Initial commit
Browse files- .gitattributes +1 -0
- README.md +487 -0
- benchmark_results.txt +56 -0
- benchmark_translations.zip +0 -0
- config.json +41 -0
- generation_config.json +16 -0
- model.safetensors +3 -0
- pytorch_model.bin +3 -0
- source.spm +3 -0
- special_tokens_map.json +1 -0
- target.spm +3 -0
- tokenizer_config.json +1 -0
- vocab.json +0 -0
.gitattributes
CHANGED
@@ -33,3 +33,4 @@ 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 |
+
*.spm filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
@@ -0,0 +1,487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: transformers
|
3 |
+
language:
|
4 |
+
- bru
|
5 |
+
- cmo
|
6 |
+
- de
|
7 |
+
- en
|
8 |
+
- es
|
9 |
+
- fr
|
10 |
+
- hoc
|
11 |
+
- jun
|
12 |
+
- kha
|
13 |
+
- km
|
14 |
+
- kxm
|
15 |
+
- mnw
|
16 |
+
- ngt
|
17 |
+
- pt
|
18 |
+
- sat
|
19 |
+
- vi
|
20 |
+
- wbm
|
21 |
+
|
22 |
+
tags:
|
23 |
+
- translation
|
24 |
+
- opus-mt-tc-bible
|
25 |
+
|
26 |
+
license: apache-2.0
|
27 |
+
model-index:
|
28 |
+
- name: opus-mt-tc-bible-big-deu_eng_fra_por_spa-aav
|
29 |
+
results:
|
30 |
+
- task:
|
31 |
+
name: Translation deu-vie
|
32 |
+
type: translation
|
33 |
+
args: deu-vie
|
34 |
+
dataset:
|
35 |
+
name: flores200-devtest
|
36 |
+
type: flores200-devtest
|
37 |
+
args: deu-vie
|
38 |
+
metrics:
|
39 |
+
- name: BLEU
|
40 |
+
type: bleu
|
41 |
+
value: 34.0
|
42 |
+
- name: chr-F
|
43 |
+
type: chrf
|
44 |
+
value: 0.53671
|
45 |
+
- task:
|
46 |
+
name: Translation eng-vie
|
47 |
+
type: translation
|
48 |
+
args: eng-vie
|
49 |
+
dataset:
|
50 |
+
name: flores200-devtest
|
51 |
+
type: flores200-devtest
|
52 |
+
args: eng-vie
|
53 |
+
metrics:
|
54 |
+
- name: BLEU
|
55 |
+
type: bleu
|
56 |
+
value: 42.4
|
57 |
+
- name: chr-F
|
58 |
+
type: chrf
|
59 |
+
value: 0.59842
|
60 |
+
- task:
|
61 |
+
name: Translation fra-vie
|
62 |
+
type: translation
|
63 |
+
args: fra-vie
|
64 |
+
dataset:
|
65 |
+
name: flores200-devtest
|
66 |
+
type: flores200-devtest
|
67 |
+
args: fra-vie
|
68 |
+
metrics:
|
69 |
+
- name: BLEU
|
70 |
+
type: bleu
|
71 |
+
value: 34.6
|
72 |
+
- name: chr-F
|
73 |
+
type: chrf
|
74 |
+
value: 0.54101
|
75 |
+
- task:
|
76 |
+
name: Translation por-vie
|
77 |
+
type: translation
|
78 |
+
args: por-vie
|
79 |
+
dataset:
|
80 |
+
name: flores200-devtest
|
81 |
+
type: flores200-devtest
|
82 |
+
args: por-vie
|
83 |
+
metrics:
|
84 |
+
- name: BLEU
|
85 |
+
type: bleu
|
86 |
+
value: 36.1
|
87 |
+
- name: chr-F
|
88 |
+
type: chrf
|
89 |
+
value: 0.54970
|
90 |
+
- task:
|
91 |
+
name: Translation spa-vie
|
92 |
+
type: translation
|
93 |
+
args: spa-vie
|
94 |
+
dataset:
|
95 |
+
name: flores200-devtest
|
96 |
+
type: flores200-devtest
|
97 |
+
args: spa-vie
|
98 |
+
metrics:
|
99 |
+
- name: BLEU
|
100 |
+
type: bleu
|
101 |
+
value: 28.1
|
102 |
+
- name: chr-F
|
103 |
+
type: chrf
|
104 |
+
value: 0.50025
|
105 |
+
- task:
|
106 |
+
name: Translation deu-vie
|
107 |
+
type: translation
|
108 |
+
args: deu-vie
|
109 |
+
dataset:
|
110 |
