xixianliao
commited on
Commit
•
21cb965
1
Parent(s):
40fde0e
Update README.md
Browse files
README.md
CHANGED
@@ -1,215 +1,215 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
datasets:
|
4 |
-
- projecte-aina/CA-ZH_Parallel_Corpus
|
5 |
-
language:
|
6 |
-
- zh
|
7 |
-
- ca
|
8 |
-
base_model:
|
9 |
-
- facebook/m2m100_1.2B
|
10 |
-
---
|
11 |
-
## Projecte Aina’s Catalan-Chinese machine translation model
|
12 |
-
|
13 |
-
## Table of Contents
|
14 |
-
<details>
|
15 |
-
<summary>Click to expand</summary>
|
16 |
-
|
17 |
-
- [Model description](#model-description)
|
18 |
-
- [Intended uses and limitations](#intended-uses-and-limitations)
|
19 |
-
- [How to use](#how-to-use)
|
20 |
-
- [Limitations and bias](#limitations-and-bias)
|
21 |
-
- [Training](#training)
|
22 |
-
- [Evaluation](#evaluation)
|
23 |
-
- [Additional information](#additional-information)
|
24 |
-
|
25 |
-
</details>
|
26 |
-
|
27 |
-
|
28 |
-
## Model description
|
29 |
-
|
30 |
-
This machine translation model is built upon the M2M100 1.2B, fine-tuned specifically for Catalan-Chinese translation.
|
31 |
-
It is trained on a combination of Catalan-Chinese datasets
|
32 |
-
totalling 94,187,858 sentence pairs. 113,305 sentence pairs were parallel data collected from the web, while the remaining 94,074,553 sentence pairs
|
33 |
-
were parallel synthetic data created using the
|
34 |
-
[Aina Project's Spanish-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-es-ca) and the [Aina Project's English-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-en-ca).
|
35 |
-
|
36 |
-
Following the fine-tuning phase, Contrastive Preference Optimization (CPO) was applied to further refine the model's outputs. CPO training involved pairs of "chosen" and "rejected" translations for a total of 4,006 sentences. These sentences were sourced from the Flores development set (997 sentences), the Flores devtest set (1,012 sentences), and the NTREX set (1,997 sentences).
|
37 |
-
|
38 |
-
The model was evaluated on the Projecte Aina's Catalan-Chinese evaluation dataset (unpublished), achieving results comparable to those of Google Translate.
|
39 |
-
|
40 |
-
## Intended uses and limitations
|
41 |
-
|
42 |
-
You can use this model for machine translation from Catalan to simplified Chinese.
|
43 |
-
|
44 |
-
## How to use
|
45 |
-
|
46 |
-
### Usage
|
47 |
-
|
48 |
-
Translate a sentence using python
|
49 |
-
```python
|
50 |
-
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
51 |
-
|
52 |
-
model_id = "projecte-aina/aina-translator-ca-zh
|
53 |
-
|
54 |
-
model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
|
55 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
56 |
-
|
57 |
-
sentence = "Benvingut al projecte Aina!"
|
58 |
-
|
59 |
-
input_ids = tokenizer(sentence, return_tensors="pt").input_ids
|
60 |
-
output_ids = model.generate(input_ids, max_length=200, num_beams=5)
|
61 |
-
|
62 |
-
generated_translation= tokenizer.decode(output_ids[0], skip_special_tokens=True, spaces_between_special_tokens = False).strip()
|
63 |
-
print(generated_translation)
|
64 |
-
#欢迎来到 Aina 项目!
|
65 |
-
```
|
66 |
-
|
67 |
-
|
68 |
-
## Limitations and bias
|
69 |
-
At the time of submission, no measures have been taken to estimate the bias and toxicity embedded in the model.
|
70 |
-
However, we are well aware that our models may be biased. We intend to conduct research in these areas in the future, and if completed, this model card will be updated.
