haj',
        # sana  (112 firings)
        ' Sana',
        # ammu  (111 firings)
        'ammu',
        # sor  (107 firings)
        ' Sor',
        # wichita  (102 firings)
        ' Wichita',
        # ak  (102 firings)
        'Ak', ' ak',
        # ide  (100 firings)
        ' ide', ' Ide',
        # ezra  (95 firings)
        ' Ezra',
        # nab  (90 firings)
        ' Nab',
        # kara  (89 firings)
        ' Kara',
        # rak  (89 firings)
        ' Rak',
        # olk  (88 firings)
        'olk',
        # osaka  (82 firings)
        ' Osaka',
        # shan  (80 firings)
        ' Shan',
        # kern  (79 firings)
        ' Kern',
        # kde  (79 firings)
        ' KDE',
        # darth  (77 firings)
        ' Darth',
        # kant  (76 firings)
        ' Kant',
        # apan  (74 firings)
        'apan',
        # awan  (72 firings)
        'awan',
        # konami  (69 firings)
        ' Konami',
        # wid  (68 firings)
        ' Wid',
        # dalton  (67 firings)
        ' Dalton',
        # kilomet  (63 firings)
        ' kilomet',
        # kahn  (61 firings)
        ' Kahn',
        # hanna  (57 firings)
        ' Hanna',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'an-family', 1

def prog_L07F02439(input_tokens):
    """45 token families, 46 surface forms (train F1 of the uncapped optimum: 0.580).
    Fires at 15,849 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to exceptional or unparalleled events"""
    # --- end external annotation ---
    fire = {
        # unprecedent  (893 firings)
        ' unprecedented',
        # rare  (799 firings)
        ' rare',
        # extraordinary  (583 firings)
        ' extraordinary',
        # unusu  (583 firings)
        ' unusual',
        # histor  (583 firings)
        ' historic', 'historic',
        # incred  (386 firings)
        ' incredible',
        # landmark  (314 firings)
        ' landmark',
        # unlimit  (285 firings)
        ' unlimited',
        # astonish  (263 firings)
        ' astonishing',
        # sweep  (258 firings)
        ' sweeping',
        # groundbreak  (231 firings)
        ' groundbreaking',
        # startl  (222 firings)
        ' startling',
        # stagger  (198 firings)
        ' staggering',
        # revolutionary  (194 firings)
        ' revolutionary',
        # unusual  (164 firings)
        ' unusually',
        # catastroph  (156 firings)
        ' catastrophic',
        # unbeliev  (154 firings)
        ' unbelievable',
        # unaccept  (145 firings)
        ' unacceptable',
        # bombshell  (127 firings)
        ' bombshell',
        # breathtak  (118 firings)
        ' breathtaking',
        # unthink  (117 firings)
        ' unthinkable',
        # supreme  (114 firings)
        ' supreme',
        # overdue  (98 firings)
        ' overdue',
        # uncommon  (95 firings)
        ' uncommon',
        # astound  (92 firings)
        ' astounding',
        # unheard  (88 firings)
        ' unheard',
        # cedent  (82 firings)
        'cedented',
        # unilater  (79 firings)
        ' unilateral',
        # surpris  (75 firings)
        'surprisingly',
        # unparallel  (69 firings)
        ' unparalleled',
        # akin  (67 firings)
        ' akin',
        # transformat  (62 firings)
        ' transformative',
        # extraordinari  (61 firings)
        ' extraordinarily',
        # unimagin  (59 firings)
        ' unimaginable',
        # shock  (56 firings)
        ' shockingly',
        # million  (55 firings)
        ' Millions',
        # unmatch  (54 firings)
        ' unmatched',
        # monument  (53 firings)
        ' monumental',
        # watersh  (52 firings)
        ' watershed',
        # ludicr  (51 firings)
        ' ludicrous',
        # inexplic  (48 firings)
        ' inexplicable',
        # unbelievab  (41 firings)
        ' unbelievably',
        # accept  (41 firings)
        'acceptable',
        # aclysm  (40 firings)
        'aclysm',
        # uncanny  (40 firings)
        ' uncanny',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'unprecedent-family', 1

def prog_L07F02440(input_tokens):
    """6 token families, 16 surface forms (train F1 of the uncapped optimum: 0.067).
    Fires at 31,204 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to step-by-step instructions or processes"""
    # --- end external annotation ---
    fire = {
        # how  (581 firings)
        ' how', 'HOW', ' HOW',
        # step  (389 firings)
        ' step', 'step', ' steps', ' Step', 'steps', ' Steps',
        # toward  (380 firings)
        ' toward', ' towards', ' Towards',
        # forward  (247 firings)
        ' forward', ' Forward',
        # clos  (168 firings)
        ' closer',
        # furth  (144 firings)
        ' further',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'how-family', 1

def prog_L07F02441(input_tokens):
    """4 token families, 6 surface forms (train F1 of the uncapped optimum: 0.058).
    Fires at 4,707 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to discussing a specific topic or subject"""
    # --- end external annotation ---
    fire = {
        # regard  (89 firings)
        'Regarding', ' Regarding', ' regards',
        # men  (61 firings)
        ' mention',
        # king  (29 firings)
        'king',
        # yeah  (18 firings)
        ' yeah',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'regard-family', 1

def prog_L07F02442(input_tokens):
    """3 token families, 4 surface forms (train F1 of the uncapped optimum: 0.050).
    Fires at 31,559 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to relationships between people"""
    # --- end external annotation ---
    fire = {
        # of  (18,652 firings)
        ' of', 'of',
        # whom  (128 firings)
        ' whom',
        # mine  (109 firings)
        ' mine',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'of-family', 1

def prog_L07F02443(input_tokens):
    """29 token families, 52 surface forms (train F1 of the uncapped optimum: 0.303).
    Fires at 107,947 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: destructive actions or events"""
    # --- end external annotation ---
    fire = {
        # down  (2,349 firings)
        ' down', ' downed',
        # kill  (1,610 firings)
        ' killed',
        # clos  (1,450 firings)
        ' closed', ' closing', ' closes', 'closed', ' Closed', ' Closing',
        # fail  (896 firings)
        ' failed', ' fail', ' fails', ' Fail',
        # shut  (825 firings)
        ' shut', ' shuts',
        # lost  (814 firings)
        ' lost',
        # sold  (713 firings)
        ' sold', ' SOLD', ' Sold',
        # chang  (628 firings)
        ' changed',
        # destroy  (589 firings)
        ' destroyed',
        # stolen  (527 firings)
        ' stolen',
        # gone  (501 firings)
        ' gone',
        # broken  (466 firings)
        ' broken',
        # miss  (447 firings)
        ' missing',
        # disappear  (444 firings)
        ' disappeared', ' disappear', ' disappearing', ' disappears',
        # injur  (388 firings)
        ' injured',
        # collaps  (386 firings)
        ' collapsed', ' collapsing', ' collapses',
        # burn  (383 firings)
        ' burned', ' burn',
        # apart  (369 firings)
        ' apart',
        # fell  (363 firings)
        ' fell',
        # fir  (363 firings)
        ' fired', ' Fired',
        # crash  (357 firings)
        ' crashed', ' crashing', ' crashes',
        # damag  (344 firings)
        ' damaged',
        # remov  (330 firings)
        ' removed',
        # sale  (325 firings)
        ' sale',
        # affect  (310 firings)
        ' affected',
        # seiz  (307 firings)
        ' seized',
        # explod  (293 firings)
        ' exploded', ' exploding', ' explodes',
        # cancell  (289 firings)
        ' cancelled',
        # hack  (282 firings)
        ' hacked',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'down-family', 1

def prog_L07F02444(input_tokens):
    """1 token families, 3 surface forms (train F1 of the uncapped optimum: 0.396).
    Fires at 241,275 of 36,786,566 positions.
    Fires at 0.66% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of professional athletes and their career details"""
    # --- end external annotation ---
    fire = {
        # the  (95,183 firings)
        'The', ' The', 'THE',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'the-family', 1

def prog_L07F02445(input_tokens):
    """5 token families, 10 surface forms (train F1 of the uncapped optimum: 0.820).
    Fires at 38,459 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: the number "Three" repeated in the text"""
    # --- end external annotation ---
    fire = {
        # three  (22,018 firings)
        ' three', 'Three', ' Three', 'three', ' THREE',
        # four  (6,504 firings)
        ' four',
        # dozen  (1,393 firings)
        ' dozens', ' Dozens',
        # trio  (431 firings)
        ' trio',
        # zen  (105 firings)
        'zens',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'three-family', 1

def prog_L07F02446(input_tokens):
    """1 token families, 1 surface forms (train F1 of the uncapped optimum: 0.237).
    Fires at 158,134 of 36,786,566 positions.
    Fires at 0.43% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to political and social issues"""
    # --- end external annotation ---
    fire = {
        # in  (78,905 firings)
        ' in',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'in-family', 1

def prog_L07F02447(input_tokens):
    """7 token families, 11 surface forms (train F1 of the uncapped optimum: 0.790).
    Fires at 4,379 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: keywords related to personal computers (PCs)"""
    # --- end external annotation ---
    fire = {
        # pc  (2,561 firings)
        ' PC', 'PC', 'pc', ' pc',
        # pcs  (264 firings)
        ' PCs',
        # arcade  (148 firings)
        ' Arcade',
        # isp  (70 firings)
        ' ISP',
        # modem  (48 firings)
        ' modem',
        # orc  (44 firings)
        ' orc', ' Orc',
        # paladin  (11 firings)
        ' Paladin',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'pc-family', 1

def prog_L07F02448(input_tokens):
    """2 token families, 3 surface forms (train F1 of the uncapped optimum: 0.188).
    Fires at 73,176 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to companies or organizations"""
    # --- end external annotation ---
    fire = {
        # and  (61,044 firings)
        ' and', ' AND',
        # &  (1,489 firings)
        ' &',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'and-family', 1