+
name: flores101-devtest
|
111 |
+
type: flores_101
|
112 |
+
args: deu vie devtest
|
113 |
+
metrics:
|
114 |
+
- name: BLEU
|
115 |
+
type: bleu
|
116 |
+
value: 33.8
|
117 |
+
- name: chr-F
|
118 |
+
type: chrf
|
119 |
+
value: 0.53381
|
120 |
+
- task:
|
121 |
+
name: Translation eng-vie
|
122 |
+
type: translation
|
123 |
+
args: eng-vie
|
124 |
+
dataset:
|
125 |
+
name: flores101-devtest
|
126 |
+
type: flores_101
|
127 |
+
args: eng vie devtest
|
128 |
+
metrics:
|
129 |
+
- name: BLEU
|
130 |
+
type: bleu
|
131 |
+
value: 42.1
|
132 |
+
- name: chr-F
|
133 |
+
type: chrf
|
134 |
+
value: 0.59621
|
135 |
+
- task:
|
136 |
+
name: Translation por-vie
|
137 |
+
type: translation
|
138 |
+
args: por-vie
|
139 |
+
dataset:
|
140 |
+
name: flores101-devtest
|
141 |
+
type: flores_101
|
142 |
+
args: por vie devtest
|
143 |
+
metrics:
|
144 |
+
- name: BLEU
|
145 |
+
type: bleu
|
146 |
+
value: 36.0
|
147 |
+
- name: chr-F
|
148 |
+
type: chrf
|
149 |
+
value: 0.54919
|
150 |
+
- task:
|
151 |
+
name: Translation spa-vie
|
152 |
+
type: translation
|
153 |
+
args: spa-vie
|
154 |
+
dataset:
|
155 |
+
name: flores101-devtest
|
156 |
+
type: flores_101
|
157 |
+
args: spa vie devtest
|
158 |
+
metrics:
|
159 |
+
- name: BLEU
|
160 |
+
type: bleu
|
161 |
+
value: 27.8
|
162 |
+
- name: chr-F
|
163 |
+
type: chrf
|
164 |
+
value: 0.49921
|
165 |
+
- task:
|
166 |
+
name: Translation deu-vie
|
167 |
+
type: translation
|
168 |
+
args: deu-vie
|
169 |
+
dataset:
|
170 |
+
name: ntrex128
|
171 |
+
type: ntrex128
|
172 |
+
args: deu-vie
|
173 |
+
metrics:
|
174 |
+
- name: BLEU
|
175 |
+
type: bleu
|
176 |
+
value: 31.4
|
177 |
+
- name: chr-F
|
178 |
+
type: chrf
|
179 |
+
value: 0.52124
|
180 |
+
- task:
|
181 |
+
name: Translation fra-vie
|
182 |
+
type: translation
|
183 |
+
args: fra-vie
|
184 |
+
dataset:
|
185 |
+
name: ntrex128
|
186 |
+
type: ntrex128
|
187 |
+
args: fra-vie
|
188 |
+
metrics:
|
189 |
+
- name: BLEU
|
190 |
+
type: bleu
|
191 |
+
value: 31.8
|
192 |
+
- name: chr-F
|
193 |
+
type: chrf
|
194 |
+
value: 0.52044
|
195 |
+
- task:
|
196 |
+
name: Translation por-vie
|
197 |
+
type: translation
|
198 |
+
args: por-vie
|
199 |
+
dataset:
|
200 |
+
name: ntrex128
|
201 |
+
type: ntrex128
|
202 |
+
args: por-vie
|
203 |
+
metrics:
|
204 |
+
- name: BLEU
|
205 |
+
type: bleu
|
206 |
+
value: 33.3
|
207 |
+
- name: chr-F
|
208 |
+
type: chrf
|
209 |
+
value: 0.53060
|
210 |
+
- task:
|
211 |
+
name: Translation spa-vie
|
212 |
+
type: translation
|
213 |
+
args: spa-vie
|
214 |
+
dataset:
|
215 |
+
name: ntrex128
|
216 |
+
type: ntrex128
|
217 |
+
args: spa-vie
|
218 |
+
metrics:
|
219 |
+
- name: BLEU
|
220 |
+
type: bleu
|
221 |
+
value: 33.4
|
222 |
+
- name: chr-F
|
223 |
+
type: chrf
|
224 |
+
value: 0.53293
|
225 |
+
- task:
|
226 |
+
name: Translation deu-vie
|
227 |
+
type: translation
|
228 |
+
args: deu-vie
|
229 |
+
dataset:
|
230 |
+
name: tatoeba-test-v2021-08-07
|
231 |
+
type: tatoeba_mt
|
232 |
+
args: deu-vie
|
233 |
