1NGRAMREAD(1) User Commands NGRAMREAD(1)
2
3
4
6 ngramread - manual page for ngramread 1.3.4
7
9 Transform text formats to fst.
10
11 Usage: ngramread [--options] [in.txt [out.fst]]
12
13 PROGRAM FLAGS:
14
15 --ARPA: type = bool, default = false
16
17 Read model from ARPA format --OOV_symbol: type = string, default
18 = "<unk>" Class label for OOV symbols --epsilon_symbol: type =
19 string, default = "<epsilon>" Label for epsilon transitions
20 --renormalize_arpa: type = bool, default = false If true, at‐
21 tempts to renormalize an unnormalized ARPA format model by nor‐
22 malizing the unigram state and recomputing the backoff weights.
23 Only used if --ARPA=true. --symbols: type = string, default =
24 "" Label symbol table
25
26 LIBRARY FLAGS:
27
28 Flags from: flags.cc
29
30 --help: type = bool, default = false
31
32 show usage information --helpshort: type = bool, default = false
33 show brief usage information --tmpdir: type = string, default =
34 "/tmp" temporary directory --v: type = int32, default = 0 ver‐
35 bosity level
36
37 Flags from: fst.cc
38
39 --fst_align: type = bool, default = false
40
41 Write FST data aligned where appropriate --fst_default_cache_gc:
42 type = bool, default = true Enable garbage collection of cache
43 --fst_default_cache_gc_limit: type = int64, default = 1048576
44 Cache byte size that triggers garbage collection
45 --fst_read_mode: type = string, default = "read" Default file
46 reading mode for mappable files --fst_verify_properties: type =
47 bool, default = false Verify FST properties queried by TestProp‐
48 erties --save_relabel_ipairs: type = string, default = "" Save
49 input relabel pairs to file --save_relabel_opairs: type =
50 string, default = "" Save output relabel pairs to file
51
52 Flags from: ngram-output.cc
53
54 --end_symbol: type = string, default = "</s>"
55
56 Class label for sentence end --start_symbol: type = string, de‐
57 fault = "<s>" Class label for sentence start
58
59 Flags from: symbol-table.cc
60
61 --fst_compat_symbols: type = bool, default = true
62
63 Require symbol tables to match when appropriate --fst_field_sep‐
64 arator: type = string, default = " " Set of characters used
65 as a separator between printed fields
66
67 Flags from: util.cc
68
69 --fst_error_fatal: type = bool, default = true
70
71 FST errors are fatal; o.w. return objects flagged as bad: e.g.,
72 FSTs: kError property set, FST weights: not a Member()
73 --ngram_error_fatal: type = bool, default = true NGram errors
74 are fatal if true; otherwise returns objects flagged as bad:
75 e.g., NGramModel::Error() is true
76
77 Flags from: weight.cc
78
79 --fst_weight_parentheses: type = string, default = ""
80
81 Characters enclosing the first weight of a printed composite
82 weight (e.g., pair weight, tuple weight and derived classes) to
83 ensure proper I/O of nested composite weights; must have size 0
84 (none) or 2 (open and close parenthesis) --fst_weight_separator:
85 type = string, default = "," Character separator between printed
86 composite weights; must be a single character
87
88
89
90ngramread 1.3.4 January 2018 NGRAMREAD(1)