KoichiYasuoka commited on
Commit
cdccc81
1 Parent(s): c3579d3

model improved

Browse files
Files changed (4) hide show
  1. config.json +67 -83
  2. pytorch_model.bin +2 -2
  3. supar.model +2 -2
  4. tokenizer.json +0 -0
config.json CHANGED
@@ -164,40 +164,39 @@
164
  "152": "NOUN|\u540d\u8a5e|_",
165
  "153": "NOUN|\u540d\u8a5e|_+ADP|\u683c\u52a9\u8a5e|_",
166
  "154": "NOUN|\u540d\u8a5e|_+NOUN|\u4f4d\u7f6e\u540d\u8a5e|_",
167
- "155": "NOUN|\u540d\u8a5e|_+PART|\u4eba\u79f0\u63a5\u8f9e|_",
168
- "156": "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_",
169
- "157": "NOUN|\u56fa\u6709\u540d\u8a5e|_",
170
- "158": "NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
171
- "159": "NUM|\u6570\u8a5e|_",
172
- "160": "PART|\u4eba\u79f0\u63a5\u8f9e|_",
173
- "161": "PART|\u4eba\u79f0\u63a5\u8f9e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
174
- "162": "PART|\u4eba\u79f0\u63a5\u8f9e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_",
175
- "163": "PART|\u4eba\u79f0\u63a5\u8f9e|_+VERB|\u4ed6\u52d5\u8a5e|_",
176
- "164": "PART|\u63a5\u5c3e\u8f9e|_",
177
- "165": "PART|\u63a5\u7d9a\u52a9\u8a5e|_",
178
- "166": "PART|\u63a5\u982d\u8f9e|_",
179
- "167": "PART|\u7d42\u52a9\u8a5e|_",
180
- "168": "PART|\u81ea\u52d5\u8a5e|_",
181
- "169": "PROPN|\u56fa\u6709\u540d\u8a5e|_",
182
- "170": "PUNCT|\u8a18\u53f7|_",
183
- "171": "SCONJ|\u4ed6\u52d5\u8a5e|_",
184
- "172": "SCONJ|\u526f\u8a5e|_",
185
- "173": "SCONJ|\u5f8c\u7f6e\u526f\u8a5e|_",
186
- "174": "SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_",
187
- "175": "SCONJ|\u63a5\u7d9a\u8a5e|_",
188
- "176": "SCONJ|\u683c\u52a9\u8a5e|_",
189
- "177": "SCONJ|\u7d42\u52a9\u8a5e|_",
190
- "178": "SYM",
191
- "179": "SYM|_|_",
192
- "180": "VERB|\u4ed6\u52d5\u8a5e|_",
193
- "181": "VERB|\u4ed6\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
194
- "182": "VERB|\u4ed6\u52d5\u8a5e|_+PART|\u63a5\u5c3e\u8f9e|_",
195
- "183": "VERB|\u5b8c\u5168\u52d5\u8a5e|_",
196
- "184": "VERB|\u81ea\u52d5\u8a5e|_",
197
- "185": "VERB|\u81ea\u52d5\u8a5e|_+AUX|\u52a9\u52d5\u8a5e|_",
198
- "186": "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u540d\u8a5e|_",
199
- "187": "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
200
- "188": "VERB|\u81ea\u52d5\u8a5e|_+VERB|\u81ea\u52d5\u8a5e|_"
201
  },
202
  "initializer_range": 0.02,
203
  "intermediate_size": 3072,
@@ -357,40 +356,39 @@
357
  "NOUN|\u540d\u8a5e|_": 152,
358
  "NOUN|\u540d\u8a5e|_+ADP|\u683c\u52a9\u8a5e|_": 153,
359
  "NOUN|\u540d\u8a5e|_+NOUN|\u4f4d\u7f6e\u540d\u8a5e|_": 154,
360
- "NOUN|\u540d\u8a5e|_+PART|\u4eba\u79f0\u63a5\u8f9e|_": 155,
361
