abdiharyadi
commited on
Commit
•
b550b6f
1
Parent(s):
917d030
Upload folder using huggingface_hub
Browse files- checkpoint-1/model/config.json +33 -0
- checkpoint-1/model/generation_config.json +7 -0
- checkpoint-1/model/model.safetensors +3 -0
- checkpoint-1/tokenizer/added_tokens.json +316 -0
- checkpoint-1/tokenizer/special_tokens_map.json +0 -0
- checkpoint-1/tokenizer/spiece.model +3 -0
- checkpoint-1/tokenizer/tokenizer.json +0 -0
- checkpoint-1/tokenizer/tokenizer_config.json +0 -0
- checkpoint-2/model/config.json +33 -0
- checkpoint-2/model/generation_config.json +7 -0
- checkpoint-2/model/model.safetensors +3 -0
- checkpoint-2/tokenizer/added_tokens.json +316 -0
- checkpoint-2/tokenizer/special_tokens_map.json +0 -0
- checkpoint-2/tokenizer/spiece.model +3 -0
- checkpoint-2/tokenizer/tokenizer.json +0 -0
- checkpoint-2/tokenizer/tokenizer_config.json +0 -0
- loss_data.tsv +3 -0
checkpoint-1/model/config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/kaggle/tmp/taufiq-indo-amr-generation-silver-3-epoch-uncased-continued/checkpoint-4/model",
|
3 |
+
"architectures": [
|
4 |
+
"T5ForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"classifier_dropout": 0.0,
|
7 |
+
"d_ff": 2048,
|
8 |
+
"d_kv": 64,
|
9 |
+
"d_model": 768,
|
10 |
+
"decoder_start_token_id": 0,
|
11 |
+
"dense_act_fn": "gelu_new",
|
12 |
+
"dropout_rate": 0.1,
|
13 |
+
"eos_token_id": 1,
|
14 |
+
"feed_forward_proj": "gated-gelu",
|
15 |
+
"gradient_checkpointing": false,
|
16 |
+
"initializer_factor": 1.0,
|
17 |
+
"is_encoder_decoder": true,
|
18 |
+
"is_gated_act": true,
|
19 |
+
"layer_norm_epsilon": 1e-06,
|
20 |
+
"model_type": "t5",
|
21 |
+
"num_decoder_layers": 12,
|
22 |
+
"num_heads": 12,
|
23 |
+
"num_layers": 12,
|
24 |
+
"output_past": true,
|
25 |
+
"pad_token_id": 0,
|
26 |
+
"relative_attention_max_distance": 128,
|
27 |
+
"relative_attention_num_buckets": 32,
|
28 |
+
"tie_word_embeddings": false,
|
29 |
+
"torch_dtype": "float32",
|
30 |
+
"transformers_version": "4.45.1",
|
31 |
+
"use_cache": true,
|
32 |
+
"vocab_size": 32414
|
33 |
+
}
|
checkpoint-1/model/generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"decoder_start_token_id": 0,
|
4 |
+
"eos_token_id": 1,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.45.1"
|
7 |
+
}
|
checkpoint-1/model/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6bbd68b20db6937599a20fb6c14e4e1d49a99f1e893e18fd6c23cab356d425b9
|
3 |
+
size 992102248
|
checkpoint-1/tokenizer/added_tokens.json
ADDED
@@ -0,0 +1,316 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"-00": 32193,
|
3 |
+
"-01": 32194,
|
4 |
+
"-02": 32195,
|
5 |
+
"-03": 32196,
|
6 |
+
"-04": 32197,
|
7 |
+
"-05": 32198,
|
8 |
+
"-06": 32199,
|
9 |
+
"-07": 32200,
|
10 |
+
"-08": 32201,
|
11 |
+
"-09": 32202,
|
12 |
+
"-33": 32203,
|
13 |
+
"-36": 32204,
|
14 |
+
"-37": 32205,
|
15 |
+
"-38": 32206,
|
16 |
+
"-39": 32207,
|
17 |
+
"-41": 32208,
|
18 |
+
"-42": 32209,
|
19 |
+
