Hugging Face's logo

al1808th
/
stanza-digphil

Model card Files
xet
Community
Albin Thörn Cleland commited on
Commit
b01c18c
·
1 Parent(s): d404cba

new models

Browse files
Files changed (11) hide show
  1. README.md +20 -8
  2. logs/latest.txt +1 -0
  3. logs/log_20251202_233139.txt +757 -0
  4. logs/log_diachronic.pt_baseline_batch16.txt +695 -0
  5. logs/log_diachronic.pt_sv_diachron_nn_20251203_001047.txt +190 -0
  6. logs/log_diachronic.pt_sv_diachron_nn_20251203_003001.txt +142 -0
  7. make_new_model.sh +17 -3
  8. prepare-train-val-test.py +5 -1
  9. saved_models/depparse/kubhist2_sv_diachron/sv_diachronic_charlm_parser.pt +3 -0
  10. saved_models/depparse/kubhist2_sv_diachron/sv_diachronic_charlm_parser_checkpoint.pt +3 -0
  11. ud-treebanks-is/is_pud-ud-test.conllu +3 -0
README.md CHANGED
@@ -1,15 +1,15 @@
1
- # Test
2
 
3
- - Train: hela den historiska maskinparsade korpusen `alanev52/Diachronic_Treebanks_DigPhil/stanza_results` plus alla fem svenska trädbanker från ud, och även de från bokmål
4
- - Dev: 10% av guld (`alanev52/Diachronic_Treebanks_DigPhil/stanza_results/validated`)
5
- - Test: 90% av guld
6
 
7
- <!-- tmux attach -t stanza ****-->
8
 
9
- Work flow:
 
 
10
 
11
  ```
12
- python prepare-train-val-test.py {language codes}
13
 
14
  source scripts/config_alvis.sh
15
 
@@ -18,11 +18,23 @@ python -m stanza.utils.datasets.prepare_depparse_treebank UD_Swedish-diachronic
18
  python -m stanza.utils.training.run_depparse UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt --batch_size 32 --dropout 0.33
19
  ```
20
 
 
 
 
 
 
 
 
 
 
 
 
 
21
  ## Pretrained vectors
22
 
23
  We use the incremental vectors up until 1880 from Henchen & Tahmasebi 2021.
24
 
25
- Jag konverterade först kubhist2-vektorerna från gensim fasttext .ft till en vanlig textfil med gensims pythonpaket, sedan använde jag stanzas konverterare till .pt:
26
 
27
  ```
28
  from stanza.models.common.pretrain import Pretrain
 
1
+ # Retraining Stanza to optimize depparse on a diachronic Swedish corpus
2
 
3
+ This repository contains forked code from the official Stanza github, with scripts that help prepare for and train models on different combinations of treebanks relevant to historical Swedish.
 
 
4
 
5
+ ## Guide
6
 
7
+ Dev/test for all models is 10/90 of our human validated gold sentences (https://github.com/alanev52/Diachronic_Treebanks_DigPhil/tree/main/parsed_data/validated)
8
+
9
+ Example work flow, training a model with pretrained vectors from kubhist2 1880 incremental and training data from Swedish and Bokmål treebanks, as well as our own machine-parsed diachronic corpus:
10
 
11
  ```
12
+ python prepare-train-val-test.py sv diachron bm
13
 