def prog_L07F02449(input_tokens):
    """43 token families, 63 surface forms (train F1 of the uncapped optimum: 0.644).
    Fires at 13,018 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of or related to computer-generated animations and animators"""
    # --- end external annotation ---
    fire = {
        # animat  (742 firings)
        ' animated', ' Animated',
        # anima  (629 firings)
        ' animation', ' Animation', 'Animation',
        # anim  (386 firings)
        ' anim', 'anim', ' Anim', 'Anim',
        # gif  (304 firings)
        ' GIF', ' gif', 'gif',
        # atar  (228 firings)
        'atar', 'atars',
        # yang  (226 firings)
        ' Yang', 'Yang',
        # inventory  (220 firings)
        ' inventory', ' Inventory', 'inventory',
        # ang  (189 firings)
        'Ang', ' ANG',
        # action  (179 firings)
        'action',
        # emmanuel  (162 firings)
        ' Emmanuel',
        # illu  (161 firings)
        ' illusion', ' Illusion', 'illusion',
        # render  (142 firings)
        ' rendering', 'rendered',
        # miy  (139 firings)
        ' Miy',
        # ui  (137 firings)
        ' UI',
        # animation  (131 firings)
        ' animations',
        # avatar  (131 firings)
        ' Avatar', ' avatar',
        # marion  (111 firings)
        ' Marion',
        # incub  (107 firings)
        ' incub',
        # mund  (100 firings)
        ' Mund',
        # amphib  (100 firings)
        ' amphib',
        # anat  (99 firings)
        ' Anat',
        # limb  (98 firings)
        ' limbs',
        # skeleton  (95 firings)
        ' skeleton',
        # pixar  (92 firings)
        ' Pixar',
        # cartoon  (88 firings)
        ' Cartoon',
        # undead  (87 firings)
        ' undead', ' Undead',
        # motion  (86 firings)
        'motion', 'Motion',
        # ike  (83 firings)
        ' Ike',
        # anatomy  (81 firings)
        ' anatomy',
        # locom  (81 firings)
        ' locom',
        # neurolog  (78 firings)
        ' neurolog',
        # cg  (76 firings)
        ' CG', 'CG',
        # script  (72 firings)
        ' scripted',
        # bend  (71 firings)
        'bender',
        # lag  (68 firings)
        ' lag',
        # rozen  (63 firings)
        'rozen',
        # erica  (55 firings)
        ' Erica',
        # cgi  (53 firings)
        ' CGI',
        # instruction  (47 firings)
        ' instructional',
        # manuel  (47 firings)
        'manuel',
        # ument  (39 firings)
        'ument',
        # ballet  (38 firings)
        ' ballet',
        # svg  (37 firings)
        ' SVG',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'animat-family', 1

def prog_L07F02450(input_tokens):
    """12 token families, 18 surface forms (train F1 of the uncapped optimum: 0.101).
    Fires at 95,647 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to global situations and settings"""
    # --- end external annotation ---
    fire = {
        # into  (2,087 firings)
        ' into',
        # come  (915 firings)
        ' come',
        # reach  (832 firings)
        ' reached', ' reach', ' reaching', ' Reached',
        # going  (739 firings)
        ' going',
        # live  (531 firings)
        ' live',
        # toward  (478 firings)
        ' toward', ' towards',
        # enter  (403 firings)
        ' entered', ' entering',
        # look  (394 firings)
        ' looking',
        # fac  (383 firings)
        ' facing', ' faced',
        # liv  (284 firings)
        ' living',
        # age  (278 firings)
        ' age',
        # talk  (265 firings)
        ' talking',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'into-family', 1

def prog_L07F02451(input_tokens):
    """16 token families, 25 surface forms (train F1 of the uncapped optimum: 0.412).
    Fires at 26,680 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: legal terms related to court proceedings"""
    # --- end external annotation ---
    fire = {
        # court  (6,175 firings)
        ' court', ' Court', ' courts', 'Court', ' COURT',
        # judge  (1,618 firings)
        ' judge', ' Judge', 'Judge',
        # committee  (785 firings)
        ' committee',
        # council  (475 firings)
        ' council', 'Council',
        # panel  (454 firings)
        ' panel',
        # jury  (438 firings)
        ' jury',
        # commis  (323 firings)
        ' commission',
        # legislature  (242 firings)
        ' legislature', ' Legislature',
        # justic  (148 firings)
        ' justices',
        # tribun  (124 firings)
        ' tribunal', ' Tribunal',
        # coron  (117 firings)
        ' coroner',
        # fcc  (111 firings)
        ' FCC',
        # appeal  (105 firings)
        ' Appeals',
        # inqu  (83 firings)
        ' inquest',
        # fda  (83 firings)
        ' FDA',
        # examin  (69 firings)
        ' examiner',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'court-family', 1

def prog_L07F02452(input_tokens):
    """43 token families, 52 surface forms (train F1 of the uncapped optimum: 0.641).
    Fires at 13,035 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: websites or online platforms"""
    # --- end external annotation ---
    fire = {
        # ty  (2,129 firings)
        'ty', 'TY', ' ty', 'Ty', ' TY',
        # typ  (316 firings)
        ' typ', ' Typ', ' typed', 'Typ', 'typ',
        # ody  (209 firings)
        'ody',
        # sy  (175 firings)
        'sy',
        # fy  (166 firings)
        'fy',
        # ety  (163 firings)
        'ety',
        # specialty  (160 firings)
        ' specialty',
        # petty  (150 firings)
        ' petty', ' Petty',
        # phil  (144 firings)
        ' Philly',
        # pu  (139 firings)
        'PU',
        # cruelty  (138 firings)
        ' cruelty',
        # 130  (137 firings)
        '130',
        # indy  (134 firings)
        ' Indy',
        # itty  (124 firings)
        'itty',
        # beside  (121 firings)
        ' beside',
        # itary  (117 firings)
        'itary',
        # ucky  (116 firings)
        'ucky',
        # tery  (113 firings)
        'tery',
        # lafayette  (112 firings)
        ' Lafayette',
        # ا  (110 firings)
        'ا',
        # sty  (108 firings)
        'sty',
        # alty  (97 firings)
        'alty',
        # tie  (96 firings)
        'ties',
        # fatty  (91 firings)
        ' fatty',
        # sul  (90 firings)
        ' Sul',
        # falsehood  (85 firings)
        ' falsehood',
        # betty  (83 firings)
        ' Betty',
        # irrit  (78 firings)
        ' irrit',
        # ocol  (73 firings)
        'ocol',
        # patty  (70 firings)
        ' Patty',
        # border  (63 firings)
        ' bordering',
        # vic  (61 firings)
        'vic',
        # typew  (56 firings)
        ' typew',
        # ippy  (54 firings)
        'ippy',
        # tymology  (51 firings)
        'tymology',
        # icia  (51 firings)
        'icia',
        # smy  (50 firings)
        ' Smy',
        # gly  (49 firings)
        'gly',
        # oddy  (47 firings)
        'oddy',
        # typo  (46 firings)
        ' typo',
        # pty  (44 firings)
        'pty',
        # cery  (40 firings)
        'cery',
        # tty  (36 firings)
        'tty',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ty-family', 1

def prog_L07F02453(input_tokens):
    """2 token families, 3 surface forms (train F1 of the uncapped optimum: 0.057).
    Fires at 24,603 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: instances where the concept of needing more time is discussed"""
    # --- end external annotation ---
    fire = {
        # to  (20,115 firings)
        ' to', ' TO',
        # delib  (116 firings)
        ' deliber',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'to-family', 1

def prog_L07F02454(input_tokens):
    """7 token families, 12 surface forms (train F1 of the uncapped optimum: 0.135).
    Fires at 44,314 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to legislation and political discourse"""
    # --- end external annotation ---
    fire = {
        # but  (4,005 firings)
        ' but', 'but',
        # and  (3,448 firings)
        'and', ' And',
        # or  (1,046 firings)
        'or', ' Or', 'Or',
        # rath  (551 firings)
        ' rather', 'rather',
        # howev  (527 firings)
        ' however',
        # perhap  (227 firings)
        ' perhaps',
        # inde  (140 firings)
        ' indeed',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'but-family', 1

def prog_L07F02455(input_tokens):
    """25 token families, 52 surface forms (train F1 of the uncapped optimum: 0.127).
    Fires at 299,634 of 36,786,566 positions.
    Fires at 0.81% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases indicating upcoming or planned actions involving individuals"""
    # --- end external annotation ---
    fire = {
        # to  (22,915 firings)
        ' to',
        # been  (8,237 firings)
        ' been',
        # has  (5,802 firings)
        ' has',
        # be  (5,591 firings)
        ' be',
        # will  (5,177 firings)
        ' will',
        # his  (4,825 firings)
        ' his',
        # made  (1,834 firings)
        ' made',
        # being  (1,823 firings)
        ' being',
        # also  (1,548 firings)
        ' also', 'also',
        # now  (1,494 firings)
        ' now',
        # join  (1,382 firings)
        ' joined', ' join', ' joining', ' joins', 'join', 'joined',
        # return  (1,368 firings)
        ' return', ' returned', ' returning', ' returns', ' Returns', ' Returning',
        # still  (1,273 firings)
        ' still',
        # appear  (1,241 firings)
        ' appeared', ' appears', ' appear', ' appearance', ' appearing', ' Appears',
        # set  (1,041 firings)
        ' set',
        # mak  (971 firings)
        ' making', 'makes',
        # become  (904 firings)
        ' become',
        # found  (850 firings)
        ' found',
        # come  (832 firings)
        ' come',
        # com  (816 firings)
        ' coming', ' Coming', 'Coming', ' Comes',
        # tak  (783 firings)
        ' taking', ' takes', ' Takes',
        # came  (782 firings)
        ' came',
        # turn  (776 firings)
        ' turned', ' turn', ' turns', ' turning',
        # already  (768 firings)
        ' already',
        # play  (758 firings)
        ' played', ' playing', 'Playing',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'to-family', 1