"-43": 32210,
|
20 |
+
"-44": 32211,
|
21 |
+
"-45": 32212,
|
22 |
+
"-46": 32213,
|
23 |
+
"-48": 32214,
|
24 |
+
"-49": 32215,
|
25 |
+
"-51": 32216,
|
26 |
+
"-52": 32217,
|
27 |
+
"-53": 32218,
|
28 |
+
"-54": 32219,
|
29 |
+
"-55": 32220,
|
30 |
+
"-56": 32221,
|
31 |
+
"-57": 32222,
|
32 |
+
"-58": 32223,
|
33 |
+
"-59": 32224,
|
34 |
+
"-61": 32225,
|
35 |
+
"-62": 32226,
|
36 |
+
"-63": 32227,
|
37 |
+
"-64": 32228,
|
38 |
+
"-65": 32229,
|
39 |
+
"-66": 32230,
|
40 |
+
"-67": 32231,
|
41 |
+
"-68": 32232,
|
42 |
+
"-69": 32233,
|
43 |
+
"-70": 32234,
|
44 |
+
"-71": 32235,
|
45 |
+
"-72": 32236,
|
46 |
+
"-73": 32237,
|
47 |
+
"-74": 32238,
|
48 |
+
"-75": 32239,
|
49 |
+
"-76": 32240,
|
50 |
+
"-77": 32241,
|
51 |
+
"-78": 32242,
|
52 |
+
"-79": 32243,
|
53 |
+
"-81": 32244,
|
54 |
+
"-82": 32245,
|
55 |
+
"-83": 32246,
|
56 |
+
"-84": 32247,
|
57 |
+
"-85": 32248,
|
58 |
+
"-86": 32249,
|
59 |
+
"-87": 32250,
|
60 |
+
"-88": 32251,
|
61 |
+
"-89": 32252,
|
62 |
+
"-90": 32253,
|
63 |
+
"-91": 32254,
|
64 |
+
"-92": 32255,
|
65 |
+
"-93": 32256,
|
66 |
+
"-94": 32257,
|
67 |
+
"-95": 32258,
|
68 |
+
"-96": 32259,
|
69 |
+
"-97": 32260,
|
70 |
+
"-98": 32261,
|
71 |
+
"-of": 32262,
|
72 |
+
":accompanier": 32126,
|
73 |
+
":age": 32127,
|
74 |
+
":arg0": 32105,
|
75 |
+
":arg1": 32106,
|
76 |
+
":arg10": 32115,
|
77 |
+
":arg11": 32116,
|
78 |
+
":arg12": 32117,
|
79 |
+
":arg13": 32118,
|
80 |
+
":arg14": 32119,
|
81 |
+
":arg15": 32120,
|
82 |
+
":arg16": 32121,
|
83 |
+
":arg17": 32122,
|
84 |
+
":arg18": 32123,
|
85 |
+
":arg19": 32124,
|
86 |
+
":arg2": 32107,
|
87 |
+
":arg20": 32125,
|
88 |
+
":arg3": 32108,
|
89 |
+
":arg4": 32109,
|
90 |
+
":arg5": 32110,
|
91 |
+
":arg6": 32111,
|
92 |
+
":arg7": 32112,
|
93 |
+
":arg8": 32113,
|
94 |
+
":arg9": 32114,
|
95 |
+
":beneficiary": 32128,
|
96 |
+
":calendar": 32129,
|
97 |
+
":cause": 32130,
|
98 |
+
":century": 32131,
|
99 |
+
":concession": 32132,
|
100 |
+
":condition": 32133,
|
101 |
+
":conj-as-if": 32134,
|
102 |
+
":consist-of": 32135,
|
103 |
+
":cost": 32136,
|
104 |
+
":day": 32137,
|
105 |
+
":dayperiod": 32138,
|
106 |
+
":decade": 32139,
|
107 |
+
":degree": 32140,
|
108 |
+
":destination": 32141,
|
109 |
+
":direction": 32142,
|
110 |
+
":domain": 32143,
|
111 |
+
":duration": 32144,
|
112 |
+
":employed-by": 32145,
|
113 |
+
":era": 32146,
|
114 |
+
":example": 32147,
|
115 |
+
":extent": 32148,
|
116 |
+
":frequency": 32149,
|
117 |
+
":instrument": 32150,
|
118 |
+
":li": 32151,
|
119 |
+
":location": 32152,
|
120 |
+
":manner": 32153,
|
121 |
+
":meaning": 32154,
|
122 |
+
":medium": 32155,
|
123 |
+
":mod": 32156,
|
124 |
+
":mode": 32157,
|
125 |
+
":month": 32158,
|
126 |
+
":name": 32159,
|
127 |
+
":op1": 32100,
|
128 |
+
":op2": 32101,
|
129 |
+
":op3": 32102,
|
130 |
+
":op4": 32103,
|
131 |
+
":op5": 32104,
|
132 |
+
":ord": 32160,
|
133 |
+
":part": 32161,
|
134 |
+
":path": 32162,
|
135 |
+
":polarity": 32163,