14
  source scripts/config_alvis.sh
15
 
 
18
  python -m stanza.utils.training.run_depparse UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt --batch_size 32 --dropout 0.33
19
  ```
20
 
21
+ All of the above can be done with a single command by using:
22
+
23
+ ```
24
+ ./make_new_model.sh {vectors} {language codes}
25
+ ```
26
+
27
+ which for the example becomes:
28
+
29
+ ```
30
+ ./make_new_model.sh diachronic.pt sv diachron bm
31
+ ```
32
+
33
  ## Pretrained vectors
34
 
35
  We use the incremental vectors up until 1880 from Henchen & Tahmasebi 2021.
36
 
37
+ Jag konverterade först kubhist2-vektorerna från `gensim` fasttext .ft till en vanlig textfil med gensims pythonpaket, sedan använde jag stanzas konverterare till .pt:
38
 
39
  ```
40
  from stanza.models.common.pretrain import Pretrain
logs/latest.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ log_diachronic.pt_sv_diachron_nn_20251203_001047.txt
logs/log_20251202_233139.txt ADDED
@@ -0,0 +1,757 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === LOGFILE: log_20251202_233139.txt ===
2
+ Language codes: sv diachron
3
+ Using pretrained model: diachronic.pt
4
+
5
+ Running: python prepare-train-val-test.py sv diachron
6
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-dev.conllu
7
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test.conllu
8
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_pud-ud-test.conllu
9
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-test.conllu
10
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test-trg.conllu
11
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-dev.conllu
12
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/ucxn_ud_swedish-talbanken.conllu
13
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-train.conllu
14
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_old-ud-test.conllu
15
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-train.conllu
16
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-test.conllu
17
+ Including DigPhil MACHINE in TRAIN (minus gold)…
18
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec330-GyllenborgC_SwenskaSpratthoken.conllu
19
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec254-CederborghF_BerattelseOmJohnHall.conllu
20
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec277-EnbomPU_MedborgeligtSkalde.conllu
21
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec268-DulciU_VitterhetsNojen3.conllu
22
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1063-spf220.conllu
23
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec397-AngeredStrandbergH_UnderSodernsSol.conllu
24
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec324-GranbergPA_Enslighetsalskaren.conllu
25
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec252-BremerF_Teckningar1.conllu
26
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec988-spf145.conllu
27
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec987-spf144.conllu
28
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec631-HasselskogN_HallaHallaGronkoping.conllu
29
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-letter141673-Stalhammar.conllu
30
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1033-spf190.conllu
31
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec25-Runius.conllu
32
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec486-SchwartzMS_BellmansSkor.conllu
33
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec452-NyblomH_FantasierFyra.conllu
34
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec613-EngstromA_StrindbergOchJag.conllu
35
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec208-Anonym_DetGrasligaMordet.conllu
36
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec639-HeidenstamV_Proletarfilosofiens.conllu
37
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1102-spf259.conllu
38
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec991-spf148.conllu
39
+ Cleaning TRAIN...
40
+ [REMOVED] sent_id=6 ERRORS=['Line 24: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 29: Invalid token ID or head', 'Token 30 has invalid head 24']
41
+ [REMOVED] sent_id=7_8 ERRORS=['Multiple roots found: [5, 10]']
42
+ [REMOVED] sent_id=30_31 ERRORS=['Multiple roots found: [3, 18]']
43
+ [REMOVED] sent_id=35 ERRORS=['Line 36: Invalid token ID or head']
44
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [1, 5]']
45
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [7, 20]']
46
+ [REMOVED] sent_id=8_9 ERRORS=['Multiple roots found: [24, 57]']
47
+ [REMOVED] sent_id=12_13 ERRORS=['Multiple roots found: [11, 16]']
48
+ [REMOVED] sent_id=124_split2 ERRORS=['Line 4: Invalid token ID or head', 'No root found', 'Token 1 has invalid head 4', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 6 has invalid head 4', 'Token 11 has invalid head 4', 'Token 15 has invalid head 4']
49
+ [REMOVED] sent_id=396 ERRORS=['Token 2: Missing form']
50
+ [REMOVED] sent_id=416 ERRORS=['Token 2: Missing form']
51
+ [REMOVED] sent_id=589 ERRORS=['Token 2: Missing form']
52
+ [REMOVED] sent_id=909 ERRORS=['Token 2: Missing form']
53
+ [REMOVED] sent_id=912 ERRORS=['Token 2: Missing form']
54
+ [REMOVED] sent_id=3_split1 ERRORS=['Multiple roots found: [4, 15, 17]']
55
+ [REMOVED] sent_id=3_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 8: Invalid token ID or head', 'Line 15: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1', 'Token 3 has invalid head 8', 'Token 4 has invalid head 8', 'Token 5 has invalid head 8', 'Token 7 has invalid head 8', 'Token 10 has invalid head 8', 'Token 13 has invalid head 8', 'Token 14 has invalid head 8']
56
+ [REMOVED] sent_id=3_4 ERRORS=['Multiple roots found: [1, 5]']
57
+ [REMOVED] sent_id=5_6 ERRORS=['Multiple roots found: [3, 24]']
58
+ [REMOVED] sent_id=11_12_13 ERRORS=['Multiple roots found: [5, 17, 25]']
59
+ [REMOVED] sent_id=119 ERRORS=['Token 2: Missing form']
60
+ [REMOVED] sent_id=179 ERRORS=['Token 2: Missing form']
61
+ [REMOVED] sent_id=188 ERRORS=['Token 2: Missing form']
62
+ [REMOVED] sent_id=223 ERRORS=['Token 2: Missing form']
63
+ [REMOVED] sent_id=268 ERRORS=['Token 2: Missing form']
64
+ [REMOVED] sent_id=325 ERRORS=['Token 2: Missing form']
65
+ [REMOVED] sent_id=388 ERRORS=['Token 2: Missing form']
66
+ [REMOVED] sent_id=399 ERRORS=['Token 2: Missing form']
67
+ [REMOVED] sent_id=475 ERRORS=['Token 2: Missing form']
68
+ [REMOVED] sent_id=505 ERRORS=['Token 2: Missing form']
69
+ [REMOVED] sent_id=520 ERRORS=['Token 2: Missing form']
70
+ [REMOVED] sent_id=562 ERRORS=['Token 2: Missing form']
71
+ [REMOVED] sent_id=669 ERRORS=['Token 2: Missing form']
72
+ [REMOVED] sent_id=711 ERRORS=['Token 2: Missing form']
73
+ [REMOVED] sent_id=731 ERRORS=['Token 2: Missing form']
74
+ [REMOVED] sent_id=867 ERRORS=['Token 2: Missing form']
75
+ [REMOVED] sent_id=884 ERRORS=['Token 2: Missing form']
76
+ [REMOVED] sent_id=923 ERRORS=['Token 2: Missing form']
77
+ [REMOVED] sent_id=939 ERRORS=['Token 2: Missing form']
78
+ [REMOVED] sent_id=1086 ERRORS=['Token 2: Missing form']
79
+ [REMOVED] sent_id=1179 ERRORS=['Token 2: Missing form']
80
+ [REMOVED] sent_id=1251 ERRORS=['Token 2: Missing form']
81
+ [REMOVED] sent_id=1345 ERRORS=['Token 2: Missing form']
82
+ [REMOVED] sent_id=1459 ERRORS=['Token 2: Missing form']
83
+ [REMOVED] sent_id=1656 ERRORS=['Token 2: Missing form']
84
+ [REMOVED] sent_id=1669 ERRORS=['Token 2: Missing form']
85
+ [REMOVED] sent_id=87_88 ERRORS=['Multiple roots found: [3, 6]']
86
+ [REMOVED] sent_id=65_split2_66_split2 ERRORS=['Line 4: Invalid token ID or head', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 5 has invalid head 4']
87
+ [REMOVED] sent_id=25 ERRORS=['Token 2: Missing form']
88
+ [REMOVED] sent_id=136 ERRORS=['Token 2: Missing form']
89
+ [REMOVED] sent_id=208 ERRORS=['Token 2: Missing form']
90
+ [REMOVED] sent_id=230 ERRORS=['Token 2: Missing form']
91
+ [REMOVED] sent_id=245 ERRORS=['Token 2: Missing form']
92
+ [REMOVED] sent_id=276 ERRORS=['Token 2: Missing form']
93
+ [REMOVED] sent_id=320 ERRORS=['Token 2: Missing form']
94
+ [REMOVED] sent_id=366 ERRORS=['Token 2: Missing form']
95
+ [REMOVED] sent_id=519 ERRORS=['Token 2: Missing form']
96
+ [REMOVED] sent_id=569 ERRORS=['Token 2: Missing form']
97
+ [REMOVED] sent_id=50_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 6: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1']
98
+ [REMOVED] sent_id=53_54 ERRORS=['Multiple roots found: [27, 91]']
99
+ [REMOVED] sent_id=55_56_57 ERRORS=['Multiple roots found: [2, 4, 13]']
100
+ [REMOVED] sent_id=17_split1 ERRORS=['Multiple roots found: [2, 14, 17]']
101
+ [REMOVED] sent_id=17_split2 ERRORS=['Line 8: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 38: Invalid token ID or head', 'No root found', 'Token 3 has invalid head 8', 'Token 7 has invalid head 8', 'Token 9 has invalid head 8', 'Token 10 has invalid head 8', 'Token 17 has invalid head 8', 'Token 22 has invalid head 25', 'Token 23 has invalid head 25', 'Token 24 has invalid head 25', 'Token 26 has invalid head 25', 'Token 27 has invalid head 25', 'Token 28 has invalid head 25']
102
+ [REMOVED] sent_id=19_split1 ERRORS=['Multiple roots found: [3, 31]']
103
+ Cleaning DEV...
104
+ [REMOVED] sent_id=33 ERRORS=['Token 15: Missing deprel']
105
+ Cleaning TEST...
106
+ Writing TRAIN → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu (66252 valid sentences)
107
+ Writing DEV → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu (9 valid sentences)
108
+ Writing TEST → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu (99 valid sentences)
109
+ Done.
110
+ Sourcing scripts/config_alvis.sh
111
+ Running stanza dataset preparation…
112
+ 2025-12-02 23:31:48 INFO: Datasets program called with:
113
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/datasets/prepare_depparse_treebank.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
114
+ 2025-12-02 23:31:48 DEBUG: Downloading resource file from https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.11.0.json
115
+
116
+ 2025-12-02 23:31:48 INFO: Downloaded file to /cephyr/users/cleland/Alvis/stanza_resources/resources.json
117
+ 2025-12-02 23:31:48 DEBUG: Processing parameter "processors"...
118
+ 2025-12-02 23:31:48 WARNING: Can not find pos: diachronic from official model list. Ignoring it.
119
+ 2025-12-02 23:31:48 INFO: Downloading these customized packages for language: sv (Swedish)...
120
+ =======================
121
+ | Processor | Package |
122
+ -----------------------
123
+ =======================
124
+
125
+ 2025-12-02 23:31:48 INFO: Finished downloading models and saved to /cephyr/users/cleland/Alvis/stanza_resources
126
+ 2025-12-02 23:31:48 INFO: Using tagger model in /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt for sv_diachronic
127
+ 2025-12-02 23:31:48 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
128
+ 2025-12-02 23:31:48 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
129
+ Augmented 192 quotes: Counter({'„”': 28, '""': 27, '「」': 20, '″″': 20, '»«': 18, '«»': 18, '《》': 17, '””': 17, '„“': 15, '““': 12})
130
+ 2025-12-02 23:31:51 INFO: Running tagger to retag /local/tmp.5437491/tmp11l0ml41/sv_diachronic.train.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu
131
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmp11l0ml41/sv_diachronic.train.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu']
132
+ 2025-12-02 23:31:51 INFO: Running tagger in predict mode
133
+ 2025-12-02 23:31:51 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
134
+ 2025-12-02 23:31:54 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
135
+ 2025-12-02 23:31:54 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
136
+ 2025-12-02 23:31:54 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
137
+ 2025-12-02 23:31:54 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
138
+ 2025-12-02 23:31:54 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
139
+ 2025-12-02 23:31:57 INFO: Loading data with batch size 250...
140
+ 2025-12-02 23:32:35 INFO: Start evaluation...
141
+ 2025-12-02 23:36:08 INFO: UPOS XPOS UFeats AllTags
142
+ 2025-12-02 23:36:08 INFO: 94.98 80.23 90.43 78.41
143
+ 2025-12-02 23:36:08 INFO: POS Tagger score: sv_diachronic 78.41
144
+ 2025-12-02 23:36:08 INFO: Running tagger to retag /local/tmp.5437491/tmp11l0ml41/sv_diachronic.dev.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu
145
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmp11l0ml41/sv_diachronic.dev.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu']
146
+ 2025-12-02 23:36:08 INFO: Running tagger in predict mode
147
+ 2025-12-02 23:36:08 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
148
+ 2025-12-02 23:36:11 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
149
+ 2025-12-02 23:36:11 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
150
+ 2025-12-02 23:36:11 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
151
+ 2025-12-02 23:36:11 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
152
+ 2025-12-02 23:36:11 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
153
+ 2025-12-02 23:36:11 INFO: Loading data with batch size 250...
154
+ 2025-12-02 23:36:11 INFO: Start evaluation...
155
+ 2025-12-02 23:36:11 INFO: UPOS XPOS UFeats AllTags
156
+ 2025-12-02 23:36:11 INFO: 88.61 79.46 83.42 74.75
157
+ 2025-12-02 23:36:11 INFO: POS Tagger score: sv_diachronic 74.75
158
+ 2025-12-02 23:36:11 INFO: Running tagger to retag /local/tmp.5437491/tmp11l0ml41/sv_diachronic.test.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu
159
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmp11l0ml41/sv_diachronic.test.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu']
160
+ 2025-12-02 23:36:11 INFO: Running tagger in predict mode
161
+ 2025-12-02 23:36:11 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
162
+ 2025-12-02 23:36:14 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
163
+ 2025-12-02 23:36:14 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
164
+ 2025-12-02 23:36:14 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
165
+ 2025-12-02 23:36:14 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
166
+ 2025-12-02 23:36:14 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
167
+ 2025-12-02 23:36:14 INFO: Loading data with batch size 250...
168
+ 2025-12-02 23:36:14 INFO: Start evaluation...
169
+ 2025-12-02 23:36:15 INFO: UPOS XPOS UFeats AllTags
170
+ 2025-12-02 23:36:15 INFO: 89.61 86.64 86.69 81.18
171
+ 2025-12-02 23:36:15 INFO: POS Tagger score: sv_diachronic 81.18
172
+ Preparing data for UD_Swedish-diachronic: sv_diachronic, sv
173
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu and writing to /local/tmp.5437491/tmp11l0ml41/sv_diachronic.train.gold.conllu
174
+ Swapped 'w1, w2' for 'w1 ,w2' 132 times
175
+ Added 506 new sentences with asdf, zzzz -> asdf,zzzz
176
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu and writing to /local/tmp.5437491/tmp11l0ml41/sv_diachronic.dev.gold.conllu
177
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu and writing to /local/tmp.5437491/tmp11l0ml41/sv_diachronic.test.gold.conllu
178
+ Running stanza dependency parser training…
179
+ 2025-12-02 23:36:29 INFO: Training program called with:
180
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/training/run_depparse.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt --batch_size 32 --dropout 0.33
181
+ 2025-12-02 23:36:29 DEBUG: UD_Swedish-diachronic: sv_diachronic
182
+ 2025-12-02 23:36:29 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
183
+ 2025-12-02 23:36:29 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
184
+ 2025-12-02 23:36:29 INFO: UD_Swedish-diachronic: saved_models/depparse/sv_diachronic_charlm_parser.pt does not exist, training new model
185
+ 2025-12-02 23:36:29 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
186
+ 2025-12-02 23:36:29 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
187
+ 2025-12-02 23:36:29 INFO: Running train depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--train_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu', '--batch_size', '5000', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'train', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '32', '--dropout', '0.33']
188
+ 2025-12-02 23:36:29 INFO: Running parser in train mode
189
+ 2025-12-02 23:36:29 INFO: Using pretrained contextualized char embedding
190
+ 2025-12-02 23:36:29 INFO: Loading data with batch size 32...
191
+ 2025-12-02 23:36:36 INFO: Train File /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu, Data Size: 66758
192
+ 2025-12-02 23:36:36 INFO: Original data size: 66758
193
+ 2025-12-02 23:36:37 INFO: Augmented data size: 66897
194
+ 2025-12-02 23:36:54 WARNING: sv_diachronic is not a known dataset. Examining the data to choose which xpos vocab to use
195
+ 2025-12-02 23:36:54 INFO: Original length = 66897
196
+ 2025-12-02 23:36:54 INFO: Filtered length = 66897
197
+ 2025-12-02 23:37:11 WARNING: Chose XPOSDescription(xpos_type=<XPOSType.XPOS: 1>, sep='|') for the xpos factory for sv_diachronic
198
+ 2025-12-02 23:37:17 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
199
+ 2025-12-02 23:37:31 DEBUG: 38850 batches created.
200
+ 2025-12-02 23:37:31 DEBUG: 9 batches created.
201
+ 2025-12-02 23:37:31 INFO: Training parser...
202
+ 2025-12-02 23:37:31 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
203
+ 2025-12-02 23:37:31 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
204
+ 2025-12-02 23:37:31 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
205
+ 2025-12-02 23:37:32 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
206
+ 2025-12-02 23:37:40 INFO: Finished STEP 20/50000, loss = 6.609651 (0.199 sec/batch), lr: 0.003000
207
+ 2025-12-02 23:37:43 INFO: Finished STEP 40/50000, loss = 5.660924 (0.183 sec/batch), lr: 0.003000
208
+ 2025-12-02 23:37:47 INFO: Finished STEP 60/50000, loss = 5.496554 (0.143 sec/batch), lr: 0.003000
209
+ 2025-12-02 23:37:50 INFO: Finished STEP 80/50000, loss = 5.735727 (0.140 sec/batch), lr: 0.003000
210
+ 2025-12-02 23:37:52 INFO: Finished STEP 100/50000, loss = 4.814171 (0.123 sec/batch), lr: 0.003000
211
+ 2025-12-02 23:37:52 INFO: Evaluating on dev set...
212
+ 2025-12-02 23:37:53 INFO: LAS MLAS BLEX
213
+ 2025-12-02 23:37:53 INFO: 35.89 25.42 27.50
214
+ 2025-12-02 23:37:53 INFO: step 100: train_loss = 8.162946, dev_score = 0.3589
215
+ 2025-12-02 23:37:53 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
216
+ 2025-12-02 23:37:53 INFO: new best model saved.
217
+ 2025-12-02 23:37:54 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
218
+ 2025-12-02 23:37:54 INFO: new model checkpoint saved.
219
+ 2025-12-02 23:37:57 INFO: Finished STEP 120/50000, loss = 4.111693 (0.122 sec/batch), lr: 0.003000
220
+ 2025-12-02 23:37:59 INFO: Finished STEP 140/50000, loss = 4.873629 (0.131 sec/batch), lr: 0.003000
221
+ 2025-12-02 23:38:02 INFO: Finished STEP 160/50000, loss = 4.031731 (0.137 sec/batch), lr: 0.003000
222
+ 2025-12-02 23:38:04 INFO: Finished STEP 180/50000, loss = 5.186968 (0.121 sec/batch), lr: 0.003000
223
+ 2025-12-02 23:38:07 INFO: Finished STEP 200/50000, loss = 4.386778 (0.120 sec/batch), lr: 0.003000
224
+ 2025-12-02 23:38:07 INFO: Evaluating on dev set...
225
+ 2025-12-02 23:38:07 INFO: LAS MLAS BLEX
226
+ 2025-12-02 23:38:07 INFO: 47.28 31.93 36.55
227
+ 2025-12-02 23:38:07 INFO: step 200: train_loss = 5.009766, dev_score = 0.4728
228
+ 2025-12-02 23:38:07 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
229
+ 2025-12-02 23:38:07 INFO: new best model saved.
230
+ 2025-12-02 23:38:08 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
231
+ 2025-12-02 23:38:08 INFO: new model checkpoint saved.
232
+ 2025-12-02 23:38:10 INFO: Finished STEP 220/50000, loss = 4.561248 (0.126 sec/batch), lr: 0.003000
233
+ 2025-12-02 23:38:13 INFO: Finished STEP 240/50000, loss = 4.298016 (0.114 sec/batch), lr: 0.003000
234
+ 2025-12-02 23:38:15 INFO: Finished STEP 260/50000, loss = 4.667103 (0.113 sec/batch), lr: 0.003000
235
+ 2025-12-02 23:38:17 INFO: Finished STEP 280/50000, loss = 5.273771 (0.110 sec/batch), lr: 0.003000
236
+ 2025-12-02 23:38:20 INFO: Finished STEP 300/50000, loss = 3.793263 (0.115 sec/batch), lr: 0.003000
237
+ 2025-12-02 23:38:20 INFO: Evaluating on dev set...
238
+ 2025-12-02 23:38:20 INFO: LAS MLAS BLEX
239
+ 2025-12-02 23:38:20 INFO: 47.77 35.52 40.59
240
+ 2025-12-02 23:38:20 INFO: step 300: train_loss = 4.689294, dev_score = 0.4777
241
+ 2025-12-02 23:38:20 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
242
+ 2025-12-02 23:38:20 INFO: new best model saved.
243
+ 2025-12-02 23:38:21 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
244
+ 2025-12-02 23:38:21 INFO: new model checkpoint saved.
245
+ 2025-12-02 23:38:23 INFO: Finished STEP 320/50000, loss = 4.029802 (0.110 sec/batch), lr: 0.003000
246
+ 2025-12-02 23:38:25 INFO: Finished STEP 340/50000, loss = 4.041455 (0.099 sec/batch), lr: 0.003000
247
+ 2025-12-02 23:38:28 INFO: Finished STEP 360/50000, loss = 3.786230 (0.097 sec/batch), lr: 0.003000
248
+ 2025-12-02 23:38:30 INFO: Finished STEP 380/50000, loss = 5.275126 (0.095 sec/batch), lr: 0.003000
249
+ 2025-12-02 23:38:32 INFO: Finished STEP 400/50000, loss = 4.217500 (0.096 sec/batch), lr: 0.003000
250
+ 2025-12-02 23:38:32 INFO: Evaluating on dev set...
251
+ 2025-12-02 23:38:32 INFO: LAS MLAS BLEX
252
+ 2025-12-02 23:38:32 INFO: 54.21 42.62 46.84
253
+ 2025-12-02 23:38:32 INFO: step 400: train_loss = 4.364881, dev_score = 0.5421
254
+ 2025-12-02 23:38:33 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
255
+ 2025-12-02 23:38:33 INFO: new best model saved.
256
+ 2025-12-02 23:38:33 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
257
+ 2025-12-02 23:38:33 INFO: new model checkpoint saved.
258
+ 2025-12-02 23:38:35 INFO: Finished STEP 420/50000, loss = 4.242122 (0.112 sec/batch), lr: 0.003000
259
+ 2025-12-02 23:38:37 INFO: Finished STEP 440/50000, loss = 6.075905 (0.101 sec/batch), lr: 0.003000
260
+ 2025-12-02 23:38:40 INFO: Finished STEP 460/50000, loss = 4.212942 (0.108 sec/batch), lr: 0.003000
261
+ 2025-12-02 23:38:42 INFO: Finished STEP 480/50000, loss = 4.493719 (0.098 sec/batch), lr: 0.003000
262
+ 2025-12-02 23:38:44 INFO: Finished STEP 500/50000, loss = 3.897269 (0.105 sec/batch), lr: 0.003000
263
+ 2025-12-02 23:38:44 INFO: Evaluating on dev set...
264
+ 2025-12-02 23:38:44 INFO: LAS MLAS BLEX
265
+ 2025-12-02 23:38:44 INFO: 55.69 46.41 48.95
266
+ 2025-12-02 23:38:44 INFO: step 500: train_loss = 4.347657, dev_score = 0.5569
267
+ 2025-12-02 23:38:44 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
268
+ 2025-12-02 23:38:44 INFO: new best model saved.
269
+ 2025-12-02 23:38:45 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
270
+ 2025-12-02 23:38:45 INFO: new model checkpoint saved.
271
+ 2025-12-02 23:38:47 INFO: Finished STEP 520/50000, loss = 5.176975 (0.099 sec/batch), lr: 0.003000
272
+ 2025-12-02 23:38:49 INFO: Finished STEP 540/50000, loss = 3.785196 (0.094 sec/batch), lr: 0.003000
273
+ 2025-12-02 23:38:51 INFO: Finished STEP 560/50000, loss = 3.917483 (0.102 sec/batch), lr: 0.003000
274
+ 2025-12-02 23:38:53 INFO: Finished STEP 580/50000, loss = 3.429865 (0.094 sec/batch), lr: 0.003000
275
+ 2025-12-02 23:38:55 INFO: Finished STEP 600/50000, loss = 3.838296 (0.099 sec/batch), lr: 0.003000
276
+ 2025-12-02 23:38:55 INFO: Evaluating on dev set...
277
+ 2025-12-02 23:38:55 INFO: LAS MLAS BLEX
278
+ 2025-12-02 23:38:55 INFO: 56.68 46.54 50.73
279
+ 2025-12-02 23:38:55 INFO: step 600: train_loss = 3.972618, dev_score = 0.5668
280
+ 2025-12-02 23:38:55 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
281
+ 2025-12-02 23:38:55 INFO: new best model saved.
282