- "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_": 156,
362
- "NOUN|\u56fa\u6709\u540d\u8a5e|_": 157,
363
- "NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 158,
364
- "NUM|\u6570\u8a5e|_": 159,
365
- "PART|\u4eba\u79f0\u63a5\u8f9e|_": 160,
366
- "PART|\u4eba\u79f0\u63a5\u8f9e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 161,
367
- "PART|\u4eba\u79f0\u63a5\u8f9e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": 162,
368
- "PART|\u4eba\u79f0\u63a5\u8f9e|_+VERB|\u4ed6\u52d5\u8a5e|_": 163,
369
- "PART|\u63a5\u5c3e\u8f9e|_": 164,
370
- "PART|\u63a5\u7d9a\u52a9\u8a5e|_": 165,
371
- "PART|\u63a5\u982d\u8f9e|_": 166,
372
- "PART|\u7d42\u52a9\u8a5e|_": 167,
373
- "PART|\u81ea\u52d5\u8a5e|_": 168,
374
- "PROPN|\u56fa\u6709\u540d\u8a5e|_": 169,
375
- "PUNCT|\u8a18\u53f7|_": 170,
376
- "SCONJ|\u4ed6\u52d5\u8a5e|_": 171,
377
- "SCONJ|\u526f\u8a5e|_": 172,
378
- "SCONJ|\u5f8c\u7f6e\u526f\u8a5e|_": 173,
379
- "SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": 174,
380
- "SCONJ|\u63a5\u7d9a\u8a5e|_": 175,
381
- "SCONJ|\u683c\u52a9\u8a5e|_": 176,
382
- "SCONJ|\u7d42\u52a9\u8a5e|_": 177,
383
- "SYM": 178,
384
- "SYM|_|_": 179,
385
- "VERB|\u4ed6\u52d5\u8a5e|_": 180,
386
- "VERB|\u4ed6\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 181,
387
- "VERB|\u4ed6\u52d5\u8a5e|_+PART|\u63a5\u5c3e\u8f9e|_": 182,
388
- "VERB|\u5b8c\u5168\u52d5\u8a5e|_": 183,
389
- "VERB|\u81ea\u52d5\u8a5e|_": 184,
390
- "VERB|\u81ea\u52d5\u8a5e|_+AUX|\u52a9\u52d5\u8a5e|_": 185,
391
- "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u540d\u8a5e|_": 186,
392
- "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 187,
393
- "VERB|\u81ea\u52d5\u8a5e|_+VERB|\u81ea\u52d5\u8a5e|_": 188
394
  },
395
  "layer_norm_eps": 1e-07,
396
  "max_position_embeddings": 512,
@@ -643,12 +641,6 @@
643
  "be"
644
  ]
645
  },
646
- "NOUN|\u540d\u8a5e|_+PART|\u4eba\u79f0\u63a5\u8f9e|_": {
647
- "Hura": [
648
- "Hur",
649
- "a"
650
- ]
651
- },
652
  "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_": {
653
  "Shiriki": [
654
  "Shiri",
@@ -809,10 +801,6 @@
809
  "ambe": [
810
  "am",
811
  "be"
812
- ],
813
- "anpe": [
814
- "an",
815
- "pe"
816
  ]
817
  },
818
  "VERB|\u81ea\u52d5\u8a5e|_+PART|\u4eba\u79f0\u63a5\u8f9e|_": {
@@ -823,10 +811,6 @@
823
  "sapash": [
824
  "sap",
825
  "ash"
826
- ],
827
- "shinotash": [
828
- "shinot",
829
- "ash"
830
  ]
831
  },
832
  "VERB|\u81ea\u52d5\u8a5e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": {
@@ -851,5 +835,5 @@
851
  "torch_dtype": "float32",
852
  "transformers_version": "4.25.1",
853
  "type_vocab_size": 0,
854
- "vocab_size": 6143
855
  }
 