|
136 |
+
":polite": 32164,
|
137 |
+
":poss": 32165,
|
138 |
+
":purpose": 32166,
|
139 |
+
":quant": 32167,
|
140 |
+
":quarter": 32168,
|
141 |
+
":range": 32169,
|
142 |
+
":relation": 32170,
|
143 |
+
":role": 32171,
|
144 |
+
":scale": 32172,
|
145 |
+
":season": 32173,
|
146 |
+
":snt0": 32187,
|
147 |
+
":snt1": 32188,
|
148 |
+
":snt2": 32189,
|
149 |
+
":snt3": 32190,
|
150 |
+
":snt4": 32191,
|
151 |
+
":snt5": 32192,
|
152 |
+
":source": 32174,
|
153 |
+
":subevent": 32175,
|
154 |
+
":subset": 32176,
|
155 |
+
":superset": 32177,
|
156 |
+
":time": 32178,
|
157 |
+
":timezone": 32179,
|
158 |
+
":topic": 32180,
|
159 |
+
":unit": 32181,
|
160 |
+
":value": 32182,
|
161 |
+
":weekday": 32183,
|
162 |
+
":wiki": 32184,
|
163 |
+
":year": 32185,
|
164 |
+
":year2": 32186,
|
165 |
+
"<pointer:0>": 32263,
|
166 |
+
"<pointer:100>": 32363,
|
167 |
+
"<pointer:101>": 32364,
|
168 |
+
"<pointer:102>": 32365,
|
169 |
+
"<pointer:103>": 32366,
|
170 |
+
"<pointer:104>": 32367,
|
171 |
+
"<pointer:105>": 32368,
|
172 |
+
"<pointer:106>": 32369,
|
173 |
+
"<pointer:107>": 32370,
|
174 |
+
"<pointer:108>": 32371,
|
175 |
+
"<pointer:109>": 32372,
|
176 |
+
"<pointer:10>": 32273,
|
177 |
+
"<pointer:110>": 32373,
|
178 |
+
"<pointer:111>": 32374,
|
179 |
+
"<pointer:112>": 32375,
|
180 |
+
"<pointer:113>": 32376,
|
181 |
+
"<pointer:114>": 32377,
|
182 |
+
"<pointer:115>": 32378,
|
183 |
+
"<pointer:116>": 32379,
|
184 |
+
"<pointer:117>": 32380,
|
185 |
+
"<pointer:118>": 32381,
|
186 |
+
"<pointer:119>": 32382,
|
187 |
+
"<pointer:11>": 32274,
|
188 |
+
"<pointer:120>": 32383,
|
189 |
+
"<pointer:121>": 32384,
|
190 |
+
"<pointer:122>": 32385,
|
191 |
+
"<pointer:123>": 32386,
|
192 |
+
"<pointer:124>": 32387,
|
193 |
+
"<pointer:125>": 32388,
|
194 |
+
"<pointer:126>": 32389,
|
195 |
+
"<pointer:127>": 32390,
|
196 |
+
"<pointer:128>": 32391,
|
197 |
+
"<pointer:129>": 32392,
|
198 |
+
"<pointer:12>": 32275,
|
199 |
+
"<pointer:130>": 32393,
|
200 |
+
"<pointer:131>": 32394,
|
201 |
+
"<pointer:132>": 32395,
|
202 |
+
"<pointer:133>": 32396,
|
203 |
+
"<pointer:134>": 32397,
|
204 |
+
"<pointer:135>": 32398,
|
205 |
+
"<pointer:136>": 32399,
|
206 |
+
"<pointer:137>": 32400,
|
207 |
+
"<pointer:138>": 32401,
|
208 |
+
"<pointer:139>": 32402,
|
209 |
+
"<pointer:13>": 32276,
|
210 |
+
"<pointer:140>": 32403,
|
211 |
+
"<pointer:141>": 32404,
|
212 |
+
"<pointer:142>": 32405,
|
213 |
+
"<pointer:143>": 32406,
|
214 |
+
"<pointer:144>": 32407,
|
215 |
+
"<pointer:145>": 32408,
|
216 |
+
"<pointer:146>": 32409,
|
217 |
+
"<pointer:147>": 32410,
|
218 |
+
"<pointer:148>": 32411,
|
219 |
+
"<pointer:149>": 32412,
|
220 |
+
"<pointer:14>": 32277,
|
221 |
+
"<pointer:150>": 32413,
|
222 |
+
"<pointer:15>": 32278,
|
223 |
+
"<pointer:16>": 32279,
|
224 |
+
"<pointer:17>": 32280,
|
225 |
+
"<pointer:18>": 32281,
|
226 |
+
"<pointer:19>": 32282,
|
227 |
+
"<pointer:1>": 32264,
|
228 |
+
