Update README.md
Browse files
README.md
CHANGED
@@ -9,19 +9,22 @@ license: apache-2.0
|
|
9 |
pipeline_tag: text-generation
|
10 |
tags:
|
11 |
- legal
|
|
|
|
|
12 |
---
|
13 |
|
14 |
# Salamandra 7B aligned EADOP Model Card
|
15 |
-
Salamandra 7B aligned EADOP is a full
|
16 |
[BSC Language Technologies Unit](https://huggingface.co/BSC-LT)'s
|
17 |
[Salamndra Instruct 7B](https://huggingface.co/BSC-LT/salamandra-7b-instruct)
|
18 |
-
model by the
|
19 |
the handling of out-of-domain Questions in a RAG instruction-following setting.
|
20 |
|
21 |
-
The model has been finetuned on a dataset
|
22 |
-
and out-of-domain user messages and
|
23 |
provide helpful information about the current Catalan legislation.
|
24 |
-
The dataset [
|
|
|
25 |
[Entitat Autònoma del Diari Oficial i de Publicacions (EADOP)](https://dogc.gencat.cat/ca/sobre-el-dogc/eadop/)
|
26 |
and it consists of user messages and assistant responses in Catalan and Spanish.
|
27 |
|
@@ -76,20 +79,22 @@ outputs = model.generate(input_ids=inputs.to(model.device), max_new_tokens=200)
|
|
76 |
|
77 |
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
78 |
```
|
|
|
79 |
Using this template, each turn is preceded by a `<|im_start|>` delimiter and the role of the entity
|
80 |
(either `user`, for content supplied by the user, or `assistant` for LLM responses), and finished with the `<|im_end|>` token.
|
81 |
|
82 |
---
|
83 |
|
84 |
## Finetuning Data
|
85 |
-
Please refer to [
|
86 |
|
87 |
|
88 |
### Author
|
89 |
This model has been finetuned by [Alinia AI](https://alinia.ai/).
|
90 |
|
|
|
91 |
### Contact
|
92 |
-
For further information, please
|
93 |
|
94 |
|
95 |
### Acknowledgements
|
|
|
9 |
pipeline_tag: text-generation
|
10 |
tags:
|
11 |
- legal
|
12 |
+
datasets:
|
13 |
+
- alinia/EADOP-RAG-out-of-domain
|
14 |
---
|
15 |
|
16 |
# Salamandra 7B aligned EADOP Model Card
|
17 |
+
Salamandra 7B aligned EADOP is a full-finetuning version of
|
18 |
[BSC Language Technologies Unit](https://huggingface.co/BSC-LT)'s
|
19 |
[Salamndra Instruct 7B](https://huggingface.co/BSC-LT/salamandra-7b-instruct)
|
20 |
+
model by the at the Barcelona Supercomputing Center focused on improving
|
21 |
the handling of out-of-domain Questions in a RAG instruction-following setting.
|
22 |
|
23 |
+
The model has been finetuned on a dataset consisting of 2,000+ human annotated in-
|
24 |
+
and out-of-domain user messages and assistant responses in the context of a chatbot that can
|
25 |
provide helpful information about the current Catalan legislation.
|
26 |
+
The dataset [alinia/EADOP-RAG-out-of-domain](https://huggingface.co/datasets/alinia/EADOP-RAG-out-of-domain)
|
27 |
+
was collected in collaboration with the
|
28 |
[Entitat Autònoma del Diari Oficial i de Publicacions (EADOP)](https://dogc.gencat.cat/ca/sobre-el-dogc/eadop/)
|
29 |
and it consists of user messages and assistant responses in Catalan and Spanish.
|
30 |
|
|
|
79 |
|
80 |
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
81 |
```
|
82 |
+
|
83 |
Using this template, each turn is preceded by a `<|im_start|>` delimiter and the role of the entity
|
84 |
(either `user`, for content supplied by the user, or `assistant` for LLM responses), and finished with the `<|im_end|>` token.
|
85 |
|
86 |
---
|
87 |
|
88 |
## Finetuning Data
|
89 |
+
Please refer to [alinia/EADOP-RAG-out-of-domain](https://huggingface.co/datasets/alinia/EADOP-RAG-out-of-domain) for the Dataset Card.
|
90 |
|
91 |
|
92 |
### Author
|
93 |
This model has been finetuned by [Alinia AI](https://alinia.ai/).
|
94 |
|
95 |
+
|
96 |
### Contact
|
97 |
+
For further information, please email [[email protected]](mailto:[email protected]).
|
98 |
|
99 |
|
100 |
### Acknowledgements
|