Migrate model card from transformers-repo
Browse filesRead announcement at https://discuss.huggingface.co/t/announcement-all-model-cards-will-be-migrated-to-hf-co-model-repos/2755
Original file history: https://github.com/huggingface/transformers/commits/master/model_cards/NeuML/bert-small-cord19/README.md
README.md
ADDED
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# BERT-Small fine-tuned on CORD-19 dataset
|
2 |
+
|
3 |
+
[BERT L6_H-512_A-8 model](https://huggingface.co/google/bert_uncased_L-6_H-512_A-8) fine-tuned on the [CORD-19 dataset](https://www.semanticscholar.org/cord19).
|
4 |
+
|
5 |
+
## CORD-19 data subset
|
6 |
+
The training data for this dataset is stored as a [Kaggle dataset](https://www.kaggle.com/davidmezzetti/cord19-qa?select=cord19.txt). The training
|
7 |
+
data is a subset of the full corpus, focusing on high-quality, study-design detected articles.
|
8 |
+
|
9 |
+
## Building the model
|
10 |
+
|
11 |
+
```bash
|
12 |
+
python run_language_modeling.py
|
13 |
+
--model_type bert
|
14 |
+
--model_name_or_path google/bert_uncased_L-6_H-512_A-8
|
15 |
+
--do_train
|
16 |
+
--mlm
|
17 |
+
--line_by_line
|
18 |
+
--block_size 512
|
19 |
+
--train_data_file cord19.txt
|
20 |
+
--per_gpu_train_batch_size 4
|
21 |
+
--learning_rate 3e-5
|
22 |
+
--num_train_epochs 3.0
|
23 |
+
--output_dir bert-small-cord19
|
24 |
+
--save_steps 0
|
25 |
+
--overwrite_output_dir
|