"<pointer:20>": 32283,
|
229 |
+
"<pointer:21>": 32284,
|
230 |
+
"<pointer:22>": 32285,
|
231 |
+
"<pointer:23>": 32286,
|
232 |
+
"<pointer:24>": 32287,
|
233 |
+
"<pointer:25>": 32288,
|
234 |
+
"<pointer:26>": 32289,
|
235 |
+
"<pointer:27>": 32290,
|
236 |
+
"<pointer:28>": 32291,
|
237 |
+
"<pointer:29>": 32292,
|
238 |
+
"<pointer:2>": 32265,
|
239 |
+
"<pointer:30>": 32293,
|
240 |
+
"<pointer:31>": 32294,
|
241 |
+
"<pointer:32>": 32295,
|
242 |
+
"<pointer:33>": 32296,
|
243 |
+
"<pointer:34>": 32297,
|
244 |
+
"<pointer:35>": 32298,
|
245 |
+
"<pointer:36>": 32299,
|
246 |
+
"<pointer:37>": 32300,
|
247 |
+
"<pointer:38>": 32301,
|
248 |
+
"<pointer:39>": 32302,
|
249 |
+
"<pointer:3>": 32266,
|
250 |
+
"<pointer:40>": 32303,
|
251 |
+
"<pointer:41>": 32304,
|
252 |
+
"<pointer:42>": 32305,
|
253 |
+
"<pointer:43>": 32306,
|
254 |
+
"<pointer:44>": 32307,
|
255 |
+
"<pointer:45>": 32308,
|
256 |
+
"<pointer:46>": 32309,
|
257 |
+
"<pointer:47>": 32310,
|
258 |
+
"<pointer:48>": 32311,
|
259 |
+
"<pointer:49>": 32312,
|
260 |
+
"<pointer:4>": 32267,
|
261 |
+
"<pointer:50>": 32313,
|
262 |
+
"<pointer:51>": 32314,
|
263 |
+
"<pointer:52>": 32315,
|
264 |
+
"<pointer:53>": 32316,
|
265 |
+
"<pointer:54>": 32317,
|
266 |
+
"<pointer:55>": 32318,
|
267 |
+
"<pointer:56>": 32319,
|
268 |
+
"<pointer:57>": 32320,
|
269 |
+
"<pointer:58>": 32321,
|
270 |
+
"<pointer:59>": 32322,
|
271 |
+
"<pointer:5>": 32268,
|
272 |
+
"<pointer:60>": 32323,
|
273 |
+
"<pointer:61>": 32324,
|
274 |
+
"<pointer:62>": 32325,
|
275 |
+
"<pointer:63>": 32326,
|
276 |
+
"<pointer:64>": 32327,
|
277 |
+
"<pointer:65>": 32328,
|
278 |
+
"<pointer:66>": 32329,
|
279 |
+
"<pointer:67>": 32330,
|
280 |
+
"<pointer:68>": 32331,
|
281 |
+
"<pointer:69>": 32332,
|
282 |
+
"<pointer:6>": 32269,
|
283 |
+
"<pointer:70>": 32333,
|
284 |
+
"<pointer:71>": 32334,
|
285 |
+
"<pointer:72>": 32335,
|
286 |
+
"<pointer:73>": 32336,
|
287 |
+
"<pointer:74>": 32337,
|
288 |
+
"<pointer:75>": 32338,
|
289 |
+
"<pointer:76>": 32339,
|
290 |
+
"<pointer:77>": 32340,
|
291 |
+
"<pointer:78>": 32341,
|
292 |
+
"<pointer:79>": 32342,
|
293 |
+
"<pointer:7>": 32270,
|
294 |
+
"<pointer:80>": 32343,
|
295 |
+
"<pointer:81>": 32344,
|
296 |
+
"<pointer:82>": 32345,
|
297 |
+
"<pointer:83>": 32346,
|
298 |
+
"<pointer:84>": 32347,
|
299 |
+
"<pointer:85>": 32348,
|
300 |
+
"<pointer:86>": 32349,
|
301 |
+
"<pointer:87>": 32350,
|
302 |
+
"<pointer:88>": 32351,
|
303 |
+
"<pointer:89>": 32352,
|
304 |
+
"<pointer:8>": 32271,
|
305 |
+
"<pointer:90>": 32353,
|
306 |
+
"<pointer:91>": 32354,
|
307 |
+
"<pointer:92>": 32355,
|
308 |
+
"<pointer:93>": 32356,
|
309 |
+
"<pointer:94>": 32357,
|
310 |
+
"<pointer:95>": 32358,
|
311 |
+
"<pointer:96>": 32359,
|
312 |
+
"<pointer:97>": 32360,
|
313 |
+
"<pointer:98>": 32361,
|
314 |
+
"<pointer:99>": 32362,
|
315 |
+
"<pointer:9>": 32272
|
316 |
+
}
|