+ 2025-12-02 23:38:56 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
283
+ 2025-12-02 23:38:56 INFO: new model checkpoint saved.
284
+ 2025-12-02 23:38:58 INFO: Finished STEP 620/50000, loss = 4.540798 (0.098 sec/batch), lr: 0.003000
285
+ 2025-12-02 23:39:00 INFO: Finished STEP 640/50000, loss = 3.424882 (0.096 sec/batch), lr: 0.003000
286
+ 2025-12-02 23:39:02 INFO: Finished STEP 660/50000, loss = 3.369433 (0.101 sec/batch), lr: 0.003000
287
+ 2025-12-02 23:39:04 INFO: Finished STEP 680/50000, loss = 3.264027 (0.101 sec/batch), lr: 0.003000
288
+ 2025-12-02 23:39:06 INFO: Finished STEP 700/50000, loss = 3.240215 (0.096 sec/batch), lr: 0.003000
289
+ 2025-12-02 23:39:06 INFO: Evaluating on dev set...
290
+ 2025-12-02 23:39:06 INFO: LAS MLAS BLEX
291
+ 2025-12-02 23:39:06 INFO: 58.42 47.90 52.94
292
+ 2025-12-02 23:39:06 INFO: step 700: train_loss = 4.115382, dev_score = 0.5842
293
+ 2025-12-02 23:39:06 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
294
+ 2025-12-02 23:39:06 INFO: new best model saved.
295
+ 2025-12-02 23:39:07 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
296
+ 2025-12-02 23:39:07 INFO: new model checkpoint saved.
297
+ 2025-12-02 23:39:09 INFO: Finished STEP 720/50000, loss = 4.084485 (0.099 sec/batch), lr: 0.003000
298
+ 2025-12-02 23:39:11 INFO: Finished STEP 740/50000, loss = 3.427663 (0.092 sec/batch), lr: 0.003000
299
+ 2025-12-02 23:39:12 INFO: Finished STEP 760/50000, loss = 3.464590 (0.094 sec/batch), lr: 0.003000
300
+ 2025-12-02 23:39:14 INFO: Finished STEP 780/50000, loss = 4.105500 (0.099 sec/batch), lr: 0.003000
301
+ 2025-12-02 23:39:16 INFO: Finished STEP 800/50000, loss = 3.914870 (0.092 sec/batch), lr: 0.003000
302
+ 2025-12-02 23:39:16 INFO: Evaluating on dev set...
303
+ 2025-12-02 23:39:17 INFO: LAS MLAS BLEX
304
+ 2025-12-02 23:39:17 INFO: 57.67 48.74 51.26
305
+ 2025-12-02 23:39:17 INFO: step 800: train_loss = 3.887074, dev_score = 0.5767
306
+ 2025-12-02 23:39:17 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
307
+ 2025-12-02 23:39:17 INFO: new model checkpoint saved.
308
+ 2025-12-02 23:39:19 INFO: Finished STEP 820/50000, loss = 6.553721 (0.090 sec/batch), lr: 0.003000
309
+ 2025-12-02 23:39:21 INFO: Finished STEP 840/50000, loss = 3.608489 (0.091 sec/batch), lr: 0.003000
310
+ 2025-12-02 23:39:23 INFO: Finished STEP 860/50000, loss = 3.870910 (0.091 sec/batch), lr: 0.003000
311
+ 2025-12-02 23:39:24 INFO: Finished STEP 880/50000, loss = 4.438069 (0.096 sec/batch), lr: 0.003000
312
+ 2025-12-02 23:39:26 INFO: Finished STEP 900/50000, loss = 4.459882 (0.088 sec/batch), lr: 0.003000
313
+ 2025-12-02 23:39:26 INFO: Evaluating on dev set...
314
+ 2025-12-02 23:39:27 INFO: LAS MLAS BLEX
315
+ 2025-12-02 23:39:27 INFO: 55.94 48.03 50.52
316
+ 2025-12-02 23:39:27 INFO: step 900: train_loss = 4.130319, dev_score = 0.5594
317
+ 2025-12-02 23:39:27 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
318
+ 2025-12-02 23:39:27 INFO: new model checkpoint saved.
319
+ 2025-12-02 23:39:29 INFO: Finished STEP 920/50000, loss = 2.915673 (0.089 sec/batch), lr: 0.003000
320
+ 2025-12-02 23:39:31 INFO: Finished STEP 940/50000, loss = 5.318740 (0.094 sec/batch), lr: 0.003000
321
+ 2025-12-02 23:39:33 INFO: Finished STEP 960/50000, loss = 4.443627 (0.084 sec/batch), lr: 0.003000
322
+ 2025-12-02 23:39:34 INFO: Finished STEP 980/50000, loss = 4.706639 (0.085 sec/batch), lr: 0.003000
323
+ 2025-12-02 23:39:36 INFO: Finished STEP 1000/50000, loss = 3.425963 (0.094 sec/batch), lr: 0.003000
324
+ 2025-12-02 23:39:36 INFO: Evaluating on dev set...
325
+ 2025-12-02 23:39:37 INFO: LAS MLAS BLEX
326
+ 2025-12-02 23:39:37 INFO: 56.44 48.12 52.30
327
+ 2025-12-02 23:39:37 INFO: step 1000: train_loss = 3.999841, dev_score = 0.5644
328
+ 2025-12-02 23:39:37 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
329
+ 2025-12-02 23:39:37 INFO: new model checkpoint saved.
330
+ 2025-12-02 23:39:39 INFO: Finished STEP 1020/50000, loss = 4.753314 (0.094 sec/batch), lr: 0.003000
331
+ 2025-12-02 23:39:41 INFO: Finished STEP 1040/50000, loss = 4.193702 (0.086 sec/batch), lr: 0.003000
332
+ 2025-12-02 23:39:43 INFO: Finished STEP 1060/50000, loss = 4.332916 (0.084 sec/batch), lr: 0.003000
333
+ 2025-12-02 23:39:44 INFO: Finished STEP 1080/50000, loss = 4.260685 (0.075 sec/batch), lr: 0.003000
334
+ 2025-12-02 23:39:46 INFO: Finished STEP 1100/50000, loss = 3.021253 (0.088 sec/batch), lr: 0.003000
335
+ 2025-12-02 23:39:46 INFO: Evaluating on dev set...
336
+ 2025-12-02 23:39:46 INFO: LAS MLAS BLEX
337
+ 2025-12-02 23:39:46 INFO: 58.66 48.52 52.74
338
+ 2025-12-02 23:39:46 INFO: step 1100: train_loss = 3.822462, dev_score = 0.5866
339
+ 2025-12-02 23:39:47 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
340
+ 2025-12-02 23:39:47 INFO: new best model saved.
341
+ 2025-12-02 23:39:47 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
342
+ 2025-12-02 23:39:47 INFO: new model checkpoint saved.
343
+ 2025-12-02 23:39:49 INFO: Finished STEP 1120/50000, loss = 4.467305 (0.085 sec/batch), lr: 0.003000
344
+ 2025-12-02 23:39:51 INFO: Finished STEP 1140/50000, loss = 2.573361 (0.089 sec/batch), lr: 0.003000
345
+ 2025-12-02 23:39:52 INFO: Finished STEP 1160/50000, loss = 4.686720 (0.082 sec/batch), lr: 0.003000
346
+ 2025-12-02 23:39:54 INFO: Finished STEP 1180/50000, loss = 6.170751 (0.078 sec/batch), lr: 0.003000
347
+ 2025-12-02 23:39:56 INFO: Finished STEP 1200/50000, loss = 3.310861 (0.078 sec/batch), lr: 0.003000
348
+ 2025-12-02 23:39:56 INFO: Evaluating on dev set...
349
+ 2025-12-02 23:39:56 INFO: LAS MLAS BLEX
350
+ 2025-12-02 23:39:56 INFO: 59.41 49.68 54.74
351
+ 2025-12-02 23:39:56 INFO: step 1200: train_loss = 3.786507, dev_score = 0.5941
352
+ 2025-12-02 23:39:56 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
353
+ 2025-12-02 23:39:56 INFO: new best model saved.
354
+ 2025-12-02 23:39:57 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
355
+ 2025-12-02 23:39:57 INFO: new model checkpoint saved.
356
+ 2025-12-02 23:39:59 INFO: Finished STEP 1220/50000, loss = 2.933485 (0.087 sec/batch), lr: 0.003000
357
+ 2025-12-02 23:40:00 INFO: Finished STEP 1240/50000, loss = 3.925108 (0.084 sec/batch), lr: 0.003000
358
+ 2025-12-02 23:40:02 INFO: Finished STEP 1260/50000, loss = 5.419237 (0.091 sec/batch), lr: 0.003000
359
+ 2025-12-02 23:40:04 INFO: Finished STEP 1280/50000, loss = 3.987858 (0.095 sec/batch), lr: 0.003000
360
+ 2025-12-02 23:40:05 INFO: Finished STEP 1300/50000, loss = 4.217699 (0.083 sec/batch), lr: 0.003000
361
+ 2025-12-02 23:40:05 INFO: Evaluating on dev set...
362
+ 2025-12-02 23:40:06 INFO: LAS MLAS BLEX
363
+ 2025-12-02 23:40:06 INFO: 59.16 49.89 53.28
364
+ 2025-12-02 23:40:06 INFO: step 1300: train_loss = 3.928305, dev_score = 0.5916
365
+ 2025-12-02 23:40:07 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
366
+ 2025-12-02 23:40:07 INFO: new model checkpoint saved.
367
+ 2025-12-02 23:40:08 INFO: Finished STEP 1320/50000, loss = 4.228859 (0.076 sec/batch), lr: 0.003000
368
+ 2025-12-02 23:40:10 INFO: Finished STEP 1340/50000, loss = 4.518141 (0.078 sec/batch), lr: 0.003000
369
+ 2025-12-02 23:40:11 INFO: Finished STEP 1360/50000, loss = 3.032430 (0.087 sec/batch), lr: 0.003000
370
+ 2025-12-02 23:40:13 INFO: Finished STEP 1380/50000, loss = 3.127702 (0.085 sec/batch), lr: 0.003000
371
+ 2025-12-02 23:40:15 INFO: Finished STEP 1400/50000, loss = 2.618764 (0.081 sec/batch), lr: 0.003000
372
+ 2025-12-02 23:40:15 INFO: Evaluating on dev set...
373
+ 2025-12-02 23:40:15 INFO: LAS MLAS BLEX
374
+ 2025-12-02 23:40:15 INFO: 59.41 50.63 54.43
375
+ 2025-12-02 23:40:15 INFO: step 1400: train_loss = 3.877352, dev_score = 0.5941
376
+ 2025-12-02 23:40:15 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
377
+ 2025-12-02 23:40:15 INFO: new best model saved.
378
+ 2025-12-02 23:40:16 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
379
+ 2025-12-02 23:40:16 INFO: new model checkpoint saved.
380
+ 2025-12-02 23:40:18 INFO: Finished STEP 1420/50000, loss = 3.439885 (0.088 sec/batch), lr: 0.003000
381
+ 2025-12-02 23:40:19 INFO: Finished STEP 1440/50000, loss = 2.931584 (0.082 sec/batch), lr: 0.003000
382
+ 2025-12-02 23:40:21 INFO: Finished STEP 1460/50000, loss = 4.533800 (0.081 sec/batch), lr: 0.003000
383
+ 2025-12-02 23:40:22 INFO: Finished STEP 1480/50000, loss = 3.510649 (0.084 sec/batch), lr: 0.003000
384
+ 2025-12-02 23:40:24 INFO: Finished STEP 1500/50000, loss = 4.086914 (0.079 sec/batch), lr: 0.003000
385
+ 2025-12-02 23:40:24 INFO: Evaluating on dev set...
386
+ 2025-12-02 23:40:24 INFO: LAS MLAS BLEX
387
+ 2025-12-02 23:40:24 INFO: 59.41 51.15 54.09
388
+ 2025-12-02 23:40:24 INFO: step 1500: train_loss = 3.849247, dev_score = 0.5941
389
+ 2025-12-02 23:40:25 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
390
+ 2025-12-02 23:40:25 INFO: new best model saved.
391
+ 2025-12-02 23:40:25 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
392
+ 2025-12-02 23:40:25 INFO: new model checkpoint saved.
393
+ 2025-12-02 23:40:27 INFO: Finished STEP 1520/50000, loss = 3.263284 (0.080 sec/batch), lr: 0.003000
394
+ 2025-12-02 23:40:28 INFO: Finished STEP 1540/50000, loss = 3.815128 (0.078 sec/batch), lr: 0.003000
395
+ 2025-12-02 23:40:30 INFO: Finished STEP 1560/50000, loss = 4.004013 (0.071 sec/batch), lr: 0.003000
396
+ 2025-12-02 23:40:31 INFO: Finished STEP 1580/50000, loss = 5.315696 (0.070 sec/batch), lr: 0.003000
397
+ 2025-12-02 23:40:33 INFO: Finished STEP 1600/50000, loss = 2.390335 (0.074 sec/batch), lr: 0.003000
398
+ 2025-12-02 23:40:33 INFO: Evaluating on dev set...
399
+ 2025-12-02 23:40:34 INFO: LAS MLAS BLEX
400
+ 2025-12-02 23:40:34 INFO: 58.91 50.00 54.24
401
+ 2025-12-02 23:40:34 INFO: step 1600: train_loss = 3.795422, dev_score = 0.5891
402
+ 2025-12-02 23:40:34 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
403
+ 2025-12-02 23:40:34 INFO: new model checkpoint saved.
404
+ 2025-12-02 23:40:36 INFO: Finished STEP 1620/50000, loss = 5.436863 (0.070 sec/batch), lr: 0.003000
405
+ 2025-12-02 23:40:37 INFO: Finished STEP 1640/50000, loss = 5.494586 (0.084 sec/batch), lr: 0.003000
406
+ 2025-12-02 23:40:39 INFO: Finished STEP 1660/50000, loss = 3.680394 (0.074 sec/batch), lr: 0.003000
407
+ 2025-12-02 23:40:40 INFO: Finished STEP 1680/50000, loss = 3.352411 (0.072 sec/batch), lr: 0.003000
408
+ 2025-12-02 23:40:42 INFO: Finished STEP 1700/50000, loss = 3.811974 (0.083 sec/batch), lr: 0.003000
409
+ 2025-12-02 23:40:42 INFO: Evaluating on dev set...
410
+ 2025-12-02 23:40:42 INFO: LAS MLAS BLEX
411
+ 2025-12-02 23:40:42 INFO: 59.41 48.64 52.41
412
+ 2025-12-02 23:40:42 INFO: step 1700: train_loss = 3.939334, dev_score = 0.5941
413
+ 2025-12-02 23:40:43 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
414
+ 2025-12-02 23:40:43 INFO: new best model saved.
415
+ 2025-12-02 23:40:43 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
416
+ 2025-12-02 23:40:43 INFO: new model checkpoint saved.
417
+ 2025-12-02 23:40:45 INFO: Finished STEP 1720/50000, loss = 3.031429 (0.079 sec/batch), lr: 0.003000
418
+ 2025-12-02 23:40:46 INFO: Finished STEP 1740/50000, loss = 3.096446 (0.076 sec/batch), lr: 0.003000
419
+ 2025-12-02 23:40:48 INFO: Finished STEP 1760/50000, loss = 3.194870 (0.081 sec/batch), lr: 0.003000
420
+ 2025-12-02 23:40:49 INFO: Finished STEP 1780/50000, loss = 2.917735 (0.075 sec/batch), lr: 0.003000
421
+ 2025-12-02 23:40:51 INFO: Finished STEP 1800/50000, loss = 4.208848 (0.072 sec/batch), lr: 0.003000
422
+ 2025-12-02 23:40:51 INFO: Evaluating on dev set...
423
+ 2025-12-02 23:40:51 INFO: LAS MLAS BLEX
424
+ 2025-12-02 23:40:51 INFO: 55.69 43.50 52.45
425
+ 2025-12-02 23:40:51 INFO: step 1800: train_loss = 3.874376, dev_score = 0.5569
426
+ 2025-12-02 23:40:52 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
427
+ 2025-12-02 23:40:52 INFO: new model checkpoint saved.
428
+ 2025-12-02 23:40:53 INFO: Finished STEP 1820/50000, loss = 7.185968 (0.065 sec/batch), lr: 0.003000
429
+ 2025-12-02 23:40:55 INFO: Finished STEP 1840/50000, loss = 4.853516 (0.081 sec/batch), lr: 0.003000
430
+ 2025-12-02 23:40:57 INFO: Finished STEP 1860/50000, loss = 3.594461 (0.075 sec/batch), lr: 0.003000
431
+ 2025-12-02 23:40:58 INFO: Finished STEP 1880/50000, loss = 4.704398 (0.067 sec/batch), lr: 0.003000
432
+ 2025-12-02 23:41:00 INFO: Finished STEP 1900/50000, loss = 3.641345 (0.071 sec/batch), lr: 0.003000
433
+ 2025-12-02 23:41:00 INFO: Evaluating on dev set...
434
+ 2025-12-02 23:41:00 INFO: LAS MLAS BLEX
435
+ 2025-12-02 23:41:00 INFO: 58.17 49.16 51.68
436
+ 2025-12-02 23:41:00 INFO: step 1900: train_loss = 3.984191, dev_score = 0.5817
437
+ 2025-12-02 23:41:01 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
438
+ 2025-12-02 23:41:01 INFO: new model checkpoint saved.
439
+ 2025-12-02 23:41:02 INFO: Finished STEP 1920/50000, loss = 3.163770 (0.075 sec/batch), lr: 0.003000
440
+ 2025-12-02 23:41:04 INFO: Finished STEP 1940/50000, loss = 2.609706 (0.077 sec/batch), lr: 0.003000
441
+ 2025-12-02 23:41:05 INFO: Finished STEP 1960/50000, loss = 3.944045 (0.083 sec/batch), lr: 0.003000
442
+ 2025-12-02 23:41:07 INFO: Finished STEP 1980/50000, loss = 4.132533 (0.079 sec/batch), lr: 0.003000
443
+ 2025-12-02 23:41:08 INFO: Finished STEP 2000/50000, loss = 4.288573 (0.076 sec/batch), lr: 0.003000
444
+ 2025-12-02 23:41:08 INFO: Evaluating on dev set...
445
+ 2025-12-02 23:41:09 INFO: LAS MLAS BLEX
446
+ 2025-12-02 23:41:09 INFO: 57.92 50.21 54.01
447
+ 2025-12-02 23:41:09 INFO: step 2000: train_loss = 3.778160, dev_score = 0.5792
448
+ 2025-12-02 23:41:09 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
449
+ 2025-12-02 23:41:09 INFO: new model checkpoint saved.
450
+ 2025-12-02 23:41:11 INFO: Finished STEP 2020/50000, loss = 4.378428 (0.078 sec/batch), lr: 0.003000
451
+ 2025-12-02 23:41:12 INFO: Finished STEP 2040/50000, loss = 3.779004 (0.069 sec/batch), lr: 0.003000
452
+ 2025-12-02 23:41:14 INFO: Finished STEP 2060/50000, loss = 6.798884 (0.069 sec/batch), lr: 0.003000
453
+ 2025-12-02 23:41:16 INFO: Finished STEP 2080/50000, loss = 2.857572 (0.078 sec/batch), lr: 0.003000
454
+ 2025-12-02 23:41:17 INFO: Finished STEP 2100/50000, loss = 2.366086 (0.081 sec/batch), lr: 0.003000
455
+ 2025-12-02 23:41:17 INFO: Evaluating on dev set...
456
+ 2025-12-02 23:41:18 INFO: LAS MLAS BLEX
457
+ 2025-12-02 23:41:18 INFO: 59.65 49.58 53.39
458
+ 2025-12-02 23:41:18 INFO: step 2100: train_loss = 3.951032, dev_score = 0.5965
459
+ 2025-12-02 23:41:18 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
460
+ 2025-12-02 23:41:18 INFO: new best model saved.
461
+ 2025-12-02 23:41:18 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
462
+ 2025-12-02 23:41:18 INFO: new model checkpoint saved.
463
+ 2025-12-02 23:41:20 INFO: Finished STEP 2120/50000, loss = 3.607057 (0.072 sec/batch), lr: 0.003000
464
+ 2025-12-02 23:41:21 INFO: Finished STEP 2140/50000, loss = 3.568948 (0.073 sec/batch), lr: 0.003000
465
+ 2025-12-02 23:41:23 INFO: Finished STEP 2160/50000, loss = 3.044675 (0.073 sec/batch), lr: 0.003000
466
+ 2025-12-02 23:41:24 INFO: Finished STEP 2180/50000, loss = 3.535823 (0.081 sec/batch), lr: 0.003000
467
+ 2025-12-02 23:41:26 INFO: Finished STEP 2200/50000, loss = 2.605264 (0.071 sec/batch), lr: 0.003000
468
+ 2025-12-02 23:41:26 INFO: Evaluating on dev set...
469
+ 2025-12-02 23:41:26 INFO: LAS MLAS BLEX
470
+ 2025-12-02 23:41:26 INFO: 58.91 50.53 53.93
471
+ 2025-12-02 23:41:26 INFO: step 2200: train_loss = 3.716230, dev_score = 0.5891
472
+ 2025-12-02 23:41:27 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
473
+ 2025-12-02 23:41:27 INFO: new model checkpoint saved.
474
+ 2025-12-02 23:41:28 INFO: Finished STEP 2220/50000, loss = 3.709403 (0.077 sec/batch), lr: 0.003000
475
+ 2025-12-02 23:41:30 INFO: Finished STEP 2240/50000, loss = 2.810772 (0.072 sec/batch), lr: 0.003000
476
+ 2025-12-02 23:41:31 INFO: Finished STEP 2260/50000, loss = 5.648521 (0.071 sec/batch), lr: 0.003000
477
+ 2025-12-02 23:41:33 INFO: Finished STEP 2280/50000, loss = 4.186255 (0.073 sec/batch), lr: 0.003000
478
+ 2025-12-02 23:41:34 INFO: Finished STEP 2300/50000, loss = 3.349550 (0.078 sec/batch), lr: 0.003000
479
+ 2025-12-02 23:41:34 INFO: Evaluating on dev set...
480
+ 2025-12-02 23:41:35 INFO: LAS MLAS BLEX
481
+ 2025-12-02 23:41:35 INFO: 58.42 48.41 52.23
482
+ 2025-12-02 23:41:35 INFO: step 2300: train_loss = 3.908552, dev_score = 0.5842
483
+ 2025-12-02 23:41:35 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
484
+ 2025-12-02 23:41:35 INFO: new model checkpoint saved.
485
+ 2025-12-02 23:41:37 INFO: Finished STEP 2320/50000, loss = 3.527339 (0.071 sec/batch), lr: 0.003000
486
+ 2025-12-02 23:41:38 INFO: Finished STEP 2340/50000, loss = 4.192890 (0.071 sec/batch), lr: 0.003000
487
+ 2025-12-02 23:41:40 INFO: Finished STEP 2360/50000, loss = 2.740113 (0.072 sec/batch), lr: 0.003000
488
+ 2025-12-02 23:41:41 INFO: Finished STEP 2380/50000, loss = 2.736833 (0.072 sec/batch), lr: 0.003000
489
+ 2025-12-02 23:41:43 INFO: Finished STEP 2400/50000, loss = 3.953286 (0.077 sec/batch), lr: 0.003000
490
+ 2025-12-02 23:41:43 INFO: Evaluating on dev set...
491
+ 2025-12-02 23:41:43 INFO: LAS MLAS BLEX
492
+ 2025-12-02 23:41:43 INFO: 59.90 50.85 53.81
493
+ 2025-12-02 23:41:43 INFO: step 2400: train_loss = 3.943743, dev_score = 0.5990
494
+ 2025-12-02 23:41:43 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
495
+ 2025-12-02 23:41:43 INFO: new best model saved.
496
+ 2025-12-02 23:41:44 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
497
+ 2025-12-02 23:41:44 INFO: new model checkpoint saved.
498
+ 2025-12-02 23:41:45 INFO: Finished STEP 2420/50000, loss = 4.857200 (0.065 sec/batch), lr: 0.003000
499
+ 2025-12-02 23:41:47 INFO: Finished STEP 2440/50000, loss = 2.306742 (0.067 sec/batch), lr: 0.003000
500
+ 2025-12-02 23:41:48 INFO: Finished STEP 2460/50000, loss = 3.168222 (0.080 sec/batch), lr: 0.003000
501
+ 2025-12-02 23:41:50 INFO: Finished STEP 2480/50000, loss = 5.061126 (0.066 sec/batch), lr: 0.003000
502
+ 2025-12-02 23:41:51 INFO: Finished STEP 2500/50000, loss = 3.081437 (0.064 sec/batch), lr: 0.003000
503
+ 2025-12-02 23:41:51 INFO: Evaluating on dev set...
504
+ 2025-12-02 23:41:52 INFO: LAS MLAS BLEX
505
+ 2025-12-02 23:41:52 INFO: 58.42 50.85 54.24
506
+ 2025-12-02 23:41:52 INFO: step 2500: train_loss = 3.834500, dev_score = 0.5842
507
+ 2025-12-02 23:41:52 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
508
+ 2025-12-02 23:41:52 INFO: new model checkpoint saved.
509
+ 2025-12-02 23:41:54 INFO: Finished STEP 2520/50000, loss = 3.241362 (0.066 sec/batch), lr: 0.003000
510
+ 2025-12-02 23:41:55 INFO: Finished STEP 2540/50000, loss = 1.888306 (0.066 sec/batch), lr: 0.003000
511
+ 2025-12-02 23:41:56 INFO: Finished STEP 2560/50000, loss = 2.893296 (0.071 sec/batch), lr: 0.003000
512
+ 2025-12-02 23:41:58 INFO: Finished STEP 2580/50000, loss = 2.557456 (0.072 sec/batch), lr: 0.003000
513
+ 2025-12-02 23:41:59 INFO: Finished STEP 2600/50000, loss = 4.244951 (0.076 sec/batch), lr: 0.003000
514
+ 2025-12-02 23:41:59 INFO: Evaluating on dev set...
515
+ 2025-12-02 23:42:00 INFO: LAS MLAS BLEX
516
+ 2025-12-02 23:42:00 INFO: 59.16 50.73 53.67
517
+ 2025-12-02 23:42:00 INFO: step 2600: train_loss = 3.933291, dev_score = 0.5916
518
+ 2025-12-02 23:42:00 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
519
+ 2025-12-02 23:42:00 INFO: new model checkpoint saved.
520
+ 2025-12-02 23:42:02 INFO: Finished STEP 2620/50000, loss = 3.008387 (0.066 sec/batch), lr: 0.003000
521
+ 2025-12-02 23:42:03 INFO: Finished STEP 2640/50000, loss = 4.180097 (0.065 sec/batch), lr: 0.003000
522
+ 2025-12-02 23:42:05 INFO: Finished STEP 2660/50000, loss = 5.295879 (0.070 sec/batch), lr: 0.003000
523
+ 2025-12-02 23:42:06 INFO: Finished STEP 2680/50000, loss = 3.755968 (0.068 sec/batch), lr: 0.003000
524
+ 2025-12-02 23:42:08 INFO: Finished STEP 2700/50000, loss = 2.576284 (0.067 sec/batch), lr: 0.003000
525
+ 2025-12-02 23:42:08 INFO: Evaluating on dev set...
526
+ 2025-12-02 23:42:08 INFO: LAS MLAS BLEX
527
+ 2025-12-02 23:42:08 INFO: 60.15 51.37 54.74
528
+ 2025-12-02 23:42:08 INFO: step 2700: train_loss = 3.876875, dev_score = 0.6015
529
+ 2025-12-02 23:42:08 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
530
+ 2025-12-02 23:42:08 INFO: new best model saved.
531
+ 2025-12-02 23:42:09 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
532
+ 2025-12-02 23:42:09 INFO: new model checkpoint saved.
533
+ 2025-12-02 23:42:10 INFO: Finished STEP 2720/50000, loss = 2.522042 (0.072 sec/batch), lr: 0.003000
534
+ 2025-12-02 23:42:12 INFO: Finished STEP 2740/50000, loss = 2.345543 (0.080 sec/batch), lr: 0.003000
535
+ 2025-12-02 23:42:13 INFO: Finished STEP 2760/50000, loss = 8.553060 (0.080 sec/batch), lr: 0.003000
536
+ 2025-12-02 23:42:15 INFO: Finished STEP 2780/50000, loss = 3.173434 (0.072 sec/batch), lr: 0.003000
537
+ 2025-12-02 23:42:16 INFO: Finished STEP 2800/50000, loss = 3.539501 (0.072 sec/batch), lr: 0.003000
538
+ 2025-12-02 23:42:16 INFO: Evaluating on dev set...
539
+ 2025-12-02 23:42:17 INFO: LAS MLAS BLEX
540
+ 2025-12-02 23:42:17 INFO: 58.66 51.05 54.01
541
+ 2025-12-02 23:42:17 INFO: step 2800: train_loss = 4.048916, dev_score = 0.5866
542
+ 2025-12-02 23:42:17 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
543
+ 2025-12-02 23:42:17 INFO: new model checkpoint saved.
544
+ 2025-12-02 23:42:19 INFO: Finished STEP 2820/50000, loss = 3.329790 (0.073 sec/batch), lr: 0.003000
545
+ 2025-12-02 23:42:20 INFO: Finished STEP 2840/50000, loss = 3.331520 (0.072 sec/batch), lr: 0.003000
546
+ 2025-12-02 23:42:22 INFO: Finished STEP 2860/50000, loss = 4.227343 (0.065 sec/batch), lr: 0.003000
547
+ 2025-12-02 23:42:23 INFO: Finished STEP 2880/50000, loss = 3.918571 (0.071 sec/batch), lr: 0.003000
548
+ 2025-12-02 23:42:25 INFO: Finished STEP 2900/50000, loss = 3.550666 (0.067 sec/batch), lr: 0.003000
549
+ 2025-12-02 23:42:25 INFO: Evaluating on dev set...
550
+ 2025-12-02 23:42:25 INFO: LAS MLAS BLEX
551
+ 2025-12-02 23:42:25 INFO: 56.68 48.02 51.36
552
+ 2025-12-02 23:42:25 INFO: step 2900: train_loss = 3.999385, dev_score = 0.5668
553
+ 2025-12-02 23:42:26 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
554
+ 2025-12-02 23:42:26 INFO: new model checkpoint saved.
555
+ 2025-12-02 23:42:27 INFO: Finished STEP 2920/50000, loss = 5.337671 (0.069 sec/batch), lr: 0.003000
556
+ 2025-12-02 23:42:28 INFO: Finished STEP 2940/50000, loss = 3.878147 (0.075 sec/batch), lr: 0.003000
557
+ 2025-12-02 23:42:30 INFO: Finished STEP 2960/50000, loss = 4.860531 (0.064 sec/batch), lr: 0.003000
558
+ 2025-12-02 23:42:31 INFO: Finished STEP 2980/50000, loss = 4.283798 (0.072 sec/batch), lr: 0.003000
559
+ 2025-12-02 23:42:33 INFO: Finished STEP 3000/50000, loss = 3.867455 (0.067 sec/batch), lr: 0.003000
560
+ 2025-12-02 23:42:33 INFO: Evaluating on dev set...
561
+ 2025-12-02 23:42:33 INFO: LAS MLAS BLEX
562
+ 2025-12-02 23:42:33 INFO: 58.91 49.48 52.83
563
+ 2025-12-02 23:42:33 INFO: step 3000: train_loss = 4.055366, dev_score = 0.5891
564
+ 2025-12-02 23:42:34 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
565
+ 2025-12-02 23:42:34 INFO: new model checkpoint saved.
566
+ 2025-12-02 23:42:35 INFO: Finished STEP 3020/50000, loss = 4.723305 (0.068 sec/batch), lr: 0.003000
567
+ 2025-12-02 23:42:37 INFO: Finished STEP 3040/50000, loss = 4.147068 (0.067 sec/batch), lr: 0.003000
568
+ 2025-12-02 23:42:38 INFO: Finished STEP 3060/50000, loss = 3.139223 (0.078 sec/batch), lr: 0.003000
569
+ 2025-12-02 23:42:40 INFO: Finished STEP 3080/50000, loss = 4.757718 (0.065 sec/batch), lr: 0.003000
570
+ 2025-12-02 23:42:41 INFO: Finished STEP 3100/50000, loss = 3.247717 (0.077 sec/batch), lr: 0.003000
571
+ 2025-12-02 23:42:41 INFO: Evaluating on dev set...
