Upload folder using huggingface_hub
Browse files- args.yaml +5 -5
- config +96 -102
- metrics +0 -0
- params.best +2 -2
- vocab.src.0.json +0 -0
args.yaml
CHANGED
@@ -51,7 +51,7 @@ local_rank: null
|
|
51 |
loglevel: INFO
|
52 |
loglevel_secondary_workers: INFO
|
53 |
max_checkpoints: null
|
54 |
-
max_num_checkpoint_not_improved:
|
55 |
max_num_epochs: null
|
56 |
max_samples: null
|
57 |
max_seconds: null
|
@@ -59,7 +59,7 @@ max_seq_len:
|
|
59 |
- 512
|
60 |
- 512
|
61 |
max_updates: null
|
62 |
-
min_num_epochs:
|
63 |
min_samples: null
|
64 |
min_updates: null
|
65 |
momentum: 0.0
|
@@ -83,11 +83,11 @@ optimizer_betas:
|
|
83 |
- 0.9
|
84 |
- 0.999
|
85 |
optimizer_eps: 1.0e-08
|
86 |
-
output: /shares/
|
87 |
overwrite_output: false
|
88 |
pad_vocab_to_multiple_of: 8
|
89 |
-
params: /shares/
|
90 |
-
prepared_data: /shares/
|
91 |
quiet: false
|
92 |
quiet_secondary_workers: false
|
93 |
seed: 1
|
|
|
51 |
loglevel: INFO
|
52 |
loglevel_secondary_workers: INFO
|
53 |
max_checkpoints: null
|
54 |
+
max_num_checkpoint_not_improved: 100
|
55 |
max_num_epochs: null
|
56 |
max_samples: null
|
57 |
max_seconds: null
|
|
|
59 |
- 512
|
60 |
- 512
|
61 |
max_updates: null
|
62 |
+
min_num_epochs: 10
|
63 |
min_samples: null
|
64 |
min_updates: null
|
65 |
momentum: 0.0
|
|
|
83 |
- 0.9
|
84 |
- 0.999
|
85 |
optimizer_eps: 1.0e-08
|
86 |
+
output: /shares/iict-sp2.ebling.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-gpt-tuned/model
|
87 |
overwrite_output: false
|
88 |
pad_vocab_to_multiple_of: 8
|
89 |
+
params: /shares/iict-sp2.ebling.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-gpt/model/params.best
|
90 |
+
prepared_data: /shares/iict-sp2.ebling.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-gpt-tuned/train_data
|
91 |
quiet: false
|
92 |
quiet_secondary_workers: false
|
93 |
seed: 1
|
config
CHANGED
@@ -2,30 +2,30 @@
|
|
2 |
config_data: !DataConfig
|
3 |
data_statistics: !DataStatistics
|
4 |
average_len_target_per_bucket:
|
5 |
-
- 6.
|
6 |
-
- 10.
|
7 |
-
- 18.
|
8 |
-
-
|
9 |
-
-
|
10 |
-
- 41.
|
11 |
-
-
|
12 |
-
- 59.
|
13 |
-
-
|
14 |
-
- 75.
|
15 |
-
-
|
16 |
-
-
|
17 |
-
-
|
18 |
-
-
|
19 |
-
- 15.857142857142856
|
20 |
-
- 8.6
|
21 |
-
- 8.0
|
22 |
-
- 6.5
|
23 |
-
- 4.5
|
24 |
-
- null
|
25 |
-
- 6.25
|
26 |
- 6.0
|
27 |
-
-
|
28 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
- null
|
30 |
- null
|
31 |
- null
|
@@ -263,80 +263,73 @@ config_data: !DataConfig
|
|
263 |
- !!python/tuple
|
264 |
- 513
|
265 |
- 513
|
266 |
-
length_ratio_mean: 1.
|
267 |
length_ratio_stats_per_bucket:
|
268 |
- !!python/tuple
|
269 |
-
- 1.
