Browse Source

BIO Slot-Tagger: Alle Slot-Typen + erweiterte SlotRegistry

Erweiterte SlotRegistry um fehlende Slot-Typen:
  query (30 Werte: Kuenstler, Songs, Suchbegriffe)
  content (18 Werte: Notiz/Erinnerung-Inhalte)
  name (14 Werte: Timer-Namen)
  person (17 Werte: Deutsche Vornamen)
  topping (17 Werte: Pizza-Belaege)
  size (12 Werte: Groessen)
  pizza_type (8 Werte)
  preset_name, alias, level, value

BIO-Tags: 34 (vorher 12) — alle Slot-Typen trainiert
GPU→CPU Fix: auto_model.cpu() fuer Device-Mismatch

Ergebnis:
  "musik von rammstein spielen" → query: "rammstein" ✅
  "wetter morgen in berlin" → date: "morgen", city: "berlin" ✅
  "timer fuenf minuten" → duration: "fuenf minuten" ✅
  "merke dir milch kaufen" → content: "milch kaufen" ✅

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
patrick 2 months ago
parent
commit
d8f28bc7a4
2 changed files with 230 additions and 0 deletions
  1. 115 0
      config/slot_lists.json
  2. 115 0
      trainer/assets/intent/slot_lists.json

+ 115 - 0
config/slot_lists.json

@@ -205,5 +205,120 @@
       "dreizehn", "vierzehn", "fuenfzehn", "zwanzig", "dreissig",
       "hundert", "tausend"
     ]
+  },
+
+  "query": {
+    "description": "Freie Suchanfragen (Musik, Notizen, Kalender)",
+    "allow_unknown": true,
+    "values": [
+      "Rammstein", "Beethoven", "Mozart", "AC DC", "Metallica",
+      "Eminem", "Die Aerzte", "Deichkind", "Cro", "Seeed",
+      "Absolute Beginner", "Peter Fox", "Herbert Groenemeyer",
+      "Bohemian Rhapsody", "Stairway to Heaven", "Smells Like Teen Spirit",
+      "Rock Musik", "Jazz", "Klassik", "Hip Hop", "Pop Musik",
+      "Entspannungsmusik", "Schlafmusik", "Partymusik",
+      "Milch kaufen", "Arzttermin", "Geburtstag planen",
+      "Meeting vorbereiten", "Einkaufen gehen"
+    ]
+  },
+
+  "content": {
+    "description": "Freier Inhalt (Notizen, Erinnerungen)",
+    "allow_unknown": true,
+    "values": [
+      "Milch kaufen", "Brot holen", "Arzttermin", "Zahnarzt",
+      "Geburtstag planen", "Einkaufen gehen", "Muell rausbringen",
+      "Blumen giessen", "Wasche waschen", "Auto tanken",
+      "Paket abholen", "Rechnung bezahlen", "Anruf zurueckrufen",
+      "Meeting vorbereiten", "Praesentation fertig machen",
+      "Kuchen backen", "Hund fuettern", "Sport machen"
+    ]
+  },
+
+  "name": {
+    "description": "Namen (Timer, Geraete, etc.)",
+    "allow_unknown": true,
+    "values": [
+      "Pizza", "Eier", "Nudeln", "Tee", "Kaffee", "Kuchen",
+      "Waschmaschine", "Trockner", "Ofen", "Grill",
+      "Training", "Pause", "Meditation", "Alarm"
+    ]
+  },
+
+  "person": {
+    "description": "Personennamen",
+    "allow_unknown": true,
+    "values": [
+      "Patrick", "Anna", "Thomas", "Julia", "Michael", "Sarah",
+      "Max", "Lisa", "Felix", "Laura", "Daniel", "Marie",
+      "Mama", "Papa", "Oma", "Opa", "Chef"
+    ]
+  },
+
+  "topping": {
+    "description": "Pizza-Belaege",
+    "allow_unknown": true,
+    "values": [
+      "Salami", "Schinken", "Pilze", "Champignons", "Paprika",
+      "Zwiebeln", "Oliven", "Ananas", "Thunfisch", "Mais",
+      "Pepperoni", "Spinat", "Rucola", "Tomaten", "Mozzarella",
+      "Gorgonzola", "Extra Kaese"
+    ]
+  },
+
+  "size": {
+    "description": "Groessen (Pizza, etc.)",
+    "allow_unknown": false,
+    "values": [
+      "klein", "kleine", "kleines",
+      "mittel", "mittlere", "mittelgross",
+      "gross", "grosse", "grosses",
+      "familien", "riesig", "xxl"
+    ]
+  },
+
+  "pizza_type": {
+    "description": "Pizza-Typen",
+    "allow_unknown": true,
+    "values": [
+      "Margherita", "Salami", "Hawaii", "Tonno",
+      "Quattro Formaggi", "Diavola", "Capricciosa", "Vegetariana"
+    ]
+  },
+
+  "preset_name": {
+    "description": "Voice-Morph Preset-Namen",
+    "allow_unknown": true,
+    "values": [
+      "Trixy Maedchen", "Maennlich", "Weiblich", "Roboter",
+      "Tief", "Hoch", "Normal", "Fluestern"
+    ]
+  },
+
+  "alias": {
+    "description": "Satellite-Aliase",
+    "allow_unknown": true,
+    "values": [
+      "Patrick", "Wohnzimmer", "Schlafzimmer", "Kueche",
+      "Buero", "Kinderzimmer"
+    ]
+  },
+
+  "level": {
+    "description": "Lautstaerke-Level",
+    "allow_unknown": true,
+    "values": [
+      "10", "20", "30", "40", "50", "60", "70", "80", "90", "100",
+      "zehn", "zwanzig", "fuenfzig", "hundert"
+    ]
+  },
+
+  "value": {
+    "description": "Generischer Zahlenwert (Intensitaet etc.)",
+    "allow_unknown": true,
+    "values": [
+      "10", "20", "30", "40", "50", "60", "70", "80", "90", "100",
+      "0.5", "0.8", "1.0"
+    ]
   }
 }

