Upload folder using huggingface_hub

args.yaml

@@ -83,11 +83,11 @@ optimizer_betas:
 - 0.9
 - 0.999
 optimizer_eps: 1.0e-08
-output: /shares/volk.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-v2/model
+output: /shares/volk.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-tuned/model
 overwrite_output: false
 pad_vocab_to_multiple_of: 8
-params: null
-prepared_data: /shares/volk.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-v2/train_data
+params: /shares/volk.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-v4/model/params.best
+prepared_data: /shares/volk.cl.uzh/amoryo/checkpoints/signwriting-translation/spoken-to-signed/target-factors-tuned/train_data
 quiet: false
 quiet_secondary_workers: false
 seed: 1
@@ -152,14 +152,14 @@ transformer_preprocess:
 - n
 update_interval: 1
 use_cpu: false
-validation_source: /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/source.txt
+validation_source: /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/source.txt
 validation_source_factors: []
-validation_target: /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/target_0.txt
+validation_target: /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/target_0.txt
 validation_target_factors:
-- /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/target_1.txt
-- /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/target_2.txt
-- /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/target_3.txt
-- /home/amoryo/sign-language/signwriting-translation/parallel/spoken-to-signed/test/target_4.txt
+- /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/target_1.txt
+- /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/target_2.txt
+- /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/target_3.txt
+- /home/amoryo/sign-language/signwriting-translation/parallel-clean/spoken-to-signed/test/target_4.txt
 weight_decay: 0.0
 weight_tying_type: trg_softmax
 word_min_count:

