Views
No views yet
1def word_to_morph_features(text):
2 prefix = "<word analyze>: "
3 inputs = tokenizer(prefix + text, return_tensors="pt", max_length=128, truncation=True)
4
5 device = model.device
6 inputs = {k: v.to(device) for k, v in inputs.items()}
7
8 outputs = model.generate(**inputs)
9 return tokenizer.decode(outputs[0], skip_special_tokens=True)
10
11
12test_words = [
13 'зезгъэщхьырт',
14 'псыхьыжауэ',
15 'хъужами',
16 'япыщIат',
17 'къыщежьэри',
18 'схъумэрт',
19 'щхьэпэщ',
20 'сылъакъуэ',
21 'къеджэмэ',
22 'гъэщам',
23 'бэракъыу',
24 'уеупщIакъым',
25 'къэзыгъэпэж',
26 'къахуэбла',
27 'иращIэнтэкъым',
28 'къыбоух',
29 'гъусари',
30 'сщитIэгъащ',
31 'дытелажьэу',
32]
33for word in test_words:
34 features = word_to_morph_features(word)
35 print(f"{word} -> {features}")1зезгъэщхьырт -> <features>: 1sg-pre-1sg-caus-know-past
2псыхьыжауэ -> <features>: water-water-adv
3хъужами -> <features>: become-past-conn
4япыщIат -> <features>: 3pl-attach-past-aff
5къыщежьэри -> <features>: hor-begin-and
6схъумэрт -> <features>: 1sg-caus-stand-epv-fut
7щхьэпэщ -> <features>: dir-ben-val-aff
8сылъакъуэ -> <features>: 1sg-run-adv
9къеджэмэ -> <features>: hor-read-cond
10гъэщам -> <features>: year-erg
11бэракъыу -> <features>: flag-adv
12уеупщIакъым -> <features>: 2sg-ben-ask-neg
13къэзыгъэпэж -> <features>: hor-rel-caus-caus-run-back
14къахуэбла -> <features>: hor-ben-loc-approach
15иращIэнтэкъым -> <features>: 3pl-ben-3pl-do-epv-fut-neg
16къыбоух -> <features>: 2sg-dir-2sg-caus-fall
17гъусари -> <features>: companion-and
18сщитIэгъащ -> <features>: 1sg-ben-1sg-put-past
19дытелажьэу -> <features>: 1pl-dir-work-adv