GRM-SFST
sfst-1.2.1
OpenGrm SFst Library
Main Page
Namespaces
Classes
Files
File List
File Members
src
grm
sfst
sfst-1.2.1
src
include
sfst
sfstlib.h
Go to the documentation of this file.
1
// Copyright 2018-2024 Google LLC
2
//
3
// Licensed under the Apache License, Version 2.0 (the 'License');
4
// you may not use this file except in compliance with the License.
5
// You may obtain a copy of the License at
6
//
7
// http://www.apache.org/licenses/LICENSE-2.0
8
//
9
// Unless required by applicable law or agreed to in writing, software
10
// distributed under the License is distributed on an 'AS IS' BASIS,
11
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
// See the License for the specific language governing permissions and
13
// limitations under the License.
14
// See sfst.opengrm.org for documentation of this library for
15
// stochastic finite-state transducers.
16
17
// SFst is a library for normalizing, sampling, combining, and
18
// approximating stochastic (or probabilistic) finite-state
19
// transducers. These are weighted finite-state transducers,
20
// represented in OpenFst library format, that have the following two
21
// properties:
22
//
23
// 1. normalized: the paths into the future from each state sum to
24
// Weight::One().
25
//
26
// 2. canonical topology:
27
// * the states are sorted by input label
28
// * there may be failure transitions (see phi-labeled transitions) but
29
// * there is at most one such transition per state
30
// * there are no failure-transition (and/or epsilon-transition) cycles
31
// * no assumption is made of general determinism or what transitions must be
32
// present on failure (unlike in a canonical n-gram model).
33
//
34
//
35
// For example, an n-gram model produced by the OpenGrm NGram Library
36
// is a stochastic FST (provided the phi_label is specified to match
37
// the backoff label, typically 0, of the n-gram model), but many
38
// other topologies are possible.
39
//
40
// If the input automaton is a transducer, then unless otherwise
41
// stated, the algorithms here only examine the input labels.
42
//
43
// Computation is done internally with Log64Weight (or sometimes
44
// SignedLog64Weight). Conversion from the input weight type is done
45
// using a fst::WeightConvert functor, pre-defined for common
46
// weight types like TropicalWeight and LogWeight.
47
48
// This convenience file includes all other SFST header files.
49
50
#ifndef NLP_GRM2_SFST_SFSTLIB_H_
51
#define NLP_GRM2_SFST_SFSTLIB_H_
52
53
#include <
sfst/equal.h
>
54
#include <
sfst/sfst.h
>
55
56
#endif // NLP_GRM2_SFST_SFSTLIB_H_
sfst.h
equal.h
Generated by
1.8.11