Commit
•
7f9c385
1
Parent(s):
a61e9a0
Upload 13 files
Browse files- .gitattributes +3 -0
- xlm-roberta-large/customized-mwt/customized-mwt.tagger.mdl +3 -0
- xlm-roberta-large/customized-mwt/customized-mwt.tokenizer.mdl +3 -0
- xlm-roberta-large/customized-mwt/customized-mwt.vocabs.json +0 -0
- xlm-roberta-large/customized-mwt/customized-mwt_lemmatizer.pt +3 -0
- xlm-roberta-large/customized-mwt/customized-mwt_mwt_expander.pt +3 -0
- xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed +3 -0
- xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed.json +1 -0
- xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed.lock +9 -0
- xlm-roberta-large/customized-mwt/pl_nkjp_by_name_ud-ud-train.txt.character +0 -0
- xlm-roberta-large/customized-mwt/preds/lemmatizer.dev.conllu +0 -0
- xlm-roberta-large/customized-mwt/preds/mwt.dev.conllu +0 -0
- xlm-roberta-large/customized-mwt/preds/tagger.dev.conllu +0 -0
- xlm-roberta-large/customized-mwt/preds/tokenizer.dev.conllu +0 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
xlm-roberta-large/customized-mwt/customized-mwt.tagger.mdl filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
xlm-roberta-large/customized-mwt/customized-mwt.tokenizer.mdl filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed filter=lfs diff=lfs merge=lfs -text
|
xlm-roberta-large/customized-mwt/customized-mwt.tagger.mdl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:520f9c4f63964432e555fb44dd5c6c1647e972040958b080c1278bfe96b4d411
|
| 3 |
+
size 81445787
|
xlm-roberta-large/customized-mwt/customized-mwt.tokenizer.mdl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:371e448d1ded01b3672f69fc9f0eaf6fd7867b0d143e4a29f2d4469148093f73
|
| 3 |
+
size 50510463
|
xlm-roberta-large/customized-mwt/customized-mwt.vocabs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
xlm-roberta-large/customized-mwt/customized-mwt_lemmatizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3bbc6e11f349a20ac2c2ecf4918bc591f694239b73e2c6524c323cae832b5b64
|
| 3 |
+
size 10685944
|
xlm-roberta-large/customized-mwt/customized-mwt_mwt_expander.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c3cbc77268be4cbf060301abc1c8d9b58b679a52258e300c7ce2de0073aab964
|
| 3 |
+
size 700167
|
xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
|
| 3 |
+
size 5069051
|
xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"url": "https://s3.amazonaws.com/models.huggingface.co/bert/xlm-roberta-large-sentencepiece.bpe.model", "etag": "\"bf25eb5120ad92ef5c7d8596b5dc4046\""}
|
xlm-roberta-large/customized-mwt/f7e58cf8eef122765ff522a4c7c0805d2fe8871ec58dcb13d0c2764ea3e4a0f3.309f0c29486cffc28e1e40a2ab0ac8f500c203fe080b95f820aa9cb58e5b84ed.lock
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
--boundary_.oOo._Rb+LyG/YBjQbp7zp3OO1g3zKg7cFN8U1
|
| 3 |
+
Content-Length: 159535
|
| 4 |
+
Content-Type: application/octet-stream
|
| 5 |
+