572
+ 2025-12-02 23:42:42 INFO: LAS MLAS BLEX
573
+ 2025-12-02 23:42:42 INFO: 59.65 51.37 54.74
574
+ 2025-12-02 23:42:42 INFO: step 3100: train_loss = 3.991039, dev_score = 0.5965
575
+ 2025-12-02 23:42:42 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
576
+ 2025-12-02 23:42:42 INFO: new model checkpoint saved.
577
+ 2025-12-02 23:42:44 INFO: Finished STEP 3120/50000, loss = 4.579876 (0.070 sec/batch), lr: 0.003000
578
+ 2025-12-02 23:42:45 INFO: Finished STEP 3140/50000, loss = 3.654508 (0.068 sec/batch), lr: 0.003000
579
+ 2025-12-02 23:42:47 INFO: Finished STEP 3160/50000, loss = 3.347531 (0.065 sec/batch), lr: 0.003000
580
+ 2025-12-02 23:42:48 INFO: Finished STEP 3180/50000, loss = 4.979382 (0.063 sec/batch), lr: 0.003000
581
+ 2025-12-02 23:42:49 INFO: Finished STEP 3200/50000, loss = 4.191813 (0.069 sec/batch), lr: 0.003000
582
+ 2025-12-02 23:42:49 INFO: Evaluating on dev set...
583
+ 2025-12-02 23:42:50 INFO: LAS MLAS BLEX
584
+ 2025-12-02 23:42:50 INFO: 56.44 47.18 50.94
585
+ 2025-12-02 23:42:50 INFO: step 3200: train_loss = 3.937036, dev_score = 0.5644
586
+ 2025-12-02 23:42:51 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
587
+ 2025-12-02 23:42:51 INFO: new model checkpoint saved.
588
+ 2025-12-02 23:42:52 INFO: Finished STEP 3220/50000, loss = 3.818825 (0.074 sec/batch), lr: 0.003000
589
+ 2025-12-02 23:42:53 INFO: Finished STEP 3240/50000, loss = 4.176326 (0.072 sec/batch), lr: 0.003000
590
+ 2025-12-02 23:42:55 INFO: Finished STEP 3260/50000, loss = 3.838692 (0.070 sec/batch), lr: 0.003000
591
+ 2025-12-02 23:42:56 INFO: Finished STEP 3280/50000, loss = 4.169544 (0.074 sec/batch), lr: 0.003000
592
+ 2025-12-02 23:42:58 INFO: Finished STEP 3300/50000, loss = 3.897673 (0.075 sec/batch), lr: 0.003000
593
+ 2025-12-02 23:42:58 INFO: Evaluating on dev set...
594
+ 2025-12-02 23:42:58 INFO: LAS MLAS BLEX
595
+ 2025-12-02 23:42:58 INFO: 58.42 48.95 52.74
596
+ 2025-12-02 23:42:58 INFO: step 3300: train_loss = 3.852795, dev_score = 0.5842
597
+ 2025-12-02 23:42:59 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
598
+ 2025-12-02 23:42:59 INFO: new model checkpoint saved.
599
+ 2025-12-02 23:43:00 INFO: Finished STEP 3320/50000, loss = 3.183586 (0.071 sec/batch), lr: 0.003000
600
+ 2025-12-02 23:43:02 INFO: Finished STEP 3340/50000, loss = 4.357607 (0.068 sec/batch), lr: 0.003000
601
+ 2025-12-02 23:43:03 INFO: Finished STEP 3360/50000, loss = 3.466043 (0.071 sec/batch), lr: 0.003000
602
+ 2025-12-02 23:43:04 INFO: Finished STEP 3380/50000, loss = 3.885072 (0.068 sec/batch), lr: 0.003000
603
+ 2025-12-02 23:43:06 INFO: Finished STEP 3400/50000, loss = 4.157225 (0.064 sec/batch), lr: 0.003000
604
+ 2025-12-02 23:43:06 INFO: Evaluating on dev set...
605
+ 2025-12-02 23:43:06 INFO: LAS MLAS BLEX
606
+ 2025-12-02 23:43:06 INFO: 57.43 48.74 51.68
607
+ 2025-12-02 23:43:06 INFO: step 3400: train_loss = 3.984100, dev_score = 0.5743
608
+ 2025-12-02 23:43:07 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
609
+ 2025-12-02 23:43:07 INFO: new model checkpoint saved.
610
+ 2025-12-02 23:43:08 INFO: Finished STEP 3420/50000, loss = 2.478911 (0.067 sec/batch), lr: 0.003000
611
+ 2025-12-02 23:43:10 INFO: Finished STEP 3440/50000, loss = 3.196608 (0.067 sec/batch), lr: 0.003000
612
+ 2025-12-02 23:43:11 INFO: Finished STEP 3460/50000, loss = 4.725269 (0.070 sec/batch), lr: 0.003000
613
+ 2025-12-02 23:43:13 INFO: Finished STEP 3480/50000, loss = 7.304494 (0.064 sec/batch), lr: 0.003000
614
+ 2025-12-02 23:43:14 INFO: Finished STEP 3500/50000, loss = 5.185205 (0.071 sec/batch), lr: 0.003000
615
+ 2025-12-02 23:43:14 INFO: Evaluating on dev set...
616
+ 2025-12-02 23:43:14 INFO: LAS MLAS BLEX
617
+ 2025-12-02 23:43:14 INFO: 59.41 50.95 54.74
618
+ 2025-12-02 23:43:14 INFO: step 3500: train_loss = 3.882499, dev_score = 0.5941
619
+ 2025-12-02 23:43:15 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
620
+ 2025-12-02 23:43:15 INFO: new model checkpoint saved.
621
+ 2025-12-02 23:43:16 INFO: Finished STEP 3520/50000, loss = 4.237047 (0.067 sec/batch), lr: 0.003000
622
+ 2025-12-02 23:43:18 INFO: Finished STEP 3540/50000, loss = 3.827991 (0.065 sec/batch), lr: 0.003000
623
+ 2025-12-02 23:43:19 INFO: Finished STEP 3560/50000, loss = 5.006863 (0.069 sec/batch), lr: 0.003000
624
+ 2025-12-02 23:43:20 INFO: Finished STEP 3580/50000, loss = 4.178025 (0.069 sec/batch), lr: 0.003000
625
+ 2025-12-02 23:43:22 INFO: Finished STEP 3600/50000, loss = 3.835759 (0.079 sec/batch), lr: 0.003000
626
+ 2025-12-02 23:43:22 INFO: Evaluating on dev set...
627
+ 2025-12-02 23:43:22 INFO: LAS MLAS BLEX
628
+ 2025-12-02 23:43:22 INFO: 58.17 49.47 54.12
629
+ 2025-12-02 23:43:22 INFO: step 3600: train_loss = 3.827928, dev_score = 0.5817
630
+ 2025-12-02 23:43:23 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
631
+ 2025-12-02 23:43:23 INFO: new model checkpoint saved.
632
+ 2025-12-02 23:43:24 INFO: Finished STEP 3620/50000, loss = 3.906770 (0.062 sec/batch), lr: 0.003000
633
+ 2025-12-02 23:43:25 INFO: Finished STEP 3640/50000, loss = 3.407995 (0.066 sec/batch), lr: 0.003000
634
+ 2025-12-02 23:43:27 INFO: Finished STEP 3660/50000, loss = 4.964592 (0.074 sec/batch), lr: 0.003000
635
+ 2025-12-02 23:43:28 INFO: Finished STEP 3680/50000, loss = 4.575313 (0.067 sec/batch), lr: 0.003000
636
+ 2025-12-02 23:43:29 INFO: Finished STEP 3700/50000, loss = 5.298435 (0.066 sec/batch), lr: 0.003000
637
+ 2025-12-02 23:43:29 INFO: Evaluating on dev set...
638
+ 2025-12-02 23:43:30 INFO: LAS MLAS BLEX
639
+ 2025-12-02 23:43:30 INFO: 57.43 48.00 52.63
640
+ 2025-12-02 23:43:30 INFO: step 3700: train_loss = 4.043876, dev_score = 0.5743
641
+ 2025-12-02 23:43:30 INFO: Training ended with 3700 steps.
642
+ 2025-12-02 23:43:30 INFO: Best dev F1 = 60.15, at iteration = 2700
643
+ 2025-12-02 23:43:31 INFO: Running dev depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '32', '--dropout', '0.33']
644
+ 2025-12-02 23:43:31 INFO: Running parser in predict mode
645
+ 2025-12-02 23:43:31 INFO: Loading model from: saved_models/depparse/sv_diachronic_charlm_parser.pt
646
+ 2025-12-02 23:43:34 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
647
+ 2025-12-02 23:43:34 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
648
+ 2025-12-02 23:43:34 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
649
+ 2025-12-02 23:43:34 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
650
+ 2025-12-02 23:43:34 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
651
+ 2025-12-02 23:43:34 INFO: Loading data with batch size 32...
652
+ 2025-12-02 23:43:34 DEBUG: 9 batches created.
653
+ 2025-12-02 23:43:35 INFO: F1 scores for each dependency:
654
+ Note that unlabeled attachment errors hurt the labeled attachment scores
655
+ acl: p 0.0000 r 0.0000 f1 0.0000 (3 actual)
656
+ acl:relcl: p 0.2500 r 0.2857 f1 0.2667 (7 actual)
657
+ advcl: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
658
+ advmod: p 0.3714 r 0.5200 f1 0.4333 (25 actual)
659
+ amod: p 0.7667 r 0.7419 f1 0.7541 (31 actual)
660
+ appos: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
661
+ aux: p 0.6667 r 0.7273 f1 0.6957 (11 actual)
662
+ case: p 0.9615 r 0.8929 f1 0.9259 (56 actual)
663
+ cc: p 0.6154 r 0.6154 f1 0.6154 (13 actual)
664
+ ccomp: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
665
+ conj: p 0.3077 r 0.6667 f1 0.4211 (12 actual)
666
+ cop: p 1.0000 r 0.3333 f1 0.5000 (3 actual)
667
+ csubj: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
668
+ det: p 0.8636 r 0.8636 f1 0.8636 (22 actual)
669
+ expl: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
670
+ iobj: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
671
+ mark: p 0.5000 r 0.5000 f1 0.5000 (12 actual)
672
+ nmod: p 0.3043 r 0.4667 f1 0.3684 (15 actual)
673
+ nmod:poss: p 1.0000 r 0.7895 f1 0.8824 (19 actual)
674
+ nsubj: p 0.4231 r 0.6471 f1 0.5116 (17 actual)
675
+ nsubj:pass: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
676
+ obj: p 0.6957 r 0.7273 f1 0.7111 (22 actual)
677
+ obl: p 0.6154 r 0.5854 f1 0.6000 (41 actual)
678
+ obl:agent: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
679
+ orphan: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
680
+ parataxis: p 0.0000 r 0.0000 f1 0.0000 (3 actual)
681
+ punct: p 0.4038 r 0.4038 f1 0.4038 (52 actual)
682
+ root: p 0.4444 r 0.4444 f1 0.4444 (9 actual)
683
+ xcomp: p 1.0000 r 0.1250 f1 0.2222 (8 actual)
684
+ 2025-12-02 23:43:35 INFO: LAS MLAS BLEX
685
+ 2025-12-02 23:43:35 INFO: 60.15 51.37 54.74
686
+ 2025-12-02 23:43:35 INFO: Parser score:
687
+ 2025-12-02 23:43:35 INFO: sv_diachronic 60.15
688
+ 2025-12-02 23:43:35 INFO: Finished running dev set on
689
+ UD_Swedish-diachronic
690
+ UAS LAS CLAS MLAS BLEX
691
+ 67.08 60.15 54.74 51.37 54.74
692
+ 2025-12-02 23:43:35 INFO: Running test depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '32', '--dropout', '0.33']
693
+ 2025-12-02 23:43:35 INFO: Running parser in predict mode
694
+ 2025-12-02 23:43:35 INFO: Loading model from: saved_models/depparse/sv_diachronic_charlm_parser.pt
695
+ 2025-12-02 23:43:38 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
696
+ 2025-12-02 23:43:38 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
697
+ 2025-12-02 23:43:38 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
698
+ 2025-12-02 23:43:38 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
699
+ 2025-12-02 23:43:38 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
700
+ 2025-12-02 23:43:38 INFO: Loading data with batch size 32...
701
+ 2025-12-02 23:43:38 DEBUG: 93 batches created.
702
+ 2025-12-02 23:43:43 INFO: F1 scores for each dependency:
703
+ Note that unlabeled attachment errors hurt the labeled attachment scores
704
+ acl: p 0.0000 r 0.0000 f1 0.0000 (32 actual)
705
+ acl:cleft: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
706
+ acl:relcl: p 0.2222 r 0.2400 f1 0.2308 (75 actual)
707
+ advcl: p 0.1463 r 0.1000 f1 0.1188 (60 actual)
708
+ advcl:relcl: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
709
+ advmod: p 0.4952 r 0.5821 f1 0.5352 (268 actual)
710
+ amod: p 0.7419 r 0.8000 f1 0.7699 (230 actual)
711
+ appos: p 0.0000 r 0.0000 f1 0.0000 (13 actual)
712
+ aux: p 0.8118 r 0.8214 f1 0.8166 (84 actual)
713
+ aux:pass: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
714
+ case: p 0.8892 r 0.8177 f1 0.8520 (373 actual)
715
+ cc: p 0.6067 r 0.5871 f1 0.5967 (155 actual)
716
+ ccomp: p 0.0000 r 0.0000 f1 0.0000 (35 actual)
717
+ compound:prt: p 0.8333 r 0.7143 f1 0.7692 (21 actual)
718
+ conj: p 0.1647 r 0.2658 f1 0.2034 (158 actual)
719
+ cop: p 0.8214 r 0.5000 f1 0.6216 (46 actual)
720
+ csubj: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
721
+ dep: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
722
+ det: p 0.7837 r 0.7837 f1 0.7837 (208 actual)
723
+ discourse: p 0.0000 r 0.0000 f1 0.0000 (7 actual)
724
+ dislocated: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
725
+ expl: p 0.3333 r 0.0909 f1 0.1429 (11 actual)
726
+ expl:pv: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
727
+ fixed: p 0.0000 r 0.0000 f1 0.0000 (8 actual)
728
+ flat: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
729
+ flat:name: p 0.0000 r 0.0000 f1 0.0000 (12 actual)
730
+ goeswith: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
731
+ iobj: p 0.0000 r 0.0000 f1 0.0000 (14 actual)
732
+ mark: p 0.6815 r 0.6993 f1 0.6903 (153 actual)
733
+ nmod: p 0.3429 r 0.4706 f1 0.3967 (102 actual)
734
+ nmod:poss: p 0.8963 r 0.8521 f1 0.8736 (142 actual)
735
+ nsubj: p 0.4986 r 0.6357 f1 0.5589 (280 actual)
736
+ nsubj:pass: p 0.0000 r 0.0000 f1 0.0000 (25 actual)
737
+ nummod: p 1.0000 r 0.2000 f1 0.3333 (10 actual)
738
+ obj: p 0.6541 r 0.6612 f1 0.6576 (183 actual)
739
+ obl: p 0.5354 r 0.5719 f1 0.5530 (278 actual)
740
+ obl:agent: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
741
+ orphan: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
742
+ parataxis: p 0.0769 r 0.0556 f1 0.0645 (18 actual)
743
+ punct: p 0.3991 r 0.4094 f1 0.4042 (425 actual)
744
+ reparandum: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
745
+ root: p 0.5455 r 0.5455 f1 0.5455 (99 actual)
746
+ vocative: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
747
+ xcomp: p 0.5714 r 0.2133 f1 0.3107 (75 actual)
748
+ 2025-12-02 23:43:43 INFO: LAS MLAS BLEX
749
+ 2025-12-02 23:43:43 INFO: 57.05 47.92 51.69
750
+ 2025-12-02 23:43:43 INFO: Parser score:
751
+ 2025-12-02 23:43:43 INFO: sv_diachronic 57.05
752
+ 2025-12-02 23:43:43 INFO: Finished running test set on
753
+ UD_Swedish-diachronic
754
+ UAS LAS CLAS MLAS BLEX
755
+ 66.17 57.05 51.69 47.92 51.69
756
+ DONE.
757
+ Full log saved to: log_20251202_233139.txt
logs/log_diachronic.pt_baseline_batch16.txt ADDED
@@ -0,0 +1,695 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === LOGFILE: logs/log_diachronic.pt_sv_20251203_000134.txt ===
2
+ Language codes: sv
3
+ Using pretrained model: diachronic.pt
4
+
5
+ Running: python prepare-train-val-test.py sv
6
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-dev.conllu
7
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test.conllu
8
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_pud-ud-test.conllu
9
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-test.conllu
10
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test-trg.conllu
11
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-dev.conllu
12
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/ucxn_ud_swedish-talbanken.conllu
13
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-train.conllu
14
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_old-ud-test.conllu
15
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-train.conllu
16
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-test.conllu
17
+ Skipping DigPhil MACHINE (diachron not requested).
18
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec330-GyllenborgC_SwenskaSpratthoken.conllu
19
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec254-CederborghF_BerattelseOmJohnHall.conllu
20
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec277-EnbomPU_MedborgeligtSkalde.conllu
21
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec268-DulciU_VitterhetsNojen3.conllu
22
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1063-spf220.conllu
23
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec397-AngeredStrandbergH_UnderSodernsSol.conllu
24
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec324-GranbergPA_Enslighetsalskaren.conllu
25
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec252-BremerF_Teckningar1.conllu
26
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec988-spf145.conllu
27
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec987-spf144.conllu
28
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec631-HasselskogN_HallaHallaGronkoping.conllu
29
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-letter141673-Stalhammar.conllu
30
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1033-spf190.conllu
31
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec25-Runius.conllu
32
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec486-SchwartzMS_BellmansSkor.conllu
33
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec452-NyblomH_FantasierFyra.conllu
34
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec613-EngstromA_StrindbergOchJag.conllu
35
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec208-Anonym_DetGrasligaMordet.conllu
36
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec639-HeidenstamV_Proletarfilosofiens.conllu
37
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1102-spf259.conllu
38
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec991-spf148.conllu
39
+ Cleaning TRAIN...
40
+ Cleaning DEV...
41
+ [REMOVED] sent_id=33 ERRORS=['Token 15: Missing deprel']
42
+ Cleaning TEST...
43
+ Writing TRAIN → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu (19820 valid sentences)
44
+ Writing DEV → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu (9 valid sentences)
45
+ Writing TEST → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu (99 valid sentences)
46
+ Done.
47
+ Sourcing scripts/config_alvis.sh
48
+ Running stanza dataset preparation…
49
+ 2025-12-03 00:01:41 INFO: Datasets program called with:
50
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/datasets/prepare_depparse_treebank.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
51
+ 2025-12-03 00:01:41 DEBUG: Downloading resource file from https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.11.0.json
52
+
53
+ 2025-12-03 00:01:41 INFO: Downloaded file to /cephyr/users/cleland/Alvis/stanza_resources/resources.json
54
+ 2025-12-03 00:01:41 DEBUG: Processing parameter "processors"...
55
+ 2025-12-03 00:01:41 WARNING: Can not find pos: diachronic from official model list. Ignoring it.
56
+ 2025-12-03 00:01:41 INFO: Downloading these customized packages for language: sv (Swedish)...
57
+ =======================
58
+ | Processor | Package |
59
+ -----------------------
60
+ =======================
61
+
62
+ 2025-12-03 00:01:41 INFO: Finished downloading models and saved to /cephyr/users/cleland/Alvis/stanza_resources
63
+ 2025-12-03 00:01:41 INFO: Using tagger model in /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt for sv_diachronic
64
+ 2025-12-03 00:01:41 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
65
+ 2025-12-03 00:01:41 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
66
+ Augmented 56 quotes: Counter({'„”': 9, '″″': 9, '""': 8, '「」': 8, '””': 5, '““': 4, '《》': 4, '»«': 3, '„“': 3, '«»': 3})
67
+ 2025-12-03 00:01:42 INFO: Running tagger to retag /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.train.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu
68
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.train.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu']
69
+ 2025-12-03 00:01:42 INFO: Running tagger in predict mode
70
+ 2025-12-03 00:01:42 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
71
+ 2025-12-03 00:01:45 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
72
+ 2025-12-03 00:01:45 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
73
+ 2025-12-03 00:01:45 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
74
+ 2025-12-03 00:01:45 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
75
+ 2025-12-03 00:01:45 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
76
+ 2025-12-03 00:01:48 INFO: Loading data with batch size 250...
77
+ 2025-12-03 00:02:02 INFO: Start evaluation...
78
+ 2025-12-03 00:03:09 INFO: UPOS XPOS UFeats AllTags
79
+ 2025-12-03 00:03:09 INFO: 97.53 61.74 92.12 59.25
80
+ 2025-12-03 00:03:09 INFO: POS Tagger score: sv_diachronic 59.25
81
+ 2025-12-03 00:03:09 INFO: Running tagger to retag /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.dev.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu
82
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.dev.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu']
83
+ 2025-12-03 00:03:09 INFO: Running tagger in predict mode
84
+ 2025-12-03 00:03:09 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
85
+ 2025-12-03 00:03:12 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
86
+ 2025-12-03 00:03:12 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
87
+ 2025-12-03 00:03:12 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
88
+ 2025-12-03 00:03:12 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
89
+ 2025-12-03 00:03:12 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
90
+ 2025-12-03 00:03:12 INFO: Loading data with batch size 250...
91
+ 2025-12-03 00:03:12 INFO: Start evaluation...
92
+ 2025-12-03 00:03:12 INFO: UPOS XPOS UFeats AllTags
93
+ 2025-12-03 00:03:12 INFO: 88.61 79.46 83.42 74.75
94
+ 2025-12-03 00:03:12 INFO: POS Tagger score: sv_diachronic 74.75
95
+ 2025-12-03 00:03:12 INFO: Running tagger to retag /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.test.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu
96
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.test.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu']
97
+ 2025-12-03 00:03:12 INFO: Running tagger in predict mode
98
+ 2025-12-03 00:03:12 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
99
+ 2025-12-03 00:03:15 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
100
+ 2025-12-03 00:03:15 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
101
+ 2025-12-03 00:03:15 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
102
+ 2025-12-03 00:03:15 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
103
+ 2025-12-03 00:03:15 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
104
+ 2025-12-03 00:03:15 INFO: Loading data with batch size 250...
105
+ 2025-12-03 00:03:15 INFO: Start evaluation...
106
+ 2025-12-03 00:03:16 INFO: UPOS XPOS UFeats AllTags
107
+ 2025-12-03 00:03:16 INFO: 89.61 86.64 86.69 81.18
108
+ 2025-12-03 00:03:16 INFO: POS Tagger score: sv_diachronic 81.18
109
+ Preparing data for UD_Swedish-diachronic: sv_diachronic, sv
110
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu and writing to /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.train.gold.conllu
111
+ Swapped 'w1, w2' for 'w1 ,w2' 122 times
112
+ Added 100 new sentences with asdf, zzzz -> asdf,zzzz
113
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu and writing to /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.dev.gold.conllu
114
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu and writing to /local/tmp.5437491/tmpr9ew7a9i/sv_diachronic.test.gold.conllu
115
+ Running stanza dependency parser training…
116
+ 2025-12-03 00:03:26 INFO: Training program called with:
117
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/training/run_depparse.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt --batch_size 16 --dropout 0.33
118
+ 2025-12-03 00:03:26 DEBUG: UD_Swedish-diachronic: sv_diachronic
119
+ 2025-12-03 00:03:26 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
120
+ 2025-12-03 00:03:26 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
121
+ 2025-12-03 00:03:26 INFO: UD_Swedish-diachronic: saved_models/depparse/sv_diachronic_charlm_parser.pt does not exist, training new model
122
+ 2025-12-03 00:03:26 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
123
+ 2025-12-03 00:03:26 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
124
+ 2025-12-03 00:03:26 INFO: Running train depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--train_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu', '--batch_size', '5000', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'train', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '16', '--dropout', '0.33']
125
+ 2025-12-03 00:03:26 INFO: Running parser in train mode
126
+ 2025-12-03 00:03:26 INFO: Using pretrained contextualized char embedding
127
+ 2025-12-03 00:03:26 INFO: Loading data with batch size 16...
128
+ 2025-12-03 00:03:28 INFO: Train File /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu, Data Size: 19920
129
+ 2025-12-03 00:03:28 INFO: Original data size: 19920
130
+ 2025-12-03 00:03:29 INFO: Augmented data size: 20844
131
+ 2025-12-03 00:03:36 WARNING: sv_diachronic is not a known dataset. Examining the data to choose which xpos vocab to use
132
+ 2025-12-03 00:03:36 INFO: Original length = 20844
133
+ 2025-12-03 00:03:36 INFO: Filtered length = 20844
134
+ 2025-12-03 00:03:42 WARNING: Chose XPOSDescription(xpos_type=<XPOSType.XPOS: 1>, sep='|') for the xpos factory for sv_diachronic
135
+ 2025-12-03 00:03:46 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
136
+ 2025-12-03 00:03:51 DEBUG: 18895 batches created.
137
+ 2025-12-03 00:03:51 DEBUG: 9 batches created.
138