def prog_L07F02456(input_tokens):
    """46 token families, 50 surface forms (train F1 of the uncapped optimum: 0.753).
    Fires at 9,768 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of the name "Ross." """
    # --- end external annotation ---
    fire = {
        # ros  (1,084 firings)
        ' Ross', 'ross', 'Ross',
        # randy  (282 firings)
        ' Randy',
        # bank  (266 firings)
        ' Banks',
        # lynn  (235 firings)
        ' Lynn',
        # rus  (207 firings)
        ' Russ', 'Russ',
        # joey  (198 firings)
        ' Joey',
        # milan  (195 firings)
        ' Milan',
        # silk  (194 firings)
        ' Silk',
        # finn  (186 firings)
        ' Finn',
        # rp  (181 firings)
        'RP', ' RP',
        # serbia  (168 firings)
        ' Serbia',
        # torr  (161 firings)
        ' Torres',
        # phill  (160 firings)
        ' Phillies',
        # raf  (145 firings)
        ' Raf',
        # randall  (137 firings)
        ' Randall',
        # chandl  (137 firings)
        ' Chandler',
        # rhod  (134 firings)
        ' Rhodes',
        # felix  (132 firings)
        ' Felix',
        # falk  (130 firings)
        ' Falk',
        # floyd  (129 firings)
        ' Floyd',
        # ray  (126 firings)
        ' Rays',
        # franc  (118 firings)
        ' Frances',
        # iss  (118 firings)
        'ISS',
        # rubin  (115 firings)
        ' Rubin',
        # monica  (107 firings)
        ' Monica',
        # ripple  (104 firings)
        ' Ripple',
        # rett  (100 firings)
        'rett',
        # sr  (98 firings)
        ' SR',
        # padr  (97 firings)
        ' Padres',
        # rossi  (84 firings)
        ' Rossi',
        # polk  (80 firings)
        ' Polk',
        # neoliber  (73 firings)
        ' neoliberal',
        # epstein  (73 firings)
        ' Epstein',
        # suffolk  (72 firings)
        ' Suffolk',
        # rm  (70 firings)
        ' RM',
        # cornwall  (67 firings)
        ' Cornwall',
        # tn  (65 firings)
        ' TN',
        # borderland  (61 firings)
        ' Borderlands',
        # nes  (52 firings)
        ' Ness',
        # butterfl  (46 firings)
        ' butterflies',
        # nz  (44 firings)
        'nz',
        # plex  (43 firings)
        ' Plex',
        # iann  (41 firings)
        'iann',
        # freebsd  (40 firings)
        ' FreeBSD',
        # ruk  (36 firings)
        ' Ruk',
        # blis  (26 firings)
        ' Bliss',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ros-family', 1

def prog_L07F02457(input_tokens):
    """14 token families, 16 surface forms (train F1 of the uncapped optimum: 0.265).
    Fires at 24,045 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: language related to agreement for receiving communications and services"""
    # --- end external annotation ---
    fire = {
        # try  (1,232 firings)
        ' try',
        # subscribe  (1,091 firings)
        ' Subscribe', ' subscribe',
        # javascript  (587 firings)
        ' Javascript', ' JavaScript',
        # subscrib  (583 firings)
        ' subscribing',
        # newslett  (493 firings)
        ' newsletters',
        # deliver  (457 firings)
        ' delivered',
        # see  (380 firings)
        ' See',
        # thank  (266 firings)
        ' Thank',
        # log  (176 firings)
        ' log',
        # regist  (176 firings)
        ' register',
        # our  (147 firings)
        ' ours',
        # inbox  (147 firings)
        ' inbox',
        # dispatch  (120 firings)
        ' Dispatch',
        # html  (85 firings)
        ' HTML',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'try-family', 1

def prog_L07F02458(input_tokens):
    """33 token families, 43 surface forms (train F1 of the uncapped optimum: 0.625).
    Fires at 9,329 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of the word "squad" in various contexts"""
    # --- end external annotation ---
    fire = {
        # squad  (1,728 firings)
        ' squad', ' Squad', ' squads',
        # outfit  (416 firings)
        ' outfit', ' outfits',
        # squ  (382 firings)
        ' Squ', 'Squ', 'squ', ' SQU',
        # nsw  (231 firings)
        ' NSW',
        # squadron  (142 firings)
        ' Squadron', ' squadron',
        # wesley  (130 firings)
        ' Wesley',
        # paulo  (130 firings)
        ' Paulo',
        # leone  (124 firings)
        ' Leone',
        # gregg  (119 firings)
        ' Gregg',
        # carlo  (102 firings)
        ' Carlo',
        # ward  (93 firings)
        ' ward',
        # xi  (91 firings)
        ' XI',
        # workshop  (89 firings)
        ' Workshop', 'workshop',
        # moh  (85 firings)
        'Moh',
        # mech  (84 firings)
        'Mech', ' Mech',
        # unmark  (61 firings)
        ' unmarked',
        # clad  (61 firings)
        ' clad',
        # bst  (59 firings)
        ' BST',
        # parish  (59 firings)
        ' parish',
        # squar  (50 firings)
        ' squares',
        # esley  (48 firings)
        'esley',
        # syndicate  (46 firings)
        ' Syndicate',
        # rolley  (45 firings)
        'rolley',
        # flank  (44 firings)
        ' flanked',
        # ensemble  (44 firings)
        ' ensemble',
        # van  (43 firings)
        ' vans',
        # param  (41 firings)
        ' Param', 'Param',
        # crew  (37 firings)
        'Crew',
        # platoon  (32 firings)
        ' platoon',
        # sov  (32 firings)
        'sov',
        # rella  (31 firings)
        'rella',
        # thigh  (26 firings)
        ' thighs',
        # alky  (25 firings)
        'alky',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'squad-family', 1

def prog_L07F02459(input_tokens):
    """29 token families, 63 surface forms (train F1 of the uncapped optimum: 0.318).
    Fires at 101,469 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: political action groups and organizations"""
    # --- end external annotation ---
    fire = {
        # cent  (916 firings)
        ' Center',
        # action  (645 firings)
        ' Action', 'Action', ' ACTION', 'ACTION',
        # campaign  (514 firings)
        ' Campaign', 'Campaign', ' campaigners', ' campaigner', 'campaign',
        # alli  (481 firings)
        ' Alliance',
        # freedom  (477 firings)
        ' Freedom', 'Freedom',
        # founda  (472 firings)
        ' Foundation', 'foundation',
        # right  (464 firings)
        ' Right', ' Rights', 'rights',
        # org  (453 firings)
        'org', 'Org',
        # associa  (443 firings)
        ' Association',
        # work  (442 firings)
        ' Workers', ' Work', ' Working', ' Worker', 'Works', 'workers', 'WORK',
        # against  (433 firings)
        ' Against', 'Against', 'against',
        # project  (401 firings)
        ' Project',
        # anti  (390 firings)
        ' Anti', 'anti', 'Anti',
        # defense  (343 firings)
        ' Defense',
        # def  (319 firings)
        ' Def', ' Defence', 'Def', ' DEF', 'DEF',
        # coali  (312 firings)
        ' Coalition',
        # women  (304 firings)
        ' Women',
        # matt  (300 firings)
        ' Matter', ' Matters',
        # occupy  (299 firings)
        ' Occupy',
        # fami  (290 firings)
        ' Family',
        # policy  (274 firings)
        ' Policy', 'Policy',
        # reform  (273 firings)
        ' Reform',
        # free  (270 firings)
        ' Free',
        # citizen  (266 firings)
        ' Citizens', ' Citizen',
        # anonym  (262 firings)
        ' Anonymous', 'Anonymous',
        # society  (262 firings)
        ' Society',
        # fund  (262 firings)
        ' Fund', 'Fund',
        # move  (259 firings)
        ' Movement', ' Move', 'Move', 'move',
        # advocacy  (255 firings)
        ' advocacy',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'cent-family', 1

def prog_L07F02460(input_tokens):
    """38 token families, 44 surface forms (train F1 of the uncapped optimum: 0.614).
    Fires at 7,315 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: occurrences of the word "warrant" in texts"""
    # --- end external annotation ---
    fire = {
        # warrant  (1,120 firings)
        ' warrant', ' warrants', ' warranted', ' Warrant', ' WARRANT',
        # suspicion  (358 firings)
        ' suspicion',
        # affidavit  (168 firings)
        ' affidavit',
        # marsh  (151 firings)
        ' Marsh', ' Marshal',
        # pardon  (142 firings)
        ' pardon',
        # fourth  (120 firings)
        ' Fourth',
        # patriot  (107 firings)
        ' Patriot',
        # ransom  (98 firings)
        ' ransom',
        # informant  (97 firings)
        ' informant', ' informants',
        # subpoena  (93 firings)
        ' subpoena',
        # wiret  (93 firings)
        ' wiret',
        # prob  (91 firings)
        ' probable',
        # magistrate  (89 firings)
        ' magistrate',
        # pard  (86 firings)
        ' pard',
        # grayson  (86 firings)
        ' Grayson',
        # law  (80 firings)
        ' lawful',
        # benson  (77 firings)
        ' Benson',
        # subpoen  (69 firings)
        ' subpoen',
        # forfeiture  (67 firings)
        ' forfeiture',
        # curfew  (64 firings)
        ' curfew',
        # wembley  (61 firings)
        ' Wembley',
        # summon  (59 firings)
        ' summons',
        # whim  (58 firings)
        ' whim',
        # pretext  (53 firings)
        ' pretext',
        # fisa  (52 firings)
        ' FISA',
        # underpin  (45 firings)
        ' underpin',
        # writ  (39 firings)
        ' writ',
        # want  (35 firings)
        ' Wanted',
        # odor  (33 firings)
        ' odor',
        # 802  (31 firings)
        ' 802',
        # arrant  (28 firings)
        'arranted',
        # neces  (28 firings)
        ' necess',
        # deem  (28 firings)
        ' deem',
        # const  (26 firings)
        ' Constable',
        # istrate  (26 firings)
        'istrate',
        # lawful  (23 firings)
        ' lawfully',
        # ield  (23 firings)
        'IELD',
        # constit  (21 firings)
        ' constit',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'warrant-family', 1

def prog_L07F02461(input_tokens):
    """39 token families, 55 surface forms (train F1 of the uncapped optimum: 0.713).
    Fires at 8,598 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases and names related to astronomy"""
    # --- end external annotation ---
    fire = {
        # ast  (1,481 firings)
        'ast', ' Ast', ' ast', 'Ast', ' Aster', ' aster', ' AST',
        # ust  (241 firings)
        'UST',
        # astro  (226 firings)
        ' Astros', ' Astro',
        # astron  (206 firings)
        ' Astron', ' astron',
        # spect  (206 firings)
        ' Spect', 'Spect',
        # ancient  (200 firings)
        ' Ancient', 'Ancient',
        # insp  (194 firings)
        'Insp', ' Insp',
        # vac  (190 firings)
        ' vac', ' Vac',
        # aston  (189 firings)
        ' Aston',
        # stro  (179 firings)
        ' Stro', 'stro',
        # est  (167 firings)
        ' EST',
        # obs  (162 firings)
        ' Obs', 'Obs',
        # adm  (136 firings)
        ' adm',
        # kazakhstan  (128 firings)
        ' Kazakhstan',
        # const  (125 firings)
        ' const',
        # hawth  (124 firings)
        ' Hawth',
        # asthma  (106 firings)
        ' asthma',
        # env  (106 firings)
        ' env',
        # superst  (104 firings)
        ' superst',
        # astronomy  (96 firings)
        ' astronomy',
        # astroph  (86 firings)
        ' astroph',
        # prest  (82 firings)
        ' Prest',
        # gast  (82 firings)
        ' Gast',
        # eld  (82 firings)
        ' Eld',
        # apost  (79 firings)
        ' apost', ' Apost',
        # advi  (77 firings)
        ' Advis', ' advis',
        # anat  (69 firings)
        ' Anat',
        # fd  (63 firings)
        ' FD',
        # explo  (55 firings)
        ' Explos',
        # inv  (54 firings)
        ' Invest',
        # camb  (51 firings)
        ' Camb',
        # aest  (46 firings)
        ' aest',
        # metast  (43 firings)
        ' metast',
        # kash  (43 firings)
        ' Kash',
        # chak  (40 firings)
        ' Chak',
        # catast  (38 firings)
        ' catast',
        # lan  (38 firings)
        ' Lans',
        # kazakh  (32 firings)
        ' Kazakh',
        # astr  (29 firings)
        ' Astral',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ast-family', 1