config

@@ -2,29 +2,30 @@
 config_data: !DataConfig
   data_statistics: !DataStatistics
     average_len_target_per_bucket:
-    - 6.043404555083469
-    - 10.519883768009015
-    - 18.59887856456265
-    - 26.278671761476776
-    - 33.190033876079
-    - 41.83887781695534
-    - 51.24769950218732
-    - 59.783405313440575
-    - 68.16214942257864
-    - 75.68928386092358
-    - 82.77250900360164
-    - 18.47619047619048
-    - 8.058823529411766
-    - 5.647058823529413
-    - 9.5
-    - 8.499999999999998
-    - 5.0
-    - 6.444444444444445
-    - null
-    - 5.0
-    - null
-    - null
-    - null
+    - 6.01988730121871
+    - 10.49707202236441
+    - 18.536005711954232
+    - 25.91453940066591
+    - 33.44474802671516
+    - 41.29017292281736
+    - 50.38165137614682
+    - 59.57928802588991
+    - 67.86491420226359
+    - 75.76032388663967
+    - 82.68249075215772
+    - 12.0
+    - 23.333333333333332
+    - 24.88888888888889
+    - 15.857142857142856
+    - 8.6
+    - 8.0
+    - 6.5
+    - 4.5
+    - null
+    - 6.25
+    - 6.0
+    - 8.0
+    - 10.0
     - null
     - null
     - null
@@ -39,7 +40,6 @@ config_data: !DataConfig
     - null
     - null
     - null
-    - 6.0
     - null
     - null
     - null
@@ -263,67 +263,79 @@ config_data: !DataConfig
     - !!python/tuple
       - 513
       - 513
-    length_ratio_mean: 1.488763422684226
+    length_ratio_mean: 1.4422861755889766
     length_ratio_stats_per_bucket:
     - !!python/tuple
-      - 1.172606268110977
-      - 0.3198943487302755
+      - 1.1255982284370691
+      - 0.3093025169709561
     - !!python/tuple
-      - 1.7758876999918463
-      - 0.6634649712813119
+      - 1.6064942041128283
+      - 0.6417946634210556
     - !!python/tuple
-      - 2.667916151313991
-      - 1.011426597835219
+      - 2.359196679910085
+      - 0.9575904838242857
     - !!python/tuple
-      - 3.173004208160318
-      - 1.618051536035044
+      - 2.9607196399710274
+      - 1.5584344652106463
     - !!python/tuple
-      - 3.3799284759654693
-      - 2.4421821481902017
+      - 2.9543389770526325
+      - 2.297144978089179
     - !!python/tuple
-      - 3.5445340678088186
-      - 2.377062423666999
+      - 3.002340450312629
+      - 2.18681603522054
     - !!python/tuple
-      - 3.870005889139779
-      - 2.8520973384152954
+      - 3.081058473996043
+      - 2.5011199006515183
     - !!python/tuple
-      - 3.6032825677516436
-      - 2.6820622916732426
+      - 2.9163894753348045
+      - 2.2789742432670086
     - !!python/tuple
-      - 3.672787931128752
-      - 2.7023842818124146
+      - 2.9722600022754504
+      - 2.230352263818367
     - !!python/tuple
-      - 3.807871945448423
-      - 2.9154974163883414
+      - 3.138369081443168
+      - 2.4620594208603235
     - !!python/tuple
-      - 3.5339756014008112
-      - 2.253656007685683
+      - 2.854253034817575
+      - 1.6376340581023332
     - !!python/tuple
-      - 0.19909941868381043
-      - 0.2555613448334357
+      - 0.12974460739238317
+      - 0.160679251677897
     - !!python/tuple
-      - 0.08097343381798354
-      - 0.02276102142801515
+      - 0.23436597326420017
+      - 0.24015947073924906
     - !!python/tuple
-      - 0.05189939313382113
-      - 0.014368277929717775
+      - 0.23457181630569815
+      - 0.2675720247460546
     - !!python/tuple
-      - 0.08091857432628224
-      - 0.052824789190560516
+      - 0.1366625632624212
+      - 0.16557849394412172
     - !!python/tuple
-      - 0.06762976914376581
-      - 0.015451689277268104
+      - 0.06941451000342352
+      - 0.02389100863634256
     - !!python/tuple
-      - 0.03788218925017831
-      - 0.0054846792633349494
+      - 0.05970149253731343
+      - 0.0
     - !!python/tuple
-      - 0.0463036569542669
-      - 0.0098489446540238
+      - 0.04609929078014184
+      - 0.017730496453900707
+    - !!python/tuple
+      - 0.03063680118474639
+      - 0.0036097741577193615
     - &id001 !!python/tuple
       - null
       - null
     - !!python/tuple
-      - 0.032679738562091505
+      - 0.038657633805228714
+      - 0.010237907973024957
+    - !!python/tuple
+      - 0.03469146238377008
+      - 0.004700847851307781
+    - !!python/tuple
+      - 0.0446927374301676
+      - 0.0
+    - !!python/tuple
+      - 0.05405405405405406
       - 0.0
     - *id001
     - *id001
@@ -342,12 +354,6 @@ config_data: !DataConfig
     - *id001
     - *id001
     - *id001
-    - !!python/tuple
-      - 0.020066889632107024
-      - 0.0
-    - *id001
-    - *id001
-    - *id001
     - *id001
     - *id001
     - *id001
@@ -372,32 +378,36 @@ config_data: !DataConfig
     - *id001
     - *id001
     - *id001
-    length_ratio_std: 0.8554135166970332
-    max_observed_len_source: 299
+    length_ratio_std: 0.8479651722336846
+    max_observed_len_source: 185
     max_observed_len_target: 87
     num_discarded: 0
-    num_sents: 1874353
+    num_sents: 425392
     num_sents_per_bucket:
-    - 1183908
-    - 558194
-    - 62420
-    - 18581
-    - 9151
-    - 6523
-    - 6629
-    - 7641
-    - 8486
-    - 7708
-    - 4998
-    - 42
-    - 17
-    - 17
-    - 4
-    - 12
+    - 245433
+    - 135930
+    - 19608
+    - 7208
+    - 3294
+    - 2371
+    - 2180
+    - 2472
+    - 2739
+    - 2470
+    - 1622
+    - 20
     - 9
     - 9
+    - 7
+    - 5
+    - 1
+    - 2
+    - 2
     - 0
+    - 4
     - 3
+    - 1
+    - 2
     - 0
     - 0
     - 0
@@ -415,10 +425,6 @@ config_data: !DataConfig
     - 0
     - 0
     - 0
-    - 1
-    - 0
-    - 0
-    - 0
     - 0
     - 0
     - 0
@@ -443,11 +449,11 @@ config_data: !DataConfig
     - 0
     - 0
     - 0
-    num_tokens_source: 11815202
-    num_tokens_target: 17626167
+    num_tokens_source: 3056735
+    num_tokens_target: 4427820
     num_unks_source: 0
     num_unks_target: 0
-    size_vocab_source: 46256
+    size_vocab_source: 8016
     size_vocab_target: 656
   eop_id: -1
   max_seq_len_source: 513
@@ -479,7 +485,7 @@ config_embed_source: !EmbeddingConfig
   factor_configs: null
   num_embed: 512
   num_factors: 1
-  vocab_size: 46256
+  vocab_size: 8016
 config_embed_target: !EmbeddingConfig
   allow_sparse_grad: false
   dropout: 0.5
@@ -531,6 +537,6 @@ dtype: float32
 lhuc: false
 neural_vocab_selection: null
 neural_vocab_selection_block_loss: false
-vocab_source_size: 46256
+vocab_source_size: 8016
 vocab_target_size: 656
 weight_tying_type: trg_softmax

params.best

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c57f2a6fd2b40d982a122389c2fdd52c8d3bf1b2527fbd673cb75302803153fa
-size 279564030
+oid sha256:fbc40d947ce0a1c03679cc4c62a4a94c11b179a485637727fc1b491a063042c4
+size 201248510