+ 2025-12-03 00:03:51 INFO: Training parser...
139
+ 2025-12-03 00:03:51 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
140
+ 2025-12-03 00:03:51 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
141
+ 2025-12-03 00:03:51 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
142
+ 2025-12-03 00:03:52 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
143
+ 2025-12-03 00:03:55 INFO: Finished STEP 20/50000, loss = 0.000000 (0.031 sec/batch), lr: 0.003000
144
+ 2025-12-03 00:03:55 INFO: Finished STEP 40/50000, loss = 2.786552 (0.032 sec/batch), lr: 0.003000
145
+ 2025-12-03 00:03:56 INFO: Finished STEP 60/50000, loss = 2.014541 (0.032 sec/batch), lr: 0.003000
146
+ 2025-12-03 00:03:57 INFO: Finished STEP 80/50000, loss = 3.657895 (0.032 sec/batch), lr: 0.003000
147
+ 2025-12-03 00:03:57 INFO: Finished STEP 100/50000, loss = 2.003451 (0.031 sec/batch), lr: 0.003000
148
+ 2025-12-03 00:03:57 INFO: Evaluating on dev set...
149
+ 2025-12-03 00:03:58 INFO: LAS MLAS BLEX
150
+ 2025-12-03 00:03:58 INFO: 1.49 0.82 1.64
151
+ 2025-12-03 00:03:58 INFO: step 100: train_loss = 5559.849305, dev_score = 0.0149
152
+ 2025-12-03 00:03:58 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
153
+ 2025-12-03 00:03:58 INFO: new best model saved.
154
+ 2025-12-03 00:03:59 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
155
+ 2025-12-03 00:03:59 INFO: new model checkpoint saved.
156
+ 2025-12-03 00:03:59 INFO: Finished STEP 120/50000, loss = 2.080929 (0.031 sec/batch), lr: 0.003000
157
+ 2025-12-03 00:04:00 INFO: Finished STEP 140/50000, loss = 1.784714 (0.032 sec/batch), lr: 0.003000
158
+ 2025-12-03 00:04:00 INFO: Finished STEP 160/50000, loss = 3.210406 (0.034 sec/batch), lr: 0.003000
159
+ 2025-12-03 00:04:01 INFO: Finished STEP 180/50000, loss = 3.367260 (0.033 sec/batch), lr: 0.003000
160
+ 2025-12-03 00:04:02 INFO: Finished STEP 200/50000, loss = 2.319421 (0.034 sec/batch), lr: 0.003000
161
+ 2025-12-03 00:04:02 INFO: Evaluating on dev set...
162
+ 2025-12-03 00:04:03 INFO: LAS MLAS BLEX
163
+ 2025-12-03 00:04:03 INFO: 4.95 4.09 4.09
164
+ 2025-12-03 00:04:03 INFO: step 200: train_loss = 2.658702, dev_score = 0.0495
165
+ 2025-12-03 00:04:03 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
166
+ 2025-12-03 00:04:03 INFO: new best model saved.
167
+ 2025-12-03 00:04:04 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
168
+ 2025-12-03 00:04:04 INFO: new model checkpoint saved.
169
+ 2025-12-03 00:04:04 INFO: Finished STEP 220/50000, loss = 2.646727 (0.035 sec/batch), lr: 0.003000
170
+ 2025-12-03 00:04:05 INFO: Finished STEP 240/50000, loss = 2.225141 (0.034 sec/batch), lr: 0.003000
171
+ 2025-12-03 00:04:06 INFO: Finished STEP 260/50000, loss = 2.439486 (0.033 sec/batch), lr: 0.003000
172
+ 2025-12-03 00:04:06 INFO: Finished STEP 280/50000, loss = 3.247507 (0.033 sec/batch), lr: 0.003000
173
+ 2025-12-03 00:04:07 INFO: Finished STEP 300/50000, loss = 3.285990 (0.031 sec/batch), lr: 0.003000
174
+ 2025-12-03 00:04:07 INFO: Evaluating on dev set...
175
+ 2025-12-03 00:04:07 INFO: LAS MLAS BLEX
176
+ 2025-12-03 00:04:07 INFO: 3.22 3.76 4.10
177
+ 2025-12-03 00:04:07 INFO: step 300: train_loss = 2.677451, dev_score = 0.0322
178
+ 2025-12-03 00:04:08 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
179
+ 2025-12-03 00:04:08 INFO: new model checkpoint saved.
180
+ 2025-12-03 00:04:09 INFO: Finished STEP 320/50000, loss = 3.132120 (0.032 sec/batch), lr: 0.003000
181
+ 2025-12-03 00:04:09 INFO: Finished STEP 340/50000, loss = 3.690158 (0.033 sec/batch), lr: 0.003000
182
+ 2025-12-03 00:04:10 INFO: Finished STEP 360/50000, loss = 3.325933 (0.035 sec/batch), lr: 0.003000
183
+ 2025-12-03 00:04:11 INFO: Finished STEP 380/50000, loss = 2.511281 (0.034 sec/batch), lr: 0.003000
184
+ 2025-12-03 00:04:11 INFO: Finished STEP 400/50000, loss = 2.801303 (0.032 sec/batch), lr: 0.003000
185
+ 2025-12-03 00:04:11 INFO: Evaluating on dev set...
186
+ 2025-12-03 00:04:12 INFO: LAS MLAS BLEX
187
+ 2025-12-03 00:04:12 INFO: 7.18 4.41 6.10
188
+ 2025-12-03 00:04:12 INFO: step 400: train_loss = 3.131062, dev_score = 0.0718
189
+ 2025-12-03 00:04:12 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
190
+ 2025-12-03 00:04:12 INFO: new best model saved.
191
+ 2025-12-03 00:04:13 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
192
+ 2025-12-03 00:04:13 INFO: new model checkpoint saved.
193
+ 2025-12-03 00:04:13 INFO: Finished STEP 420/50000, loss = 2.566988 (0.036 sec/batch), lr: 0.003000
194
+ 2025-12-03 00:04:14 INFO: Finished STEP 440/50000, loss = 3.315584 (0.033 sec/batch), lr: 0.003000
195
+ 2025-12-03 00:04:15 INFO: Finished STEP 460/50000, loss = 4.081317 (0.034 sec/batch), lr: 0.003000
196
+ 2025-12-03 00:04:15 INFO: Finished STEP 480/50000, loss = 3.782612 (0.032 sec/batch), lr: 0.003000
197
+ 2025-12-03 00:04:16 INFO: Finished STEP 500/50000, loss = 3.673034 (0.032 sec/batch), lr: 0.003000
198
+ 2025-12-03 00:04:16 INFO: Evaluating on dev set...
199
+ 2025-12-03 00:04:17 INFO: LAS MLAS BLEX
200
+ 2025-12-03 00:04:17 INFO: 12.38 5.35 5.74
201
+ 2025-12-03 00:04:17 INFO: step 500: train_loss = 2.978255, dev_score = 0.1238
202
+ 2025-12-03 00:04:17 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
203
+ 2025-12-03 00:04:17 INFO: new best model saved.
204
+ 2025-12-03 00:04:17 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
205
+ 2025-12-03 00:04:17 INFO: new model checkpoint saved.
206
+ 2025-12-03 00:04:18 INFO: Finished STEP 520/50000, loss = 2.025372 (0.031 sec/batch), lr: 0.003000
207
+ 2025-12-03 00:04:19 INFO: Finished STEP 540/50000, loss = 2.984106 (0.037 sec/batch), lr: 0.003000
208
+ 2025-12-03 00:04:20 INFO: Finished STEP 560/50000, loss = 2.716340 (0.035 sec/batch), lr: 0.003000
209
+ 2025-12-03 00:04:20 INFO: Finished STEP 580/50000, loss = 2.306678 (0.033 sec/batch), lr: 0.003000
210
+ 2025-12-03 00:04:21 INFO: Finished STEP 600/50000, loss = 4.008360 (0.035 sec/batch), lr: 0.003000
211
+ 2025-12-03 00:04:21 INFO: Evaluating on dev set...
212
+ 2025-12-03 00:04:21 INFO: LAS MLAS BLEX
213
+ 2025-12-03 00:04:21 INFO: 19.31 6.94 8.48
214
+ 2025-12-03 00:04:21 INFO: step 600: train_loss = 3.075830, dev_score = 0.1931
215
+ 2025-12-03 00:04:22 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
216
+ 2025-12-03 00:04:22 INFO: new best model saved.
217
+ 2025-12-03 00:04:22 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
218
+ 2025-12-03 00:04:22 INFO: new model checkpoint saved.
219
+ 2025-12-03 00:04:23 INFO: Finished STEP 620/50000, loss = 4.868780 (0.034 sec/batch), lr: 0.003000
220
+ 2025-12-03 00:04:24 INFO: Finished STEP 640/50000, loss = 2.631726 (0.033 sec/batch), lr: 0.003000
221
+ 2025-12-03 00:04:24 INFO: Finished STEP 660/50000, loss = 2.899960 (0.035 sec/batch), lr: 0.003000
222
+ 2025-12-03 00:04:25 INFO: Finished STEP 680/50000, loss = 2.961294 (0.035 sec/batch), lr: 0.003000
223
+ 2025-12-03 00:04:26 INFO: Finished STEP 700/50000, loss = 3.054440 (0.035 sec/batch), lr: 0.003000
224
+ 2025-12-03 00:04:26 INFO: Evaluating on dev set...
225
+ 2025-12-03 00:04:26 INFO: LAS MLAS BLEX
226
+ 2025-12-03 00:04:26 INFO: 18.32 8.55 12.22
227
+ 2025-12-03 00:04:26 INFO: step 700: train_loss = 3.234651, dev_score = 0.1832
228
+ 2025-12-03 00:04:27 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
229
+ 2025-12-03 00:04:27 INFO: new model checkpoint saved.
230
+ 2025-12-03 00:04:27 INFO: Finished STEP 720/50000, loss = 3.635537 (0.034 sec/batch), lr: 0.003000
231
+ 2025-12-03 00:04:28 INFO: Finished STEP 740/50000, loss = 3.526119 (0.034 sec/batch), lr: 0.003000
232
+ 2025-12-03 00:04:29 INFO: Finished STEP 760/50000, loss = 2.225143 (0.035 sec/batch), lr: 0.003000
233
+ 2025-12-03 00:04:29 INFO: Finished STEP 780/50000, loss = 3.277092 (0.034 sec/batch), lr: 0.003000
234
+ 2025-12-03 00:04:30 INFO: Finished STEP 800/50000, loss = 4.215595 (0.037 sec/batch), lr: 0.003000
235
+ 2025-12-03 00:04:30 INFO: Evaluating on dev set...
236
+ 2025-12-03 00:04:31 INFO: LAS MLAS BLEX
237
+ 2025-12-03 00:04:31 INFO: 24.50 10.20 12.24
238
+ 2025-12-03 00:04:31 INFO: step 800: train_loss = 3.332544, dev_score = 0.2450
239
+ 2025-12-03 00:04:31 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
240
+ 2025-12-03 00:04:31 INFO: new best model saved.
241
+ 2025-12-03 00:04:31 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
242
+ 2025-12-03 00:04:31 INFO: new model checkpoint saved.
243
+ 2025-12-03 00:04:32 INFO: Finished STEP 820/50000, loss = 2.911338 (0.035 sec/batch), lr: 0.003000
244
+ 2025-12-03 00:04:33 INFO: Finished STEP 840/50000, loss = 3.892157 (0.035 sec/batch), lr: 0.003000
245
+ 2025-12-03 00:04:34 INFO: Finished STEP 860/50000, loss = 4.334442 (0.035 sec/batch), lr: 0.003000
246
+ 2025-12-03 00:04:34 INFO: Finished STEP 880/50000, loss = 4.049129 (0.035 sec/batch), lr: 0.003000
247
+ 2025-12-03 00:04:35 INFO: Finished STEP 900/50000, loss = 3.475506 (0.034 sec/batch), lr: 0.003000
248
+ 2025-12-03 00:04:35 INFO: Evaluating on dev set...
249
+ 2025-12-03 00:04:35 INFO: LAS MLAS BLEX
250
+ 2025-12-03 00:04:35 INFO: 23.02 8.61 11.35
251
+ 2025-12-03 00:04:35 INFO: step 900: train_loss = 3.328322, dev_score = 0.2302
252
+ 2025-12-03 00:04:36 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
253
+ 2025-12-03 00:04:36 INFO: new model checkpoint saved.
254
+ 2025-12-03 00:04:37 INFO: Finished STEP 920/50000, loss = 3.209418 (0.032 sec/batch), lr: 0.003000
255
+ 2025-12-03 00:04:37 INFO: Finished STEP 940/50000, loss = 2.384030 (0.032 sec/batch), lr: 0.003000
256
+ 2025-12-03 00:04:38 INFO: Finished STEP 960/50000, loss = 2.963456 (0.033 sec/batch), lr: 0.003000
257
+ 2025-12-03 00:04:39 INFO: Finished STEP 980/50000, loss = 2.940459 (0.037 sec/batch), lr: 0.003000
258
+ 2025-12-03 00:04:39 INFO: Finished STEP 1000/50000, loss = 4.433736 (0.035 sec/batch), lr: 0.003000
259
+ 2025-12-03 00:04:39 INFO: Evaluating on dev set...
260
+ 2025-12-03 00:04:40 INFO: LAS MLAS BLEX
261
+ 2025-12-03 00:04:40 INFO: 28.47 14.06 17.19
262
+ 2025-12-03 00:04:40 INFO: step 1000: train_loss = 3.266401, dev_score = 0.2847
263
+ 2025-12-03 00:04:40 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
264
+ 2025-12-03 00:04:40 INFO: new best model saved.
265
+ 2025-12-03 00:04:41 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
266
+ 2025-12-03 00:04:41 INFO: new model checkpoint saved.
267
+ 2025-12-03 00:04:41 INFO: Finished STEP 1020/50000, loss = 3.496134 (0.037 sec/batch), lr: 0.003000
268
+ 2025-12-03 00:04:42 INFO: Finished STEP 1040/50000, loss = 3.133940 (0.038 sec/batch), lr: 0.003000
269
+ 2025-12-03 00:04:43 INFO: Finished STEP 1060/50000, loss = 2.935402 (0.036 sec/batch), lr: 0.003000
270
+ 2025-12-03 00:04:44 INFO: Finished STEP 1080/50000, loss = 3.080907 (0.036 sec/batch), lr: 0.003000
271
+ 2025-12-03 00:04:44 INFO: Finished STEP 1100/50000, loss = 3.513608 (0.035 sec/batch), lr: 0.003000
272
+ 2025-12-03 00:04:44 INFO: Evaluating on dev set...
273
+ 2025-12-03 00:04:45 INFO: LAS MLAS BLEX
274
+ 2025-12-03 00:04:45 INFO: 24.50 6.39 7.58
275
+ 2025-12-03 00:04:45 INFO: step 1100: train_loss = 3.337082, dev_score = 0.2450
276
+ 2025-12-03 00:04:45 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
277
+ 2025-12-03 00:04:45 INFO: new model checkpoint saved.
278
+ 2025-12-03 00:04:46 INFO: Finished STEP 1120/50000, loss = 3.239853 (0.037 sec/batch), lr: 0.003000
279
+ 2025-12-03 00:04:47 INFO: Finished STEP 1140/50000, loss = 2.648186 (0.037 sec/batch), lr: 0.003000
280
+ 2025-12-03 00:04:48 INFO: Finished STEP 1160/50000, loss = 2.736733 (0.034 sec/batch), lr: 0.003000
281
+ 2025-12-03 00:04:48 INFO: Finished STEP 1180/50000, loss = 4.440369 (0.035 sec/batch), lr: 0.003000
282
+ 2025-12-03 00:04:49 INFO: Finished STEP 1200/50000, loss = 3.411217 (0.034 sec/batch), lr: 0.003000
283
+ 2025-12-03 00:04:49 INFO: Evaluating on dev set...
284
+ 2025-12-03 00:04:49 INFO: LAS MLAS BLEX
285
+ 2025-12-03 00:04:49 INFO: 28.71 11.49 13.07
286
+ 2025-12-03 00:04:49 INFO: step 1200: train_loss = 3.418644, dev_score = 0.2871
287
+ 2025-12-03 00:04:50 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
288
+ 2025-12-03 00:04:50 INFO: new best model saved.
289
+ 2025-12-03 00:04:50 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
290
+ 2025-12-03 00:04:50 INFO: new model checkpoint saved.
291
+ 2025-12-03 00:04:51 INFO: Finished STEP 1220/50000, loss = 5.032012 (0.036 sec/batch), lr: 0.003000
292
+ 2025-12-03 00:04:52 INFO: Finished STEP 1240/50000, loss = 2.865893 (0.036 sec/batch), lr: 0.003000
293
+ 2025-12-03 00:04:52 INFO: Finished STEP 1260/50000, loss = 6.096026 (0.037 sec/batch), lr: 0.003000
294
+ 2025-12-03 00:04:53 INFO: Finished STEP 1280/50000, loss = 2.903123 (0.036 sec/batch), lr: 0.003000
295
+ 2025-12-03 00:04:54 INFO: Finished STEP 1300/50000, loss = 3.038009 (0.037 sec/batch), lr: 0.003000
296
+ 2025-12-03 00:04:54 INFO: Evaluating on dev set...
297
+ 2025-12-03 00:04:54 INFO: LAS MLAS BLEX
298
+ 2025-12-03 00:04:54 INFO: 33.17 18.04 19.61
299
+ 2025-12-03 00:04:54 INFO: step 1300: train_loss = 3.347368, dev_score = 0.3317
300
+ 2025-12-03 00:04:55 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
301
+ 2025-12-03 00:04:55 INFO: new best model saved.
302
+ 2025-12-03 00:04:55 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
303
+ 2025-12-03 00:04:55 INFO: new model checkpoint saved.
304
+ 2025-12-03 00:04:56 INFO: Finished STEP 1320/50000, loss = 1.732098 (0.036 sec/batch), lr: 0.003000
305
+ 2025-12-03 00:04:57 INFO: Finished STEP 1340/50000, loss = 2.579418 (0.034 sec/batch), lr: 0.003000
306
+ 2025-12-03 00:04:57 INFO: Finished STEP 1360/50000, loss = 2.314942 (0.034 sec/batch), lr: 0.003000
307
+ 2025-12-03 00:04:58 INFO: Finished STEP 1380/50000, loss = 3.584379 (0.035 sec/batch), lr: 0.003000
308
+ 2025-12-03 00:04:59 INFO: Finished STEP 1400/50000, loss = 3.826652 (0.032 sec/batch), lr: 0.003000
309
+ 2025-12-03 00:04:59 INFO: Evaluating on dev set...
310
+ 2025-12-03 00:04:59 INFO: LAS MLAS BLEX
311
+ 2025-12-03 00:04:59 INFO: 35.40 22.09 24.90
312
+ 2025-12-03 00:04:59 INFO: step 1400: train_loss = 3.463270, dev_score = 0.3540
313
+ 2025-12-03 00:04:59 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
314
+ 2025-12-03 00:04:59 INFO: new best model saved.
315
+ 2025-12-03 00:05:00 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
316
+ 2025-12-03 00:05:00 INFO: new model checkpoint saved.
317
+ 2025-12-03 00:05:01 INFO: Finished STEP 1420/50000, loss = 1.796135 (0.036 sec/batch), lr: 0.003000
318
+ 2025-12-03 00:05:01 INFO: Finished STEP 1440/50000, loss = 4.020522 (0.034 sec/batch), lr: 0.003000
319
+ 2025-12-03 00:05:02 INFO: Finished STEP 1460/50000, loss = 4.122330 (0.034 sec/batch), lr: 0.003000
320
+ 2025-12-03 00:05:03 INFO: Finished STEP 1480/50000, loss = 3.105598 (0.034 sec/batch), lr: 0.003000
321
+ 2025-12-03 00:05:03 INFO: Finished STEP 1500/50000, loss = 2.779820 (0.035 sec/batch), lr: 0.003000
322
+ 2025-12-03 00:05:03 INFO: Evaluating on dev set...
323
+ 2025-12-03 00:05:04 INFO: LAS MLAS BLEX
324
+ 2025-12-03 00:05:04 INFO: 34.90 19.88 21.87
325
+ 2025-12-03 00:05:04 INFO: step 1500: train_loss = 3.723689, dev_score = 0.3490
326
+ 2025-12-03 00:05:04 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
327
+ 2025-12-03 00:05:04 INFO: new model checkpoint saved.
328
+ 2025-12-03 00:05:05 INFO: Finished STEP 1520/50000, loss = 2.801723 (0.034 sec/batch), lr: 0.003000
329
+ 2025-12-03 00:05:06 INFO: Finished STEP 1540/50000, loss = 2.692841 (0.036 sec/batch), lr: 0.003000
330
+ 2025-12-03 00:05:07 INFO: Finished STEP 1560/50000, loss = 5.184442 (0.036 sec/batch), lr: 0.003000
331
+ 2025-12-03 00:05:07 INFO: Finished STEP 1580/50000, loss = 3.971715 (0.034 sec/batch), lr: 0.003000
332
+ 2025-12-03 00:05:08 INFO: Finished STEP 1600/50000, loss = 1.744106 (0.033 sec/batch), lr: 0.003000
333
+ 2025-12-03 00:05:08 INFO: Evaluating on dev set...
334
+ 2025-12-03 00:05:08 INFO: LAS MLAS BLEX
335
+ 2025-12-03 00:05:08 INFO: 33.66 18.33 19.52
336
+ 2025-12-03 00:05:08 INFO: step 1600: train_loss = 3.424519, dev_score = 0.3366
337
+ 2025-12-03 00:05:09 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
338
+ 2025-12-03 00:05:09 INFO: new model checkpoint saved.
339
+ 2025-12-03 00:05:10 INFO: Finished STEP 1620/50000, loss = 6.497481 (0.032 sec/batch), lr: 0.003000
340
+ 2025-12-03 00:05:10 INFO: Finished STEP 1640/50000, loss = 2.378686 (0.034 sec/batch), lr: 0.003000
341
+ 2025-12-03 00:05:11 INFO: Finished STEP 1660/50000, loss = 2.704184 (0.035 sec/batch), lr: 0.003000
342
+ 2025-12-03 00:05:12 INFO: Finished STEP 1680/50000, loss = 3.217742 (0.035 sec/batch), lr: 0.003000
343
+ 2025-12-03 00:05:12 INFO: Finished STEP 1700/50000, loss = 3.253656 (0.034 sec/batch), lr: 0.003000
344
+ 2025-12-03 00:05:12 INFO: Evaluating on dev set...
345
+ 2025-12-03 00:05:13 INFO: LAS MLAS BLEX
346
+ 2025-12-03 00:05:13 INFO: 36.14 20.12 21.73
347
+ 2025-12-03 00:05:13 INFO: step 1700: train_loss = 3.412348, dev_score = 0.3614
348
+ 2025-12-03 00:05:13 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
349
+ 2025-12-03 00:05:13 INFO: new best model saved.
350
+ 2025-12-03 00:05:14 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
351
+ 2025-12-03 00:05:14 INFO: new model checkpoint saved.
352
+ 2025-12-03 00:05:14 INFO: Finished STEP 1720/50000, loss = 5.873159 (0.033 sec/batch), lr: 0.003000
353
+ 2025-12-03 00:05:15 INFO: Finished STEP 1740/50000, loss = 2.128465 (0.035 sec/batch), lr: 0.003000
354
+ 2025-12-03 00:05:16 INFO: Finished STEP 1760/50000, loss = 2.606138 (0.035 sec/batch), lr: 0.003000
355
+ 2025-12-03 00:05:16 INFO: Finished STEP 1780/50000, loss = 2.763207 (0.032 sec/batch), lr: 0.003000
356
+ 2025-12-03 00:05:17 INFO: Finished STEP 1800/50000, loss = 1.491771 (0.035 sec/batch), lr: 0.003000
357
+ 2025-12-03 00:05:17 INFO: Evaluating on dev set...
358
+ 2025-12-03 00:05:18 INFO: LAS MLAS BLEX
359
+ 2025-12-03 00:05:18 INFO: 33.42 17.50 19.48
360
+ 2025-12-03 00:05:18 INFO: step 1800: train_loss = 3.318562, dev_score = 0.3342
361
+ 2025-12-03 00:05:18 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
362
+ 2025-12-03 00:05:18 INFO: new model checkpoint saved.
363
+ 2025-12-03 00:05:19 INFO: Finished STEP 1820/50000, loss = 5.831316 (0.035 sec/batch), lr: 0.003000
364
+ 2025-12-03 00:05:20 INFO: Finished STEP 1840/50000, loss = 3.740195 (0.033 sec/batch), lr: 0.003000
365
+ 2025-12-03 00:05:20 INFO: Finished STEP 1860/50000, loss = 2.089522 (0.037 sec/batch), lr: 0.003000
366
+ 2025-12-03 00:05:21 INFO: Finished STEP 1880/50000, loss = 2.194854 (0.038 sec/batch), lr: 0.003000
367
+ 2025-12-03 00:05:22 INFO: Finished STEP 1900/50000, loss = 2.892240 (0.036 sec/batch), lr: 0.003000
368
+ 2025-12-03 00:05:22 INFO: Evaluating on dev set...
369
+ 2025-12-03 00:05:22 INFO: LAS MLAS BLEX
370
+ 2025-12-03 00:05:22 INFO: 33.42 21.90 23.97
371
+ 2025-12-03 00:05:22 INFO: step 1900: train_loss = 3.385874, dev_score = 0.3342
372
+ 2025-12-03 00:05:23 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
373
+ 2025-12-03 00:05:23 INFO: new model checkpoint saved.
374
+ 2025-12-03 00:05:23 INFO: Finished STEP 1920/50000, loss = 3.134588 (0.033 sec/batch), lr: 0.003000
375
+ 2025-12-03 00:05:24 INFO: Finished STEP 1940/50000, loss = 3.754439 (0.037 sec/batch), lr: 0.003000
376
+ 2025-12-03 00:05:25 INFO: Finished STEP 1960/50000, loss = 5.545552 (0.036 sec/batch), lr: 0.003000
377
+ 2025-12-03 00:05:26 INFO: Finished STEP 1980/50000, loss = 2.778515 (0.035 sec/batch), lr: 0.003000
378
+ 2025-12-03 00:05:26 INFO: Finished STEP 2000/50000, loss = 2.672247 (0.036 sec/batch), lr: 0.003000
379
+ 2025-12-03 00:05:26 INFO: Evaluating on dev set...
380
+ 2025-12-03 00:05:27 INFO: LAS MLAS BLEX
381
+ 2025-12-03 00:05:27 INFO: 35.15 20.90 22.54
382
+ 2025-12-03 00:05:27 INFO: step 2000: train_loss = 3.640794, dev_score = 0.3515
383
+ 2025-12-03 00:05:27 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
384
+ 2025-12-03 00:05:27 INFO: new model checkpoint saved.
385
+ 2025-12-03 00:05:28 INFO: Finished STEP 2020/50000, loss = 2.810196 (0.035 sec/batch), lr: 0.003000
386
+ 2025-12-03 00:05:29 INFO: Finished STEP 2040/50000, loss = 1.569465 (0.033 sec/batch), lr: 0.003000
387
+ 2025-12-03 00:05:29 INFO: Finished STEP 2060/50000, loss = 3.126494 (0.034 sec/batch), lr: 0.003000
388
+ 2025-12-03 00:05:30 INFO: Finished STEP 2080/50000, loss = 2.120464 (0.034 sec/batch), lr: 0.003000
389
+ 2025-12-03 00:05:31 INFO: Finished STEP 2100/50000, loss = 4.094249 (0.036 sec/batch), lr: 0.003000
390
+ 2025-12-03 00:05:31 INFO: Evaluating on dev set...
391
+ 2025-12-03 00:05:31 INFO: LAS MLAS BLEX
392
+ 2025-12-03 00:05:31 INFO: 31.68 19.60 20.40
393
+ 2025-12-03 00:05:31 INFO: step 2100: train_loss = 3.533487, dev_score = 0.3168
394
+ 2025-12-03 00:05:32 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
395
+ 2025-12-03 00:05:32 INFO: new model checkpoint saved.
396
+ 2025-12-03 00:05:33 INFO: Finished STEP 2120/50000, loss = 2.959285 (0.035 sec/batch), lr: 0.003000
397
+ 2025-12-03 00:05:33 INFO: Finished STEP 2140/50000, loss = 3.287594 (0.033 sec/batch), lr: 0.003000
398
+ 2025-12-03 00:05:34 INFO: Finished STEP 2160/50000, loss = 4.220065 (0.034 sec/batch), lr: 0.003000
399
+ 2025-12-03 00:05:35 INFO: Finished STEP 2180/50000, loss = 3.364312 (0.034 sec/batch), lr: 0.003000
400
+ 2025-12-03 00:05:35 INFO: Finished STEP 2200/50000, loss = 4.725568 (0.034 sec/batch), lr: 0.003000
401
+ 2025-12-03 00:05:35 INFO: Evaluating on dev set...
402
+ 2025-12-03 00:05:36 INFO: LAS MLAS BLEX
403
+ 2025-12-03 00:05:36 INFO: 30.45 15.95 16.77
404
+ 2025-12-03 00:05:36 INFO: step 2200: train_loss = 3.346921, dev_score = 0.3045
405
+ 2025-12-03 00:05:36 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
406
+ 2025-12-03 00:05:36 INFO: new model checkpoint saved.
407
+ 2025-12-03 00:05:37 INFO: Finished STEP 2220/50000, loss = 2.388415 (0.034 sec/batch), lr: 0.003000
408
+ 2025-12-03 00:05:38 INFO: Finished STEP 2240/50000, loss = 5.593441 (0.033 sec/batch), lr: 0.003000
409
+ 2025-12-03 00:05:39 INFO: Finished STEP 2260/50000, loss = 3.226351 (0.035 sec/batch), lr: 0.003000
410
+ 2025-12-03 00:05:39 INFO: Finished STEP 2280/50000, loss = 2.496039 (0.037 sec/batch), lr: 0.003000
411
+ 2025-12-03 00:05:40 INFO: Finished STEP 2300/50000, loss = 3.316859 (0.037 sec/batch), lr: 0.003000
412
+ 2025-12-03 00:05:40 INFO: Evaluating on dev set...
413
+ 2025-12-03 00:05:40 INFO: LAS MLAS BLEX
414
+ 2025-12-03 00:05:40 INFO: 36.39 21.67 23.33
415
+ 2025-12-03 00:05:40 INFO: step 2300: train_loss = 3.677127, dev_score = 0.3639
416
+ 2025-12-03 00:05:41 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
417
+ 2025-12-03 00:05:41 INFO: new best model saved.
418
+ 2025-12-03 00:05:41 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
419
+ 2025-12-03 00:05:41 INFO: new model checkpoint saved.
420
+ 2025-12-03 00:05:42 INFO: Finished STEP 2320/50000, loss = 4.229099 (0.034 sec/batch), lr: 0.003000
421
+ 2025-12-03 00:05:43 INFO: Finished STEP 2340/50000, loss = 4.050467 (0.035 sec/batch), lr: 0.003000
422
+ 2025-12-03 00:05:43 INFO: Finished STEP 2360/50000, loss = 4.414966 (0.037 sec/batch), lr: 0.003000
423
+ 2025-12-03 00:05:44 INFO: Finished STEP 2380/50000, loss = 2.851241 (0.037 sec/batch), lr: 0.003000
424
+ 2025-12-03 00:05:45 INFO: Finished STEP 2400/50000, loss = 4.186286 (0.034 sec/batch), lr: 0.003000
425
+ 2025-12-03 00:05:45 INFO: Evaluating on dev set...
426
+ 2025-12-03 00:05:45 INFO: LAS MLAS BLEX
427
+ 2025-12-03 00:05:45 INFO: 38.86 21.95 23.60
428
+ 2025-12-03 00:05:45 INFO: step 2400: train_loss = 3.684238, dev_score = 0.3886
429
+ 2025-12-03 00:05:46 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
430
+ 2025-12-03 00:05:46 INFO: new best model saved.
431