|
270 |
-
- 0.
|
271 |
-
- !!python/tuple
|
272 |
-
- 1.6064942041128283
|
273 |
-
- 0.6417946634210556
|
274 |
-
- !!python/tuple
|
275 |
-
- 2.359196679910085
|
276 |
-
- 0.9575904838242857
|
277 |
- !!python/tuple
|
278 |
-
-
|
279 |
-
-
|
280 |
- !!python/tuple
|
281 |
-
- 2.
|
282 |
-
-
|
283 |
- !!python/tuple
|
284 |
-
-
|
285 |
-
-
|
286 |
- !!python/tuple
|
287 |
-
- 3.
|
288 |
-
- 2.
|
289 |
- !!python/tuple
|
290 |
-
-
|
291 |
-
- 2.
|
292 |
- !!python/tuple
|
293 |
-
-
|
294 |
-
- 2.
|
295 |
- !!python/tuple
|
296 |
-
- 3.
|
297 |
-
- 2.
|
298 |
- !!python/tuple
|
299 |
-
-
|
300 |
-
-
|
301 |
- !!python/tuple
|
302 |
-
-
|
303 |
-
-
|
304 |
- !!python/tuple
|
305 |
-
-
|
306 |
-
-
|
307 |
- !!python/tuple
|
308 |
-
- 0.
|
309 |
-
- 0.
|
310 |
- !!python/tuple
|
311 |
-
- 0.
|
312 |
-
- 0.
|
313 |
- !!python/tuple
|
314 |
-
- 0.
|
315 |
-
- 0.
|
316 |
- !!python/tuple
|
317 |
-
- 0.
|
318 |
- 0.0
|
319 |
- !!python/tuple
|
320 |
-
- 0.
|
321 |
-
- 0.
|
322 |
- !!python/tuple
|
323 |
-
- 0.
|
324 |
-
- 0.
|
325 |
-
- &id001 !!python/tuple
|
326 |
-
- null
|
327 |
-
- null
|
328 |
- !!python/tuple
|
329 |
-
- 0.
|
330 |
-
- 0.
|
331 |
- !!python/tuple
|
332 |
-
- 0.
|
333 |
-
- 0.004700847851307781
|
334 |
-
- !!python/tuple
|
335 |
-
- 0.0446927374301676
|
336 |
- 0.0
|
337 |
- !!python/tuple
|
338 |
-
- 0.
|
339 |
- 0.0
|
|
|
|
|
|
|
|
|
|
|
340 |
- *id001
|
341 |
- *id001
|
342 |
- *id001
|
@@ -378,36 +371,37 @@ config_data: !DataConfig
|
|
378 |
- *id001
|
379 |
- *id001
|
380 |
- *id001
|
381 |
-
|
382 |
-
|
|
|
383 |
max_observed_len_target: 87
|
384 |
num_discarded: 0
|
385 |
-
num_sents:
|
386 |
num_sents_per_bucket:
|
387 |
-
-
|
388 |
-
-
|
389 |
-
-
|
390 |
-
-
|
391 |
-
-
|
392 |
-
-
|
393 |
-
-
|
394 |
-
-
|
395 |
-
-
|
396 |
-
-
|
397 |
-
-
|
398 |
-
-
|
399 |
-
-
|
400 |
-
-
|
401 |
-
- 7
|
402 |
-
- 5
|
403 |
-
- 1
|
404 |
-
- 2
|
405 |
- 2
|
406 |
-
- 0
|
407 |
- 4
|
408 |
-
-
|
409 |
-
-
|
410 |
- 2
|
|
|
|
|
|
|
|
|
|
|
411 |
- 0
|
412 |
- 0
|
413 |
- 0
|
@@ -449,11 +443,11 @@ config_data: !DataConfig
|
|
449 |
- 0
|
450 |
- 0
|
451 |
- 0
|
452 |
-
num_tokens_source:
|
453 |
-
num_tokens_target:
|
454 |
num_unks_source: 0
|
455 |
num_unks_target: 0
|
456 |
-
size_vocab_source:
|
457 |
size_vocab_target: 656
|
458 |
eop_id: -1
|
459 |
max_seq_len_source: 513
|
@@ -485,7 +479,7 @@ config_embed_source: !EmbeddingConfig
|
|
485 |
factor_configs: null
|
486 |
num_embed: 512
|