|
71 |
-
|
72 |
-
## Training
|
73 |
-
|
74 |
-
### Training data
|
75 |
-
|
76 |
-
The Catalan-Chinese data collected from the web was a combination of the following datasets:
|
77 |
-
|
78 |
-
| Dataset | Sentences before cleaning |
|
79 |
-
|-------------------|----------------|
|
80 |
-
| OpenSubtitles | 139,300 |
|
81 |
-
| WikiMatrix | 90,643 |
|
82 |
-
| Wikipedia | 68,623|
|
83 |
-
| **Total** | **298,566** |
|
84 |
-
|
85 |
-
94,074,553 sentence pairs of synthetic parallel data were created from the following Spanish-Chinese datasets and English-Chinese datasets:
|
86 |
-
|
87 |
-
**Spanish-Chinese:**
|
88 |
-
|
89 |
-
| Dataset | Sentences before cleaning |
|
90 |
-
|-------------------|----------------|
|
91 |
-
| NLLB |24,051,233|
|
92 |
-
| UNPC | 17,599,223 |
|
93 |
-
| MultiUN | 9,847,770 |
|
94 |
-
| OpenSubtitles | 9,319,658 |
|
95 |
-
| MultiParaCrawl | 3,410,087 |
|
96 |
-
| MultiCCAligned | 3,006,694 |
|
97 |
-
| WikiMatrix | 1,214,322 |
|
98 |
-
| News Commentary | 375,982 |
|
99 |
-
| Tatoeba | 9,404 |
|
100 |
-
| **Total** | **68,834,373** |
|
101 |
-
|
102 |
-
**English-Chinese:**
|
103 |
-
|
104 |
-
| Dataset | Sentences before cleaning |
|
105 |
-
|-------------------|----------------|
|
106 |
-
| NLLB |71,383,325|
|
107 |
-
| CCAligned | 15,181,415 |
|
108 |
-
| Paracrawl | 14,170,869|
|
109 |
-
| WikiMatrix | 2,595,119|
|
110 |
-
| **Total** | **103,330,728** |
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
### Training procedure
|
115 |
-
|
116 |
-
### Data preparation
|
117 |
-
|
118 |
-
**Catalan-Chinese parallel data**
|
119 |
-
|
120 |
-
The Chinese side of all datasets were first processed using the [Hanzi Identifier](https://github.com/tsroten/hanzidentifier) to detect Traditional Chinese, which was subsequently converted to Simplified Chinese using [OpenCC](https://github.com/BYVoid/OpenCC).
|
121 |
-
|
122 |
-
All data was then filtered according to two specific criteria:
|
123 |
-
|
124 |
-
- Alignment: sentence level alignments were calculated using [LaBSE](https://huggingface.co/sentence-transformers/LaBSE) and sentence pairs with a score below 0.75 were discarded.
|
125 |
-
|
126 |
-
- Language identification: the probability of being the target language was calculated using [Lingua.py](https://github.com/pemistahl/lingua-py) and sentences with a language probability score below 0.5 were discarded.
|
127 |
-
|
128 |
-
Next, Spanish data was translated into Catalan using the Aina Project's [Spanish-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-es-ca), while English data was translated into Catalan using the Aina Project's [English-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-en-ca).
|
129 |
-
|
130 |
-
The filtered and translated datasets are then concatenated and deduplicated to form a final corpus of 94,187,858.
|
131 |
-
|
132 |
-
**Catalan-Chinese Contrastive Preference Optimization dataset**
|
133 |
-
|
134 |
-
The CPO dataset is built by comparing the quality of translations across four distinct sources:
|
135 |
-
|
136 |
-
- Reference translation: Chinese sentences from Flores test set, Flores devtest set, and NTREX dataset.
|
137 |
-
- aina-translator-ca-zh: A specialized bilingual model for Catalan-Chinese translations.
|
138 |
-
- Google Translate: A widely-used general-purpose machine translation system.
|
139 |
-
- OpenAI GPT-4: A large-scale language model capable of performing a wide range of tasks in conversational settings, including high-quality translation.