+ 2025-12-03 00:05:46 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
432
+ 2025-12-03 00:05:46 INFO: new model checkpoint saved.
433
+ 2025-12-03 00:05:47 INFO: Finished STEP 2420/50000, loss = 1.748532 (0.037 sec/batch), lr: 0.003000
434
+ 2025-12-03 00:05:47 INFO: Finished STEP 2440/50000, loss = 3.137299 (0.034 sec/batch), lr: 0.003000
435
+ 2025-12-03 00:05:48 INFO: Finished STEP 2460/50000, loss = 3.227959 (0.035 sec/batch), lr: 0.003000
436
+ 2025-12-03 00:05:49 INFO: Finished STEP 2480/50000, loss = 2.614409 (0.039 sec/batch), lr: 0.003000
437
+ 2025-12-03 00:05:50 INFO: Finished STEP 2500/50000, loss = 7.968997 (0.037 sec/batch), lr: 0.003000
438
+ 2025-12-03 00:05:50 INFO: Evaluating on dev set...
439
+ 2025-12-03 00:05:50 INFO: LAS MLAS BLEX
440
+ 2025-12-03 00:05:50 INFO: 35.64 18.18 21.21
441
+ 2025-12-03 00:05:50 INFO: step 2500: train_loss = 3.618949, dev_score = 0.3564
442
+ 2025-12-03 00:05:51 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
443
+ 2025-12-03 00:05:51 INFO: new model checkpoint saved.
444
+ 2025-12-03 00:05:51 INFO: Finished STEP 2520/50000, loss = 2.584875 (0.033 sec/batch), lr: 0.003000
445
+ 2025-12-03 00:05:52 INFO: Finished STEP 2540/50000, loss = 2.403082 (0.033 sec/batch), lr: 0.003000
446
+ 2025-12-03 00:05:53 INFO: Finished STEP 2560/50000, loss = 2.680532 (0.037 sec/batch), lr: 0.003000
447
+ 2025-12-03 00:05:54 INFO: Finished STEP 2580/50000, loss = 3.621366 (0.036 sec/batch), lr: 0.003000
448
+ 2025-12-03 00:05:54 INFO: Finished STEP 2600/50000, loss = 2.256649 (0.035 sec/batch), lr: 0.003000
449
+ 2025-12-03 00:05:54 INFO: Evaluating on dev set...
450
+ 2025-12-03 00:05:55 INFO: LAS MLAS BLEX
451
+ 2025-12-03 00:05:55 INFO: 39.11 24.37 26.47
452
+ 2025-12-03 00:05:55 INFO: step 2600: train_loss = 3.325648, dev_score = 0.3911
453
+ 2025-12-03 00:05:55 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
454
+ 2025-12-03 00:05:55 INFO: new best model saved.
455
+ 2025-12-03 00:05:56 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
456
+ 2025-12-03 00:05:56 INFO: new model checkpoint saved.
457
+ 2025-12-03 00:05:56 INFO: Finished STEP 2620/50000, loss = 5.835835 (0.035 sec/batch), lr: 0.003000
458
+ 2025-12-03 00:05:57 INFO: Finished STEP 2640/50000, loss = 3.077916 (0.035 sec/batch), lr: 0.003000
459
+ 2025-12-03 00:05:58 INFO: Finished STEP 2660/50000, loss = 4.020663 (0.036 sec/batch), lr: 0.003000
460
+ 2025-12-03 00:05:59 INFO: Finished STEP 2680/50000, loss = 4.354639 (0.033 sec/batch), lr: 0.003000
461
+ 2025-12-03 00:05:59 INFO: Finished STEP 2700/50000, loss = 2.966883 (0.035 sec/batch), lr: 0.003000
462
+ 2025-12-03 00:05:59 INFO: Evaluating on dev set...
463
+ 2025-12-03 00:06:00 INFO: LAS MLAS BLEX
464
+ 2025-12-03 00:06:00 INFO: 42.33 29.11 31.19
465
+ 2025-12-03 00:06:00 INFO: step 2700: train_loss = 3.449816, dev_score = 0.4233
466
+ 2025-12-03 00:06:00 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser.pt
467
+ 2025-12-03 00:06:00 INFO: new best model saved.
468
+ 2025-12-03 00:06:01 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
469
+ 2025-12-03 00:06:01 INFO: new model checkpoint saved.
470
+ 2025-12-03 00:06:01 INFO: Finished STEP 2720/50000, loss = 4.358669 (0.036 sec/batch), lr: 0.003000
471
+ 2025-12-03 00:06:02 INFO: Finished STEP 2740/50000, loss = 1.960829 (0.037 sec/batch), lr: 0.003000
472
+ 2025-12-03 00:06:03 INFO: Finished STEP 2760/50000, loss = 3.342113 (0.034 sec/batch), lr: 0.003000
473
+ 2025-12-03 00:06:03 INFO: Finished STEP 2780/50000, loss = 1.876661 (0.036 sec/batch), lr: 0.003000
474
+ 2025-12-03 00:06:04 INFO: Finished STEP 2800/50000, loss = 1.878595 (0.040 sec/batch), lr: 0.003000
475
+ 2025-12-03 00:06:04 INFO: Evaluating on dev set...
476
+ 2025-12-03 00:06:05 INFO: LAS MLAS BLEX
477
+ 2025-12-03 00:06:05 INFO: 32.43 17.26 20.21
478
+ 2025-12-03 00:06:05 INFO: step 2800: train_loss = 3.684543, dev_score = 0.3243
479
+ 2025-12-03 00:06:05 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
480
+ 2025-12-03 00:06:05 INFO: new model checkpoint saved.
481
+ 2025-12-03 00:06:06 INFO: Finished STEP 2820/50000, loss = 7.423325 (0.033 sec/batch), lr: 0.003000
482
+ 2025-12-03 00:06:07 INFO: Finished STEP 2840/50000, loss = 3.846512 (0.038 sec/batch), lr: 0.003000
483
+ 2025-12-03 00:06:07 INFO: Finished STEP 2860/50000, loss = 4.674613 (0.035 sec/batch), lr: 0.003000
484
+ 2025-12-03 00:06:08 INFO: Finished STEP 2880/50000, loss = 5.598680 (0.034 sec/batch), lr: 0.003000
485
+ 2025-12-03 00:06:09 INFO: Finished STEP 2900/50000, loss = 4.584077 (0.038 sec/batch), lr: 0.003000
486
+ 2025-12-03 00:06:09 INFO: Evaluating on dev set...
487
+ 2025-12-03 00:06:10 INFO: LAS MLAS BLEX
488
+ 2025-12-03 00:06:10 INFO: 37.13 21.10 23.21
489
+ 2025-12-03 00:06:10 INFO: step 2900: train_loss = 3.578357, dev_score = 0.3713
490
+ 2025-12-03 00:06:10 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
491
+ 2025-12-03 00:06:10 INFO: new model checkpoint saved.
492
+ 2025-12-03 00:06:11 INFO: Finished STEP 2920/50000, loss = 2.098350 (0.035 sec/batch), lr: 0.003000
493
+ 2025-12-03 00:06:11 INFO: Finished STEP 2940/50000, loss = 7.716115 (0.036 sec/batch), lr: 0.003000
494
+ 2025-12-03 00:06:12 INFO: Finished STEP 2960/50000, loss = 3.773098 (0.035 sec/batch), lr: 0.003000
495
+ 2025-12-03 00:06:13 INFO: Finished STEP 2980/50000, loss = 2.773946 (0.036 sec/batch), lr: 0.003000
496
+ 2025-12-03 00:06:14 INFO: Finished STEP 3000/50000, loss = 3.093710 (0.039 sec/batch), lr: 0.003000
497
+ 2025-12-03 00:06:14 INFO: Evaluating on dev set...
498
+ 2025-12-03 00:06:14 INFO: LAS MLAS BLEX
499
+ 2025-12-03 00:06:14 INFO: 38.12 20.92 23.85
500
+ 2025-12-03 00:06:14 INFO: step 3000: train_loss = 3.502959, dev_score = 0.3812
501
+ 2025-12-03 00:06:15 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
502
+ 2025-12-03 00:06:15 INFO: new model checkpoint saved.
503
+ 2025-12-03 00:06:16 INFO: Finished STEP 3020/50000, loss = 2.860210 (0.039 sec/batch), lr: 0.003000
504
+ 2025-12-03 00:06:16 INFO: Finished STEP 3040/50000, loss = 3.387549 (0.036 sec/batch), lr: 0.003000
505
+ 2025-12-03 00:06:17 INFO: Finished STEP 3060/50000, loss = 3.077563 (0.036 sec/batch), lr: 0.003000
506
+ 2025-12-03 00:06:18 INFO: Finished STEP 3080/50000, loss = 2.989982 (0.033 sec/batch), lr: 0.003000
507
+ 2025-12-03 00:06:19 INFO: Finished STEP 3100/50000, loss = 3.402043 (0.038 sec/batch), lr: 0.003000
508
+ 2025-12-03 00:06:19 INFO: Evaluating on dev set...
509
+ 2025-12-03 00:06:19 INFO: LAS MLAS BLEX
510
+ 2025-12-03 00:06:19 INFO: 37.38 22.04 24.08
511
+ 2025-12-03 00:06:19 INFO: step 3100: train_loss = 3.450367, dev_score = 0.3738
512
+ 2025-12-03 00:06:20 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
513
+ 2025-12-03 00:06:20 INFO: new model checkpoint saved.
514
+ 2025-12-03 00:06:20 INFO: Finished STEP 3120/50000, loss = 3.050076 (0.034 sec/batch), lr: 0.003000
515
+ 2025-12-03 00:06:21 INFO: Finished STEP 3140/50000, loss = 3.250776 (0.037 sec/batch), lr: 0.003000
516
+ 2025-12-03 00:06:22 INFO: Finished STEP 3160/50000, loss = 3.034782 (0.034 sec/batch), lr: 0.003000
517
+ 2025-12-03 00:06:22 INFO: Finished STEP 3180/50000, loss = 2.803900 (0.033 sec/batch), lr: 0.003000
518
+ 2025-12-03 00:06:23 INFO: Finished STEP 3200/50000, loss = 3.034755 (0.038 sec/batch), lr: 0.003000
519
+ 2025-12-03 00:06:23 INFO: Evaluating on dev set...
520
+ 2025-12-03 00:06:24 INFO: LAS MLAS BLEX
521
+ 2025-12-03 00:06:24 INFO: 36.63 19.29 21.80
522
+ 2025-12-03 00:06:24 INFO: step 3200: train_loss = 3.493535, dev_score = 0.3663
523
+ 2025-12-03 00:06:24 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
524
+ 2025-12-03 00:06:24 INFO: new model checkpoint saved.
525
+ 2025-12-03 00:06:25 INFO: Finished STEP 3220/50000, loss = 2.634099 (0.036 sec/batch), lr: 0.003000
526
+ 2025-12-03 00:06:26 INFO: Finished STEP 3240/50000, loss = 7.289287 (0.036 sec/batch), lr: 0.003000
527
+ 2025-12-03 00:06:27 INFO: Finished STEP 3260/50000, loss = 3.269890 (0.039 sec/batch), lr: 0.003000
528
+ 2025-12-03 00:06:27 INFO: Finished STEP 3280/50000, loss = 3.867342 (0.039 sec/batch), lr: 0.003000
529
+ 2025-12-03 00:06:28 INFO: Finished STEP 3300/50000, loss = 2.824182 (0.038 sec/batch), lr: 0.003000
530
+ 2025-12-03 00:06:28 INFO: Evaluating on dev set...
531
+ 2025-12-03 00:06:29 INFO: LAS MLAS BLEX
532
+ 2025-12-03 00:06:29 INFO: 36.63 20.37 21.62
533
+ 2025-12-03 00:06:29 INFO: step 3300: train_loss = 3.637132, dev_score = 0.3663
534
+ 2025-12-03 00:06:29 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
535
+ 2025-12-03 00:06:29 INFO: new model checkpoint saved.
536
+ 2025-12-03 00:06:30 INFO: Finished STEP 3320/50000, loss = 2.506874 (0.036 sec/batch), lr: 0.003000
537
+ 2025-12-03 00:06:31 INFO: Finished STEP 3340/50000, loss = 2.887654 (0.034 sec/batch), lr: 0.003000
538
+ 2025-12-03 00:06:31 INFO: Finished STEP 3360/50000, loss = 3.862370 (0.037 sec/batch), lr: 0.003000
539
+ 2025-12-03 00:06:32 INFO: Finished STEP 3380/50000, loss = 4.608120 (0.034 sec/batch), lr: 0.003000
540
+ 2025-12-03 00:06:33 INFO: Finished STEP 3400/50000, loss = 3.780773 (0.039 sec/batch), lr: 0.003000
541
+ 2025-12-03 00:06:33 INFO: Evaluating on dev set...
542
+ 2025-12-03 00:06:34 INFO: LAS MLAS BLEX
543
+ 2025-12-03 00:06:34 INFO: 38.12 21.21 23.70
544
+ 2025-12-03 00:06:34 INFO: step 3400: train_loss = 3.844786, dev_score = 0.3812
545
+ 2025-12-03 00:06:34 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
546
+ 2025-12-03 00:06:34 INFO: new model checkpoint saved.
547
+ 2025-12-03 00:06:35 INFO: Finished STEP 3420/50000, loss = 3.224651 (0.036 sec/batch), lr: 0.003000
548
+ 2025-12-03 00:06:36 INFO: Finished STEP 3440/50000, loss = 4.342725 (0.037 sec/batch), lr: 0.003000
549
+ 2025-12-03 00:06:36 INFO: Finished STEP 3460/50000, loss = 15.964293 (0.039 sec/batch), lr: 0.003000
550
+ 2025-12-03 00:06:37 INFO: Finished STEP 3480/50000, loss = 9.027036 (0.032 sec/batch), lr: 0.003000
551
+ 2025-12-03 00:06:38 INFO: Finished STEP 3500/50000, loss = 3.789010 (0.034 sec/batch), lr: 0.003000
552
+ 2025-12-03 00:06:38 INFO: Evaluating on dev set...
553
+ 2025-12-03 00:06:38 INFO: LAS MLAS BLEX
554
+ 2025-12-03 00:06:38 INFO: 41.58 26.92 31.06
555
+ 2025-12-03 00:06:38 INFO: step 3500: train_loss = 3.767138, dev_score = 0.4158
556
+ 2025-12-03 00:06:39 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
557
+ 2025-12-03 00:06:39 INFO: new model checkpoint saved.
558
+ 2025-12-03 00:06:40 INFO: Finished STEP 3520/50000, loss = 6.266014 (0.035 sec/batch), lr: 0.003000
559
+ 2025-12-03 00:06:40 INFO: Finished STEP 3540/50000, loss = 2.677692 (0.036 sec/batch), lr: 0.003000
560
+ 2025-12-03 00:06:41 INFO: Finished STEP 3560/50000, loss = 2.750907 (0.037 sec/batch), lr: 0.003000
561
+ 2025-12-03 00:06:42 INFO: Finished STEP 3580/50000, loss = 2.846963 (0.036 sec/batch), lr: 0.003000
562
+ 2025-12-03 00:06:43 INFO: Finished STEP 3600/50000, loss = 3.642042 (0.038 sec/batch), lr: 0.003000
563
+ 2025-12-03 00:06:43 INFO: Evaluating on dev set...
564
+ 2025-12-03 00:06:43 INFO: LAS MLAS BLEX
565
+ 2025-12-03 00:06:43 INFO: 40.59 27.00 30.80
566
+ 2025-12-03 00:06:43 INFO: step 3600: train_loss = 3.839358, dev_score = 0.4059
567
+ 2025-12-03 00:06:44 INFO: Model saved to saved_models/depparse/sv_diachronic_charlm_parser_checkpoint.pt
568
+ 2025-12-03 00:06:44 INFO: new model checkpoint saved.
569
+ 2025-12-03 00:06:44 INFO: Finished STEP 3620/50000, loss = 3.241040 (0.039 sec/batch), lr: 0.003000
570
+ 2025-12-03 00:06:45 INFO: Finished STEP 3640/50000, loss = 2.835604 (0.036 sec/batch), lr: 0.003000
571
+ 2025-12-03 00:06:46 INFO: Finished STEP 3660/50000, loss = 2.748945 (0.036 sec/batch), lr: 0.003000
572
+ 2025-12-03 00:06:47 INFO: Finished STEP 3680/50000, loss = 3.752511 (0.039 sec/batch), lr: 0.003000
573
+ 2025-12-03 00:06:47 INFO: Finished STEP 3700/50000, loss = 2.389401 (0.036 sec/batch), lr: 0.003000
574
+ 2025-12-03 00:06:47 INFO: Evaluating on dev set...
575
+ 2025-12-03 00:06:48 INFO: LAS MLAS BLEX
576
+ 2025-12-03 00:06:48 INFO: 34.16 19.70 27.41
577
+ 2025-12-03 00:06:48 INFO: step 3700: train_loss = 3.949752, dev_score = 0.3416
578
+ 2025-12-03 00:06:48 INFO: Training ended with 3700 steps.
579
+ 2025-12-03 00:06:48 INFO: Best dev F1 = 42.33, at iteration = 2700
580
+ 2025-12-03 00:06:48 INFO: Running dev depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '16', '--dropout', '0.33']
581
+ 2025-12-03 00:06:48 INFO: Running parser in predict mode
582
+ 2025-12-03 00:06:48 INFO: Loading model from: saved_models/depparse/sv_diachronic_charlm_parser.pt
583
+ 2025-12-03 00:06:51 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
584
+ 2025-12-03 00:06:51 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
585
+ 2025-12-03 00:06:51 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
586
+ 2025-12-03 00:06:51 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
587
+ 2025-12-03 00:06:51 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
588
+ 2025-12-03 00:06:51 INFO: Loading data with batch size 16...
589
+ 2025-12-03 00:06:51 DEBUG: 9 batches created.
590
+ 2025-12-03 00:06:52 INFO: F1 scores for each dependency:
591
+ Note that unlabeled attachment errors hurt the labeled attachment scores
592
+ acl: p 0.0000 r 0.0000 f1 0.0000 (3 actual)
593
+ acl:relcl: p 0.0000 r 0.0000 f1 0.0000 (7 actual)
594
+ advcl: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
595
+ advmod: p 0.3243 r 0.4800 f1 0.3871 (25 actual)
596
+ amod: p 0.7857 r 0.7097 f1 0.7458 (31 actual)
597
+ appos: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
598
+ aux: p 0.6000 r 0.5455 f1 0.5714 (11 actual)
599
+ case: p 0.7857 r 0.7857 f1 0.7857 (56 actual)
600
+ cc: p 0.5000 r 0.5385 f1 0.5185 (13 actual)
601
+ ccomp: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
602
+ conj: p 0.0000 r 0.0000 f1 0.0000 (12 actual)
603
+ cop: p 0.0000 r 0.0000 f1 0.0000 (3 actual)
604
+ csubj: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
605
+ det: p 0.8636 r 0.8636 f1 0.8636 (22 actual)
606
+ expl: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
607
+ iobj: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
608
+ mark: p 1.0000 r 0.1667 f1 0.2857 (12 actual)
609
+ nmod: p 0.1311 r 0.5333 f1 0.2105 (15 actual)
610
+ nmod:poss: p 0.0000 r 0.0000 f1 0.0000 (19 actual)
611
+ nsubj: p 0.1569 r 0.4706 f1 0.2353 (17 actual)
612
+ nsubj:pass: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
613
+ obj: p 0.8333 r 0.2273 f1 0.3571 (22 actual)
614
+ obl: p 0.2778 r 0.3659 f1 0.3158 (41 actual)
615
+ obl:agent: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
616
+ orphan: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
617
+ parataxis: p 0.0000 r 0.0000 f1 0.0000 (3 actual)
618
+ punct: p 0.3462 r 0.3462 f1 0.3462 (52 actual)
619
+ root: p 0.4444 r 0.4444 f1 0.4444 (9 actual)
620
+ xcomp: p 0.0000 r 0.0000 f1 0.0000 (8 actual)
621
+ 2025-12-03 00:06:52 INFO: LAS MLAS BLEX
622
+ 2025-12-03 00:06:52 INFO: 42.33 29.11 31.19
623
+ 2025-12-03 00:06:52 INFO: Parser score:
624
+ 2025-12-03 00:06:52 INFO: sv_diachronic 42.33
625
+ 2025-12-03 00:06:52 INFO: Finished running dev set on
626
+ UD_Swedish-diachronic
627
+ UAS LAS CLAS MLAS BLEX
628
+ 56.19 42.33 31.19 29.11 31.19
629
+ 2025-12-03 00:06:52 INFO: Running test depparse for UD_Swedish-diachronic with args ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--eval_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--batch_size', '16', '--dropout', '0.33']
630
+ 2025-12-03 00:06:52 INFO: Running parser in predict mode
631
+ 2025-12-03 00:06:52 INFO: Loading model from: saved_models/depparse/sv_diachronic_charlm_parser.pt
632
+ 2025-12-03 00:06:55 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
633
+ 2025-12-03 00:06:55 DEBUG: Depparse model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
634
+ 2025-12-03 00:06:55 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
635
+ 2025-12-03 00:06:55 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
636
+ 2025-12-03 00:06:55 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
637
+ 2025-12-03 00:06:55 INFO: Loading data with batch size 16...
638
+ 2025-12-03 00:06:55 DEBUG: 98 batches created.
639
+ 2025-12-03 00:07:00 INFO: F1 scores for each dependency:
640
+ Note that unlabeled attachment errors hurt the labeled attachment scores
641
+ acl: p 0.0000 r 0.0000 f1 0.0000 (32 actual)
642
+ acl:cleft: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
643
+ acl:relcl: p 0.0000 r 0.0000 f1 0.0000 (75 actual)
644
+ advcl: p 0.0000 r 0.0000 f1 0.0000 (60 actual)
645
+ advcl:relcl: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
646
+ advmod: p 0.3673 r 0.5373 f1 0.4364 (268 actual)
647
+ amod: p 0.7559 r 0.7000 f1 0.7269 (230 actual)
648
+ appos: p 0.0000 r 0.0000 f1 0.0000 (13 actual)
649
+ aux: p 0.5926 r 0.7619 f1 0.6667 (84 actual)
650
+ aux:pass: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
651
+ case: p 0.7021 r 0.7265 f1 0.7141 (373 actual)
652
+ cc: p 0.5075 r 0.4387 f1 0.4706 (155 actual)
653
+ ccomp: p 0.0000 r 0.0000 f1 0.0000 (35 actual)
654
+ compound:prt: p 0.0000 r 0.0000 f1 0.0000 (21 actual)
655
+ conj: p 0.0000 r 0.0000 f1 0.0000 (158 actual)
656
+ cop: p 0.7143 r 0.1087 f1 0.1887 (46 actual)
657
+ csubj: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
658
+ dep: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
659
+ det: p 0.7980 r 0.7788 f1 0.7883 (208 actual)
660
+ discourse: p 0.0000 r 0.0000 f1 0.0000 (7 actual)
661
+ dislocated: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
662
+ expl: p 0.0000 r 0.0000 f1 0.0000 (11 actual)
663
+ expl:pv: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
664
+ fixed: p 0.0000 r 0.0000 f1 0.0000 (8 actual)
665
+ flat: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
666
+ flat:name: p 0.0000 r 0.0000 f1 0.0000 (12 actual)
667
+ goeswith: p 0.0000 r 0.0000 f1 0.0000 (2 actual)
668
+ iobj: p 0.0000 r 0.0000 f1 0.0000 (14 actual)
669
+ mark: p 0.8333 r 0.2614 f1 0.3980 (153 actual)
670
+ nmod: p 0.1290 r 0.4706 f1 0.2025 (102 actual)
671
+ nmod:poss: p 0.0000 r 0.0000 f1 0.0000 (142 actual)
672
+ nsubj: p 0.2644 r 0.5071 f1 0.3476 (280 actual)
673
+ nsubj:pass: p 0.0000 r 0.0000 f1 0.0000 (25 actual)
674
+ nummod: p 0.0000 r 0.0000 f1 0.0000 (10 actual)
675
+ obj: p 0.4595 r 0.0929 f1 0.1545 (183 actual)
676
+ obl: p 0.1677 r 0.4029 f1 0.2368 (278 actual)
677
+ obl:agent: p 0.0000 r 0.0000 f1 0.0000 (4 actual)
678
+ orphan: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
679
+ parataxis: p 0.0000 r 0.0000 f1 0.0000 (18 actual)
680
+ punct: p 0.3357 r 0.3365 f1 0.3361 (425 actual)
681
+ reparandum: p 0.0000 r 0.0000 f1 0.0000 (1 actual)
682
+ root: p 0.4444 r 0.4444 f1 0.4444 (99 actual)
683
+ vocative: p 0.0000 r 0.0000 f1 0.0000 (5 actual)
684
+ xcomp: p 0.0000 r 0.0000 f1 0.0000 (75 actual)
685
+ 2025-12-03 00:07:00 INFO: LAS MLAS BLEX
686
+ 2025-12-03 00:07:00 INFO: 39.53 27.99 30.21
687
+ 2025-12-03 00:07:00 INFO: Parser score:
688
+ 2025-12-03 00:07:00 INFO: sv_diachronic 39.53
689
+ 2025-12-03 00:07:00 INFO: Finished running test set on
690
+ UD_Swedish-diachronic
691
+ UAS LAS CLAS MLAS BLEX
692
+ 56.75 39.53 30.21 27.99 30.21
693
+ DONE.
694
+ Full log saved to: logs/log_diachronic.pt_sv_20251203_000134.txt
695
+ Symlink updated: logs/latest.txt → log_diachronic.pt_sv_20251203_000134.txt
logs/log_diachronic.pt_sv_diachron_nn_20251203_001047.txt ADDED
@@ -0,0 +1,190 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === LOGFILE: logs/log_diachronic.pt_sv_diachron_nn_20251203_001047.txt ===
2
+ Language codes: sv diachron nn
3
+ Using pretrained model: diachronic.pt
4
+
5
+ Running: python prepare-train-val-test.py sv diachron nn
6
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-dev.conllu
7
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test.conllu
8
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_pud-ud-test.conllu
9
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-test.conllu
10
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test-trg.conllu
11
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-dev.conllu
12
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/ucxn_ud_swedish-talbanken.conllu
13
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-train.conllu
14
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_old-ud-test.conllu
15
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-train.conllu
16
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-test.conllu
17
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-test.conllu
18
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-train.conllu
19
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-dev.conllu
20
+ Including DigPhil MACHINE in TRAIN (minus gold)…
21
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec330-GyllenborgC_SwenskaSpratthoken.conllu
22
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec254-CederborghF_BerattelseOmJohnHall.conllu
23
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec277-EnbomPU_MedborgeligtSkalde.conllu
24
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec268-DulciU_VitterhetsNojen3.conllu
25
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1063-spf220.conllu
26
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec397-AngeredStrandbergH_UnderSodernsSol.conllu
27
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec324-GranbergPA_Enslighetsalskaren.conllu
28
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec252-BremerF_Teckningar1.conllu
29
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec988-spf145.conllu
30
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec987-spf144.conllu
31
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec631-HasselskogN_HallaHallaGronkoping.conllu
32
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-letter141673-Stalhammar.conllu
33
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1033-spf190.conllu
34
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec25-Runius.conllu
35
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec486-SchwartzMS_BellmansSkor.conllu
36
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec452-NyblomH_FantasierFyra.conllu
37
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec613-EngstromA_StrindbergOchJag.conllu
38
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec208-Anonym_DetGrasligaMordet.conllu
39
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec639-HeidenstamV_Proletarfilosofiens.conllu
40
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1102-spf259.conllu
41
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec991-spf148.conllu
42
+ Cleaning TRAIN...
43
+ [REMOVED] sent_id=6 ERRORS=['Line 24: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 29: Invalid token ID or head', 'Token 30 has invalid head 24']
44
+ [REMOVED] sent_id=7_8 ERRORS=['Multiple roots found: [5, 10]']
45
+ [REMOVED] sent_id=30_31 ERRORS=['Multiple roots found: [3, 18]']
46
+ [REMOVED] sent_id=35 ERRORS=['Line 36: Invalid token ID or head']
47
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [1, 5]']
48
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [7, 20]']
49
+ [REMOVED] sent_id=8_9 ERRORS=['Multiple roots found: [24, 57]']
50
+ [REMOVED] sent_id=12_13 ERRORS=['Multiple roots found: [11, 16]']
51
+ [REMOVED] sent_id=124_split2 ERRORS=['Line 4: Invalid token ID or head', 'No root found', 'Token 1 has invalid head 4', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 6 has invalid head 4', 'Token 11 has invalid head 4', 'Token 15 has invalid head 4']
52
+ [REMOVED] sent_id=396 ERRORS=['Token 2: Missing form']
53
+ [REMOVED] sent_id=416 ERRORS=['Token 2: Missing form']
54
+ [REMOVED] sent_id=589 ERRORS=['Token 2: Missing form']
55
+ [REMOVED] sent_id=909 ERRORS=['Token 2: Missing form']