487 |
num_factors: 1
|
488 |
-
vocab_size:
|
489 |
config_embed_target: !EmbeddingConfig
|
490 |
allow_sparse_grad: false
|
491 |
dropout: 0.5
|
@@ -537,6 +531,6 @@ dtype: float32
|
|
537 |
lhuc: false
|
538 |
neural_vocab_selection: null
|
539 |
neural_vocab_selection_block_loss: false
|
540 |
-
vocab_source_size:
|
541 |
vocab_target_size: 656
|
542 |
weight_tying_type: trg_softmax
|
|
|
2 |
config_data: !DataConfig
|
3 |
data_statistics: !DataStatistics
|
4 |
average_len_target_per_bucket:
|
5 |
+
- 6.011277314861252
|
6 |
+
- 10.39671010650862
|
7 |
+
- 18.504618481112797
|
8 |
+
- 26.041977712863865
|
9 |
+
- 32.48293857888395
|
10 |
+
- 41.60258525852575
|
11 |
+
- 51.3669121514993
|
12 |
+
- 59.76513060097686
|
13 |
+
- 68.25156398104271
|
14 |
+
- 75.67976141505562
|
15 |
+
- 83.09476309226905
|
16 |
+
- 21.1875
|
17 |
+
- 7.749999999999999
|
18 |
+
- 5.833333333333333
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
- 6.0
|
20 |
+
- 7.0
|
21 |
+
- 7.25
|
22 |
+
- 6.333333333333333
|
23 |
+
- 7.0
|
24 |
+
- 5.0
|
25 |
+
- null
|
26 |
+
- null
|
27 |
+
- null
|
28 |
+
- null
|
29 |
- null
|
30 |
- null
|
31 |
- null
|
|
|
263 |
- !!python/tuple
|
264 |
- 513
|
265 |
- 513
|
266 |
+
length_ratio_mean: 1.4462363190719616
|
267 |
length_ratio_stats_per_bucket:
|
268 |
- !!python/tuple
|
269 |
+
- 1.13785261776882
|
270 |
+
- 0.30617192612461125
|
|
|
|
|
|
|
|
|
|
|
|
|
271 |
- !!python/tuple
|
272 |
+
- 1.6532001104346803
|
273 |
+
- 0.6379238313847178
|
274 |
- !!python/tuple
|
275 |
+
- 2.471066932954014
|
276 |
+
- 0.9542933578872858
|
277 |
- !!python/tuple
|
278 |
+
- 2.9657919123449297
|
279 |
+
- 1.5793440289847793
|
280 |
- !!python/tuple
|
281 |
+
- 3.096168086847911
|
282 |
+
- 2.441325976657657
|
283 |
- !!python/tuple
|
284 |
+
- 3.268248874962792
|
285 |
+
- 2.2403195790511785
|
286 |
- !!python/tuple
|
287 |
+
- 3.453956732311208
|
288 |
+
- 2.518686961607376
|
289 |
- !!python/tuple
|
290 |
+
- 3.2456624469723585
|
291 |
+
- 2.2619507905455962
|
292 |
- !!python/tuple
|
293 |
+
- 3.322785538319874
|
294 |
+
- 2.1624153548695157
|
295 |
- !!python/tuple
|
296 |
+
- 3.489342008082259
|
297 |
+
- 2.469578333618206
|
298 |
- !!python/tuple
|
299 |
+
- 3.278259980416589
|
300 |
+
- 1.6906831932492183
|
301 |
- !!python/tuple
|
302 |
+
- 0.23045117210041233
|
303 |
+
- 0.26501940303452104
|
304 |
- !!python/tuple
|
305 |
+
- 0.07818400556098729
|
306 |
+
- 0.027479091254378488
|
307 |