+
metrics:
|
234 |
+
- name: BLEU
|
235 |
+
type: bleu
|
236 |
+
value: 25.6
|
237 |
+
- name: chr-F
|
238 |
+
type: chrf
|
239 |
+
value: 0.45795
|
240 |
+
- task:
|
241 |
+
name: Translation eng-vie
|
242 |
+
type: translation
|
243 |
+
args: eng-vie
|
244 |
+
dataset:
|
245 |
+
name: tatoeba-test-v2021-08-07
|
246 |
+
type: tatoeba_mt
|
247 |
+
args: eng-vie
|
248 |
+
metrics:
|
249 |
+
- name: BLEU
|
250 |
+
type: bleu
|
251 |
+
value: 39.4
|
252 |
+
- name: chr-F
|
253 |
+
type: chrf
|
254 |
+
value: 0.56461
|
255 |
+
- task:
|
256 |
+
name: Translation fra-vie
|
257 |
+
type: translation
|
258 |
+
args: fra-vie
|
259 |
+
dataset:
|
260 |
+
name: tatoeba-test-v2021-08-07
|
261 |
+
type: tatoeba_mt
|
262 |
+
args: fra-vie
|
263 |
+
metrics:
|
264 |
+
- name: BLEU
|
265 |
+
type: bleu
|
266 |
+
value: 35.2
|
267 |
+
- name: chr-F
|
268 |
+
type: chrf
|
269 |
+
value: 0.52806
|
270 |
+
- task:
|
271 |
+
name: Translation multi-multi
|
272 |
+
type: translation
|
273 |
+
args: multi-multi
|
274 |
+
dataset:
|
275 |
+
name: tatoeba-test-v2020-07-28-v2023-09-26
|
276 |
+
type: tatoeba_mt
|
277 |
+
args: multi-multi
|
278 |
+
metrics:
|
279 |
+
- name: BLEU
|
280 |
+
type: bleu
|
281 |
+
value: 22.9
|
282 |
+
- name: chr-F
|
283 |
+
type: chrf
|
284 |
+
value: 0.40649
|
285 |
+
- task:
|
286 |
+
name: Translation spa-vie
|
287 |
+
type: translation
|
288 |
+
args: spa-vie
|
289 |
+
dataset:
|
290 |
+
name: tatoeba-test-v2021-08-07
|
291 |
+
type: tatoeba_mt
|
292 |
+
args: spa-vie
|
293 |
+
metrics:
|
294 |
+
- name: BLEU
|
295 |
+
type: bleu
|
296 |
+
value: 34.2
|
297 |
+
- name: chr-F
|
298 |
+
type: chrf
|
299 |
+
value: 0.52131
|
300 |
+
---
|
301 |
+
# opus-mt-tc-bible-big-deu_eng_fra_por_spa-aav
|
302 |
+
|
303 |
+
## Table of Contents
|
304 |
+
- [Model Details](#model-details)
|
305 |
+
- [Uses](#uses)
|
306 |
+
- [Risks, Limitations and Biases](#risks-limitations-and-biases)
|
307 |
+
- [How to Get Started With the Model](#how-to-get-started-with-the-model)
|
308 |
+
- [Training](#training)
|
309 |
+
- [Evaluation](#evaluation)
|
310 |
+
- [Citation Information](#citation-information)
|
311 |
+
- [Acknowledgements](#acknowledgements)
|
312 |
+
|
313 |
+
## Model Details
|
314 |
+
|
315 |
+
Neural machine translation model for translating from unknown (deu+eng+fra+por+spa) to Austro-Asiatic languages (aav).
|
316 |
+
|
317 |
+
This model is part of the [OPUS-MT project](https://github.com/Helsinki-NLP/Opus-MT), an effort to make neural machine translation models widely available and accessible for many languages in the world. All models are originally trained using the amazing framework of [Marian NMT](https://marian-nmt.github.io/), an efficient NMT implementation written in pure C++. The models have been converted to pyTorch using the transformers library by huggingface. Training data is taken from [OPUS](https://opus.nlpl.eu/) and training pipelines use the procedures of [OPUS-MT-train](https://github.com/Helsinki-NLP/Opus-MT-train).