|
140 |
-
|
141 |
-
To evaluate the quality of translations without relying on human annotations, we employ two reference-free evaluation models:
|
142 |
-
|
143 |
-
- [Unbabel/wmt23-cometkiwi-da-xxl](https://huggingface.co/Unbabel/wmt23-cometkiwi-da-xxl)
|
144 |
-
- [Unbabel/XCOMET-XXL](https://huggingface.co/Unbabel/XCOMET-XXL)
|
145 |
-
|
146 |
-
These models provide direct assessment scores for each translation. The scores from both models are averaged to determine the relative quality of each translation. Based on this evaluation, the highest-scoring ("chosen") and lowest-scoring ("rejected") translations are identified for each source sentence, forming contrastive pairs. The CPO dataset comprises a total of 4,006 such pairs of "chosen" and "rejected" translations.
|
147 |
-
|
148 |
-
|
149 |
-
#### Training
|
150 |
-
|
151 |
-
The training was executed on NVIDIA GPUs utilizing the Hugging Face Transformers framework.
|
152 |
-
The model was trained for 245,000 updates.
|
153 |
-
|
154 |
-
Following fine-tuning on the M2M100 1.2B model, Contrastive Preference Optimization (CPO) was performed using our CPO dataset and the Hugging Face CPO Trainer. This phase involved 1,500 updates.
|
155 |
-
|
156 |
-
## Evaluation
|
157 |
-
|
158 |
-
### Variable and metrics
|
159 |
-
|
160 |
-
Below are the evaluation results on the Projecte Aina's Catalan-Chinese test set (unpublished), compared to Google Translate for the CA-ZH direction. The evaluation was conducted using [`tower-eval`](https://github.com/deep-spin/tower-eval) following the standard setting (beam search with beam size 5, limiting the translation length to 200 tokens). We report the following metrics:
|
161 |
-
|
162 |
-
- BLEU: Sacrebleu implementation, version:2.4.0
|
163 |
-
- ChrF: Sacrebleu implementation.
|
164 |
-
- Comet: Model checkpoint: "Unbabel/wmt22-comet-da".
|
165 |
-
- Comet-kiwi: Model checkpoint: "Unbabel/wmt22-cometkiwi-da".
|
166 |
-
|
167 |
-
|
168 |
-
### Evaluation results
|
169 |
-
|
170 |
-
Below are the evaluation results on the machine translation from Catalan to Chinese compared to [Google Translate](https://translate.google.com/):
|
171 |
-
|
172 |
-
|
173 |
-
#### Projecte Aina's Catalan-Chinese evaluation dataset
|
174 |
-
|
175 |
-
| | Bleu ↑ | ChrF ↑ | Comet ↑ | Comet-kiwi ↑ |
|
176 |
-
|:-----------------------|-------:|------:|-------:|--------:|
|
177 |
-
| aina-translator-ca-zh
|
178 |
-
| Google Translate | **44.64** | **41.15** | **0.87** | 0.80 |
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
## Additional information
|
183 |
-
|
184 |
-
### Author
|
185 |
-
The Language Technologies Unit from Barcelona Supercomputing Center.
|
186 |
-
|
187 |
-
### Contact
|
188 |
-
For further information, please send an email to <[email protected]>.
|
189 |
-
|
190 |
-
### Copyright
|
191 |
-
Copyright(c) 2023 by Language Technologies Unit, Barcelona Supercomputing Center.
|
192 |
-
|
193 |
-
### License
|
194 |
-
[Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
195 |
-
|
196 |
-
### Funding
|
197 |
-
This work has been promoted and financed by the Generalitat de Catalunya through the [Aina project](https://projecteaina.cat/).
|
198 |
-
|
199 |
-
### Disclaimer
|
200 |
-
|
201 |
-
<details>
|
202 |
-
<summary>Click to expand</summary>
|
203 |
-
|
204 |
-
The model published in this repository is intended for a generalist purpose and is available to third parties under a permissive Apache License, Version 2.0.
|
205 |
-
|
206 |
-
Be aware that the model may have biases and/or any other undesirable distortions.