X-File-MD5: 77df3b5c74c029c559e998bd59b1672f
|
| 6 |
+
X-File-Mtime: 1665901652
|
| 7 |
+
X-File-Path: /97- 8 Martyna Wiącek/models/trankit/ud/xlm-roberta-large/nkjp_by_name/xlm-roberta-large/customized-mwt/customized-mwt.vocabs.json
|
| 8 |
+
|
| 9 |
+
{"lemma": {"0": 0}, "upos": {"_": 0, "VERB": 1, "NOUN": 2, "ADP": 3, "PUNCT": 4, "NUM": 5, "SCONJ": 6, "PART": 7, "ADV": 8, "ADJ": 9, "PRON": 10, "CCONJ": 11, "DET": 12, "PROPN": 13, "AUX": 14, "X": 15, "INTJ": 16, "SYM": 17}, "xpos": {"_": 0, "praet:sg:m1:perf": 1, "subst:pl:acc:n:pt": 2, "prep:gen:nwok": 3, "subst:sg:gen:n:ncol": 4, "interp": 5, "num:pl:acc:m3:congr:ncol": 6, "subst:pl:acc:m3": 7, "subst:sg:acc:m3": 8, "subst:sg:acc:f": 9, "comp": 10, "inf:perf": 11, "part": 12, "adv:pos": 13, "ppas:pl:nom:n:perf:aff": 14, "prep:loc": 15, "subst:pl:loc:n:pt": 16, "ppron3:sg:acc:f:ter:akc:npraep": 17, "conj": 18, "adj:sg:loc:f:pos": 19, "subst:sg:loc:f": 20, "prep:inst": 21, "subst:pl:inst:m3": 22, "adv": 23, "praet:pl:m3:imperf": 24, "prep:loc:nwok": 25, "adj:sg:loc:n:pos": 26, "subst:sg:loc:n:ncol": 27, "adj:pl:nom:m3:pos": 28, "subst:pl:nom:m3": 29, "adj:pl:gen:m3:pos": 30, "praet:sg:n:imperf": 31, "ppron3:sg:gen:m1:ter:akc:npraep": 32, "subst:sg:gen:m3": 33, "subst:sg:inst:m1": 34, "subst:sg:gen:f": 35, "subst:sg:gen:m1": 36, "fin:sg:ter:imperf": 37, "subst:sg:nom:n:ncol": 38, "subst:sg:nom:m3": 39, "prep:gen": 40, "subst:pl:gen:m3": 41, "subst:sg:nom:f": 42, "adj:sg:nom:m3:pos": 43, "adj:sg:inst:m3:pos": 44, "subst:sg:inst:m3": 45, "subst:sg:inst:n:ncol": 46, "ppas:sg:inst:n:perf:aff": 47, "adj:sg:gen:f:pos": 48, "adj:sg:acc:f:pos": 49, "adj:sg:gen:n:pos": 50, "subst:sg:nom:m1": 51, "adjp:gen": 52, "praet:sg:m1:imperf": 53, "aglt:sg:pri:imperf:wok": 54, "adj:sg:loc:m3:pos": 55, "subst:sg:loc:m3": 56, "subst:sg:acc:n:ncol": 57, "fin:sg:pri:imperf": 58, "fin:sg:ter:perf": 59, "ppron3:sg:gen:f:ter:akc:npraep": 60, "ger:sg:gen:n:imperf:aff": 61, "subst:sg:inst:f": 62, "adj:sg:nom:f:pos": 63, "prep:acc:nwok": 64, "subst:pl:acc:f": 65, "adj:pl:acc:m3:pos": 66, "subst:pl:gen:m1": 67, "fin:pl:ter:imperf": 68, "subst:pl:nom:m1": 69, "adj:pl:nom:m1:pos": 70, "prep:acc": 71, "subst:pl:nom:n:ncol": 72, "adj:pl:nom:n:pos": 73, "subst:pl:gen:f": 74, "adj:sg:acc:m3:pos": 75, "pred": 76, "subst:pl:nom:f": 77, "prep:loc:wok": 78, "adj:pl:loc:f:pos": 79, "subst:pl:loc:f": 80, "adj:pl:dat:m1:pos": 81, "subst:pl:dat:m1": 82, "inf:imperf": 83, "adj:sg:acc:n:sup": 84, "adj:pl:loc:m3:pos": 85, "subst:pl:loc:m3": 86, "subst:pl:dat:n:pt": 87, "adv:com": 88, "ppron3:pl:gen:n:ter:akc:praep": 89, "prep:inst:nwok": 90, "fin:pl:pri:imperf": 91, "bedzie:sg:ter:imperf": 92, "pact:sg:inst:f:imperf:aff": 93, "adj:sg:nom:n:pos": 94, "pact:sg:nom:f:imperf:aff": 95, "num:pl:gen:f:congr:ncol": 96, "adj:pl:gen:f:sup": 97, "adj:pl:gen:f:pos": 98, "prep:gen:wok": 99, "ger:sg:acc:n:perf:aff": 100, "adj:sg:inst:f:pos": 101, "adj:sg:gen:m1:pos": 