1NGRAMSPLIT(1)                    User Commands                   NGRAMSPLIT(1)
2
3
4

NAME

6       ngramsplit - manual page for ngramsplit 1.3.14
7

DESCRIPTION

9       Split an n-gram model using context patterns.
10
11              Usage: ngramsplit [--options] in_fst [out_fsts_prefix]
12
13       PROGRAM FLAGS:
14
15       --backoff_label: type = int64_t, default = 0
16
17              Backoff  label --complete: type = bool, default = false Complete
18              partial models --contexts: type = std::string, default = "" Con‐
19              text  patterns file --far_type: type = std::string, default = ""
20              Type of far to compile (not FAR if empty):  one  of:  "default",
21              "stlist",  "sttable"  --method:  type  =  std::string, default =
22              "count_split"   One   of:    "count_split",    "histogram_split"
23              --norm_eps:  type  = double, default = 0.001 Normalization check
24              epsilon
25
26       LIBRARY FLAGS:
27
28       Flags from: compile-strings.cc
29
30       --far_field_separator: type = std::string, default = "
31              "
32
33              Set of characters used as a separator between printed fields
34
35       Flags from: flags.cc
36
37       --help: type = bool, default = false
38
39              show usage information --helpshort: type = bool, default = false
40              show  brief  usage information --tmpdir: type = std::string, de‐
41              fault = "/tmp" temporary directory --v: type = int32_t,  default
42              = 0 verbosity level
43
44       Flags from: fst.cc
45
46       --fst_align: type = bool, default = false
47
48              Write FST data aligned where appropriate --fst_default_cache_gc:
49              type = bool, default = true Enable garbage collection  of  cache
50              --fst_default_cache_gc_limit:  type = int64_t, default = 1048576
51              Cache   byte   size    that    triggers    garbage    collection
52              --fst_read_mode:  type  =  std::string, default = "read" Default
53              file reading mode for  mappable  files  --fst_verify_properties:
54              type  =  bool,  default = false Verify FST properties queried by
55              TestProperties --save_relabel_ipairs: type  =  std::string,  de‐
56              fault  =  ""  Save  input  relabel  pairs  to  file --save_rela‐
57              bel_opairs: type = std::string, default = "" Save output relabel
58              pairs to file
59
60       Flags from: ngram-output.cc
61
62       --end_symbol: type = std::string, default = "</s>"
63
64              Class label for sentence end --start_symbol: type = std::string,
65              default = "<s>" Class label for sentence start
66
67       Flags from: symbol-table.cc
68
69       --fst_compat_symbols: type = bool, default = true
70
71              Require symbol tables to match when appropriate --fst_field_sep‐
72              arator: type = std::string, default = "         " Set of charac‐
73              ters used as a separator between printed fields
74
75       Flags from: util.cc
76
77       --fst_error_fatal: type = bool, default = true
78
79              FST errors are fatal; o.w. return objects flagged as bad:  e.g.,
80              FSTs:   kError   property  set,  FST  weights:  not  a  Member()
81              --ngram_error_fatal: type = bool, default =  true  NGram  errors
82              are  fatal  if  true;  otherwise returns objects flagged as bad:
83              e.g., NGramModel::Error() is true
84
85       Flags from: weight.cc
86
87       --fst_weight_parentheses: type = std::string, default = ""
88
89              Characters enclosing the first weight  of  a  printed  composite
90              weight  (e.g., pair weight, tuple weight and derived classes) to
91              ensure proper I/O of nested composite weights; must have size  0
92              (none) or 2 (open and close parenthesis) --fst_weight_separator:
93              type = std::string, default = ","  Character  separator  between
94              printed composite weights; must be a single character
95
96
97
98ngramsplit 1.3.14                February 2022                   NGRAMSPLIT(1)