def prog_L07F02462(input_tokens):
    """47 token families, 59 surface forms (train F1 of the uncapped optimum: 0.554).
    Fires at 14,636 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: terms related to a specific topic or entity identified as "ider" """
    # --- end external annotation ---
    fire = {
        # ider  (1,061 firings)
        'ider', 'iders', 'IDER', 'idered',
        # patrol  (216 firings)
        ' Patrol',
        # iper  (207 firings)
        'iper', 'ipers',
        # ader  (195 firings)
        'ader',
        # redesign  (159 firings)
        ' redesign',
        # york  (152 firings)
        ' Yorker',
        # iker  (148 firings)
        'ikers', 'iker',
        # chie  (143 firings)
        'chie',
        # iding  (142 firings)
        'iding',
        # acon  (134 firings)
        'acon', 'acons',
        # enlist  (133 firings)
        ' enlisted',
        # prose  (116 firings)
        ' prose',
        # ayer  (114 firings)
        'ayer',
        # abel  (113 firings)
        'abel',
        # izen  (113 firings)
        'izen', 'izens',
        # cunningham  (112 firings)
        ' Cunningham',
        # ect  (104 firings)
        'ected',
        # fter  (102 firings)
        'fter',
        # idel  (99 firings)
        'idel',
        # propag  (98 firings)
        ' propag',
        # iden  (96 firings)
        'iden',
        # bau  (95 firings)
        ' Bauer',
        # jed  (91 firings)
        ' Jed',
        # oper  (89 firings)
        'opers',
        # rid  (88 firings)
        ' Rider', ' Riders', 'rider',
        # ecc  (80 firings)
        ' Ecc',
        # ubs  (79 firings)
        'ubs',
        # cler  (79 firings)
        ' cler', 'cler',
        # thorn  (77 firings)
        ' thorn',
        # →  (77 firings)
        ' →', '→',
        # cid  (75 firings)
        ' cider',
        # endra  (74 firings)
        'endra',
        # cer  (74 firings)
        'cers',
        # eat  (69 firings)
        ' Eater',
        # fitzgerald  (69 firings)
        ' Fitzgerald',
        # ilot  (67 firings)
        'ilot',
        # cic  (62 firings)
        ' Cic',
        # afer  (62 firings)
        'afer',
        # ned  (58 firings)
        ' Ned',
        # ection  (56 firings)
        'ection',
        # annotation  (46 firings)
        ' annotations', ' Annotations',
        # slid  (42 firings)
        ' slider',
        # idan  (42 firings)
        'idan',
        # ern  (41 firings)
        'erning',
        # riend  (41 firings)
        'riend',
        # cog  (40 firings)
        ' cog',
        # annot  (39 firings)
        ' annot',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'ider-family', 1

def prog_L07F02463(input_tokens):
    """16 token families, 35 surface forms (train F1 of the uncapped optimum: 0.097).
    Fires at 93,697 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to natural environments, specifically forests"""
    # --- end external annotation ---
    fire = {
        # want  (1,152 firings)
        ' want', ' wants', ' wanting', ' Want', 'want',
        # going  (1,119 firings)
        ' going', ' Going',
        # accord  (941 firings)
        ' according', ' According', 'according',
        # need  (821 firings)
        ' need', ' needed',
        # seem  (539 firings)
        ' seems', ' seemed', ' Seems',
        # try  (482 firings)
        ' trying', ' Trying',
        # like  (477 firings)
        ' likely', 'likely',
        # decid  (381 firings)
        ' decided', ' decides',
        # able  (351 firings)
        ' able',
        # add  (310 firings)
        ' added', 'Adding', ' Adding', 'added',
        # abil  (307 firings)
        ' ability',
        # expect  (296 firings)
        ' expected',
        # guilty  (293 firings)
        ' guilty',
        # tri  (293 firings)
        ' tried', ' tries',
        # continue  (277 firings)
        ' continue', 'continue',
        # fail  (275 firings)
        ' failed', ' failing',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'want-family', 1

def prog_L07F02464(input_tokens):
    """34 token families, 44 surface forms (train F1 of the uncapped optimum: 0.654).
    Fires at 11,174 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to competitive interactions or conflict, especially in a gaming context"""
    # --- end external annotation ---
    fire = {
        # opponent  (1,791 firings)
        ' opponents', ' opponent',
        # competitor  (548 firings)
        ' competitors', ' competitor',
        # matchup  (389 firings)
        ' matchup', ' matchups',
        # challeng  (346 firings)
        ' challenger', ' challengers',
        # oppos  (333 firings)
        ' opposing',
        # foe  (247 firings)
        ' foe', ' foes',
        # aj  (212 firings)
        ' Aj',
        # batt  (163 firings)
        ' batter', ' batters',
        # onent  (155 firings)
        'onents', 'onent',
        # assailant  (134 firings)
        ' assailant', ' assailants',
        # opp  (129 firings)
        ' Opp',
        # contestant  (108 firings)
        ' contestants', ' contestant',
        # unbeaten  (98 firings)
        ' unbeaten',
        # defens  (95 firings)
        ' defenses',
        # unsuccess  (90 firings)
        ' unsuccessful',
        # detractor  (89 firings)
        ' detractors',
        # obstacl  (85 firings)
        ' obstacles',
        # adversar  (84 firings)
        ' adversaries',
        # nem  (82 firings)
        ' nem',
        # azerbaijan  (78 firings)
        ' Azerbaijan',
        # atm  (78 firings)
        ' ATM',
        # interview  (75 firings)
        ' interviewer',
        # accus  (72 firings)
        ' accuser',
        # emulator  (71 firings)
        ' emulator',
        # antagonist  (67 firings)
        ' antagonist', ' antagonists',
        # undefeat  (67 firings)
        ' undefeated',
        # adversary  (66 firings)
        ' adversary',
        # formid  (62 firings)
        ' formidable',
        # hitt  (56 firings)
        ' hitters',
        # advers  (55 firings)
        ' adversity',
        # spectator  (53 firings)
        ' spectators',
        # ulator  (50 firings)
        'ulators',
        # concede  (39 firings)
        ' concede',
        # mat  (35 firings)
        ' mater',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'opponent-family', 1

def prog_L07F02465(input_tokens):
    """32 token families, 66 surface forms (train F1 of the uncapped optimum: 0.235).
    Fires at 208,569 of 36,786,566 positions.
    Fires at 0.57% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: specific events, organizations, and legislative terms"""
    # --- end external annotation ---
    fire = {
        # )  (3,396 firings)
        ')',
        # committee  (2,333 firings)
        ' Committee', ' committee', 'committee',
        # confer  (2,138 firings)
        ' conference', ' Conference', 'conference',
        # ser  (1,479 firings)
        ' series', ' Series', 'series',
        # cent  (1,471 firings)
        ' Center',
        # office  (1,413 firings)
        ' Office', ' office',
        # court  (1,322 firings)
        ' Court',
        # com  (1,222 firings)
        'com',
        # meet  (1,219 firings)
        ' meeting', ' meetings', ' Meeting',
        # act  (1,115 firings)
        ' Act',
        # program  (1,067 firings)
        ' program', ' Program', ' Programs', ' PROGRAM',
        # district  (1,041 firings)
        ' District', ' district', ' districts', ' DISTRICT',
        # website  (970 firings)
        ' website', ' Website',
        # council  (966 firings)
        ' Council', ' council',
        # magazine  (950 firings)
        ' magazine', ' Magazine',
        # elec  (911 firings)
        ' election', 'election',
        # championship  (906 firings)
        ' Championship', ' Championships', ' championship', ' championships',
        # cup  (873 firings)
        ' Cup', ' Cups',
        # league  (814 firings)
        ' League',
        # agency  (811 firings)
        ' agency', ' Agency',
        # award  (787 firings)
        ' Awards', ' Award', ' awards',
        # school  (779 firings)
        ' School',
        # depart  (772 firings)
        ' Department',
        # page  (710 firings)
        ' page',
        # associa  (706 firings)
        ' Association',
        # store  (702 firings)
        ' store', ' Store', 'store',
        # board  (697 firings)
        ' Board', ' board', ' Boards',
        # festiv  (683 firings)
        ' Festival', ' festival',
        # blog  (663 firings)
        ' blog', ' Blog',
        # group  (629 firings)
        ' Group', ' GROUP',
        # sta  (598 firings)
        ' station', ' Station',
        # show  (547 firings)
        ' Show',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return ')-family', 1

def prog_L07F02466(input_tokens):
    """21 token families, 26 surface forms (train F1 of the uncapped optimum: 0.203).
    Fires at 53,025 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: references to organizations, specific groups, and committees"""
    # --- end external annotation ---
    fire = {
        # secur  (789 firings)
        ' Security',
        # league  (694 firings)
        ' League',
        # park  (626 firings)
        ' Park', ' parks', ' Parks',
        # committee  (600 firings)
        ' Committee',
        # agency  (531 firings)
        ' Agency',
        # day  (474 firings)
        ' Day',
        # conven  (375 firings)
        ' Convention',
        # associa  (337 firings)
        ' Association',
        # anthem  (319 firings)
        ' anthem', ' Anthem',
        # institute  (317 firings)
        ' Institute',
        # space  (302 firings)
        ' Space',
        # health  (281 firings)
        ' Health',
        # hockey  (267 firings)
        ' Hockey',
        # intellig  (252 firings)
        ' Intelligence',
        # advis  (236 firings)
        ' adviser', ' Adviser',
        # football  (235 firings)
        ' Football',
        # defense  (213 firings)
        ' Defense',
        # guard  (207 firings)
        ' Guard', ' Guards',
        # academy  (152 firings)
        ' Academy',
        # review  (144 firings)
        ' Review',
        # rifle  (139 firings)
        ' Rifle',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'secur-family', 1