checkpoint-1/tokenizer/special_tokens_map.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-1/tokenizer/spiece.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70d33eab49f262358f962dbf38433dec85c44b71cb05f4b0e23f439c45209218
|
3 |
+
size 776904
|
checkpoint-1/tokenizer/tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-1/tokenizer/tokenizer_config.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-2/model/config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/kaggle/tmp/taufiq-indo-amr-generation-silver-3-epoch-uncased-continued/checkpoint-4/model",
|
3 |
+
"architectures": [
|
4 |
+
"T5ForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"classifier_dropout": 0.0,
|
7 |
+
"d_ff": 2048,
|
8 |
+
"d_kv": 64,
|
9 |
+
"d_model": 768,
|
10 |
+
"decoder_start_token_id": 0,
|
11 |
+
"dense_act_fn": "gelu_new",
|
12 |
+
"dropout_rate": 0.1,
|
13 |
+
"eos_token_id": 1,
|
14 |
+
"feed_forward_proj": "gated-gelu",
|
15 |
+
"gradient_checkpointing": false,
|
16 |
+
"initializer_factor": 1.0,
|
17 |
+
"is_encoder_decoder": true,
|
18 |
+
"is_gated_act": true,
|
19 |
+
"layer_norm_epsilon": 1e-06,
|
20 |
+
"model_type": "t5",
|
21 |
+
"num_decoder_layers": 12,
|
22 |
+
"num_heads": 12,
|
23 |
+
"num_layers": 12,
|
24 |
+
"output_past": true,
|
25 |
+
"pad_token_id": 0,
|
26 |
+
"relative_attention_max_distance": 128,
|
27 |
+
"relative_attention_num_buckets": 32,
|
28 |
+
"tie_word_embeddings": false,
|
29 |
+
"torch_dtype": "float32",
|
30 |
+
"transformers_version": "4.45.1",
|
31 |
+
"use_cache": true,
|
32 |
+
"vocab_size": 32414
|
33 |
+
}
|
checkpoint-2/model/generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"decoder_start_token_id": 0,
|
4 |
+
"eos_token_id": 1,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.45.1"
|
7 |
+
}
|
checkpoint-2/model/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6fe32eed5ac66b9c2a27b1bccd44e0c5fce232cd1f619d3cfe21a599740abc35
|
3 |
+
size 992102248
|
checkpoint-2/tokenizer/added_tokens.json
ADDED
@@ -0,0 +1,316 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"-00": 32193,
|
3 |
+
"-01": 32194,
|
4 |
+
"-02": 32195,
|
5 |
+
"-03": 32196,
|
6 |
+
"-04": 32197,
|
7 |
+
"-05": 32198,
|
8 |
+
"-06": 32199,
|
9 |
+
"-07": 32200,
|
10 |
+
"-08": 32201,
|
11 |
+
"-09": 32202,
|
12 |
+
"-33": 32203,
|
13 |
+
"-36": 32204,
|
14 |
+
"-37": 32205,
|
15 |
+
"-38": 32206,
|
16 |
+
"-39": 32207,
|
17 |
+
"-41": 32208,
|
18 |
+
"-42": 32209,
|
19 |
+
"-43": 32210,
|
20 |
+
"-44": 32211,
|
21 |
+
"-45": 32212,
|
22 |
+
"-46": 32213,
|
23 |
+
"-48": 32214,
|
24 |
+
"-49": 32215,
|
25 |
+
"-51": 32216,
|
26 |
+
"-52": 32217,
|
27 |
+
"-53": 32218,
|
28 |
+
"-54": 32219,
|
29 |
+
"-55": 32220,
|
30 |
+
"-56": 32221,
|
31 |
+
"-57": 32222,
|
32 |
+
"-58": 32223,
|
33 |
+
"-59": 32224,
|
34 |
+
"-61": 32225,
|
35 |
+
"-62": 32226,
|
36 |
+
"-63": 32227,
|
37 |
+
"-64": 32228,
|
38 |
+
"-65": 32229,
|
39 |
+
"-66": 32230,
|
40 |
+
"-67": 32231,
|
41 |
+
"-68": 32232,
|
42 |
+
"-69": 32233,
|
43 |
