kiansheik commited on
Commit
d8a6d49
1 Parent(s): c34680d

add object direct

Browse files
added_tokens.json CHANGED
@@ -136,6 +136,7 @@
136
  "[OBJECT:2ps]": 32179,
137
  "[OBJECT:3p:MONOSYLLABIC]": 32167,
138
  "[OBJECT:3p]": 32137,
 
139
  "[OBJECT:MUTUAL]": 32106,
140
  "[OBJECT:REFLEXIVE]": 32148,
141
  "[OBJECT_MARKER:3p:DEFAULT]": 32114,
@@ -154,6 +155,7 @@
154
  "[SUBJECT:2pp]": 32172,
155
  "[SUBJECT:2ps:OBJECT_1P]": 32178,
156
  "[SUBJECT:2ps]": 32173,
 
157
  "[SUBJECT:3p]": 32113,
158
  "[SUBJECT_PREFIX:1ppe]": 32189,
159
  "[SUBJECT_PREFIX:1ppi]": 32107,
 
136
  "[OBJECT:2ps]": 32179,
137
  "[OBJECT:3p:MONOSYLLABIC]": 32167,
138
  "[OBJECT:3p]": 32137,
139
+ "[OBJECT:DIRECT]": 32194,
140
  "[OBJECT:MUTUAL]": 32106,
141
  "[OBJECT:REFLEXIVE]": 32148,
142
  "[OBJECT_MARKER:3p:DEFAULT]": 32114,
 
155
  "[SUBJECT:2pp]": 32172,
156
  "[SUBJECT:2ps:OBJECT_1P]": 32178,
157
  "[SUBJECT:2ps]": 32173,
158
+ "[SUBJECT:3p:DIRECT]": 32193,
159
  "[SUBJECT:3p]": 32113,
160
  "[SUBJECT_PREFIX:1ppe]": 32189,
161
  "[SUBJECT_PREFIX:1ppi]": 32107,
config.json CHANGED
@@ -57,5 +57,5 @@
57
  "torch_dtype": "float32",
58
  "transformers_version": "4.38.1",
59
  "use_cache": true,
60
- "vocab_size": 32193
61
  }
 
57
  "torch_dtype": "float32",
58
  "transformers_version": "4.38.1",
59
  "use_cache": true,
60
+ "vocab_size": 32195
61
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f79b9b04086111aba561d6b4686c472571fb1cdebd00e0436b11f6a6b72d9ed6
3
- size 242175016
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de19e80020a884acdd3e0d50703c96debfcfa822fdb1ada4bf5e33d5b8e3e2f7
3
+ size 242179112
special_tokens_map.json CHANGED
@@ -106,21 +106,21 @@
106
  "single_word": false
107
  },
108
  {
109
- "content": "[IMPERATIVE_PREFIX:2pp]",
110
  "lstrip": false,
111
  "normalized": false,
112
  "rstrip": false,
113
  "single_word": false
114
  },
115
  {
116
- "content": "o",
117
  "lstrip": false,
118
  "normalized": false,
119
  "rstrip": false,
120
  "single_word": false
121
  },
122
  {
123
- "content": "[SUBJECT_PREFIX:1ppi]",
124
  "lstrip": false,
125
  "normalized": false,
126
  "rstrip": false,
@@ -309,14 +309,14 @@
309
  "single_word": false
310
  },
311
  {
312
- "content": "gûi",
313
  "lstrip": false,
314
  "normalized": false,
315
  "rstrip": false,
316
  "single_word": false
317
  },
318
  {
319
- "content": "[GERUND_SUBJECT_PREFIX:2ps]",
320
  "lstrip": false,
321
  "normalized": false,
322
  "rstrip": false,
@@ -350,6 +350,13 @@
350
  "rstrip": false,
351
  "single_word": false
352
  },
 
 
 
 
 
 
 
353
  {
354
  "content": "mo",
355
  "lstrip": false,
@@ -553,6 +560,13 @@
553
  "rstrip": false,
554
  "single_word": false
555
  },
 
 
 
 
 
 
 
556
  {
557
  "content": "peîepé",
558
  "lstrip": false,
 
106
  "single_word": false
107
  },
108
  {
109
+ "content": "o",
110
  "lstrip": false,
111
  "normalized": false,
112
  "rstrip": false,
113
  "single_word": false
114
  },
115
  {
116
+ "content": "[SUBJECT_PREFIX:1ppi]",
117
  "lstrip": false,
118
  "normalized": false,
119
  "rstrip": false,
120
  "single_word": false
121
  },
122
  {
123
+ "content": "[IMPERATIVE_PREFIX:2pp]",
124
  "lstrip": false,
125
  "normalized": false,
126
  "rstrip": false,
 