def prog_L07F02467(input_tokens):
    """25 token families, 89 surface forms (train F1 of the uncapped optimum: 0.225).
    Fires at 235,679 of 36,786,566 positions.
    Fires at 0.64% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions or quotes involving personal actions or statements"""
    # --- end external annotation ---
    fire = {
        # said  (8,091 firings)
        ' said', ' Said', 'said',
        # about  (5,998 firings)
        ' about', ' About', ' ABOUT',
        # say  (5,621 firings)
        ' says', ' say', ' saying', ' Says', ' Say', ' Saying', 'say', ' SAY',
        # told  (4,061 firings)
        ' told', 'told',
        # call  (3,837 firings)
        ' called', ' calling', ' call', ' calls', 'Calling', ' Calls', ' Calling',
        # announc  (2,546 firings)
        ' announced', ' announcing', ' announces',
        # ask  (2,464 firings)
        ' asked', ' ask', ' asking', ' asks',
        # claim  (2,324 firings)
        ' claimed', ' claims', ' claim', ' claiming', ' Claims',
        # talk  (1,256 firings)
        ' talked', ' talking', ' talks', 'Talking', ' Talks', ' Talking',
        # look  (1,187 firings)
        ' looking', ' looked',
        # tell  (1,161 firings)
        ' tell', ' tells', ' telling',
        # speak  (1,118 firings)
        ' speak', ' speaking', ' speaks', 'Speaking', ' Speaking', ' Speak',
        # describ  (1,012 firings)
        ' described', ' describes', ' describing', 'described',
        # reveal  (963 firings)
        ' revealed', ' revealing',
        # accus  (923 firings)
        ' accused', ' accusing', ' accuses',
        # threaten  (828 firings)
        ' threatened', ' threatening', ' threaten', ' threatens',
        # decid  (821 firings)
        ' decided', ' decides', ' deciding',
        # want  (770 firings)
        ' wants', ' wanting', ' WANT', 'want',
        # believ  (758 firings)
        ' believes', ' believed', ' believing',
        # spoke  (720 firings)
        ' spoke',
        # believe  (719 firings)
        ' believe',
        # gave  (699 firings)
        ' gave',
        # den  (676 firings)
        ' denied', ' den', ' denies', ' denial',
        # hop  (641 firings)
        ' hoping', ' hopes', ' hoped',
        # complain  (608 firings)
        ' complained', ' complaining', ' complain', ' complains',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'said-family', 1

def prog_L07F02468(input_tokens):
    """4 token families, 15 surface forms (train F1 of the uncapped optimum: 0.827).
    Fires at 27,748 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases indicating the beginning of an action or process"""
    # --- end external annotation ---
    fire = {
        # start  (17,246 firings)
        ' started', ' start', ' starting', ' starts', 'Starting', ' Starting', ' Start', 'Start', 'started', ' Starts', ' Started', ' START',
        # began  (5,634 firings)
        ' began',
        # begun  (823 firings)
        ' begun',
        # found  (89 firings)
        'founded',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'start-family', 1

def prog_L07F02469(input_tokens):
    """49 token families, 61 surface forms (train F1 of the uncapped optimum: 0.704).
    Fires at 60,662 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: concepts related to darkness and negative emotions"""
    # --- end external annotation ---
    fire = {
        # delay  (796 firings)
        ' delay', ' delays', ' Delay',
        # failure  (645 firings)
        ' failure', ' Failure',
        # resigna  (558 firings)
        ' resignation',
        # specula  (549 firings)
        ' speculation',
        # tragedy  (506 firings)
        ' tragedy',
        # departure  (502 firings)
        ' departure',
        # abs  (501 firings)
        ' absence',
        # sil  (500 firings)
        ' silence', ' Silence',
        # ines  (476 firings)
        'iness',
        # confu  (465 firings)
        ' confusion',
        # vulner  (412 firings)
        ' vulnerability',
        # uncertainty  (390 firings)
        ' uncertainty',
        # depres  (388 firings)
        ' depression',
        # chao  (372 firings)
        ' chaos',
        # dark  (372 firings)
        ' darkness', ' Darkness', ' DARK',
        # dis  (359 firings)
        ' disability', ' Disability',
        # frustra  (348 firings)
        ' frustration',
        # disappoint  (336 firings)
        ' disappointment',
        # complex  (332 firings)
        ' complexity', ' complexities',
        # inabil  (281 firings)
        ' inability',
        # nightmare  (272 firings)
        ' nightmare', ' Nightmare',
        # setback  (265 firings)
        ' setback', ' setbacks',
        # excite  (261 firings)
        ' excitement',
        # revela  (254 firings)
        ' revelation',
        # isola  (252 firings)
        ' isolation',
        # confronta  (242 firings)
        ' confrontation',
        # limitation  (238 firings)
        ' limitations',
        # mad  (236 firings)
        ' madness', ' Madness',
        # refus  (231 firings)
        ' refusal',
        # obses  (228 firings)
        ' obsession',
        # aggres  (224 firings)
        ' aggression',
        # disrup  (208 firings)
        ' disruption',
        # turmoil  (200 firings)
        ' turmoil',
        # ignor  (199 firings)
        ' ignorance',
        # nonsense  (198 firings)
        ' nonsense',
        # necess  (194 firings)
        ' necessity', ' necessities',
        # less  (191 firings)
        'lessness',
        # weak  (190 firings)
        ' weakness',
        # reflec  (190 firings)
        ' reflection',
        # disabil  (189 firings)
        ' disabilities',
        # complication  (181 firings)
        ' complications',
        # concus  (181 firings)
        ' concussion',
        # catastrophe  (179 firings)
        ' catastrophe',
        # vulnerabil  (178 firings)
        ' vulnerabilities',
        # secrecy  (175 firings)
        ' secrecy',
        # embarrass  (173 firings)
        ' embarrassment',
        # disappear  (173 firings)
        ' disappearance',
        # extremism  (170 firings)
        ' extremism',
        # conges  (167 firings)
        ' congestion',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'delay-family', 1

def prog_L07F02470(input_tokens):
    """21 token families, 30 surface forms (train F1 of the uncapped optimum: 0.125).
    Fires at 153,414 of 36,786,566 positions.
    Fires at 0.42% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases that suggest uncertainty or speculation"""
    # --- end external annotation ---
    fire = {
        # not  (3,260 firings)
        ' not', 'Not', ' Not',
        # when  (1,759 firings)
        ' when',
        # ?  (1,701 firings)
        '?',
        # what  (1,632 firings)
        ' what',
        # so  (1,319 firings)
        'So', ' So',
        # well  (869 firings)
        ' Well', 'Well',
        # howev  (823 firings)
        ' however', 'However',
        # why  (806 firings)
        ' why',
        # happen  (774 firings)
        ' happened', ' happen', ' happening', ' happens',
        # again  (773 firings)
        ' again',
        # case  (726 firings)
        ' case',
        # least  (677 firings)
        ' least',
        # that  (652 firings)
        'That',
        # enough  (651 firings)
        ' enough',
        # ques  (583 firings)
        ' question',
        # true  (557 firings)
        ' true',
        # exact  (549 firings)
        ' exactly', ' exact',
        # came  (525 firings)
        ' came',
        # com  (471 firings)
        ' comes',
        # wasn  (414 firings)
        ' wasn',
        # doesn  (393 firings)
        ' doesn',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'not-family', 1

def prog_L07F02471(input_tokens):
    """2 token families, 6 surface forms (train F1 of the uncapped optimum: 0.518).
    Fires at 15,892 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases indicating justification or reasoning"""
    # --- end external annotation ---
    fire = {
        # because  (9,526 firings)
        ' because', ' Because', 'Because', 'because',
        # cause  (50 firings)
        'cause', 'Cause',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'because-family', 1

def prog_L07F02472(input_tokens):
    """25 token families, 54 surface forms (train F1 of the uncapped optimum: 0.144).
    Fires at 316,872 of 36,786,566 positions.
    Fires at 0.86% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: instances of action or change in a governmental or organizational context"""
    # --- end external annotation ---
    fire = {
        # to  (22,424 firings)
        ' to',
        # been  (4,885 firings)
        ' been', 'been',
        # had  (3,207 firings)
        ' had',
        # its  (2,657 firings)
        ' its',
        # get  (1,934 firings)
        ' get', ' gets', ' Gets',
        # want  (1,783 firings)
        ' wants', ' want', ' wanted',
        # receiv  (1,359 firings)
        ' received', ' receiving', 'received',
        # made  (1,348 firings)
        ' made',
        # report  (1,312 firings)
        ' reported', ' reporting', 'Reporting',
        # found  (1,287 firings)
        ' found',
        # won  (1,280 firings)
        ' won',
        # got  (1,229 firings)
        ' got',
        # announc  (1,188 firings)
        ' announced', ' announces',
        # need  (1,135 firings)
        ' need', ' needs', ' needed', ' needing', 'needs',
        # take  (1,090 firings)
        ' take', 'take',
        # came  (1,089 firings)
        ' came',
        # come  (1,063 firings)
        ' come', 'come',
        # start  (1,060 firings)
        ' started', ' start', 'started',
        # releas  (961 firings)
        ' released', ' releasing',
        # und  (943 firings)
        ' under', ' undes',
        # took  (907 firings)
        ' took',
        # already  (886 firings)
        ' already', 'Already',
        # fac  (871 firings)
        ' facing', ' faced', ' faces',
        # ask  (868 firings)
        ' asked', ' asking', ' asks',
        # join  (850 firings)
        ' joined', ' join', ' joining', ' joins', 'joined',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'to-family', 1

def prog_L07F02473(input_tokens):
    """2 token families, 2 surface forms (train F1 of the uncapped optimum: 0.182).
    Fires at 62,725 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: dates and times in a specific format"""
    # --- end external annotation ---
    fire = {
        # at  (17,941 firings)
        ' at',
        # file  (262 firings)
        ' file',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'at-family', 1