- !!python/tuple
|
308 |
+
- 0.05445908159715249
|
309 |
+
- 0.013332869808672716
|
310 |
- !!python/tuple
|
311 |
+
- 0.05042016806722689
|
312 |
- 0.0
|
313 |
- !!python/tuple
|
314 |
+
- 0.05658536585365853
|
315 |
+
- 0.016585365853658534
|
316 |
- !!python/tuple
|
317 |
+
- 0.05537665840439907
|
318 |
+
- 0.021182928316626336
|
|
|
|
|
|
|
319 |
- !!python/tuple
|
320 |
+
- 0.04568795432892477
|
321 |
+
- 0.009015856562239487
|
322 |
- !!python/tuple
|
323 |
+
- 0.047619047619047616
|
|
|
|
|
|
|
324 |
- 0.0
|
325 |
- !!python/tuple
|
326 |
+
- 0.032679738562091505
|
327 |
- 0.0
|
328 |
+
- &id001 !!python/tuple
|
329 |
+
- null
|
330 |
+
- null
|
331 |
+
- *id001
|
332 |
+
- *id001
|
333 |
- *id001
|
334 |
- *id001
|
335 |
- *id001
|
|
|
371 |
- *id001
|
372 |
- *id001
|
373 |
- *id001
|
374 |
+
- *id001
|
375 |
+
length_ratio_std: 0.8476478699398606
|
376 |
+
max_observed_len_source: 153
|
377 |
max_observed_len_target: 87
|
378 |
num_discarded: 0
|
379 |
+
num_sents: 779247
|
380 |
num_sents_per_bucket:
|
381 |
+
- 478128
|
382 |
+
- 233503
|
383 |
+
- 27823
|
384 |
+
- 9243
|
385 |
+
- 4982
|
386 |
+
- 3636
|
387 |
+
- 3802
|
388 |
+
- 4709
|
389 |
+
- 5275
|
390 |
+
- 4862
|
391 |
+
- 3208
|
392 |
+
- 32
|
393 |
+
- 8
|
394 |
+
- 12
|
|
|
|
|
|
|
|
|
395 |
- 2
|
|
|
396 |
- 4
|
397 |
+
- 8
|
398 |
+
- 6
|
399 |
- 2
|
400 |
+
- 2
|
401 |
+
- 0
|
402 |
+
- 0
|
403 |
+
- 0
|
404 |
+
- 0
|
405 |
- 0
|
406 |
- 0
|
407 |
- 0
|
|
|
443 |
- 0
|
444 |
- 0
|
445 |
- 0
|
446 |
+
num_tokens_source: 5305413
|
447 |
+
num_tokens_target: 7842731
|
448 |
num_unks_source: 0
|
449 |
num_unks_target: 0
|
450 |
+
size_vocab_source: 45864
|
451 |
size_vocab_target: 656
|
452 |
eop_id: -1
|
453 |
max_seq_len_source: 513
|
|
|
479 |
factor_configs: null
|
480 |
num_embed: 512
|
481 |
num_factors: 1
|
482 |
+
vocab_size: 45864
|
483 |
config_embed_target: !EmbeddingConfig
|
484 |
allow_sparse_grad: false
|
485 |
dropout: 0.5
|
|
|
531 |
lhuc: false
|
532 |
neural_vocab_selection: null
|
533 |
neural_vocab_selection_block_loss: false
|
534 |
+
vocab_source_size: 45864
|
535 |
vocab_target_size: 656
|
536 |
weight_tying_type: trg_softmax
|
metrics
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
params.best
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c3cbabac4d28a48fbef952e0155e471556d607dda6700a919f08b44b49299597
|
3 |
+
size 278761214
|
vocab.src.0.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|