56
+ [REMOVED] sent_id=912 ERRORS=['Token 2: Missing form']
57
+ [REMOVED] sent_id=3_split1 ERRORS=['Multiple roots found: [4, 15, 17]']
58
+ [REMOVED] sent_id=3_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 8: Invalid token ID or head', 'Line 15: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1', 'Token 3 has invalid head 8', 'Token 4 has invalid head 8', 'Token 5 has invalid head 8', 'Token 7 has invalid head 8', 'Token 10 has invalid head 8', 'Token 13 has invalid head 8', 'Token 14 has invalid head 8']
59
+ [REMOVED] sent_id=3_4 ERRORS=['Multiple roots found: [1, 5]']
60
+ [REMOVED] sent_id=5_6 ERRORS=['Multiple roots found: [3, 24]']
61
+ [REMOVED] sent_id=11_12_13 ERRORS=['Multiple roots found: [5, 17, 25]']
62
+ [REMOVED] sent_id=119 ERRORS=['Token 2: Missing form']
63
+ [REMOVED] sent_id=179 ERRORS=['Token 2: Missing form']
64
+ [REMOVED] sent_id=188 ERRORS=['Token 2: Missing form']
65
+ [REMOVED] sent_id=223 ERRORS=['Token 2: Missing form']
66
+ [REMOVED] sent_id=268 ERRORS=['Token 2: Missing form']
67
+ [REMOVED] sent_id=325 ERRORS=['Token 2: Missing form']
68
+ [REMOVED] sent_id=388 ERRORS=['Token 2: Missing form']
69
+ [REMOVED] sent_id=399 ERRORS=['Token 2: Missing form']
70
+ [REMOVED] sent_id=475 ERRORS=['Token 2: Missing form']
71
+ [REMOVED] sent_id=505 ERRORS=['Token 2: Missing form']
72
+ [REMOVED] sent_id=520 ERRORS=['Token 2: Missing form']
73
+ [REMOVED] sent_id=562 ERRORS=['Token 2: Missing form']
74
+ [REMOVED] sent_id=669 ERRORS=['Token 2: Missing form']
75
+ [REMOVED] sent_id=711 ERRORS=['Token 2: Missing form']
76
+ [REMOVED] sent_id=731 ERRORS=['Token 2: Missing form']
77
+ [REMOVED] sent_id=867 ERRORS=['Token 2: Missing form']
78
+ [REMOVED] sent_id=884 ERRORS=['Token 2: Missing form']
79
+ [REMOVED] sent_id=923 ERRORS=['Token 2: Missing form']
80
+ [REMOVED] sent_id=939 ERRORS=['Token 2: Missing form']
81
+ [REMOVED] sent_id=1086 ERRORS=['Token 2: Missing form']
82
+ [REMOVED] sent_id=1179 ERRORS=['Token 2: Missing form']
83
+ [REMOVED] sent_id=1251 ERRORS=['Token 2: Missing form']
84
+ [REMOVED] sent_id=1345 ERRORS=['Token 2: Missing form']
85
+ [REMOVED] sent_id=1459 ERRORS=['Token 2: Missing form']
86
+ [REMOVED] sent_id=1656 ERRORS=['Token 2: Missing form']
87
+ [REMOVED] sent_id=1669 ERRORS=['Token 2: Missing form']
88
+ [REMOVED] sent_id=87_88 ERRORS=['Multiple roots found: [3, 6]']
89
+ [REMOVED] sent_id=65_split2_66_split2 ERRORS=['Line 4: Invalid token ID or head', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 5 has invalid head 4']
90
+ [REMOVED] sent_id=25 ERRORS=['Token 2: Missing form']
91
+ [REMOVED] sent_id=136 ERRORS=['Token 2: Missing form']
92
+ [REMOVED] sent_id=208 ERRORS=['Token 2: Missing form']
93
+ [REMOVED] sent_id=230 ERRORS=['Token 2: Missing form']
94
+ [REMOVED] sent_id=245 ERRORS=['Token 2: Missing form']
95
+ [REMOVED] sent_id=276 ERRORS=['Token 2: Missing form']
96
+ [REMOVED] sent_id=320 ERRORS=['Token 2: Missing form']
97
+ [REMOVED] sent_id=366 ERRORS=['Token 2: Missing form']
98
+ [REMOVED] sent_id=519 ERRORS=['Token 2: Missing form']
99
+ [REMOVED] sent_id=569 ERRORS=['Token 2: Missing form']
100
+ [REMOVED] sent_id=50_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 6: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1']
101
+ [REMOVED] sent_id=53_54 ERRORS=['Multiple roots found: [27, 91]']
102
+ [REMOVED] sent_id=55_56_57 ERRORS=['Multiple roots found: [2, 4, 13]']
103
+ [REMOVED] sent_id=17_split1 ERRORS=['Multiple roots found: [2, 14, 17]']
104
+ [REMOVED] sent_id=17_split2 ERRORS=['Line 8: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 38: Invalid token ID or head', 'No root found', 'Token 3 has invalid head 8', 'Token 7 has invalid head 8', 'Token 9 has invalid head 8', 'Token 10 has invalid head 8', 'Token 17 has invalid head 8', 'Token 22 has invalid head 25', 'Token 23 has invalid head 25', 'Token 24 has invalid head 25', 'Token 26 has invalid head 25', 'Token 27 has invalid head 25', 'Token 28 has invalid head 25']
105
+ [REMOVED] sent_id=19_split1 ERRORS=['Multiple roots found: [3, 31]']
106
+ Cleaning DEV...
107
+ [REMOVED] sent_id=33 ERRORS=['Token 15: Missing deprel']
108
+ Cleaning TEST...
109
+ Writing TRAIN → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu (83827 valid sentences)
110
+ Writing DEV → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu (9 valid sentences)
111
+ Writing TEST → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu (99 valid sentences)
112
+ Done.
113
+ Sourcing scripts/config_alvis.sh
114
+ Running stanza dataset preparation…
115
+ 2025-12-03 00:10:57 INFO: Datasets program called with:
116
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/datasets/prepare_depparse_treebank.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
117
+ 2025-12-03 00:10:57 DEBUG: Downloading resource file from https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.11.0.json
118
+
119
+ 2025-12-03 00:10:57 INFO: Downloaded file to /cephyr/users/cleland/Alvis/stanza_resources/resources.json
120
+ 2025-12-03 00:10:57 DEBUG: Processing parameter "processors"...
121
+ 2025-12-03 00:10:57 WARNING: Can not find pos: diachronic from official model list. Ignoring it.
122
+ 2025-12-03 00:10:57 INFO: Downloading these customized packages for language: sv (Swedish)...
123
+ =======================
124
+ | Processor | Package |
125
+ -----------------------
126
+ =======================
127
+
128
+ 2025-12-03 00:10:57 INFO: Finished downloading models and saved to /cephyr/users/cleland/Alvis/stanza_resources
129
+ 2025-12-03 00:10:57 INFO: Using tagger model in /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt for sv_diachronic
130
+ 2025-12-03 00:10:57 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
131
+ 2025-12-03 00:10:57 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
132
+ Augmented 291 quotes: Counter({'″″': 35, '„”': 34, '「」': 33, '””': 32, '«»': 30, '»«': 29, '《》': 25, '„“': 25, '""': 24, '““': 24})
133
+ 2025-12-03 00:11:01 INFO: Running tagger to retag /local/tmp.5437491/tmprwphhavn/sv_diachronic.train.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu
134
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmprwphhavn/sv_diachronic.train.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu']
135
+ 2025-12-03 00:11:01 INFO: Running tagger in predict mode
136
+ 2025-12-03 00:11:01 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
137
+ 2025-12-03 00:11:04 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
138
+ 2025-12-03 00:11:04 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
139
+ 2025-12-03 00:11:04 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
140
+ 2025-12-03 00:11:04 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
141
+ 2025-12-03 00:11:06 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
142
+ 2025-12-03 00:11:08 INFO: Loading data with batch size 250...
143
+ 2025-12-03 00:11:57 INFO: Start evaluation...
144
+ 2025-12-03 00:16:22 INFO: UPOS XPOS UFeats AllTags
145
+ 2025-12-03 00:16:22 INFO: 87.98 62.42 77.22 61.01
146
+ 2025-12-03 00:16:22 INFO: POS Tagger score: sv_diachronic 61.01
147
+ 2025-12-03 00:16:23 INFO: Running tagger to retag /local/tmp.5437491/tmprwphhavn/sv_diachronic.dev.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu
148
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmprwphhavn/sv_diachronic.dev.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.dev.in.conllu']
149
+ 2025-12-03 00:16:23 INFO: Running tagger in predict mode
150
+ 2025-12-03 00:16:23 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
151
+ 2025-12-03 00:16:25 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
152
+ 2025-12-03 00:16:25 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
153
+ 2025-12-03 00:16:25 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
154
+ 2025-12-03 00:16:25 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
155
+ 2025-12-03 00:16:25 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
156
+ 2025-12-03 00:16:25 INFO: Loading data with batch size 250...
157
+ 2025-12-03 00:16:26 INFO: Start evaluation...
158
+ 2025-12-03 00:16:26 INFO: UPOS XPOS UFeats AllTags
159
+ 2025-12-03 00:16:26 INFO: 88.61 79.46 83.42 74.75
160
+ 2025-12-03 00:16:26 INFO: POS Tagger score: sv_diachronic 74.75
161
+ 2025-12-03 00:16:26 INFO: Running tagger to retag /local/tmp.5437491/tmprwphhavn/sv_diachronic.test.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu
162
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmprwphhavn/sv_diachronic.test.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.test.in.conllu']
163
+ 2025-12-03 00:16:26 INFO: Running tagger in predict mode
164
+ 2025-12-03 00:16:26 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
165
+ 2025-12-03 00:16:28 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
166
+ 2025-12-03 00:16:28 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
167
+ 2025-12-03 00:16:28 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
168
+ 2025-12-03 00:16:28 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
169
+ 2025-12-03 00:16:29 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
170
+ 2025-12-03 00:16:29 INFO: Loading data with batch size 250...
171
+ 2025-12-03 00:16:29 INFO: Start evaluation...
172
+ 2025-12-03 00:16:29 INFO: UPOS XPOS UFeats AllTags
173
+ 2025-12-03 00:16:29 INFO: 89.61 86.64 86.69 81.18
174
+ 2025-12-03 00:16:29 INFO: POS Tagger score: sv_diachronic 81.18
175
+ Preparing data for UD_Swedish-diachronic: sv_diachronic, sv
176
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu and writing to /local/tmp.5437491/tmprwphhavn/sv_diachronic.train.gold.conllu
177
+ Swapped 'w1, w2' for 'w1 ,w2' 287 times
178
+ Added 675 new sentences with asdf, zzzz -> asdf,zzzz
179
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu and writing to /local/tmp.5437491/tmprwphhavn/sv_diachronic.dev.gold.conllu
180
+ Reading from /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu and writing to /local/tmp.5437491/tmprwphhavn/sv_diachronic.test.gold.conllu
181
+ Running stanza dependency parser training…
182
+ 2025-12-03 00:16:47 INFO: Training program called with:
183
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/training/run_depparse.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt --batch_size 16 --dropout 0.33
184
+ 2025-12-03 00:16:47 DEBUG: UD_Swedish-diachronic: sv_diachronic
185
+ 2025-12-03 00:16:47 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
186
+ 2025-12-03 00:16:47 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
187
+ 2025-12-03 00:16:47 INFO: UD_Swedish-diachronic: saved_models/depparse/sv_diachronic_charlm_parser.pt exists, skipping!
188
+ DONE.
189
+ Full log saved to: logs/log_diachronic.pt_sv_diachron_nn_20251203_001047.txt
190
+ Symlink updated: logs/latest.txt → log_diachronic.pt_sv_diachron_nn_20251203_001047.txt
logs/log_diachronic.pt_sv_diachron_nn_20251203_003001.txt ADDED
@@ -0,0 +1,142 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === LOGFILE: logs/log_diachronic.pt_sv_diachron_nn_20251203_003001.txt ===
2
+ Language codes: sv diachron nn
3
+ Using pretrained model: diachronic.pt
4
+
5
+ Running: python prepare-train-val-test.py sv diachron nn
6
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-dev.conllu
7
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test.conllu
8
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_pud-ud-test.conllu
9
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-test.conllu
10
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_swell-ud-test-trg.conllu
11
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-dev.conllu
12
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/ucxn_ud_swedish-talbanken.conllu
13
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_talbanken-ud-train.conllu
14
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_old-ud-test.conllu
15
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-train.conllu
16
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-sv/sv_lines-ud-test.conllu
17
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-test.conllu
18
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-train.conllu
19
+ Reading: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud-treebanks-nn/no_nynorsk-ud-dev.conllu
20
+ Including DigPhil MACHINE in TRAIN (minus gold)…
21
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec330-GyllenborgC_SwenskaSpratthoken.conllu
22
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec254-CederborghF_BerattelseOmJohnHall.conllu
23
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec277-EnbomPU_MedborgeligtSkalde.conllu
24
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec268-DulciU_VitterhetsNojen3.conllu
25
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1063-spf220.conllu
26
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec397-AngeredStrandbergH_UnderSodernsSol.conllu
27
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec324-GranbergPA_Enslighetsalskaren.conllu
28
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec252-BremerF_Teckningar1.conllu
29
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec988-spf145.conllu
30
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec987-spf144.conllu
31
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec631-HasselskogN_HallaHallaGronkoping.conllu
32
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-letter141673-Stalhammar.conllu
33
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1033-spf190.conllu
34
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec25-Runius.conllu
35
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec486-SchwartzMS_BellmansSkor.conllu
36
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec452-NyblomH_FantasierFyra.conllu
37
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec613-EngstromA_StrindbergOchJag.conllu
38
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec208-Anonym_DetGrasligaMordet.conllu
39
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec639-HeidenstamV_Proletarfilosofiens.conllu
40
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec1102-spf259.conllu
41
+ Reading GOLD: /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/alanev_raw_files/diachron-validated/svediakorp-sec991-spf148.conllu
42
+ Cleaning TRAIN...
43
+ [REMOVED] sent_id=6 ERRORS=['Line 24: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 29: Invalid token ID or head', 'Token 30 has invalid head 24']
44
+ [REMOVED] sent_id=7_8 ERRORS=['Multiple roots found: [5, 10]']
45
+ [REMOVED] sent_id=30_31 ERRORS=['Multiple roots found: [3, 18]']
46
+ [REMOVED] sent_id=35 ERRORS=['Line 36: Invalid token ID or head']
47
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [1, 5]']
48
+ [REMOVED] sent_id=2_3 ERRORS=['Multiple roots found: [7, 20]']
49
+ [REMOVED] sent_id=8_9 ERRORS=['Multiple roots found: [24, 57]']
50
+ [REMOVED] sent_id=12_13 ERRORS=['Multiple roots found: [11, 16]']
51
+ [REMOVED] sent_id=124_split2 ERRORS=['Line 4: Invalid token ID or head', 'No root found', 'Token 1 has invalid head 4', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 6 has invalid head 4', 'Token 11 has invalid head 4', 'Token 15 has invalid head 4']
52
+ [REMOVED] sent_id=396 ERRORS=['Token 2: Missing form']
53
+ [REMOVED] sent_id=416 ERRORS=['Token 2: Missing form']
54
+ [REMOVED] sent_id=589 ERRORS=['Token 2: Missing form']
55
+ [REMOVED] sent_id=909 ERRORS=['Token 2: Missing form']
56
+ [REMOVED] sent_id=912 ERRORS=['Token 2: Missing form']
57
+ [REMOVED] sent_id=3_split1 ERRORS=['Multiple roots found: [4, 15, 17]']
58
+ [REMOVED] sent_id=3_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 8: Invalid token ID or head', 'Line 15: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1', 'Token 3 has invalid head 8', 'Token 4 has invalid head 8', 'Token 5 has invalid head 8', 'Token 7 has invalid head 8', 'Token 10 has invalid head 8', 'Token 13 has invalid head 8', 'Token 14 has invalid head 8']
59
+ [REMOVED] sent_id=3_4 ERRORS=['Multiple roots found: [1, 5]']
60
+ [REMOVED] sent_id=5_6 ERRORS=['Multiple roots found: [3, 24]']
61
+ [REMOVED] sent_id=11_12_13 ERRORS=['Multiple roots found: [5, 17, 25]']
62
+ [REMOVED] sent_id=119 ERRORS=['Token 2: Missing form']
63
+ [REMOVED] sent_id=179 ERRORS=['Token 2: Missing form']
64
+ [REMOVED] sent_id=188 ERRORS=['Token 2: Missing form']
65
+ [REMOVED] sent_id=223 ERRORS=['Token 2: Missing form']
66
+ [REMOVED] sent_id=268 ERRORS=['Token 2: Missing form']
67
+ [REMOVED] sent_id=325 ERRORS=['Token 2: Missing form']
68
+ [REMOVED] sent_id=388 ERRORS=['Token 2: Missing form']
69
+ [REMOVED] sent_id=399 ERRORS=['Token 2: Missing form']
70
+ [REMOVED] sent_id=475 ERRORS=['Token 2: Missing form']
71
+ [REMOVED] sent_id=505 ERRORS=['Token 2: Missing form']
72
+ [REMOVED] sent_id=520 ERRORS=['Token 2: Missing form']
73
+ [REMOVED] sent_id=562 ERRORS=['Token 2: Missing form']
74
+ [REMOVED] sent_id=669 ERRORS=['Token 2: Missing form']
75
+ [REMOVED] sent_id=711 ERRORS=['Token 2: Missing form']
76
+ [REMOVED] sent_id=731 ERRORS=['Token 2: Missing form']
77
+ [REMOVED] sent_id=867 ERRORS=['Token 2: Missing form']
78
+ [REMOVED] sent_id=884 ERRORS=['Token 2: Missing form']
79
+ [REMOVED] sent_id=923 ERRORS=['Token 2: Missing form']
80
+ [REMOVED] sent_id=939 ERRORS=['Token 2: Missing form']
81
+ [REMOVED] sent_id=1086 ERRORS=['Token 2: Missing form']
82
+ [REMOVED] sent_id=1179 ERRORS=['Token 2: Missing form']
83
+ [REMOVED] sent_id=1251 ERRORS=['Token 2: Missing form']
84
+ [REMOVED] sent_id=1345 ERRORS=['Token 2: Missing form']
85
+ [REMOVED] sent_id=1459 ERRORS=['Token 2: Missing form']
86
+ [REMOVED] sent_id=1656 ERRORS=['Token 2: Missing form']
87
+ [REMOVED] sent_id=1669 ERRORS=['Token 2: Missing form']
88
+ [REMOVED] sent_id=87_88 ERRORS=['Multiple roots found: [3, 6]']
89
+ [REMOVED] sent_id=65_split2_66_split2 ERRORS=['Line 4: Invalid token ID or head', 'Token 2 has invalid head 4', 'Token 3 has invalid head 4', 'Token 5 has invalid head 4']
90
+ [REMOVED] sent_id=25 ERRORS=['Token 2: Missing form']
91
+ [REMOVED] sent_id=136 ERRORS=['Token 2: Missing form']
92
+ [REMOVED] sent_id=208 ERRORS=['Token 2: Missing form']
93
+ [REMOVED] sent_id=230 ERRORS=['Token 2: Missing form']
94
+ [REMOVED] sent_id=245 ERRORS=['Token 2: Missing form']
95
+ [REMOVED] sent_id=276 ERRORS=['Token 2: Missing form']
96
+ [REMOVED] sent_id=320 ERRORS=['Token 2: Missing form']
97
+ [REMOVED] sent_id=366 ERRORS=['Token 2: Missing form']
98
+ [REMOVED] sent_id=519 ERRORS=['Token 2: Missing form']
99
+ [REMOVED] sent_id=569 ERRORS=['Token 2: Missing form']
100
+ [REMOVED] sent_id=50_split2 ERRORS=['Line 1: Invalid token ID or head', 'Line 6: Invalid token ID or head', 'No root found', 'Token 2 has invalid head 1']
101
+ [REMOVED] sent_id=53_54 ERRORS=['Multiple roots found: [27, 91]']
102
+ [REMOVED] sent_id=55_56_57 ERRORS=['Multiple roots found: [2, 4, 13]']
103
+ [REMOVED] sent_id=17_split1 ERRORS=['Multiple roots found: [2, 14, 17]']
104
+ [REMOVED] sent_id=17_split2 ERRORS=['Line 8: Invalid token ID or head', 'Line 25: Invalid token ID or head', 'Line 38: Invalid token ID or head', 'No root found', 'Token 3 has invalid head 8', 'Token 7 has invalid head 8', 'Token 9 has invalid head 8', 'Token 10 has invalid head 8', 'Token 17 has invalid head 8', 'Token 22 has invalid head 25', 'Token 23 has invalid head 25', 'Token 24 has invalid head 25', 'Token 26 has invalid head 25', 'Token 27 has invalid head 25', 'Token 28 has invalid head 25']
105
+ [REMOVED] sent_id=19_split1 ERRORS=['Multiple roots found: [3, 31]']
106
+ Cleaning DEV...
107
+ [REMOVED] sent_id=33 ERRORS=['Token 15: Missing deprel']
108
+ Cleaning TEST...
109
+ Writing TRAIN → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-train.conllu (83827 valid sentences)
110
+ Writing DEV → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-dev.conllu (9 valid sentences)
111
+ Writing TEST → /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/ud/UD_Swedish-diachronic/sv_diachronic-ud-test.conllu (99 valid sentences)
112
+ Done.
113
+ Sourcing scripts/config_alvis.sh
114
+ Running stanza dataset preparation…
115
+ 2025-12-03 00:30:11 INFO: Datasets program called with:
116
+ /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/stanza/utils/datasets/prepare_depparse_treebank.py UD_Swedish-diachronic --wordvec_pretrain_file /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
117
+ 2025-12-03 00:30:11 DEBUG: Downloading resource file from https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.11.0.json
118
+
119
+ 2025-12-03 00:30:11 INFO: Downloaded file to /cephyr/users/cleland/Alvis/stanza_resources/resources.json
120
+ 2025-12-03 00:30:11 DEBUG: Processing parameter "processors"...
121
+ 2025-12-03 00:30:11 WARNING: Can not find pos: diachronic from official model list. Ignoring it.
122
+ 2025-12-03 00:30:11 INFO: Downloading these customized packages for language: sv (Swedish)...
123
+ =======================
124
+ | Processor | Package |
125
+ -----------------------
126
+ =======================
127
+
128
+ 2025-12-03 00:30:11 INFO: Finished downloading models and saved to /cephyr/users/cleland/Alvis/stanza_resources
129
+ 2025-12-03 00:30:11 INFO: Using tagger model in /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt for sv_diachronic
130
+ 2025-12-03 00:30:11 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt for forward charlm
131
+ 2025-12-03 00:30:11 INFO: Using model /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt for backward charlm
132
+ Augmented 291 quotes: Counter({'″″': 35, '„”': 34, '「」': 33, '””': 32, '«»': 30, '»«': 29, '《》': 25, '„“': 25, '""': 24, '““': 24})
133
+ 2025-12-03 00:30:15 INFO: Running tagger to retag /local/tmp.5437491/tmp6nhkre69/sv_diachronic.train.gold.conllu to /mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu
134
+ Args: ['--wordvec_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain', '--lang', 'sv', '--shorthand', 'sv_diachronic', '--mode', 'predict', '--save_dir', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pos', '--save_name', 'diachronic.pt', '--wordvec_pretrain_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt', '--charlm', '--charlm_shorthand', 'sv_conll17', '--charlm_forward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt', '--charlm_backward_file', '/cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt', '--eval_file', '/local/tmp.5437491/tmp6nhkre69/sv_diachronic.train.gold.conllu', '--output_file', '/mimer/NOBACKUP/groups/dionysus/cleland/stanza-digphil/data/depparse/sv_diachronic.train.in.conllu']
135
+ 2025-12-03 00:30:15 INFO: Running tagger in predict mode
136
+ 2025-12-03 00:30:15 INFO: Loading model from: /cephyr/users/cleland/Alvis/stanza_resources/sv/pos/diachronic.pt
137
+ 2025-12-03 00:30:18 DEBUG: Loaded pretrain from /cephyr/users/cleland/Alvis/stanza_resources/sv/pretrain/diachronic.pt
138
+ 2025-12-03 00:30:18 DEBUG: POS model loading charmodels: /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt and /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
139
+ 2025-12-03 00:30:18 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/forward_charlm/conll17.pt
140
+ 2025-12-03 00:30:18 DEBUG: Loading charlm from /cephyr/users/cleland/Alvis/stanza_resources/sv/backward_charlm/conll17.pt
141
+ 2025-12-03 00:30:18 DEBUG: Building Adam with lr=0.003000, betas=(0.9, 0.95), eps=0.000001
142
+ 2025-12-03 00:30:21 INFO: Loading data with batch size 250...
make_new_model.sh CHANGED
@@ -18,11 +18,19 @@ LANGCODES=("$@")
18
  # ========================================
19
  # 0. SET UP LOGGING
20
  # ========================================
 