+
"-70": 32234,
|
44 |
+
"-71": 32235,
|
45 |
+
"-72": 32236,
|
46 |
+
"-73": 32237,
|
47 |
+
"-74": 32238,
|
48 |
+
"-75": 32239,
|
49 |
+
"-76": 32240,
|
50 |
+
"-77": 32241,
|
51 |
+
"-78": 32242,
|
52 |
+
"-79": 32243,
|
53 |
+
"-81": 32244,
|
54 |
+
"-82": 32245,
|
55 |
+
"-83": 32246,
|
56 |
+
"-84": 32247,
|
57 |
+
"-85": 32248,
|
58 |
+
"-86": 32249,
|
59 |
+
"-87": 32250,
|
60 |
+
"-88": 32251,
|
61 |
+
"-89": 32252,
|
62 |
+
"-90": 32253,
|
63 |
+
"-91": 32254,
|
64 |
+
"-92": 32255,
|
65 |
+
"-93": 32256,
|
66 |
+
"-94": 32257,
|
67 |
+
"-95": 32258,
|
68 |
+
"-96": 32259,
|
69 |
+
"-97": 32260,
|
70 |
+
"-98": 32261,
|
71 |
+
"-of": 32262,
|
72 |
+
":accompanier": 32126,
|
73 |
+
":age": 32127,
|
74 |
+
":arg0": 32105,
|
75 |
+
":arg1": 32106,
|
76 |
+
":arg10": 32115,
|
77 |
+
":arg11": 32116,
|
78 |
+
":arg12": 32117,
|
79 |
+
":arg13": 32118,
|
80 |
+
":arg14": 32119,
|
81 |
+
":arg15": 32120,
|
82 |
+
":arg16": 32121,
|
83 |
+
":arg17": 32122,
|
84 |
+
":arg18": 32123,
|
85 |
+
":arg19": 32124,
|
86 |
+
":arg2": 32107,
|
87 |
+
":arg20": 32125,
|
88 |
+
":arg3": 32108,
|
89 |
+
":arg4": 32109,
|
90 |
+
":arg5": 32110,
|
91 |
+
":arg6": 32111,
|
92 |
+
":arg7": 32112,
|
93 |
+
":arg8": 32113,
|
94 |
+
":arg9": 32114,
|
95 |
+
":beneficiary": 32128,
|
96 |
+
":calendar": 32129,
|
97 |
+
":cause": 32130,
|
98 |
+
":century": 32131,
|
99 |
+
":concession": 32132,
|
100 |
+
":condition": 32133,
|
101 |
+
":conj-as-if": 32134,
|
102 |
+
":consist-of": 32135,
|
103 |
+
":cost": 32136,
|
104 |
+
":day": 32137,
|
105 |
+
":dayperiod": 32138,
|
106 |
+
":decade": 32139,
|
107 |
+
":degree": 32140,
|
108 |
+
":destination": 32141,
|
109 |
+
":direction": 32142,
|
110 |
+
":domain": 32143,
|
111 |
+
":duration": 32144,
|
112 |
+
":employed-by": 32145,
|
113 |
+
":era": 32146,
|
114 |
+
":example": 32147,
|
115 |
+
":extent": 32148,
|
116 |
+
":frequency": 32149,
|
117 |
+
":instrument": 32150,
|
118 |
+
":li": 32151,
|
119 |
+
":location": 32152,
|
120 |
+
":manner": 32153,
|
121 |
+
":meaning": 32154,
|
122 |
+
":medium": 32155,
|
123 |
+
":mod": 32156,
|
124 |
+
":mode": 32157,
|
125 |
+
":month": 32158,
|
126 |
+
":name": 32159,
|
127 |
+
":op1": 32100,
|
128 |
+
":op2": 32101,
|
129 |
+
":op3": 32102,
|
130 |
+
":op4": 32103,
|
131 |
+
":op5": 32104,
|
132 |
+
":ord": 32160,
|
133 |
+
":part": 32161,
|
134 |
+
":path": 32162,
|
135 |
+
":polarity": 32163,
|
136 |
+
":polite": 32164,
|
137 |
+
":poss": 32165,
|
138 |
+
":purpose": 32166,
|
139 |
+
":quant": 32167,
|
140 |
+
":quarter": 32168,
|
141 |
+
":range": 32169,
|
142 |
+
":relation": 32170,
|
143 |
+
":role": 32171,
|
144 |
+
":scale": 32172,
|
145 |
+
":season": 32173,
|
146 |
+
":snt0": 32187,
|
147 |
+
":snt1": 32188,
|
148 |
+
":snt2": 32189,
|
149 |
+
":snt3": 32190,
|
150 |
+
":snt4": 32191,
|
151 |
+
":snt5": 32192,
|
152 |
+
":source": 32174,
|
153 |
+
":subevent": 32175,
|
154 |
+
":subset": 32176,
|
155 |
+
