Commit
•
f5778e7
1
Parent(s):
ff621fc
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,46 +1,46 @@
|
|
| 1 |
---
|
| 2 |
-
tags:
|
| 3 |
-
- generated_from_keras_callback
|
| 4 |
model-index:
|
| 5 |
- name: medieval-it5-base
|
| 6 |
results: []
|
|
|
|
|
|
|
| 7 |
---
|
| 8 |
|
| 9 |
-
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
| 10 |
-
probably proofread and complete it, then remove this comment. -->
|
| 11 |
-
|
| 12 |
# medieval-it5-base
|
| 13 |
|
| 14 |
-
This model
|
| 15 |
-
It achieves the following results on the evaluation set:
|
| 16 |
-
|
| 17 |
|
| 18 |
-
|
| 19 |
|
| 20 |
-
More information needed
|
| 21 |
|
| 22 |
-
##
|
| 23 |
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
## Training procedure
|
| 31 |
|
| 32 |
-
|
| 33 |
|
| 34 |
-
|
| 35 |
-
- optimizer: None
|
| 36 |
-
- training_precision: float32
|
| 37 |
|
| 38 |
-
|
| 39 |
|
|
|
|
| 40 |
|
|
|
|
| 41 |
|
| 42 |
### Framework versions
|
| 43 |
|
| 44 |
-
- Transformers 4.26.
|
| 45 |
-
- TensorFlow 2.11.0
|
| 46 |
- Tokenizers 0.13.2
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
model-index:
|
| 3 |
- name: medieval-it5-base
|
| 4 |
results: []
|
| 5 |
+
language:
|
| 6 |
+
- it
|
| 7 |
---
|
| 8 |
|
|
|
|
|
|
|
|
|
|
| 9 |
# medieval-it5-base
|
| 10 |
|
| 11 |
+
This model is a version of [gsarti/it5-base](https://huggingface.co/gsarti/it5-base) fine-tuned on a dataset called [ita2medieval](https://huggingface.co/datasets/leobertolazzi/ita2medieval). The Dataset contains sentences from medieval italian along with paraphrases in contemporary italian (approximately 6.5k pairs in total).
|
|
|
|
|
|
|
| 12 |
|
| 13 |
+
The fine-tuning task is text-style-tansfer from contemporary to medieval italian.
|
| 14 |
|
|
|
|
| 15 |
|
| 16 |
+
## Using the model
|
| 17 |
|
| 18 |
+
```
|
| 19 |
+
from transformers import AutoTokenzier, AutoModelForSeq2SeqLM
|
| 20 |
+
tokenizer = AutoTokenizer.from_pretrained("leobertolazzi/medieval-it5-base")
|
| 21 |
+
model = AutoModelForSeq2SeqLM.from_pretrained("leobertolazzi/medieval-it5-base")
|
| 22 |
+
```
|
| 23 |
|
| 24 |
+
Flax and Tensorflow versions of the model are also available:
|
| 25 |
+
```
|
| 26 |
+
from transformers import FlaxT5ForConditionalGeneration, TFT5ForConditionalGeneration
|
| 27 |
+
model_flax = FlaxT5ForConditionalGeneration.from_pretrained("leobertolazzi/medieval-it5-base")
|
| 28 |
+
model_tf = TFT5ForConditionalGeneration.from_pretrained("leobertolazzi/medieval-it5-base")
|
| 29 |
+
```
|
| 30 |
|
| 31 |
## Training procedure
|
| 32 |
|
| 33 |
+
The code used for the fine-tuning is available in this [repo](https://github.com/leobertolazzi/medievalIT5)
|
| 34 |
|
| 35 |
+
## Intended uses & limitations
|
|
|
|
|
|
|
| 36 |
|
| 37 |
+
The biggest limitation for this project is the size of the ita2medieval dataset. In fact, it consists only of 6.5K sentence pairs whereas [gsarti/it5-small](https://huggingface.co/gsarti/it5-base) has 220M parameters.
|
| 38 |
|
| 39 |
+
For this reason the results can be far from perfect, but some nice style translations can also be obtained.
|
| 40 |
|
| 41 |
+
It would be nice to expand ita2medieval with text and paraphrases from more medieval italian authors!
|
| 42 |
|
| 43 |
### Framework versions
|
| 44 |
|
| 45 |
+
- Transformers 4.26.0
|
|
|
|
| 46 |
- Tokenizers 0.13.2
|