+ 115 - 0
trainer/assets/intent/slot_lists.json

@@ -205,5 +205,120 @@
       "dreizehn", "vierzehn", "fuenfzehn", "zwanzig", "dreissig",
       "hundert", "tausend"
     ]
+  },
+
+  "query": {
+    "description": "Freie Suchanfragen (Musik, Notizen, Kalender)",
+    "allow_unknown": true,
+    "values": [
+      "Rammstein", "Beethoven", "Mozart", "AC DC", "Metallica",
+      "Eminem", "Die Aerzte", "Deichkind", "Cro", "Seeed",
+      "Absolute Beginner", "Peter Fox", "Herbert Groenemeyer",
+      "Bohemian Rhapsody", "Stairway to Heaven", "Smells Like Teen Spirit",
+      "Rock Musik", "Jazz", "Klassik", "Hip Hop", "Pop Musik",
+      "Entspannungsmusik", "Schlafmusik", "Partymusik",
+      "Milch kaufen", "Arzttermin", "Geburtstag planen",
+      "Meeting vorbereiten", "Einkaufen gehen"
+    ]
+  },
+
+  "content": {
+    "description": "Freier Inhalt (Notizen, Erinnerungen)",
+    "allow_unknown": true,
+    "values": [
+      "Milch kaufen", "Brot holen", "Arzttermin", "Zahnarzt",
+      "Geburtstag planen", "Einkaufen gehen", "Muell rausbringen",
+      "Blumen giessen", "Wasche waschen", "Auto tanken",
+      "Paket abholen", "Rechnung bezahlen", "Anruf zurueckrufen",
+      "Meeting vorbereiten", "Praesentation fertig machen",
+      "Kuchen backen", "Hund fuettern", "Sport machen"
+    ]
+  },
+
+  "name": {
+    "description": "Namen (Timer, Geraete, etc.)",
+    "allow_unknown": true,
+    "values": [
+      "Pizza", "Eier", "Nudeln", "Tee", "Kaffee", "Kuchen",
+      "Waschmaschine", "Trockner", "Ofen", "Grill",
+      "Training", "Pause", "Meditation", "Alarm"
+    ]
+  },
+
+  "person": {
+    "description": "Personennamen",
+    "allow_unknown": true,
+    "values": [
+      "Patrick", "Anna", "Thomas", "Julia", "Michael", "Sarah",
+      "Max", "Lisa", "Felix", "Laura", "Daniel", "Marie",
+      "Mama", "Papa", "Oma", "Opa", "Chef"
+    ]
+  },
+
+  "topping": {
+    "description": "Pizza-Belaege",
+    "allow_unknown": true,
+    "values": [
+      "Salami", "Schinken", "Pilze", "Champignons", "Paprika",
+      "Zwiebeln", "Oliven", "Ananas", "Thunfisch", "Mais",
+      "Pepperoni", "Spinat", "Rucola", "Tomaten", "Mozzarella",
+      "Gorgonzola", "Extra Kaese"
+    ]
+  },
+
+  "size": {
+    "description": "Groessen (Pizza, etc.)",
+    "allow_unknown": false,
+    "values": [
+      "klein", "kleine", "kleines",
+      "mittel", "mittlere", "mittelgross",
+      "gross", "grosse", "grosses",
+      "familien", "riesig", "xxl"
+    ]
+  },
+
+  "pizza_type": {
+    "description": "Pizza-Typen",
+    "allow_unknown": true,
+    "values": [
+      "Margherita", "Salami", "Hawaii", "Tonno",
+      "Quattro Formaggi", "Diavola", "Capricciosa", "Vegetariana"
+    ]
+  },
+
+  "preset_name": {
+    "description": "Voice-Morph Preset-Namen",
+    "allow_unknown": true,
+    "values": [
+      "Trixy Maedchen", "Maennlich", "Weiblich", "Roboter",
+      "Tief", "Hoch", "Normal", "Fluestern"
+    ]
+  },
+
+  "alias": {
+    "description": "Satellite-Aliase",
+    "allow_unknown": true,
+    "values": [
+      "Patrick", "Wohnzimmer", "Schlafzimmer", "Kueche",
+      "Buero", "Kinderzimmer"
+    ]
+  },
+
+  "level": {
+    "description": "Lautstaerke-Level",
+    "allow_unknown": true,
+    "values": [
+      "10", "20", "30", "40", "50", "60", "70", "80", "90", "100",
+      "zehn", "zwanzig", "fuenfzig", "hundert"
+    ]
+  },
+
+  "value": {
+    "description": "Generischer Zahlenwert (Intensitaet etc.)",
+    "allow_unknown": true,
+    "values": [
+      "10", "20", "30", "40", "50", "60", "70", "80", "90", "100",
+      "0.5", "0.8", "1.0"
+    ]
   }
 }