update model
Browse files
mirror_outputs/Mirror_Pretrain_AllExcluded_2/ckpt/SchemaGuidedInstructBertModel.best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5e4e49b84688218d2e60b1123682e313478cd796739dbe61a035890600e324c0
|
3 |
+
size 5229689445
|
mirror_outputs/Mirror_Pretrain_AllExcluded_2/task_params.yaml
ADDED
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
_config_info:
|
2 |
+
create_time: '2023-06-16 12:37:14'
|
3 |
+
use_default_base_config: true
|
4 |
+
config_filepath:
|
5 |
+
- /data/tzhu/Mirror/conf/Pretrain_excluded.yaml
|
6 |
+
task_type: SchemaGuidedInstructBertTask
|
7 |
+
task_name: Mirror_Pretrain_AllExcluded_2
|
8 |
+
comment: ~~content as label, (start, end + 1) span
|
9 |
+
output_dir: mirror_outputs
|
10 |
+
task_dir: mirror_outputs/Mirror_Pretrain_AllExcluded_2
|
11 |
+
dump_cache_dir: mirror_outputs/Mirror_Pretrain_AllExcluded_2/cache
|
12 |
+
regenerate_cache: false
|
13 |
+
data_dir: resources/Mirror/v1.4_sampled_v3/merged/all_excluded
|
14 |
+
train_filepath: resources/Mirror/v1.4_sampled_v3/merged/all_excluded/train.jsonl
|
15 |
+
dev_filepath: resources/Mirror/v1.4_sampled_v3/merged/all_excluded/dev.jsonl
|
16 |
+
test_filepath: resources/Mirror/v1.4_sampled_v3/merged/all_excluded/test.jsonl
|
17 |
+
random_seed: 1227
|
18 |
+
num_epochs: 3
|
19 |
+
num_steps: -1
|
20 |
+
warmup_proportion: 0.1
|
21 |
+
epoch_patience: -1
|
22 |
+
step_patience: -1
|
23 |
+
batch_size: 64
|
24 |
+
learning_rate: 2.0e-05
|
25 |
+
max_grad_norm: 1.0
|
26 |
+
skip_train: false
|
27 |
+
debug_mode: false
|
28 |
+
grad_accum_steps: 1
|
29 |
+
resumed_training_path: null
|
30 |
+
step_eval_interval: 10000
|
31 |
+
epoch_eval_interval: 1
|
32 |
+
eval_on_data:
|
33 |
+
- train
|
34 |
+
select_best_on_data: train
|
35 |
+
select_best_by_key: loss
|
36 |
+
best_metric_field: micro.f1
|
37 |
+
save_every_ckpt: true
|
38 |
+
save_best_ckpt: true
|
39 |
+
final_eval_on_test: false
|
40 |
+
main_process_logging: true
|
41 |
+
max_seq_len: 512
|
42 |
+
label_span: tag
|
43 |
+
mode: span
|
44 |
+
stream_mode: false
|
45 |
+
plm_dir: /data/tzhu/PLM/microsoft--deberta-v3-large
|
46 |
+
base_model_path: null
|
47 |
+
train_batch_size: 8
|
48 |
+
eval_batch_size: 8
|
49 |
+
other_learning_rate: 0.0001
|
50 |
+
weight_decay: 0.1
|
51 |
+
dropout: 0.3
|
52 |
+
use_rope: true
|
53 |
+
biaffine_size: 512
|
54 |
+
include_package:
|
55 |
+
- src.task
|
56 |
+
use_default_base_config: true
|