":superset": 32177,
|
156 |
+
":time": 32178,
|
157 |
+
":timezone": 32179,
|
158 |
+
":topic": 32180,
|
159 |
+
":unit": 32181,
|
160 |
+
":value": 32182,
|
161 |
+
":weekday": 32183,
|
162 |
+
":wiki": 32184,
|
163 |
+
":year": 32185,
|
164 |
+
":year2": 32186,
|
165 |
+
"<pointer:0>": 32263,
|
166 |
+
"<pointer:100>": 32363,
|
167 |
+
"<pointer:101>": 32364,
|
168 |
+
"<pointer:102>": 32365,
|
169 |
+
"<pointer:103>": 32366,
|
170 |
+
"<pointer:104>": 32367,
|
171 |
+
"<pointer:105>": 32368,
|
172 |
+
"<pointer:106>": 32369,
|
173 |
+
"<pointer:107>": 32370,
|
174 |
+
"<pointer:108>": 32371,
|
175 |
+
"<pointer:109>": 32372,
|
176 |
+
"<pointer:10>": 32273,
|
177 |
+
"<pointer:110>": 32373,
|
178 |
+
"<pointer:111>": 32374,
|
179 |
+
"<pointer:112>": 32375,
|
180 |
+
"<pointer:113>": 32376,
|
181 |
+
"<pointer:114>": 32377,
|
182 |
+
"<pointer:115>": 32378,
|
183 |
+
"<pointer:116>": 32379,
|
184 |
+
"<pointer:117>": 32380,
|
185 |
+
"<pointer:118>": 32381,
|
186 |
+
"<pointer:119>": 32382,
|
187 |
+
"<pointer:11>": 32274,
|
188 |
+
"<pointer:120>": 32383,
|
189 |
+
"<pointer:121>": 32384,
|
190 |
+
"<pointer:122>": 32385,
|
191 |
+
"<pointer:123>": 32386,
|
192 |
+
"<pointer:124>": 32387,
|
193 |
+
"<pointer:125>": 32388,
|
194 |
+
"<pointer:126>": 32389,
|
195 |
+
"<pointer:127>": 32390,
|
196 |
+
"<pointer:128>": 32391,
|
197 |
+
"<pointer:129>": 32392,
|
198 |
+
"<pointer:12>": 32275,
|
199 |
+
"<pointer:130>": 32393,
|
200 |
+
"<pointer:131>": 32394,
|
201 |
+
"<pointer:132>": 32395,
|
202 |
+
"<pointer:133>": 32396,
|
203 |
+
"<pointer:134>": 32397,
|
204 |
+
"<pointer:135>": 32398,
|
205 |
+
"<pointer:136>": 32399,
|
206 |
+
"<pointer:137>": 32400,
|
207 |
+
"<pointer:138>": 32401,
|
208 |
+
"<pointer:139>": 32402,
|
209 |
+
"<pointer:13>": 32276,
|
210 |
+
"<pointer:140>": 32403,
|
211 |
+
"<pointer:141>": 32404,
|
212 |
+
"<pointer:142>": 32405,
|
213 |
+
"<pointer:143>": 32406,
|
214 |
+
"<pointer:144>": 32407,
|
215 |
+
"<pointer:145>": 32408,
|
216 |
+
"<pointer:146>": 32409,
|
217 |
+
"<pointer:147>": 32410,
|
218 |
+
"<pointer:148>": 32411,
|
219 |
+
"<pointer:149>": 32412,
|
220 |
+
"<pointer:14>": 32277,
|
221 |
+
"<pointer:150>": 32413,
|
222 |
+
"<pointer:15>": 32278,
|
223 |
+
"<pointer:16>": 32279,
|
224 |
+
"<pointer:17>": 32280,
|
225 |
+
"<pointer:18>": 32281,
|
226 |
+
"<pointer:19>": 32282,
|
227 |
+
"<pointer:1>": 32264,
|
228 |
+
"<pointer:20>": 32283,
|
229 |
+
"<pointer:21>": 32284,
|
230 |
+
"<pointer:22>": 32285,
|
231 |
+
"<pointer:23>": 32286,
|
232 |
+
"<pointer:24>": 32287,
|
233 |
+
"<pointer:25>": 32288,
|
234 |
+
"<pointer:26>": 32289,
|
235 |
+
"<pointer:27>": 32290,
|
236 |
+
"<pointer:28>": 32291,
|
237 |
+
"<pointer:29>": 32292,
|
238 |
+
"<pointer:2>": 32265,
|
239 |
+
"<pointer:30>": 32293,
|
240 |
+
"<pointer:31>": 32294,
|
241 |
+
"<pointer:32>": 32295,
|
242 |
+
"<pointer:33>": 32296,
|
243 |
+
"<pointer:34>": 32297,
|
244 |
+
"<pointer:35>": 32298,