def prog_L07F02474(input_tokens):
    """37 token families, 41 surface forms (train F1 of the uncapped optimum: 0.657).
    Fires at 4,629 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to neutrality or being impartial"""
    # --- end external annotation ---
    fire = {
        # neutr  (513 firings)
        ' neutral', 'neutral', ' Neutral',
        # balanc  (222 firings)
        ' balanced', 'balanced', ' Balanced',
        # light  (207 firings)
        ' lighter',
        # albion  (111 firings)
        ' Albion',
        # →  (99 firings)
        ' →',
        # benign  (83 firings)
        ' benign',
        # neut  (77 firings)
        ' neut',
        # kosh  (77 firings)
        ' kosher',
        # clean  (75 firings)
        ' cleaner',
        # saturat  (70 firings)
        ' saturated',
        # impart  (67 firings)
        ' impartial',
        # bland  (67 firings)
        ' bland',
        # indifferent  (62 firings)
        ' indifferent',
        # subdu  (59 firings)
        ' subdued',
        # dull  (58 firings)
        ' dull',
        # tranquil  (51 firings)
        ' tranquil',
        # aturat  (46 firings)
        'aturated',
        # mut  (46 firings)
        ' muted',
        # mute  (45 firings)
        ' mute',
        # acid  (42 firings)
        ' acidic',
        # unaffect  (40 firings)
        ' unaffected',
        # inert  (39 firings)
        ' inert',
        # unbias  (39 firings)
        ' unbiased',
        # valkyrie  (33 firings)
        ' Valkyrie',
        # equilibrium  (31 firings)
        ' equilibrium',
        # nas  (30 firings)
        ' nasal',
        # annihil  (29 firings)
        ' annihil',
        # nek  (29 firings)
        ' Nek',
        # ategor  (25 firings)
        'ategor',
        # vulcan  (24 firings)
        ' Vulcan',
        # nost  (22 firings)
        'nostic',
        # vinegar  (20 firings)
        ' vinegar',
        # utr  (17 firings)
        'utral',
        # irgin  (16 firings)
        'irgin',
        # イ  (16 firings)
        'イ',
        # turret  (14 firings)
        ' turrets',
        # passive  (13 firings)
        ' passively',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'neutr-family', 1

def prog_L07F02475(input_tokens):
    """7 token families, 13 surface forms (train F1 of the uncapped optimum: 0.604).
    Fires at 15,204 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mention of political campaigns and election-related activities"""
    # --- end external annotation ---
    fire = {
        # run  (4,963 firings)
        ' run', ' runs', 'run',
        # runn  (3,662 firings)
        ' running', ' Running', 'Running',
        # ran  (1,187 firings)
        ' ran',
        # candidacy  (270 firings)
        ' candidacy',
        # campaign  (238 firings)
        ' campaigning', ' campaigned',
        # field  (117 firings)
        ' fielded', ' fielding',
        # petro  (59 firings)
        ' Petro',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'run-family', 1

def prog_L07F02476(input_tokens):
    """36 token families, 38 surface forms (train F1 of the uncapped optimum: 0.436).
    Fires at 29,377 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of specific geographical locations, with a focus on South Africa"""
    # --- end external annotation ---
    fire = {
        # korea  (1,128 firings)
        ' Korea',
        # africa  (1,068 firings)
        ' Africa',
        # carolina  (962 firings)
        ' Carolina',
        # korean  (552 firings)
        ' Korean', ' Koreans',
        # african  (533 firings)
        ' African', ' Africans',
        # hi  (450 firings)
        'HI',
        # pradesh  (437 firings)
        ' Pradesh',
        # kingdom  (406 firings)
        ' Kingdom',
        # zealand  (366 firings)
        ' Zealand',
        # wal  (292 firings)
        ' Wales',
        # seoul  (268 firings)
        ' Seoul',
        # dakota  (259 firings)
        ' Dakota',
        # ver  (236 firings)
        'VER',
        # scotia  (232 firings)
        ' Scotia',
        # burg  (231 firings)
        'burg',
        # side  (217 firings)
        ' Side',
        # lanka  (208 firings)
        ' Lanka',
        # sudan  (194 firings)
        ' Sudan',
        # sa  (178 firings)
        ' SA',
        # zimbabwe  (177 firings)
        ' Zimbabwe',
        # johann  (171 firings)
        ' Johannes',
        # charleston  (171 firings)
        ' Charleston',
        # ataka  (134 firings)
        'ataka',
        # leone  (130 firings)
        ' Leone',
        # salvador  (125 firings)
        ' Salvador',
        # southampton  (123 firings)
        ' Southampton',
        # atoon  (105 firings)
        'atoon',
        # beng  (102 firings)
        ' Bengal',
        # paulo  (101 firings)
        ' Paulo',
        # mandela  (95 firings)
        ' Mandela',
        # nadu  (90 firings)
        ' Nadu',
        # andan  (87 firings)
        'andan',
        # ind  (86 firings)
        ' Indies',
        # pret  (81 firings)
        ' Pret',
        # apartheid  (81 firings)
        ' apartheid',
        # rica  (76 firings)
        ' Rica',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'korea-family', 1

def prog_L07F02477(input_tokens):
    """39 token families, 43 surface forms (train F1 of the uncapped optimum: 0.380).
    Fires at 125,929 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to planning and coordination"""
    # --- end external annotation ---
    fire = {
        # being  (3,637 firings)
        ' being',
        # work  (3,225 firings)
        ' working',
        # look  (2,945 firings)
        ' looking',
        # going  (2,842 firings)
        ' going',
        # try  (2,420 firings)
        ' trying',
        # tak  (1,827 firings)
        ' taking',
        # com  (1,553 firings)
        ' coming',
        # gett  (1,545 firings)
        ' getting',
        # doing  (1,520 firings)
        ' doing',
        # runn  (1,324 firings)
        ' running',
        # fac  (1,275 firings)
        ' facing',
        # plann  (1,236 firings)
        ' planning',
        # investigat  (1,053 firings)
        ' investigating',
        # talk  (961 firings)
        ' talking',
        # seek  (945 firings)
        ' seeking',
        # prepar  (926 firings)
        ' preparing', ' prepares',
        # consider  (918 firings)
        ' considering',
        # start  (842 firings)
        ' starting',
        # call  (807 firings)
        ' calling',
        # mov  (760 firings)
        ' moving',
        # fight  (744 firings)
        ' fighting',
        # head  (727 firings)
        ' heading', ' headed', 'heading',
        # hop  (684 firings)
        ' hoping',
        # ask  (658 firings)
        ' asking',
        # hold  (645 firings)
        ' holding',
        # think  (633 firings)
        ' thinking', 'thinking',
        # offer  (601 firings)
        ' offering',
        # wait  (600 firings)
        ' waiting',
        # push  (597 firings)
        ' pushing',
        # report  (505 firings)
        ' reporting',
        # speak  (492 firings)
        ' speaking',
        # develop  (485 firings)
        ' developing',
        # struggl  (446 firings)
        ' struggling',
        # launch  (394 firings)
        ' launching',
        # continu  (381 firings)
        ' continuing',
        # return  (372 firings)
        ' returning',
        # attempt  (364 firings)
        ' attempting',
        # suing  (362 firings)
        ' suing',
        # search  (326 firings)
        ' searching',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'being-family', 1

def prog_L07F02478(input_tokens):
    """42 token families, 53 surface forms (train F1 of the uncapped optimum: 0.731).
    Fires at 7,307 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: medical conditions related to hypoglycemia"""
    # --- end external annotation ---
    fire = {
        # hyp  (589 firings)
        ' hyp', 'hyp', ' Hyp', 'Hyp', ' hypers', 'Hyper', 'hyper',
        # hype  (419 firings)
        ' hype',
        # sup  (286 firings)
        ' Sup', 'Sup', 'sup', ' sup',
        # tayyip  (206 firings)
        ' Tayyip',
        # alleg  (205 firings)
        ' Alleg',
        # hypothesi  (187 firings)
        ' hypothesis',
        # dup  (178 firings)
        ' dup', ' Dup',
        # psych  (157 firings)
        ' psych',
        # geoff  (138 firings)
        ' Geoff',
        # unconsci  (137 firings)
        ' unconscious',
        # dmitry  (109 firings)
        ' Dmitry',
        # ips  (107 firings)
        ' Ips',
        # overweight  (103 firings)
        ' overweight',
        # scept  (93 firings)
        ' scept',
        # olig  (87 firings)
        ' olig',
        # pseud  (83 firings)
        ' pseud',
        # opp  (80 firings)
        ' opp',
        # postp  (79 firings)
        ' postp',
        # ide  (77 firings)
        ' ide',
        # parap  (76 firings)
        ' parap',
        # euph  (74 firings)
        ' euph',
        # mim  (68 firings)
        ' mim',
        # supp  (66 firings)
        ' supp',
        # ily  (64 firings)
        ' Ily',
        # typ  (60 firings)
        'Typ', ' Typ',
        # oxy  (57 firings)
        ' oxy',
        # suk  (57 firings)
        ' Suk',
        # overhe  (56 firings)
        ' overhe',
        # ht  (53 firings)
        ' HT',
        # hor  (46 firings)
        ' hors',
        # hypnot  (45 firings)
        ' hypnot',
        # dysph  (38 firings)
        ' dysph',
        # pse  (37 firings)
        ' Pse',
        # under  (37 firings)
        ' underest',
        # ————————————————  (35 firings)
        '————————————————',
        # ana  (34 firings)
        ' ana',
        # oct  (31 firings)
        'oct',
        # invi  (28 firings)
        ' invis',
        # hypot  (26 firings)
        ' hypot',
        # hypothes  (26 firings)
        ' hypotheses',
        # ove  (23 firings)
        ' ove',
        # morp  (23 firings)
        ' morp',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'hyp-family', 1

def prog_L07F02479(input_tokens):
    """3 token families, 4 surface forms (train F1 of the uncapped optimum: 0.052).
    Fires at 14,381 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to increasing or intensifying actions or characteristics"""
    # --- end external annotation ---
    fire = {
        # and  (376 firings)
        'and',
        # often  (224 firings)
        ' often',
        # human  (90 firings)
        ' Human', 'human',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'and-family', 1

def prog_L07F02480(input_tokens):
    """7 token families, 13 surface forms (train F1 of the uncapped optimum: 0.167).
    Fires at 129,487 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to raising awareness about social issues or causes"""
    # --- end external annotation ---
    fire = {
        # about  (12,137 firings)
        ' about', ' About', ' ABOUT',
        # against  (2,877 firings)
        ' against',
        # surround  (457 firings)
        ' surrounding',
        # addres  (452 firings)
        ' address',
        # highlight  (390 firings)
        ' highlight', ' highlighting', ' highlighted',
        # regard  (373 firings)
        ' regarding',
        # defend  (359 firings)
        ' defend', ' defending', ' defends',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'about-family', 1