|
207 |
-
|
208 |
-
When third parties deploy or provide systems and/or services to other parties using this model (or any system based on it)
|
209 |
-
or become users of the model, they should note that it is their responsibility to mitigate the risks arising from its use and,
|
210 |
-
in any event, to comply with applicable regulations, including regulations regarding the use of Artificial Intelligence.
|
211 |
-
|
212 |
-
In no event shall the owner and creator of the model (Barcelona Supercomputing Center)
|
213 |
-
be liable for any results arising from the use made by third parties.
|
214 |
-
|
215 |
-
</details>
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
datasets:
|
4 |
+
- projecte-aina/CA-ZH_Parallel_Corpus
|
5 |
+
language:
|
6 |
+
- zh
|
7 |
+
- ca
|
8 |
+
base_model:
|
9 |
+
- facebook/m2m100_1.2B
|
10 |
+
---
|
11 |
+
## Projecte Aina’s Catalan-Chinese machine translation model
|
12 |
+
|
13 |
+
## Table of Contents
|
14 |
+
<details>
|
15 |
+
<summary>Click to expand</summary>
|
16 |
+
|
17 |
+
- [Model description](#model-description)
|
18 |
+
- [Intended uses and limitations](#intended-uses-and-limitations)
|
19 |
+
- [How to use](#how-to-use)
|
20 |
+
- [Limitations and bias](#limitations-and-bias)
|
21 |
+
- [Training](#training)
|
22 |
+
- [Evaluation](#evaluation)
|
23 |
+
- [Additional information](#additional-information)
|
24 |
+
|
25 |
+
</details>
|
26 |
+
|
27 |
+
|
28 |
+
## Model description
|
29 |
+
|
30 |
+
This machine translation model is built upon the M2M100 1.2B, fine-tuned specifically for Catalan-Chinese translation.
|
31 |
+
It is trained on a combination of Catalan-Chinese datasets
|
32 |
+
totalling 94,187,858 sentence pairs. 113,305 sentence pairs were parallel data collected from the web, while the remaining 94,074,553 sentence pairs
|
33 |
+
were parallel synthetic data created using the
|
34 |
+
[Aina Project's Spanish-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-es-ca) and the [Aina Project's English-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-en-ca).
|
35 |
+
|
36 |
+
Following the fine-tuning phase, Contrastive Preference Optimization (CPO) was applied to further refine the model's outputs. CPO training involved pairs of "chosen" and "rejected" translations for a total of 4,006 sentences. These sentences were sourced from the Flores development set (997 sentences), the Flores devtest set (1,012 sentences), and the NTREX set (1,997 sentences).
|
37 |
+
|
38 |
+
The model was evaluated on the Projecte Aina's Catalan-Chinese evaluation dataset (unpublished), achieving results comparable to those of Google Translate.
|
39 |
+
|
40 |
+
## Intended uses and limitations
|
41 |
+
|
42 |
+
You can use this model for machine translation from Catalan to simplified Chinese.
|
43 |
+
|
44 |
+
## How to use
|
45 |
+
|
46 |
+
### Usage
|
47 |
+
|
48 |
+
Translate a sentence using python
|
49 |
+
```python
|
50 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
51 |
+
|
52 |
+
model_id = "projecte-aina/aina-translator-ca-zh"
|
53 |
+
|
54 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
|
55 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
56 |
+
|
57 |
+
sentence = "Benvingut al projecte Aina!"
|
58 |
+
|
59 |
+
input_ids = tokenizer(sentence, return_tensors="pt").input_ids
|
60 |
+
output_ids = model.generate(input_ids, max_length=200, num_beams=5)
|
61 |
+
|
62 |
+
generated_translation= tokenizer.decode(output_ids[0], skip_special_tokens=True, spaces_between_special_tokens = False).strip()
|
63 |
+
print(generated_translation)
|
64 |
+
#欢迎来到 Aina 项目!
|
65 |
+
```
|
66 |
+
|
67 |
+
|
68 |
+
## Limitations and bias
|
69 |
+
At the time of submission, no measures have been taken to estimate the bias and toxicity embedded in the model.
