Commit
•
3db57d2
1
Parent(s):
e4f4d97
hyperparameters added
Browse files- hparams.yaml +42 -0
hparams.yaml
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
accumulate_grad_batches: 1
|
| 2 |
+
adafactor: false
|
| 3 |
+
adam_epsilon: 1.0e-08
|
| 4 |
+
attention_dropout: null
|
| 5 |
+
cache_dir: null
|
| 6 |
+
config_name: ''
|
| 7 |
+
deterministic: false
|
| 8 |
+
do_eval: false
|
| 9 |
+
do_test: false
|
| 10 |
+
do_train: true
|
| 11 |
+
dropout_rate: null
|
| 12 |
+
eval_batch_size: 16
|
| 13 |
+
eval_dataset_path: ''
|
| 14 |
+
force_reinit: false
|
| 15 |
+
fp16: true
|
| 16 |
+
gpus: -1
|
| 17 |
+
gradient_clip_val: 1.0
|
| 18 |
+
learning_rate: 1.0e-05
|
| 19 |
+
lr_scheduler: linear
|
| 20 |
+
max_epochs: 10
|
| 21 |
+
max_history: 1
|
| 22 |
+
max_seq_length: 0
|
| 23 |
+
min_delta: 0.0
|
| 24 |
+
model_name_or_path: roberta-large
|
| 25 |
+
num_workers: 8
|
| 26 |
+
output_dir: ./checkpoints/roberta-large-faithcritic
|
| 27 |
+
overwrite_output_dir: false
|
| 28 |
+
pad_to_multiple_of: 8
|
| 29 |
+
patience: 5
|
| 30 |
+
predict_dataset_path: null
|
| 31 |
+
save_last: false
|
| 32 |
+
seed: 42
|
| 33 |
+
test_dataset_path: null
|
| 34 |
+
test_task: FaithCritic
|
| 35 |
+
tokenizer_name: null
|
| 36 |
+
train_batch_size: 16
|
| 37 |
+
train_dataset_path: ''
|
| 38 |
+
train_task: FaithCritic
|
| 39 |
+
val_check_interval: 0.5
|
| 40 |
+
warmup_ratio: 0.08
|
| 41 |
+
warmup_steps: 0
|
| 42 |
+
weight_decay: 0.1
|