def prog_L07F02481(input_tokens):
    """18 token families, 24 surface forms (train F1 of the uncapped optimum: 0.091).
    Fires at 37,232 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of professional associations or organizations"""
    # --- end external annotation ---
    fire = {
        # of  (9,587 firings)
        ' of', ' Of', ' OF',
        # (  (1,859 firings)
        ' (',
        # associa  (347 firings)
        ' Association',
        # nation  (346 firings)
        ' National',
        # state  (318 firings)
        ' State',
        # america  (194 firings)
        ' America',
        # against  (187 firings)
        ' Against',
        # de  (174 firings)
        ' de',
        # represent  (159 firings)
        ' representing', ' represents', 'represented',
        # party  (143 firings)
        ' Party',
        # ya  (128 firings)
        'ya',
        # getty  (124 firings)
        ' Getty',
        # intern  (118 firings)
        ' International', 'International',
        # council  (113 firings)
        ' Council',
        # women  (111 firings)
        ' Women',
        # mi  (95 firings)
        'mi',
        # organ  (95 firings)
        ' Organization', ' Organisation',
        # od  (95 firings)
        'od',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'of-family', 1

def prog_L07F02482(input_tokens):
    """1 token families, 3 surface forms (train F1 of the uncapped optimum: 0.115).
    Fires at 88,006 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: instances where someone knows something or the concept of knowledge"""
    # --- end external annotation ---
    fire = {
        # the  (83,959 firings)
        ' the', 'the', ' THE',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'the-family', 1

def prog_L07F02483(input_tokens):
    """20 token families, 39 surface forms (train F1 of the uncapped optimum: 0.400).
    Fires at 38,698 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to denial or refutation"""
    # --- end external annotation ---
    fire = {
        # confirm  (3,499 firings)
        ' confirmed', ' confirm', ' confirms', ' confirming',
        # den  (1,458 firings)
        ' denied', ' denies', ' denial',
        # admitt  (1,058 firings)
        ' admitted', ' admitting',
        # reject  (762 firings)
        ' rejected', ' reject', ' rejecting', ' rejects',
        # insist  (673 firings)
        ' insisted', ' insists', ' insist',
        # defend  (645 firings)
        ' defended', ' defending', ' defends',
        # deny  (630 firings)
        ' deny', ' denying',
        # refus  (492 firings)
        ' refused',
        # acknowledg  (362 firings)
        ' acknowledged', ' acknowledges',
        # maintain  (304 firings)
        ' maintained', ' maintains',
        # dismiss  (211 firings)
        ' dismissed',
        # declin  (207 firings)
        ' declined',
        # disput  (203 firings)
        ' disputed',
        # apolog  (189 firings)
        ' apologized', ' apologised',
        # admit  (185 firings)
        ' admits',
        # regret  (175 firings)
        ' regret', ' regrets',
        # reiterat  (171 firings)
        ' reiterated',
        # condemn  (150 firings)
        ' condemned', ' condemns',
        # apologize  (130 firings)
        ' apologize',
        # clarif  (106 firings)
        ' clarified',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'confirm-family', 1

def prog_L07F02484(input_tokens):
    """1 token families, 2 surface forms (train F1 of the uncapped optimum: 0.447).
    Fires at 22,730 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: specific time references"""
    # --- end external annotation ---
    fire = {
        # when  (14,236 firings)
        ' when', 'when',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'when-family', 1

def prog_L07F02485(input_tokens):
    """3 token families, 4 surface forms (train F1 of the uncapped optimum: 0.104).
    Fires at 26,581 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to nostalgia for old-fashioned things"""
    # --- end external annotation ---
    fire = {
        # -  (18,852 firings)
        '-',
        # sw  (130 firings)
        'sw', 'SW',
        # school  (101 firings)
        'school',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return '--family', 1

def prog_L07F02486(input_tokens):
    """3 token families, 3 surface forms (train F1 of the uncapped optimum: 0.030).
    Fires at 34,327 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: code examples and explanations"""
    # --- end external annotation ---
    fire = {
        # :  (2,135 firings)
        ':',
        # example  (136 firings)
        ' example',
        # .,  (127 firings)
        '.,',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return ':-family', 1

def prog_L07F02487(input_tokens):
    """48 token families, 63 surface forms (train F1 of the uncapped optimum: 0.613).
    Fires at 44,030 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of legislative actions or processes"""
    # --- end external annotation ---
    fire = {
        # polit  (1,934 firings)
        ' political',
        # leg  (1,098 firings)
        ' legal',
        # legislat  (940 firings)
        ' legislative', ' Legislative',
        # histor  (786 firings)
        ' historical', ' Historical',
        # creat  (772 firings)
        ' creative',
        # regulatory  (645 firings)
        ' regulatory', ' Regulatory',
        # diplomat  (629 firings)
        ' diplomatic',
        # parliamentary  (628 firings)
        ' parliamentary', ' Parliamentary',
        # elector  (561 firings)
        ' electoral',
        # judic  (549 firings)
        ' judicial', ' Judicial', 'judicial',
        # educ  (544 firings)
        ' educational', ' Educational',
        # scientif  (465 firings)
        ' scientific',
        # academ  (464 firings)
        ' academic',
        # investigat  (450 firings)
        ' investigative', ' Investigative',
        # cultur  (428 firings)
        ' cultural',
        # agricultur  (394 firings)
        ' agricultural', ' Agricultural',
        # mus  (392 firings)
        ' musical',
        # athlet  (387 firings)
        ' athletic',
        # literary  (375 firings)
        ' literary', ' Literary',
        # editor  (362 firings)
        ' editorial', ' Editorial',
        # technolog  (359 firings)
        ' technological',
        # administrat  (330 firings)
        ' administrative', ' Administrative',
        # roy  (321 firings)
        ' royal',
        # statist  (307 firings)
        ' statistical', ' Statistical',
        # nav  (296 firings)
        ' naval',
        # sport  (296 firings)
        ' sporting',
        # olog  (272 firings)
        'ological',
        # promotion  (263 firings)
        ' promotional',
        # artist  (247 firings)
        ' artistic',
        # resident  (246 firings)
        ' residential',
        # fiction  (236 firings)
        ' fictional',
        # tact  (223 firings)
        ' tactical', ' Tactical',
        # automot  (221 firings)
        ' automotive',
        # disciplinary  (193 firings)
        ' disciplinary',
        # municip  (192 firings)
        ' municipal',
        # oper  (191 firings)
        ' operational',
        # architectur  (179 firings)
        ' architectural',
        # mayor  (176 firings)
        ' mayoral',
        # pharmaceut  (174 firings)
        ' pharmaceutical',
        # cinemat  (171 firings)
        ' cinematic',
        # territor  (167 firings)
        ' territorial',
        # dietary  (167 firings)
        ' dietary', ' Dietary',
        # charit  (166 firings)
        ' charitable',
        # gubernator  (157 firings)
        ' gubernatorial',
        # archaeolog  (145 firings)
        ' archaeological',
        # surg  (143 firings)
        ' surgical',
        # theatr  (141 firings)
        ' theatrical',
        # government  (132 firings)
        ' governmental',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'polit-family', 1

def prog_L07F02488(input_tokens):
    """6 token families, 7 surface forms (train F1 of the uncapped optimum: 0.700).
    Fires at 19,509 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: proper nouns associated with locations and people"""
    # --- end external annotation ---
    fire = {
        # k  (13,508 firings)
        'K', ' K',
        # mck  (262 firings)
        ' McK',
        # king  (221 firings)
        'King',
        # ke  (76 firings)
        'Ke',
        # kr  (68 firings)
        'KR',
        # kc  (65 firings)
        'KC',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'k-family', 1

def prog_L07F02489(input_tokens):
    """31 token families, 34 surface forms (train F1 of the uncapped optimum: 0.639).
    Fires at 10,074 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: mentions of downtown areas and activities within a city center"""
    # --- end external annotation ---
    fire = {
        # downtown  (1,978 firings)
        ' downtown', ' Downtown',
        # town  (532 firings)
        'town',
        # plaza  (283 firings)
        ' Plaza', ' plaza',
        # mainland  (201 firings)
        ' mainland',
        # redevelop  (167 firings)
        ' redevelopment', ' redevelop',
        # hometown  (166 firings)
        ' hometown',
        # front  (165 firings)
        'front',
        # mont  (146 firings)
        'mont',
        # gta  (137 firings)
        ' GTA',
        # waterfront  (126 firings)
        ' waterfront',
        # stown  (126 firings)
        'stown',
        # skyline  (123 firings)
        ' skyline',
        # revit  (117 firings)
        ' revital',
        # histor  (116 firings)
        ' Historic',
        # storefront  (112 firings)
        ' storefront',
        # northeast  (111 firings)
        ' northeast',
        # upt  (90 firings)
        ' upt',
        # chinatown  (82 firings)
        ' Chinatown',
        # ancouv  (82 firings)
        'ancouver',
        # cbd  (82 firings)
        ' CBD',
        # disneyland  (80 firings)
        ' Disneyland',
        # bustl  (77 firings)
        ' bustling',
        # rundown  (67 firings)
        ' rundown',
        # city  (67 firings)
        'city',
        # harbour  (55 firings)
        ' harbour',
        # courtyard  (52 firings)
        ' courtyard',
        # owntown  (51 firings)
        'owntown',
        # recon  (47 firings)
        ' recon',
        # street  (46 firings)
        'street',
        # mart  (38 firings)
        ' mart',
        # upstair  (34 firings)
        ' upstairs',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'downtown-family', 1

def prog_L07F02490(input_tokens):
    """1 token families, 1 surface forms (train F1 of the uncapped optimum: 0.298).
    Fires at 55,357 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: prepositions followed by a place or event"""
    # --- end external annotation ---
    fire = {
        # for  (45,727 firings)
        ' for',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'for-family', 1

