Pilota model for hotel reviews

A model for Pilota trained with Hotel Review SCUD

• scud
  • Fine tuned model of t5-base-japanese-web (with Byte-fallback, 8K)
  • The original model is distributed in the Apache License 2.0
• scorer
  • Fine tuned model of LINE DistilBERT Japanese
  • The original model is distributed in the Apache License 2.0

Usage

1. Install Pilota

2. Prepare inputs

   • Command

     echo -e '平和記念公園へも徒歩圏内で駐車場もあり、便利なホテルです。' | python -m pilota.convert.plain2request | tee input.jsonl
     

   • Output

     {"context":null,"utterance":"平和記念園へも徒歩圏内で駐車場もあり、便利なホテルです。","sentences":null,"meta":{}}
     

3. Feed it to Pilota

   • Command

     pilota -m megagonlabs/pilota_hotel_review --batch_size 1 --outlen 60 --nbest 1 --beam 5 < input.jsonl
     

   • Output (Formatted by jq .)

     [
       {
         "scuds_nbest": [
           [
             "平和記念園へ▔徒歩圏内だ。",
             "駐車場が▔ある。",
             "便利なホテルだ。"
           ]
         ],
         "original_ranks": [
           0
         ],
         "scores": [
           0.9465934991836548
         ],
         "scores_detail": [
           {
             "OK": 0.8219783306121826,
             "incorrect_none": 0.016287358477711678,
             "lack": 0.09869803488254547,
             "limited": 0.05470234900712967,
             "non_fluent": 0.04080141708254814,
             "untruth": 0.008109904825687408
           }
         ],
         "sentence": "平和記念園へも徒歩圏内で駐車場もあり、便利なホテルです。"
       }
     ]
     

License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)