309
  "single_word": false
310
  },
311
  {
312
+ "content": "[GERUND_SUBJECT_PREFIX:2ps]",
313
  "lstrip": false,
314
  "normalized": false,
315
  "rstrip": false,
316
  "single_word": false
317
  },
318
  {
319
+ "content": "gûi",
320
  "lstrip": false,
321
  "normalized": false,
322
  "rstrip": false,
 
350
  "rstrip": false,
351
  "single_word": false
352
  },
353
+ {
354
+ "content": "[SUBJECT:3p:DIRECT]",
355
+ "lstrip": false,
356
+ "normalized": false,
357
+ "rstrip": false,
358
+ "single_word": false
359
+ },
360
  {
361
  "content": "mo",
362
  "lstrip": false,
 
560
  "rstrip": false,
561
  "single_word": false
562
  },
563
+ {
564
+ "content": "[OBJECT:DIRECT]",
565
+ "lstrip": false,
566
+ "normalized": false,
567
+ "rstrip": false,
568
+ "single_word": false
569
+ },
570
  {
571
  "content": "peîepé",
572
  "lstrip": false,
tokenizer_config.json CHANGED
@@ -1656,6 +1656,22 @@
1656
  "rstrip": false,
1657
  "single_word": false,
1658
  "special": true
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1659
  }
1660
  },
1661
  "additional_special_tokens": [
@@ -1674,9 +1690,9 @@
1674
  "îa",
1675
  "umẽ",
1676
  "[GERUND_SUBJECT_PREFIX:1ps]",
1677
- "[IMPERATIVE_PREFIX:2pp]",
1678
  "o",
1679
  "[SUBJECT_PREFIX:1ppi]",
 
1680
  "xe",
1681
  "[SUBJECT_PREFIX:1ps]",
1682
  "[PLURIFORM_PREFIX:R]",
@@ -1703,12 +1719,13 @@
1703
  "[GERUND_SUBJECT_PREFIX:2pp]",
1704
  "[CIRCUMSTANTIAL_SUFFIX:CONSONANT_ENDING]",
1705
  "[GERUND_SUFFIX:CLASS_2:ORAL_VOWEL_ENDING]",
1706
- "gûi",
1707
  "[GERUND_SUBJECT_PREFIX:2ps]",
 
1708
  "ixé",
1709
  "namo",
1710
  "[OBJECT:3p]",
1711
  "ere",
 
1712
  "mo",
1713
  "[GERUND_SUFFIX:CLASS_2:NASAL_VOWEL_ENDING]",
1714
  "[SUBJECT:1ppe]",
@@ -1738,6 +1755,7 @@
1738
  "ta",
1739
  "[SUBJECT_PREFIX:2ps]",
1740
  "a'e",
 
1741
  "peîepé",
1742
  "a",
1743
  "[GERUND_SUBJECT_PREFIX:1ppe]",
 
1656
  "rstrip": false,
1657
  "single_word": false,
1658
  "special": true
1659
+ },
1660
+ "32193": {
1661
+ "content": "[SUBJECT:3p:DIRECT]",
1662
+ "lstrip": false,
1663
+ "normalized": false,
1664
+ "rstrip": false,
1665
+ "single_word": false,
1666
+ "special": true
1667
+ },
1668
+ "32194": {
1669
+ "content": "[OBJECT:DIRECT]",
1670
+ "lstrip": false,
1671
+ "normalized": false,
1672
+ "rstrip": false,
1673
+ "single_word": false,
1674
+ "special": true
1675
  }
1676
  },
1677
  "additional_special_tokens": [
 
1690
  "îa",
1691
  "umẽ",
1692
  "[GERUND_SUBJECT_PREFIX:1ps]",
 
1693
  "o",
1694
  "[SUBJECT_PREFIX:1ppi]",
1695
+ "[IMPERATIVE_PREFIX:2pp]",
1696
  "xe",
1697
  "[SUBJECT_PREFIX:1ps]",
1698
  "[PLURIFORM_PREFIX:R]",
 
1719
  "[GERUND_SUBJECT_PREFIX:2pp]",
1720
  "[CIRCUMSTANTIAL_SUFFIX:CONSONANT_ENDING]",
1721
  "[GERUND_SUFFIX:CLASS_2:ORAL_VOWEL_ENDING]",
 
1722
  "[GERUND_SUBJECT_PREFIX:2ps]",
1723
+ "gûi",
1724
  "ixé",
1725
  "namo",
1726
  "[OBJECT:3p]",
1727
  "ere",
1728
+ "[SUBJECT:3p:DIRECT]",
1729
  "mo",
1730
  "[GERUND_SUFFIX:CLASS_2:NASAL_VOWEL_ENDING]",
1731
  "[SUBJECT:1ppe]",
 
1755
  "ta",
1756
  "[SUBJECT_PREFIX:2ps]",
1757
  "a'e",
1758
+ "[OBJECT:DIRECT]",
1759
  "peîepé",
1760
  "a",
1761
  "[GERUND_SUBJECT_PREFIX:1ppe]",