|
318 |
+
**Model Description:**
|
319 |
+
- **Developed by:** Language Technology Research Group at the University of Helsinki
|
320 |
+
- **Model Type:** Translation (transformer-big)
|
321 |
+
- **Release**: 2024-05-29
|
322 |
+
- **License:** Apache-2.0
|
323 |
+
- **Language(s):**
|
324 |
+
- Source Language(s): deu eng fra por spa
|
325 |
+
- Target Language(s): bru cmo hoc jun kha khm kxm mnw ngt sat vie wbm
|
326 |
+
- Valid Target Language Labels: >>aem<< >>alk<< >>aml<< >>asr<< >>bbh<< >>bdq<< >>bfw<< >>bgk<< >>bgl<< >>bix<< >>biy<< >>blr<< >>brb<< >>bru<< >>brv<< >>btq<< >>caq<< >>cbn<< >>cdz<< >>cma<< >>cmo<< >>cog<< >>crv<< >>crw<< >>cua<< >>cwg<< >>dnu<< >>ekl<< >>gaq<< >>gbj<< >>hal<< >>hld<< >>hnu<< >>hoc<< >>hoc_Wara<< >>hre<< >>huo<< >>irr<< >>jah<< >>jeh<< >>jhi<< >>jun<< >>juy<< >>kdt<< >>kfp<< >>kfq<< >>kha<< >>khf<< >>khm<< >>khr<< >>kjg<< >>kjm<< >>knq<< >>kns<< >>kpm<< >>krr<< >>krv<< >>ksz<< >>kta<< >>ktv<< >>kuf<< >>kxm<< >>kxy<< >>lbn<< >>lbo<< >>lcp<< >>lnh<< >>lwl<< >>lyg<< >>mef<< >>mhe<< >>mjx<< >>mlf<< >>mmj<< >>mml<< >>mng<< >>mnn<< >>mnq<< >>mnw<< >>moo<< >>mqt<< >>mra<< >>mtq<< >>mzt<< >>ncb<< >>ncq<< >>nev<< >>ngt<< >>ngt_Latn<< >>nik<< >>nuo<< >>nyl<< >>omx<< >>oog<< >>oyb<< >>pac<< >>pbv<< >>pcb<< >>pce<< >>pcj<< >>phg<< >>pkt<< >>pll<< >>ply<< >>pnx<< >>prk<< >>prt<< >>puo<< >>rbb<< >>ren<< >>ril<< >>rka<< >>rmx<< >>sat<< >>sat_Latn<< >>sbo<< >>scb<< >>scq<< >>sct<< >>sea<< >>sed<< >>sii<< >>smu<< >>spu<< >>sqq<< >>srb<< >>ssm<< >>sss<< >>stg<< >>sti<< >>stt<< >>stu<< >>syo<< >>sza<< >>szc<< >>tdf<< >>tdr<< >>tea<< >>tef<< >>thm<< >>tkz<< >>tlq<< >>tmo<< >>tnz<< >>tou<< >>tpu<< >>trd<< >>tth<< >>tto<< >>tyh<< >>unr<< >>uuu<< >>vie<< >>vwa<< >>wbm<< >>xao<< >>xkk<< >>xnh<< >>xxx<< >>yin<< >>zng<<
|
327 |
+
- **Original Model**: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/deu+eng+fra+por+spa-aav/opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.zip)
|
328 |
+
- **Resources for more information:**
|
329 |
+
- [OPUS-MT dashboard](https://opus.nlpl.eu/dashboard/index.php?pkg=opusmt&test=all&scoreslang=all&chart=standard&model=Tatoeba-MT-models/deu%2Beng%2Bfra%2Bpor%2Bspa-aav/opusTCv20230926max50%2Bbt%2Bjhubc_transformer-big_2024-05-29)
|
330 |
+
- [OPUS-MT-train GitHub Repo](https://github.com/Helsinki-NLP/OPUS-MT-train)
|
331 |
+
- [More information about MarianNMT models in the transformers library](https://huggingface.co/docs/transformers/model_doc/marian)
|
332 |
+
- [Tatoeba Translation Challenge](https://github.com/Helsinki-NLP/Tatoeba-Challenge/)
|
333 |
+
- [HPLT bilingual data v1 (as part of the Tatoeba Translation Challenge dataset)](https://hplt-project.org/datasets/v1)
|
334 |
+
- [A massively parallel Bible corpus](https://aclanthology.org/L14-1215/)
|
335 |
+
|
336 |
+
This is a multilingual translation model with multiple target languages. A sentence initial language token is required in the form of `>>id<<` (id = valid target language ID), e.g. `>>bru<<`
|
337 |
+
|
338 |
+
## Uses
|
339 |
+
|
340 |
+
This model can be used for translation and text-to-text generation.
|
341 |
+
|
342 |
+
## Risks, Limitations and Biases
|
343 |
+
|
344 |
+
**CONTENT WARNING: Readers should be aware that the model is trained on various public data sets that may contain content that is disturbing, offensive, and can propagate historical and current stereotypes.**
|
345 |
+
|
346 |
+
Significant research has explored bias and fairness issues with language models (see, e.g., [Sheng et al. (2021)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)).
|
347 |
+
|
348 |
+
## How to Get Started With the Model
|
349 |
+
|
350 |
+
A short example code:
|
351 |
+
|
352 |
+
```python
|
353 |
+
from transformers import MarianMTModel, MarianTokenizer
|
354 |
+
|
355 |
+
src_text = [
|
356 |
+
">>khm<< Der Junge wirft einen Stein.",
|
357 |
+
">>vie<< ¿Y tú?"