164
  "152": "NOUN|\u540d\u8a5e|_",
165
  "153": "NOUN|\u540d\u8a5e|_+ADP|\u683c\u52a9\u8a5e|_",
166
  "154": "NOUN|\u540d\u8a5e|_+NOUN|\u4f4d\u7f6e\u540d\u8a5e|_",
167
+ "155": "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_",
168
+ "156": "NOUN|\u56fa\u6709\u540d\u8a5e|_",
169
+ "157": "NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
170
+ "158": "NUM|\u6570\u8a5e|_",
171
+ "159": "PART|\u4eba\u79f0\u63a5\u8f9e|_",
172
+ "160": "PART|\u4eba\u79f0\u63a5\u8f9e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
173
+ "161": "PART|\u4eba\u79f0\u63a5\u8f9e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_",
174
+ "162": "PART|\u4eba\u79f0\u63a5\u8f9e|_+VERB|\u4ed6\u52d5\u8a5e|_",
175
+ "163": "PART|\u63a5\u5c3e\u8f9e|_",
176
+ "164": "PART|\u63a5\u7d9a\u52a9\u8a5e|_",
177
+ "165": "PART|\u63a5\u982d\u8f9e|_",
178
+ "166": "PART|\u7d42\u52a9\u8a5e|_",
179
+ "167": "PART|\u81ea\u52d5\u8a5e|_",
180
+ "168": "PROPN|\u56fa\u6709\u540d\u8a5e|_",
181
+ "169": "PUNCT|\u8a18\u53f7|_",
182
+ "170": "SCONJ|\u4ed6\u52d5\u8a5e|_",
183
+ "171": "SCONJ|\u526f\u8a5e|_",
184
+ "172": "SCONJ|\u5f8c\u7f6e\u526f\u8a5e|_",
185
+ "173": "SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_",
186
+ "174": "SCONJ|\u63a5\u7d9a\u8a5e|_",
187
+ "175": "SCONJ|\u683c\u52a9\u8a5e|_",
188
+ "176": "SCONJ|\u7d42\u52a9\u8a5e|_",
189
+ "177": "SYM",
190
+ "178": "SYM|_|_",
191
+ "179": "VERB|\u4ed6\u52d5\u8a5e|_",
192
+ "180": "VERB|\u4ed6\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
193
+ "181": "VERB|\u4ed6\u52d5\u8a5e|_+PART|\u63a5\u5c3e\u8f9e|_",
194
+ "182": "VERB|\u5b8c\u5168\u52d5\u8a5e|_",
195
+ "183": "VERB|\u81ea\u52d5\u8a5e|_",
196
+ "184": "VERB|\u81ea\u52d5\u8a5e|_+AUX|\u52a9\u52d5\u8a5e|_",
197
+ "185": "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u540d\u8a5e|_",
198
+ "186": "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_",
199
+ "187": "VERB|\u81ea\u52d5\u8a5e|_+VERB|\u81ea\u52d5\u8a5e|_"
 