def prog_L07F02491(input_tokens):
    """21 token families, 89 surface forms (train F1 of the uncapped optimum: 0.237).
    Fires at 222,483 of 36,786,566 positions.
    Fires at 0.60% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to data analysis and database operations"""
    # --- end external annotation ---
    fire = {
        # data  (3,567 firings)
        ' data', ' Data', 'Data', 'data', ' DATA',
        # bas  (1,730 firings)
        ' based', 'based',
        # system  (1,689 firings)
        ' system', ' System',
        # informa  (1,612 firings)
        ' information', ' Information',
        # tool  (1,363 firings)
        ' tool', ' tools', ' Tool', ' Tools', 'Tools', 'tools',
        # using  (1,272 firings)
        ' using', 'Using', ' Using',
        # collect  (1,131 firings)
        ' collect', ' collected', ' collecting', ' collects', ' Collect', 'collect',
        # track  (1,101 firings)
        ' track', ' tracking', ' tracks', 'tracking', ' tracked', ' tracker', ' Tracker', ' Tracking', 'Tracker', 'Track', ' Tracks',
        # view  (1,035 firings)
        ' view', ' View', 'View', 'view', ' viewing', ' viewer',
        # releas  (1,033 firings)
        ' released',
        # search  (1,015 firings)
        ' search', 'Search', ' Search', ' searching', ' searches', 'search',
        # form  (810 firings)
        ' form', ' forms', 'form', ' Form', 'Form', ' Forms', 'forms', ' FORM',
        # online  (800 firings)
        ' online', 'online',
        # analysi  (729 firings)
        ' analysis', ' Analysis', 'analysis',
        # database  (720 firings)
        ' database', ' Database', 'database', 'Database',
        # map  (679 firings)
        ' map', ' maps', ' Maps', ' Map', 'Map', 'map', 'maps',
        # comput  (663 firings)
        ' computer', ' computational', ' comput', 'Computer', ' computed',
        # display  (627 firings)
        ' display', ' displays', ' displayed', ' displaying', ' Display',
        # site  (608 firings)
        ' site',
        # provid  (601 firings)
        ' provides', ' providing', ' Provides',
        # provide  (565 firings)
        ' provide',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'data-family', 1

def prog_L07F02492(input_tokens):
    """4 token families, 4 surface forms (train F1 of the uncapped optimum: 0.087).
    Fires at 47,608 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: details about people being taken to hospitals or medical facilities"""
    # --- end external annotation ---
    fire = {
        # to  (31,735 firings)
        ' to',
        # taken  (435 firings)
        ' taken',
        # outside  (303 firings)
        ' outside',
        # onto  (138 firings)
        ' onto',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'to-family', 1

def prog_L07F02493(input_tokens):
    """15 token families, 52 surface forms (train F1 of the uncapped optimum: 0.105).
    Fires at 416,622 of 36,786,566 positions.
    Fires at 1.13% of positions, and
    token rules do not serve features this dense: median test F1 0.086
    across the 37,173 features above 0.39% of positions, against 0.763 for
    the 76,283 below 0.016%. Read this fire set as the best available
    approximation, not as what the feature does."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to self-improvement and personal development"""
    # --- end external annotation ---
    fire = {
        # you  (9,429 firings)
        ' you', ' You', 'You', 'YOU',
        # are  (5,515 firings)
        ' are', ' Are', 'Are',
        # can  (3,682 firings)
        ' can',
        # when  (3,101 firings)
        ' when', ' When', 'When', 'when',
        # your  (2,791 firings)
        ' your', ' Your', 'Your', ' yours', ' YOUR',
        # if  (2,787 firings)
        ' if', ' If', 'If',
        # there  (2,029 firings)
        ' there', ' There',
        # most  (1,697 firings)
        ' most', ' Most', 'Most',
        # these  (1,510 firings)
        ' these', ' These', 'These', 'these',
        # make  (1,302 firings)
        ' make', ' Make', 'Make', ' MAKE',
        # know  (1,228 firings)
        ' know', ' knowing', ' knows', ' Knowing', 'Know', 'know', 'Knowing', ' KNOW',
        # even  (1,216 firings)
        ' even', ' Even', 'Even', 'even',
        # don  (1,176 firings)
        ' don', ' Don', 'Don', 'DON', ' DON',
        # !  (1,150 firings)
        '!',
        # get  (1,042 firings)
        ' get',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'you-family', 1

def prog_L07F02494(input_tokens):
    """10 token families, 21 surface forms (train F1 of the uncapped optimum: 0.138).
    Fires at 38,173 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases related to addressing or acknowledging an audience"""
    # --- end external annotation ---
    fire = {
        # !  (2,676 firings)
        '!', ' !',
        # everyone  (587 firings)
        ' everyone', ' Everyone',
        # hello  (404 firings)
        'Hello', ' Hello', ' hello',
        # guy  (239 firings)
        ' guys', ' Guys',
        # welcome  (206 firings)
        ' welcome', ' Welcome',
        # hi  (204 firings)
        'Hi', ' Hi', ' hi',
        # hey  (187 firings)
        'Hey', ' Hey', ' hey',
        # read  (176 firings)
        ' Reader', ' Readers',
        # reet  (158 firings)
        'reetings',
        # folk  (148 firings)
        ' folks',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return '!-family', 1

def prog_L07F02495(input_tokens):
    """28 token families, 36 surface forms (train F1 of the uncapped optimum: 0.367).
    Fires at 39,203 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to significant life events and changes"""
    # --- end external annotation ---
    fire = {
        # friend  (752 firings)
        'friendly', ' friendly', ' Friendly',
        # winn  (697 firings)
        'winning', 'winner',
        # free  (426 firings)
        'free',
        # break  (403 firings)
        'breaking', 'breaker', ' breaker',
        # relat  (363 firings)
        'related',
        # new  (296 firings)
        'new',
        # chang  (289 firings)
        'changing', ' chang',
        # end  (270 firings)
        'ending',
        # siz  (266 firings)
        'sized',
        # wide  (266 firings)
        'wide',
        # level  (259 firings)
        'level',
        # threaten  (207 firings)
        'threatening',
        # specif  (170 firings)
        'specific', 'Specific',
        # grow  (158 firings)
        'growing',
        # landmark  (157 firings)
        ' landmark',
        # high  (156 firings)
        'high',
        # look  (147 firings)
        'looking',
        # profile  (146 firings)
        'profile',
        # sav  (144 firings)
        'saving',
        # reach  (139 firings)
        'reaching',
        # torn  (135 firings)
        'torn',
        # bind  (135 firings)
        ' binding', 'binding',
        # need  (134 firings)
        'needed',
        # worthy  (132 firings)
        'worthy',
        # edge  (123 firings)
        'edge',
        # sett  (112 firings)
        'setting',
        # proof  (108 firings)
        'proof',
        # cost  (106 firings)
        'cost',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'friend-family', 1

def prog_L07F02496(input_tokens):
    """8 token families, 9 surface forms (train F1 of the uncapped optimum: 0.082).
    Fires at 57,459 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: parents or parental figures expressed in various contexts and perspectives"""
    # --- end external annotation ---
    fire = {
        # year  (892 firings)
        'year',
        # form  (470 firings)
        ' former',
        # fan  (307 firings)
        ' fan',
        # memb  (302 firings)
        ' member',
        # time  (238 firings)
        'time',
        # someone  (177 firings)
        ' someone',
        # student  (175 firings)
        ' student',
        # own  (150 firings)
        ' owner', 'owner',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'year-family', 1

def prog_L07F02497(input_tokens):
    """10 token families, 12 surface forms (train F1 of the uncapped optimum: 0.125).
    Fires at 53,687 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: locations and organizations related to ports"""
    # --- end external annotation ---
    fire = {
        # o  (1,029 firings)
        'o',
        # is  (435 firings)
        'IS',
        # h  (360 firings)
        'h',
        # qu  (235 firings)
        'qu',
        # to  (229 firings)
        'TO',
        # land  (162 firings)
        'LAND', 'lands',
        # ="  (152 firings)
        '="',
        # uk  (149 firings)
        'uk',
        # port  (146 firings)
        'port', ' Port',
        # ou  (139 firings)
        'ou',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'o-family', 1

def prog_L07F02498(input_tokens):
    """3 token families, 3 surface forms (train F1 of the uncapped optimum: 0.039).
    Fires at 13,707 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: phrases indicating a refusal to give information or comment"""
    # --- end external annotation ---
    fire = {
        # not  (1,608 firings)
        ' not',
        # com  (96 firings)
        ' comment',
        # wouldn  (49 firings)
        ' wouldn',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'not-family', 1

def prog_L07F02499(input_tokens):
    """25 token families, 58 surface forms (train F1 of the uncapped optimum: 0.266).
    Fires at 57,969 of 36,786,566 positions."""
    # --- begin external annotation (ignored at runtime) ---
    """prev SAE description: words related to suppressing or quelling actions or emotions"""
    # --- end external annotation ---
    fire = {
        # down  (1,655 firings)
        ' down', ' DOWN',
        # off  (1,273 firings)
        ' off',
        # ock  (1,045 firings)
        'ocked',
        # block  (873 firings)
        ' block', ' blocking', ' blocked',
        # control  (783 firings)
        ' control', 'control', ' CONTROL',
        # stop  (683 firings)
        ' stop',
        # prevent  (339 firings)
        ' prevent', ' preventing', ' prevented',
        # reduce  (330 firings)
        ' reduce',
        # combat  (298 firings)
        ' combat', ' combating',
        # ash  (292 firings)
        'ashing', 'ashed', 'ashes',
        # slow  (285 firings)
        ' slow', ' slowing', ' slowed', ' slows',
        # halt  (268 firings)
        ' halt', ' halted', ' halting',
        # quiet  (256 firings)
        ' quiet', ' quieter', ' Quiet',
        # sil  (254 firings)
        ' silence', 'sil', ' Silence',
        # suppres  (246 firings)
        ' suppress', ' suppression',
        # curb  (245 firings)
        ' curb',
        # limit  (241 firings)
        ' limit', ' limiting',
        # uff  (219 firings)
        'uff', 'uffed', 'uffs',
        # contain  (218 firings)
        ' contain', ' containment',
        # controll  (218 firings)
        ' controlling', ' controlled',
        # crush  (205 firings)
        ' crush', ' crushed', ' crushing',
        # delay  (198 firings)
        ' delay', ' delaying',
        # calm  (179 firings)
        ' calm', ' calmed', ' calming',
        # disrupt  (151 firings)
        ' disrupt', ' disrupting', ' disrupted',
        # unleash  (150 firings)
        ' unleashed', ' unleash',
    }
    token = input_tokens[-1]
    if token not in fire:
        return "no match", 0
    return 'down-family', 1