|
358 |
+
]
|
359 |
+
|
360 |
+
model_name = "pytorch-models/opus-mt-tc-bible-big-deu_eng_fra_por_spa-aav"
|
361 |
+
tokenizer = MarianTokenizer.from_pretrained(model_name)
|
362 |
+
model = MarianMTModel.from_pretrained(model_name)
|
363 |
+
translated = model.generate(**tokenizer(src_text, return_tensors="pt", padding=True))
|
364 |
+
|
365 |
+
for t in translated:
|
366 |
+
print( tokenizer.decode(t, skip_special_tokens=True) )
|
367 |
+
|
368 |
+
# expected output:
|
369 |
+
# ក្មេងប្រុស នោះ យក ដុំ ថ្ម គប់ ។
|
370 |
+
# Còn anh thì sao?
|
371 |
+
```
|
372 |
+
|
373 |
+
You can also use OPUS-MT models with the transformers pipelines, for example:
|
374 |
+
|
375 |
+
```python
|
376 |
+
from transformers import pipeline
|
377 |
+
pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-bible-big-deu_eng_fra_por_spa-aav")
|
378 |
+
print(pipe(">>khm<< Der Junge wirft einen Stein."))
|
379 |
+
|
380 |
+
# expected output: ក្មេងប្រុស នោះ យក ដុំ ថ្ម គប់ ។
|
381 |
+
```
|
382 |
+
|
383 |
+
## Training
|
384 |
+
|
385 |
+
- **Data**: opusTCv20230926max50+bt+jhubc ([source](https://github.com/Helsinki-NLP/Tatoeba-Challenge))
|
386 |
+
- **Pre-processing**: SentencePiece (spm32k,spm32k)
|
387 |
+
- **Model Type:** transformer-big
|
388 |
+
- **Original MarianNMT Model**: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/deu+eng+fra+por+spa-aav/opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.zip)
|
389 |
+
- **Training Scripts**: [GitHub Repo](https://github.com/Helsinki-NLP/OPUS-MT-train)
|
390 |
+
|
391 |
+
## Evaluation
|
392 |
+
|
393 |
+
* [Model scores at the OPUS-MT dashboard](https://opus.nlpl.eu/dashboard/index.php?pkg=opusmt&test=all&scoreslang=all&chart=standard&model=Tatoeba-MT-models/deu%2Beng%2Bfra%2Bpor%2Bspa-aav/opusTCv20230926max50%2Bbt%2Bjhubc_transformer-big_2024-05-29)
|
394 |
+
* test set translations: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/deu+eng+fra+por+spa-aav/opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.test.txt)
|
395 |
+
* test set scores: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/deu+eng+fra+por+spa-aav/opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.eval.txt)
|
396 |
+
* benchmark results: [benchmark_results.txt](benchmark_results.txt)
|
397 |
+
* benchmark output: [benchmark_translations.zip](benchmark_translations.zip)
|
398 |
+
|
399 |
+
| langpair | testset | chr-F | BLEU | #sent | #words |
|
400 |
+
|----------|---------|-------|-------|-------|--------|
|
401 |
+
| deu-vie | tatoeba-test-v2021-08-07 | 0.45795 | 25.6 | 400 | 3768 |
|
402 |
+
| eng-hoc | tatoeba-test-v2021-08-07 | 6.438 | 0.2 | 660 | 2591 |
|
403 |
+
| eng-kha | tatoeba-test-v2021-08-07 | 5.741 | 0.0 | 1314 | 9269 |
|
404 |
+
| eng-vie | tatoeba-test-v2021-08-07 | 0.56461 | 39.4 | 2500 | 24427 |
|
405 |
+
| fra-vie | tatoeba-test-v2021-08-07 | 0.52806 | 35.2 | 1299 | 13219 |
|
406 |
+
| spa-vie | tatoeba-test-v2021-08-07 | 0.52131 | 34.2 | 594 | 4740 |
|
407 |
+
| deu-vie | flores101-devtest | 0.53381 | 33.8 | 1012 | 33331 |
|
408 |
+
| eng-khm | flores101-devtest | 0.42302 | 1.3 | 1012 | 7006 |
|
409 |
+
| eng-vie | flores101-devtest | 0.59621 | 42.1 | 1012 | 33331 |
|
410 |
+
| fra-khm | flores101-devtest | 0.40042 | 2.2 | 1012 | 7006 |
|
411 |
+
| por-khm | flores101-devtest | 0.40585 | 2.1 | 1012 | 7006 |
|
412 |
+
| por-vie | flores101-devtest | 0.54919 | 36.0 | 1012 | 33331 |
|
413 |
+
| spa-vie | flores101-devtest | 0.49921 | 27.8 | 1012 | 33331 |
|
414 |
+
| deu-vie | flores200-devtest | 0.53671 | 34.0 | 1012 | 33331 |
|
415 |
+
| eng-khm | flores200-devtest | 0.42148 | 1.3 | 1012 | 7006 |
|
416 |
+
| eng-vie | flores200-devtest | 0.59842 | 42.4 | 1012 | 33331 |
|
417 |
+
| fra-vie | flores200-devtest | 0.54101 | 34.6 | 1012 | 33331 |
|
418 |
+