|
245 |
+
"<pointer:36>": 32299,
|
246 |
+
"<pointer:37>": 32300,
|
247 |
+
"<pointer:38>": 32301,
|
248 |
+
"<pointer:39>": 32302,
|
249 |
+
"<pointer:3>": 32266,
|
250 |
+
"<pointer:40>": 32303,
|
251 |
+
"<pointer:41>": 32304,
|
252 |
+
"<pointer:42>": 32305,
|
253 |
+
"<pointer:43>": 32306,
|
254 |
+
"<pointer:44>": 32307,
|
255 |
+
"<pointer:45>": 32308,
|
256 |
+
"<pointer:46>": 32309,
|
257 |
+
"<pointer:47>": 32310,
|
258 |
+
"<pointer:48>": 32311,
|
259 |
+
"<pointer:49>": 32312,
|
260 |
+
"<pointer:4>": 32267,
|
261 |
+
"<pointer:50>": 32313,
|
262 |
+
"<pointer:51>": 32314,
|
263 |
+
"<pointer:52>": 32315,
|
264 |
+
"<pointer:53>": 32316,
|
265 |
+
"<pointer:54>": 32317,
|
266 |
+
"<pointer:55>": 32318,
|
267 |
+
"<pointer:56>": 32319,
|
268 |
+
"<pointer:57>": 32320,
|
269 |
+
"<pointer:58>": 32321,
|
270 |
+
"<pointer:59>": 32322,
|
271 |
+
"<pointer:5>": 32268,
|
272 |
+
"<pointer:60>": 32323,
|
273 |
+
"<pointer:61>": 32324,
|
274 |
+
"<pointer:62>": 32325,
|
275 |
+
"<pointer:63>": 32326,
|
276 |
+
"<pointer:64>": 32327,
|
277 |
+
"<pointer:65>": 32328,
|
278 |
+
"<pointer:66>": 32329,
|
279 |
+
"<pointer:67>": 32330,
|
280 |
+
"<pointer:68>": 32331,
|
281 |
+
"<pointer:69>": 32332,
|
282 |
+
"<pointer:6>": 32269,
|
283 |
+
"<pointer:70>": 32333,
|
284 |
+
"<pointer:71>": 32334,
|
285 |
+
"<pointer:72>": 32335,
|
286 |
+
"<pointer:73>": 32336,
|
287 |
+
"<pointer:74>": 32337,
|
288 |
+
"<pointer:75>": 32338,
|
289 |
+
"<pointer:76>": 32339,
|
290 |
+
"<pointer:77>": 32340,
|
291 |
+
"<pointer:78>": 32341,
|
292 |
+
"<pointer:79>": 32342,
|
293 |
+
"<pointer:7>": 32270,
|
294 |
+
"<pointer:80>": 32343,
|
295 |
+
"<pointer:81>": 32344,
|
296 |
+
"<pointer:82>": 32345,
|
297 |
+
"<pointer:83>": 32346,
|
298 |
+
"<pointer:84>": 32347,
|
299 |
+
"<pointer:85>": 32348,
|
300 |
+
"<pointer:86>": 32349,
|
301 |
+
"<pointer:87>": 32350,
|
302 |
+
"<pointer:88>": 32351,
|
303 |
+
"<pointer:89>": 32352,
|
304 |
+
"<pointer:8>": 32271,
|
305 |
+
"<pointer:90>": 32353,
|
306 |
+
"<pointer:91>": 32354,
|
307 |
+
"<pointer:92>": 32355,
|
308 |
+
"<pointer:93>": 32356,
|
309 |
+
"<pointer:94>": 32357,
|
310 |
+
"<pointer:95>": 32358,
|
311 |
+
"<pointer:96>": 32359,
|
312 |
+
"<pointer:97>": 32360,
|
313 |
+
"<pointer:98>": 32361,
|
314 |
+
"<pointer:99>": 32362,
|
315 |
+
"<pointer:9>": 32272
|
316 |
+
}
|
checkpoint-2/tokenizer/special_tokens_map.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-2/tokenizer/spiece.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70d33eab49f262358f962dbf38433dec85c44b71cb05f4b0e23f439c45209218
|
3 |
+
size 776904
|
checkpoint-2/tokenizer/tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-2/tokenizer/tokenizer_config.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
loss_data.tsv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
train_loss val_loss
|
2 |
+
0.8372968805502741 1.2302920717374355
|
3 |
+
0.7675029515004483 1.2640078090943758
|