added trained model
Browse files- dataset-configs.json +20 -0
- eventnet_model.pt +3 -0
- params-config.json +73 -0
dataset-configs.json
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"FR_PAR": {
|
3 |
+
"dev_data_path": "data/frpar/fold_21/dev_set.tsv",
|
4 |
+
"tasks": {
|
5 |
+
"frpar": {
|
6 |
+
"layers_to_use": [
|
7 |
+
-1
|
8 |
+
],
|
9 |
+
"loss_weight": 1,
|
10 |
+
"metric": "multi_acc",
|
11 |
+
"topn": 1,
|
12 |
+
"threshold": 0.7,
|
13 |
+
"column_idx": 1,
|
14 |
+
"task_type": "multiseq"
|
15 |
+
}
|
16 |
+
},
|
17 |
+
"train_data_path": "data/frpar/fold_21/training_set.tsv",
|
18 |
+
"word_idx": 0
|
19 |
+
}
|
20 |
+
}
|
eventnet_model.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cc0ac9068c8f072dcd097b88740d5de59ba4e024a927af0856bc93225b9fc1a4
|
3 |
+
size 464631800
|
params-config.json
ADDED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"batching": {
|
3 |
+
"batch_size": 64,
|
4 |
+
"max_tokens": 1024,
|
5 |
+
"sampling_smoothing": 1,
|
6 |
+
"sort_by_size": true
|
7 |
+
},
|
8 |
+
"decoders": {
|
9 |
+
"classification": {},
|
10 |
+
"default_decoder": {
|
11 |
+
"layers_to_use": [
|
12 |
+
-1
|
13 |
+
],
|
14 |
+
"loss_weight": 1,
|
15 |
+
"metric": "accuracy",
|
16 |
+
"topn": 1
|
17 |
+
},
|
18 |
+
"dependency": {
|
19 |
+
"arc_representation_dim": 768,
|
20 |
+
"metric": "las",
|
21 |
+
"tag_representation_dim": 256
|
22 |
+
},
|
23 |
+
"mlm": {
|
24 |
+
"metric": "perplexity"
|
25 |
+
},
|
26 |
+
"multiclas": {
|
27 |
+
"metric": "multi_acc",
|
28 |
+
"threshold": 0.7
|
29 |
+
},
|
30 |
+
"multiseq": {
|
31 |
+
"metric": "multi_acc",
|
32 |
+
"threshold": 0.7
|
33 |
+
},
|
34 |
+
"regression": {
|
35 |
+
"metric": "avg_dist"
|
36 |
+
},
|
37 |
+
"seq": {},
|
38 |
+
"seq_bio": {
|
39 |
+
"metric": "span_f1"
|
40 |
+
},
|
41 |
+
"string2string": {},
|
42 |
+
"tok": {
|
43 |
+
"pre_split": true
|
44 |
+
}
|
45 |
+
},
|
46 |
+
"default_dec_dataset_embeds_dim": 12,
|
47 |
+
"encoder": {
|
48 |
+
"dropout": 0.2,
|
49 |
+
"max_input_length": 64,
|
50 |
+
"update_weights_encoder": true
|
51 |
+
},
|
52 |
+
"random_seed": 8446,
|
53 |
+
"training": {
|
54 |
+
"keep_top_n": 1,
|
55 |
+
"learning_rate_scheduler": {
|
56 |
+
"cut_frac": 0.3,
|
57 |
+
"decay_factor": 0.38,
|
58 |
+
"discriminative_fine_tuning": true,
|
59 |
+
"gradual_unfreezing": true
|
60 |
+
},
|
61 |
+
"num_epochs": 50,
|
62 |
+
"optimizer": {
|
63 |
+
"betas": [
|
64 |
+
0.9,
|
65 |
+
0.99
|
66 |
+
],
|
67 |
+
"correct_bias": false,
|
68 |
+
"lr": 0.0015,
|
69 |
+
"weight_decay": 0.01
|
70 |
+
}
|
71 |
+
},
|
72 |
+
"transformer_model": "dbmdz/bert-base-italian-xxl-cased"
|
73 |
+
}
|