200
  },
201
  "initializer_range": 0.02,
202
  "intermediate_size": 3072,
 
356
  "NOUN|\u540d\u8a5e|_": 152,
357
  "NOUN|\u540d\u8a5e|_+ADP|\u683c\u52a9\u8a5e|_": 153,
358
  "NOUN|\u540d\u8a5e|_+NOUN|\u4f4d\u7f6e\u540d\u8a5e|_": 154,
359
+ "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_": 155,
360
+ "NOUN|\u56fa\u6709\u540d\u8a5e|_": 156,
361
+ "NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 157,
362
+ "NUM|\u6570\u8a5e|_": 158,
363
+ "PART|\u4eba\u79f0\u63a5\u8f9e|_": 159,
364
+ "PART|\u4eba\u79f0\u63a5\u8f9e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 160,
365
+ "PART|\u4eba\u79f0\u63a5\u8f9e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": 161,
366
+ "PART|\u4eba\u79f0\u63a5\u8f9e|_+VERB|\u4ed6\u52d5\u8a5e|_": 162,
367
+ "PART|\u63a5\u5c3e\u8f9e|_": 163,
368
+ "PART|\u63a5\u7d9a\u52a9\u8a5e|_": 164,
369
+ "PART|\u63a5\u982d\u8f9e|_": 165,
370
+ "PART|\u7d42\u52a9\u8a5e|_": 166,
371
+ "PART|\u81ea\u52d5\u8a5e|_": 167,
372
+ "PROPN|\u56fa\u6709\u540d\u8a5e|_": 168,
373
+ "PUNCT|\u8a18\u53f7|_": 169,
374
+ "SCONJ|\u4ed6\u52d5\u8a5e|_": 170,
375
+ "SCONJ|\u526f\u8a5e|_": 171,
376
+ "SCONJ|\u5f8c\u7f6e\u526f\u8a5e|_": 172,
377
+ "SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": 173,
378
+ "SCONJ|\u63a5\u7d9a\u8a5e|_": 174,
379
+ "SCONJ|\u683c\u52a9\u8a5e|_": 175,
380
+ "SCONJ|\u7d42\u52a9\u8a5e|_": 176,
381
+ "SYM": 177,
382
+ "SYM|_|_": 178,
383
+ "VERB|\u4ed6\u52d5\u8a5e|_": 179,
384
+ "VERB|\u4ed6\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 180,
385
+ "VERB|\u4ed6\u52d5\u8a5e|_+PART|\u63a5\u5c3e\u8f9e|_": 181,
386
+ "VERB|\u5b8c\u5168\u52d5\u8a5e|_": 182,
387
+ "VERB|\u81ea\u52d5\u8a5e|_": 183,
388
+ "VERB|\u81ea\u52d5\u8a5e|_+AUX|\u52a9\u52d5\u8a5e|_": 184,
389
+ "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u540d\u8a5e|_": 185,
390
+ "VERB|\u81ea\u52d5\u8a5e|_+NOUN|\u5f62\u5f0f\u540d\u8a5e|_": 186,
391
+ "VERB|\u81ea\u52d5\u8a5e|_+VERB|\u81ea\u52d5\u8a5e|_": 187
 
392
  },
393
  "layer_norm_eps": 1e-07,
394
  "max_position_embeddings": 512,
 
641
  "be"
642
  ]
643
  },
 
 
 
 
 
 
644
  "NOUN|\u540d\u8a5e|_+VERB|\u4ed6\u52d5\u8a5e|_": {
645
  "Shiriki": [
646
  "Shiri",
 
801
  "ambe": [
802
  "am",
803
  "be"
 
 
 
 
804
  ]
805
  },
806
  "VERB|\u81ea\u52d5\u8a5e|_+PART|\u4eba\u79f0\u63a5\u8f9e|_": {
 
811
  "sapash": [
812
  "sap",
813
  "ash"
 
 
 
 
814
  ]
815
  },
816
  "VERB|\u81ea\u52d5\u8a5e|_+SCONJ|\u63a5\u7d9a\u52a9\u8a5e|_": {
 
835
  "torch_dtype": "float32",
836
  "transformers_version": "4.25.1",
837
  "type_vocab_size": 0,
838
+ "vocab_size": 6255
839
  }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0b4c5885682bfe051b418876efa5c49806188eb904f7e1224a118bd6261584e8
3
- size 419610764
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bcc050993150c056a0ba07648e5995f1520e68c31149b9d5c24136787593419
3
+ size 419951756
supar.model CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6b06cf8b5048465adb6844602fa20bb4ca5ea2c1183522f99b5433b1c8c89b76
3
- size 464368260
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d19bd0d4c5f5b54716e99ac55ee472bb2e6f523c0ade3a9052e60c15b2fb6ab
3
+ size 464714397
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff