Commit
•
d583491
1
Parent(s):
83d2a3e
Hezar: Upload training files
Browse files
train/dataset_config.yaml
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: sequence_labeling
|
| 2 |
+
config_type: dataset
|
| 3 |
+
task: sequence_labeling
|
| 4 |
+
path: hezarai/lscp-pos-500k
|
| 5 |
+
tokenizer_path: hezarai/roberta-base-fa
|
| 6 |
+
tags_field: pos_tags
|
| 7 |
+
tokens_field: tokens
|
| 8 |
+
ignore_index: -100
|
| 9 |
+
label_all_tokens: true
|