102, "adj:sg:gen:m3:pos": 103, "praet:sg:m1:perf:nagl": 104, "praet:sg:n:perf": 105, "adj:sg:inst:f:sup": 106, "praet:pl:f:imperf": 107, "praet:sg:f:perf": 108, "ppron3:pl:acc:f:ter:akc:npraep": 109, "xxs:nom": 110, "siebie:inst": 111, "praet:sg:f:imperf": 112, "adj:sg:nom:m1:pos": 113, "pcon:imperf": 114, "adj:sg:dat:f:pos": 115, "subst:sg:dat:f": 116, "subst:pl:gen:n:pt": 117, "pact:pl:inst:m3:imperf:aff": 118, "prep:nom": 119, "adj:pl:nom:f:pos": 120, "prep:inst:wok": 121, "ppas:sg:inst:m3:perf:aff": 122, "ger:sg:nom:n:imperf:aff": 123, "imps:imperf": 124, "subst:pl:loc:m1": 125, "subst:sg:loc:m1": 126, "subst:sg:gen:m2": 127, "praet:sg:m3:perf": 128, "subst:pl:gen:n:ncol": 129, "subst:pl:acc:n:col": 130, "xxx": 131, "ppas:sg:nom:f:perf:aff": 132, "ger:sg:inst:n:perf:aff": 133, "siebie:loc": 134, "adj:pl:acc:n:pos": 135, "subst:pl:acc:n:ncol": 136, "siebie:acc": 137, "ppas:pl:inst:n:perf:aff": 138, "subst:pl:inst:n:col": 139, "adj:pl:loc:n:pos": 140, "subst:pl:loc:n:col": 141, "praet:sg:m3:imperf": 142, "ppron3:sg:acc:m3:ter:nakc:npraep": 143, "ppron3:sg:gen:m3:ter:akc:npraep": 144, "subst:pl:inst:n:ncol": 145, "adj:sg:acc:n:pos": 146, "subst:sg:dat:m3": 147, "ppas:pl:nom:f:imperf:aff": 148, "adj:sg:inst:n:pos": 149, "praet:pl:n:imperf": 150, "ppas:sg:nom:m3:perf:aff": 151, "num:sg:acc:m3:rec": 152, "praet:sg:m2:imperf": 153, "subst:sg:nom:m2": 154, "ppas:pl:nom:m3:imperf:aff": 155, "subst:pl:acc:m1": 156, "ppron3:sg:dat:f:ter:akc:npraep": 157, "subst:pl:inst:f": 158, "pact:pl:nom:f:imperf:aff": 159, "praet:pl:m1:imperf": 160, "praet:pl:f:perf": 161, "fin:sg:sec:imperf": 162, "impt:sg:sec:imperf": 163, "siebie:gen": 164, "ppas:pl:acc:m3:perf:aff": 165, "adj:sg:nom:f:com": 166, "prep:dat": 167, "adj:sg:dat:n:pos": 168, "praet:sg:m2:perf": 169, "praet:pl:m1:perf": 170, "ger:sg:loc:n:perf:aff": 171, "pact:pl:nom:m1:imperf:aff": 172, "adja": 173, "adj:sg:nom:m2:pos": 174, "subst:pl:gen:m2": 175, "adj:sg:inst:m3:sup": 176, "siebie:dat": 177, "ppron3:sg:dat:m1:ter:nakc:npraep": 178, "adj:pl:acc:f:pos": 179, "subst:pl:nom:m2": 180, "praet:pl:m2:imperf": 181, "pact:pl:acc:n:imperf:aff": 182, "ppron3:pl:gen:m1:ter:akc:npraep": 183, "ppron3:pl:dat:m1:ter:akc:praep": 184, "ppas:sg:nom:m2:imperf:aff": 185, "ppas:sg:nom:f:imperf:aff": 186, "praet:pl:m3:perf": 187, "ppas:pl:nom:m3:perf:aff": 188, "num:pl:acc:m3:rec:ncol": 189, "ppron3:pl:inst:m1:ter:akc:praep": 190, "praet:sg:m1:imperf:nagl": 191, "ppas:sg:inst:f:perf:aff": 192, "adj:pl:nom:m2:pos": 193, "pact:pl:nom:m2:imperf:aff": 194, "adj:pl:gen:n:pos": 195, "ppron3:pl:dat:m1:ter:akc:npraep": 196, "subst:pl:nom:n:col": 197, "ppas:pl:nom:f:perf:aff": 198, "ppron3:pl:nom:m1:ter:akc:npraep": 199, "frag": 200, "pant:perf": 201, "subst:sg:acc:m1": 202, "subst:sg:acc:m2": 203, "pact:sg:acc:f:imperf:aff": 204, "ppron3:pl:acc:m1:ter:akc:npraep": 205, "ppas:pl:acc:f:perf:aff": 206, "adj:pl:acc:m2:pos": 207, "subst:pl:acc:m2": 208, "pact:sg:nom:n:imperf:aff": 209, "ppron12:pl:dat:m1:pri": 210, "ppron12:sg:dat:m1:pri:nakc": 211, "ppron3:sg:dat:m1:ter:akc:npraep": 212, "fin:sg:pri:perf": 213, "subst:sg:voc:m1": 214, "fin:pl:pri:perf": 215, "ppron12:pl:loc:m1:pri": 216, "ppron3:pl:gen:m1:ter:akc:praep": 217, "ppron12:pl:nom:m1:pri": 218, "pact:pl:nom:n:imperf:aff": 219, "adj:sg:acc:m1:pos": 220, "ppron3:sg:inst:m3:ter:akc:praep": 221, "ppas:pl:nom:m2:perf:aff": 222, "ppas:sg:acc:m2:perf:aff": 223, "ppron3:sg:acc:m1:ter:nakc:npraep": 224, "adj:sg:inst:m1:pos": 225, "ppron12:sg:acc:m1:pri:akc": 226, "aglt:sg:pri:imperf:nwok": 227, "adj:sg:gen:m3:sup": 228, "ppron12:sg:nom:m1:pri": 229, "ger:sg:gen:n:perf:aff": 230, "adv:sup": 231, "ppron3:sg:acc:n:ter:akc:praep": 232, "adj:pl:acc:m1:pos": 233, "pact:pl:acc:m1:imperf:aff": 234, "adj:pl:inst:n:pos": 235, "ppas:pl:nom:m1:perf:aff": 236, "adj:pl:gen:m1:pos": 237, "ppas:pl:gen:f:perf:aff": 238, "ppas:sg:acc:n:perf:aff": 239, "subst:pl:inst:m2": 240, "adj:sg:dat:m1:pos": 241, "subst:sg:dat:m1": 242, "ppron3:sg:inst:m1:ter:akc:praep": 243, "adj:sg:voc:f:pos": 244, "subst:sg:voc:f": 245, "subst:pl:nom:n:pt": 246, "subst:pl:inst:m1": 247, "fin:pl:ter:perf": 248, "aglt:pl:pri:imperf:nwok": 249, "subst:pl:inst:n:pt": 250, "ppron12:pl:nom:m1:sec": 251, "num:pl:nom:m1:congr:ncol": 252, "ppas:sg:nom:m1:perf:aff": 253, "praet:pl:n:perf": 254, "adj:pl:inst:f:pos": 255, "ppron3:sg:gen:m1:ter:nakc:praep": 256, "ppas:sg:acc:m3:perf:aff": 257, "ppas:sg:acc:f:perf:aff": 258, "adjc": 259, "ppron3:sg:nom:m1:ter:akc:npraep": 260, "ppron3:sg:acc:f:ter:akc:praep": 261, "adj:pl:loc:m1:pos": 262, "ppron3:sg:gen:m3:ter:nakc:npraep": 263, "fin:sg:sec:perf": 264, "num:pl:nom:m3:congr:ncol": 265, "ppas:sg:nom:n:perf:aff": 266, "subst:sg:gen:n:col": 267, "ppas:pl:gen:m3:perf:aff": 268, "adj:sg:voc:m1:pos": 269, "ppron3:sg:nom:f:ter:akc:npraep": 270, "ppron3:sg:loc:f:ter:akc:praep": 271, "interj": 272, "adj:sg:gen:m2:pos": 273, "ger:sg:loc:n:imperf:aff": 274, "imps:perf": 275, "adj:pl:nom:f:com": 276, "pact:sg:acc:m3:imperf:aff": 277, "ppas:pl:acc:m3:imperf:aff": 278, "ppron3:sg:acc:m1:ter:akc:praep": 279, "adj:pl:gen:m2:pos": 280, "pact:pl:acc:m3:imperf:aff": 281, "pact:pl:nom:m3:imperf:aff": 282, "ppas:sg:gen:n:perf:aff": 283, "subst:pl:nom:m1:pt": 284, "
|
xlm-roberta-large/customized-mwt/pl_nkjp_by_name_ud-ud-train.txt.character
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
xlm-roberta-large/customized-mwt/preds/lemmatizer.dev.conllu
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
xlm-roberta-large/customized-mwt/preds/mwt.dev.conllu
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
xlm-roberta-large/customized-mwt/preds/tagger.dev.conllu
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
xlm-roberta-large/customized-mwt/preds/tokenizer.dev.conllu
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|