Add gitlab issue/merge request templates
[src/snips-model-agl.git] / model / probabilistic_intent_parser / intent_parser.json
1 {
2   "config": {
3     "intent_classifier_config": {
4       "data_augmentation_config": {
5         "add_builtin_entities_examples": false,
6         "max_unknown_words": null,
7         "min_utterances": 20,
8         "noise_factor": 5,
9         "unknown_word_prob": 0.0,
10         "unknown_words_replacement_string": null
11       },
12       "featurizer_config": {
13         "added_cooccurrence_feature_ratio": 0.0,
14         "cooccurrence_vectorizer_config": {
15           "filter_stop_words": true,
16           "keep_order": true,
17           "unit_name": "cooccurrence_vectorizer",
18           "unknown_words_replacement_string": null,
19           "window_size": null
20         },
21         "pvalue_threshold": 0.4,
22         "tfidf_vectorizer_config": {
23           "unit_name": "tfidf_vectorizer",
24           "use_stemming": false,
25           "word_clusters_name": null
26         },
27         "unit_name": "featurizer"
28       },
29       "noise_reweight_factor": 1,
30       "unit_name": "log_reg_intent_classifier"
31     },
32     "slot_filler_config": {
33       "crf_args": {
34         "algorithm": "lbfgs",
35         "c1": 0.1,
36         "c2": 0.1
37       },
38       "data_augmentation_config": {
39         "add_builtin_entities_examples": true,
40         "capitalization_ratio": 0.2,
41         "min_utterances": 200
42       },
43       "feature_factory_configs": [
44         {
45           "args": {
46             "common_words_gazetteer_name": "top_10000_words_stemmed",
47             "n": 1,
48             "use_stemming": true
49           },
50           "factory_name": "ngram",
51           "offsets": [
52             -2,
53             -1,
54             0,
55             1,
56             2
57           ]
58         },
59         {
60           "args": {
61             "common_words_gazetteer_name": "top_10000_words_stemmed",
62             "n": 2,
63             "use_stemming": true
64           },
65           "factory_name": "ngram",
66           "offsets": [
67             -2,
68             1
69           ]
70         },
71         {
72           "args": {},
73           "factory_name": "is_digit",
74           "offsets": [
75             -1,
76             0,
77             1
78           ]
79         },
80         {
81           "args": {},
82           "factory_name": "is_first",
83           "offsets": [
84             -2,
85             -1,
86             0
87           ]
88         },
89         {
90           "args": {},
91           "factory_name": "is_last",
92           "offsets": [
93             0,
94             1,
95             2
96           ]
97         },
98         {
99           "args": {
100             "n": 1
101           },
102           "factory_name": "shape_ngram",
103           "offsets": [
104             0
105           ]
106         },
107         {
108           "args": {
109             "n": 2
110           },
111           "factory_name": "shape_ngram",
112           "offsets": [
113             -1,
114             0
115           ]
116         },
117         {
118           "args": {
119             "n": 3
120           },
121           "factory_name": "shape_ngram",
122           "offsets": [
123             -1
124           ]
125         },
126         {
127           "args": {
128             "entity_filter": {
129               "automatically_extensible": false
130             },
131             "tagging_scheme_code": 2,
132             "use_stemming": true
133           },
134           "factory_name": "entity_match",
135           "offsets": [
136             -2,
137             -1,
138             0
139           ]
140         },
141         {
142           "args": {
143             "entity_filter": {
144               "automatically_extensible": true
145             },
146             "tagging_scheme_code": 2,
147             "use_stemming": true
148           },
149           "drop_out": 0.5,
150           "factory_name": "entity_match",
151           "offsets": [
152             -2,
153             -1,
154             0
155           ]
156         },
157         {
158           "args": {
159             "tagging_scheme_code": 1
160           },
161           "factory_name": "builtin_entity_match",
162           "offsets": [
163             -2,
164             -1,
165             0
166           ]
167         },
168         {
169           "args": {
170             "cluster_name": "brown_clusters",
171             "use_stemming": false
172           },
173           "factory_name": "word_cluster",
174           "offsets": [
175             -2,
176             -1,
177             0,
178             1
179           ]
180         }
181       ],
182       "tagging_scheme": 1,
183       "unit_name": "crf_slot_filler"
184     },
185     "unit_name": "probabilistic_intent_parser"
186   },
187   "slot_fillers": [
188     {
189       "intent": "HVACFanSpeed",
190       "slot_filler_name": "slot_filler_0"
191     },
192     {
193       "intent": "HVACTemperature",
194       "slot_filler_name": "slot_filler_1"
195     },
196     {
197       "intent": "VolumeControl",
198       "slot_filler_name": "slot_filler_2"
199     }
200   ]
201 }