| por-khm | flores200-devtest | 0.40832 | 1.9 | 1012 | 7006 |
|
419 |
+
| por-vie | flores200-devtest | 0.54970 | 36.1 | 1012 | 33331 |
|
420 |
+
| spa-vie | flores200-devtest | 0.50025 | 28.1 | 1012 | 33331 |
|
421 |
+
| deu-khm | ntrex128 | 0.44903 | 3.5 | 1997 | 15866 |
|
422 |
+
| deu-vie | ntrex128 | 0.52124 | 31.4 | 1997 | 64655 |
|
423 |
+
| eng-khm | ntrex128 | 0.50494 | 1.6 | 1997 | 15866 |
|
424 |
+
| eng-vie | ntrex128 | 3.831 | 0.0 | 1997 | 64655 |
|
425 |
+
| fra-khm | ntrex128 | 0.43841 | 2.4 | 1997 | 15866 |
|
426 |
+
| fra-vie | ntrex128 | 0.52044 | 31.8 | 1997 | 64655 |
|
427 |
+
| por-khm | ntrex128 | 0.46655 | 2.5 | 1997 | 15866 |
|
428 |
+
| por-vie | ntrex128 | 0.53060 | 33.3 | 1997 | 64655 |
|
429 |
+
| spa-khm | ntrex128 | 0.46443 | 2.7 | 1997 | 15866 |
|
430 |
+
| spa-vie | ntrex128 | 0.53293 | 33.4 | 1997 | 64655 |
|
431 |
+
| eng-khm | tico19-test | 0.47806 | 2.5 | 2100 | 15810 |
|
432 |
+
| fra-khm | tico19-test | 3.268 | 1.0 | 2100 | 15810 |
|
433 |
+
| por-khm | tico19-test | 3.900 | 1.1 | 2100 | 15810 |
|
434 |
+
| spa-khm | tico19-test | 3.784 | 1.0 | 2100 | 15810 |
|
435 |
+
|
436 |
+
## Citation Information
|
437 |
+
|
438 |
+
* Publications: [Democratizing neural machine translation with OPUS-MT](https://doi.org/10.1007/s10579-023-09704-w) and [OPUS-MT – Building open translation services for the World](https://aclanthology.org/2020.eamt-1.61/) and [The Tatoeba Translation Challenge – Realistic Data Sets for Low Resource and Multilingual MT](https://aclanthology.org/2020.wmt-1.139/) (Please, cite if you use this model.)
|
439 |
+
|
440 |
+
```bibtex
|
441 |
+
@article{tiedemann2023democratizing,
|
442 |
+
title={Democratizing neural machine translation with {OPUS-MT}},
|
443 |
+
author={Tiedemann, J{\"o}rg and Aulamo, Mikko and Bakshandaeva, Daria and Boggia, Michele and Gr{\"o}nroos, Stig-Arne and Nieminen, Tommi and Raganato, Alessandro and Scherrer, Yves and Vazquez, Raul and Virpioja, Sami},
|
444 |
+
journal={Language Resources and Evaluation},
|
445 |
+
number={58},
|
446 |
+
pages={713--755},
|
447 |
+
year={2023},
|
448 |
+
publisher={Springer Nature},
|
449 |
+
issn={1574-0218},
|
450 |
+
doi={10.1007/s10579-023-09704-w}
|
451 |
+
}
|
452 |
+
|
453 |
+
@inproceedings{tiedemann-thottingal-2020-opus,
|
454 |
+
title = "{OPUS}-{MT} {--} Building open translation services for the World",
|
455 |
+
author = {Tiedemann, J{\"o}rg and Thottingal, Santhosh},
|
456 |
+
booktitle = "Proceedings of the 22nd Annual Conference of the European Association for Machine Translation",
|
457 |
+
month = nov,
|
458 |
+
year = "2020",
|
459 |
+
address = "Lisboa, Portugal",
|
460 |
+
publisher = "European Association for Machine Translation",
|
461 |
+
url = "https://aclanthology.org/2020.eamt-1.61",
|
462 |
+
pages = "479--480",
|
463 |
+
}
|
464 |
+
|
465 |
+
@inproceedings{tiedemann-2020-tatoeba,
|
466 |
+
title = "The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}",
|
467 |
+
author = {Tiedemann, J{\"o}rg},
|
468 |
+
booktitle = "Proceedings of the Fifth Conference on Machine Translation",
|
469 |
+
month = nov,
|
470 |
+
year = "2020",
|
471 |
+
address = "Online",
|
472 |
+
publisher = "Association for Computational Linguistics",
|
473 |
+
url = "https://aclanthology.org/2020.wmt-1.139",
|
474 |
+
pages = "1174--1182",
|
475 |
+
}
|
476 |
+
```
|
477 |
+
|
478 |
+
## Acknowledgements
|
479 |
+
|
480 |
+
The work is supported by the [HPLT project](https://hplt-project.org/), funded by the European Union’s Horizon Europe research and innovation programme under grant agreement No 101070350. We are also grateful for the generous computational resources and IT infrastructure provided by [CSC -- IT Center for Science](https://www.csc.fi/), Finland, and the [EuroHPC supercomputer LUMI](https://www.lumi-supercomputer.eu/).
|
481 |
+
|
482 |
+
## Model conversion info
|
483 |
+
|
484 |
+
* transformers version: 4.45.1
|
485 |
+
* OPUS-MT git hash: 0882077
|
486 |
+
* port time: Tue Oct 8 00:06:48 EEST 2024
|
487 |
+
* port machine: LM0-400-22516.local
|
benchmark_results.txt
ADDED
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
multi-multi tatoeba-test-v2020-07-28-v2023-09-26 0.40649 22.9 9540 66166
|
2 |
+
deu-khm flores101-devtest 0.38800 2.3 1012 7006
|
3 |
+
deu-vie flores101-devtest 0.53381 33.8 1012 33331
|
4 |
+
eng-khm flores101-devtest 0.42302 1.3 1012 7006
|
5 |
+
eng-vie flores101-devtest 0.59621 42.1 1012 33331
|
6 |
+
fra-khm flores101-devtest 0.40042 2.2 1012 7006
|
7 |
+
por-khm flores101-devtest 0.40585 2.1 1012 7006
|
8 |
+
por-vie flores101-devtest 0.54919 36.0 1012 33331
|
9 |
+
spa-khm flores101-devtest 0.36844 1.3 1012 7006
|
10 |
+
spa-vie flores101-devtest 0.49921 27.8 1012 33331
|
11 |
+
deu-khm flores200-devtest 0.38953 2.4 1012 7006
|
12 |
+
deu-sat flores200-devtest 0.13449 0.1 1012 25693
|
13 |
+
deu-vie flores200-devtest 0.53671 34.0 1012 33331
|
14 |
+
eng-khm flores200-devtest 0.42148 1.3 1012 7006
|
15 |
+
eng-sat flores200-devtest 0.355 0.0 1012 25693
|
16 |
+
eng-vie flores200-devtest 0.59842 42.4 1012 33331
|
17 |
+
fra-khm flores200-devtest 0.39960 2.3 1012 7006
|
18 |
+
fra-sat flores200-devtest 0.13630 0.1 1012 25693
|
19 |
+
fra-vie flores200-devtest 0.54101 34.6 1012 33331
|
20 |
+
por-khm flores200-devtest 0.40832 1.9 1012 7006
|
21 |
+
por-sat flores200-devtest 0.13503 0.1 1012 25693
|
22 |
+
por-vie flores200-devtest 0.54970 36.1 1012 33331
|
23 |
+
spa-khm flores200-devtest 0.36868 1.4 1012 7006
|
24 |
+
spa-sat flores200-devtest 0.13742 0.1 1012 25693
|
25 |
+
spa-vie flores200-devtest 0.50025 28.1 1012 33331
|
26 |
+
eng-khm newstest2020 0.35104 0.9 2320 15454
|
27 |
+
deu-khm ntrex128 0.44903 3.5 1997 15866
|
28 |
+
deu-vie ntrex128 0.52124 31.4 1997 64655
|
29 |
+
eng-khm ntrex128 0.50494 1.6 1997 15866
|
30 |
+
eng-vie ntrex128 3.831 0.0 1997 64655
|
31 |
+
fra-khm ntrex128 0.43841 2.4 1997 15866
|
32 |
+
fra-vie ntrex128 0.52044 31.8 1997 64655
|
33 |
+
por-khm ntrex128 0.46655 2.5 1997 15866
|
34 |
+
por-vie ntrex128 0.53060 33.3 1997 64655
|
35 |
+
spa-khm ntrex128 0.46443 2.7 1997 15866
|
36 |
+
spa-vie ntrex128 0.53293 33.4 1997 64655
|
37 |
+
eng-khm tatoeba-test-v2020-07-28 0.34565 0.2 752 1737
|
38 |
+
spa-khm tatoeba-test-v2020-07-28 0.37159 0.2 1472 3391
|
39 |
+
spa-vie tatoeba-test-v2020-07-28 0.52030 34.2 582 4654
|
40 |
+
deu-vie tatoeba-test-v2021-03-30 0.45756 25.4 401 3775
|
41 |
+
eng-hoc tatoeba-test-v2021-03-30 6.373 0.2 661 2594
|
42 |
+
eng-khm tatoeba-test-v2021-03-30 0.34549 0.2 754 1741
|
43 |
+
spa-khm tatoeba-test-v2021-03-30 0.37159 0.2 1472 3391
|
44 |
+
spa-vie tatoeba-test-v2021-03-30 0.52010 34.3 604 4824
|
45 |
+
deu-vie tatoeba-test-v2021-08-07 0.45795 25.6 400 3768
|
46 |
+
eng-hoc tatoeba-test-v2021-08-07 6.438 0.2 660 2591
|
47 |
+
eng-kha tatoeba-test-v2021-08-07 5.741 0.0 1314 9269
|
48 |
+
eng-khm tatoeba-test-v2021-08-07 0.33850 0.2 726 1692
|
49 |
+
eng-vie tatoeba-test-v2021-08-07 0.56461 39.4 2500 24427
|
50 |
+
fra-vie tatoeba-test-v2021-08-07 0.52806 35.2 1299 13219
|
51 |
+
spa-khm tatoeba-test-v2021-08-07 0.37427 0.2 1448 3343