 
 
21
  timestamp=$(date +"%Y%m%d_%H%M%S")
22
- LOGFILE="log_${timestamp}.txt"
23
 
24
- # Redirect both stdout + stderr to tee
25
- # (Everything printed to terminal AND saved)
 
 
 
 
 
 
26
  exec > >(tee -a "$LOGFILE") 2>&1
27
 
28
  echo "=== LOGFILE: $LOGFILE ==="
@@ -60,3 +68,9 @@ python -m stanza.utils.training.run_depparse UD_Swedish-diachronic \
60
 
61
  echo "DONE."
62
  echo "Full log saved to: $LOGFILE"
 
 
 
 
 
 
 
18
  # ========================================
19
  # 0. SET UP LOGGING
20
  # ========================================
21
+
22
+ mkdir -p logs
23
+
24
  timestamp=$(date +"%Y%m%d_%H%M%S")
 
25
 
26
+ # Join language codes: sv_nn_da
27
+ LANG_JOINED=$(printf "_%s" "${LANGCODES[@]}")
28
+ LANG_JOINED="${LANG_JOINED:1}"
29
+
30
+ # Build log filename
31
+ LOGFILE="logs/log_${PRETRAINED_MODEL}_${LANG_JOINED}_${timestamp}.txt"
32
+
33
+ # Redirect output to tee
34
  exec > >(tee -a "$LOGFILE") 2>&1
35
 
36
  echo "=== LOGFILE: $LOGFILE ==="
 
68
 
69
  echo "DONE."
70
  echo "Full log saved to: $LOGFILE"
71
+
72
+ # ========================================
73
+ # 5. UPDATE 'latest.txt' SYMLINK
74
+ # ========================================
75
+ ln -sf "$(basename "$LOGFILE")" logs/latest.txt
76
+ echo "Symlink updated: logs/latest.txt → $(basename "$LOGFILE")"
prepare-train-val-test.py CHANGED
@@ -2,7 +2,7 @@
2
 
3
  '''
4
  Accepted language codes:
5
- sv, diachron, bm, nn, dk, de_lit
6
 
7
  Example arg:
8
  python prepare-train-val-test.py sv diachron bm nn dk de_lit
@@ -28,6 +28,7 @@ NORSKA_PROJEKT = BASE / "ud-treebanks-bm"
28
  NYNORSKA_PROJEKT = BASE / "ud-treebanks-nn"
29
  DANSKA_PROJEKT = BASE / "ud-treebanks-dk"
30
  TYSKA_PROJEKT = BASE / "ud-treebanks-de_lit"
 
31
 
32
  DIGPHIL_MACHINE = BASE / "alanev_raw_files/diachron"
33
  DIGPHIL_GOLD = BASE / "alanev_raw_files/diachron-validated"
@@ -202,6 +203,9 @@ if "nn" in ud_treebank_groups_used:
202
  if "dk" in ud_treebank_groups_used:
203
  train_sentences.extend(load_from_treebank_dir(DANSKA_PROJEKT))
204
 
 
 
 
205
  if "de_lit" in ud_treebank_groups_used:
206
  train_sentences.extend(load_from_treebank_dir(TYSKA_PROJEKT))
207
 
 
2
 
3
  '''
4
  Accepted language codes:
5
+ sv, diachron, bm, nn, dk, de_lit, is
6
 
7
  Example arg:
8
  python prepare-train-val-test.py sv diachron bm nn dk de_lit
 
28
  NYNORSKA_PROJEKT = BASE / "ud-treebanks-nn"
29
  DANSKA_PROJEKT = BASE / "ud-treebanks-dk"
30
  TYSKA_PROJEKT = BASE / "ud-treebanks-de_lit"
31
+ ICELANDIC_PROJEKT = BASE / "ud-treebanks-is"
32
 
33
  DIGPHIL_MACHINE = BASE / "alanev_raw_files/diachron"
34
  DIGPHIL_GOLD = BASE / "alanev_raw_files/diachron-validated"
 
203
  if "dk" in ud_treebank_groups_used:
204
  train_sentences.extend(load_from_treebank_dir(DANSKA_PROJEKT))
205
 
206
+ if "is" in ud_treebank_groups_used:
207
+ train_sentences.extend(load_from_treebank_dir(ICELANDIC_PROJEKT))
208
+
209
  if "de_lit" in ud_treebank_groups_used:
210
  train_sentences.extend(load_from_treebank_dir(TYSKA_PROJEKT))
211
 
saved_models/depparse/kubhist2_sv_diachron/sv_diachronic_charlm_parser.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37ba70d57a2103f24fc47e323d1a251bca892ee098ac1a4846d327286650ca32
3
+ size 145281953
saved_models/depparse/kubhist2_sv_diachron/sv_diachronic_charlm_parser_checkpoint.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6480cb09a94c891b24ef9ca638f5fa35daac18b9fe9d578b0d1c23e82b34fc51
3
+ size 434726762
ud-treebanks-is/is_pud-ud-test.conllu ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a024997a765ba1c5350f999a421b7faef43f7ebc0412b1378305ab7b91a8df52
3
+ size 1555630