|
52 |
+
spa-vie tatoeba-test-v2021-08-07 0.52131 34.2 594 4740
|
53 |
+
eng-khm tico19-test 0.47806 2.5 2100 15810
|
54 |
+
fra-khm tico19-test 3.268 1.0 2100 15810
|
55 |
+
por-khm tico19-test 3.900 1.1 2100 15810
|
56 |
+
spa-khm tico19-test 3.784 1.0 2100 15810
|
benchmark_translations.zip
ADDED
File without changes
|
config.json
ADDED
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "pytorch-models/opus-mt-tc-bible-big-deu_eng_fra_por_spa-aav",
|
3 |
+
"activation_dropout": 0.0,
|
4 |
+
"activation_function": "relu",
|
5 |
+
"architectures": [
|
6 |
+
"MarianMTModel"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.0,
|
9 |
+
"bos_token_id": 0,
|
10 |
+
"classifier_dropout": 0.0,
|
11 |
+
"d_model": 1024,
|
12 |
+
"decoder_attention_heads": 16,
|
13 |
+
"decoder_ffn_dim": 4096,
|
14 |
+
"decoder_layerdrop": 0.0,
|
15 |
+
"decoder_layers": 6,
|
16 |
+
"decoder_start_token_id": 57538,
|
17 |
+
"decoder_vocab_size": 57539,
|
18 |
+
"dropout": 0.1,
|
19 |
+
"encoder_attention_heads": 16,
|
20 |
+
"encoder_ffn_dim": 4096,
|
21 |
+
"encoder_layerdrop": 0.0,
|
22 |
+
"encoder_layers": 6,
|
23 |
+
"eos_token_id": 943,
|
24 |
+
"forced_eos_token_id": null,
|
25 |
+
"init_std": 0.02,
|
26 |
+
"is_encoder_decoder": true,
|
27 |
+
"max_length": null,
|
28 |
+
"max_position_embeddings": 1024,
|
29 |
+
"model_type": "marian",
|
30 |
+
"normalize_embedding": false,
|
31 |
+
"num_beams": null,
|
32 |
+
"num_hidden_layers": 6,
|
33 |
+
"pad_token_id": 57538,
|
34 |
+
"scale_embedding": true,
|
35 |
+
"share_encoder_decoder_embeddings": true,
|
36 |
+
"static_position_embeddings": true,
|
37 |
+
"torch_dtype": "float32",
|
38 |
+
"transformers_version": "4.45.1",
|
39 |
+
"use_cache": true,
|
40 |
+
"vocab_size": 57539
|
41 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bad_words_ids": [
|
4 |
+
[
|
5 |
+
57538
|
6 |
+
]
|
7 |
+
],
|
8 |
+
"bos_token_id": 0,
|
9 |
+
"decoder_start_token_id": 57538,
|
10 |
+
"eos_token_id": 943,
|
11 |
+
"forced_eos_token_id": 943,
|
12 |
+
"max_length": 512,
|
13 |
+
"num_beams": 4,
|
14 |
+
"pad_token_id": 57538,
|
15 |
+
"transformers_version": "4.45.1"
|
16 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f43caecd953c319c06057ae84727f24a5441f39e463ae8d63799e9a0af8b1ba5
|
3 |
+
size 941369020
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:24d6d94fd32087fe201bd6338b5703f8d5ff3ac1776c170faee97935156ec170
|
3 |
+
size 941420293
|
source.spm
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:42cc250d301ef587fcb7af9c7c5d4651ccb4f9a69deb0ebcc29f9e55e3bb8ca6
|
3 |
+
size 803648
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
|
target.spm
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cb53816a56d71f4a8b3ccf5f3e6ada62b1107acae9c90b7b4c792108a9bc520e
|
3 |
+
size 888156
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"source_lang": "deu+eng+fra+por+spa", "target_lang": "aav", "unk_token": "<unk>", "eos_token": "</s>", "pad_token": "<pad>", "model_max_length": 512, "sp_model_kwargs": {}, "separate_vocabs": false, "special_tokens_map_file": null, "name_or_path": "marian-models/opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29/deu+eng+fra+por+spa-aav", "tokenizer_class": "MarianTokenizer"}
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|