|
70 |
+
However, we are well aware that our models may be biased. We intend to conduct research in these areas in the future, and if completed, this model card will be updated.
|
71 |
+
|
72 |
+
## Training
|
73 |
+
|
74 |
+
### Training data
|
75 |
+
|
76 |
+
The Catalan-Chinese data collected from the web was a combination of the following datasets:
|
77 |
+
|
78 |
+
| Dataset | Sentences before cleaning |
|
79 |
+
|-------------------|----------------|
|
80 |
+
| OpenSubtitles | 139,300 |
|
81 |
+
| WikiMatrix | 90,643 |
|
82 |
+
| Wikipedia | 68,623|
|
83 |
+
| **Total** | **298,566** |
|
84 |
+
|
85 |
+
94,074,553 sentence pairs of synthetic parallel data were created from the following Spanish-Chinese datasets and English-Chinese datasets:
|
86 |
+
|
87 |
+
**Spanish-Chinese:**
|
88 |
+
|
89 |
+
| Dataset | Sentences before cleaning |
|
90 |
+
|-------------------|----------------|
|
91 |
+
| NLLB |24,051,233|
|
92 |
+
| UNPC | 17,599,223 |
|
93 |
+
| MultiUN | 9,847,770 |
|
94 |
+
| OpenSubtitles | 9,319,658 |
|
95 |
+
| MultiParaCrawl | 3,410,087 |
|
96 |
+
| MultiCCAligned | 3,006,694 |
|
97 |
+
| WikiMatrix | 1,214,322 |
|
98 |
+
| News Commentary | 375,982 |
|
99 |
+
| Tatoeba | 9,404 |
|
100 |
+
| **Total** | **68,834,373** |
|
101 |
+
|
102 |
+
**English-Chinese:**
|
103 |
+
|
104 |
+
| Dataset | Sentences before cleaning |
|
105 |
+
|-------------------|----------------|
|
106 |
+
| NLLB |71,383,325|
|
107 |
+
| CCAligned | 15,181,415 |
|
108 |
+
| Paracrawl | 14,170,869|
|
109 |
+
| WikiMatrix | 2,595,119|
|
110 |
+
| **Total** | **103,330,728** |
|
111 |
+
|
112 |
+
|
113 |
+
|
114 |
+
### Training procedure
|
115 |
+
|
116 |
+
### Data preparation
|
117 |
+
|
118 |
+
**Catalan-Chinese parallel data**
|
119 |
+
|
120 |
+
The Chinese side of all datasets were first processed using the [Hanzi Identifier](https://github.com/tsroten/hanzidentifier) to detect Traditional Chinese, which was subsequently converted to Simplified Chinese using [OpenCC](https://github.com/BYVoid/OpenCC).
|
121 |
+
|
122 |
+
All data was then filtered according to two specific criteria:
|
123 |
+
|
124 |
+
- Alignment: sentence level alignments were calculated using [LaBSE](https://huggingface.co/sentence-transformers/LaBSE) and sentence pairs with a score below 0.75 were discarded.
|
125 |
+
|
126 |
+
- Language identification: the probability of being the target language was calculated using [Lingua.py](https://github.com/pemistahl/lingua-py) and sentences with a language probability score below 0.5 were discarded.
|
127 |
+
|
128 |
+
Next, Spanish data was translated into Catalan using the Aina Project's [Spanish-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-es-ca), while English data was translated into Catalan using the Aina Project's [English-Catalan machine translation model](https://huggingface.co/projecte-aina/aina-translator-en-ca).
|
129 |
+
|
130 |
+
The filtered and translated datasets are then concatenated and deduplicated to form a final corpus of 94,187,858.
|
131 |
+
|
132 |
+
**Catalan-Chinese Contrastive Preference Optimization dataset**
|
133 |
+
|
134 |
+
The CPO dataset is built by comparing the quality of translations across four distinct sources:
|
135 |
+
|
136 |
+
- Reference translation: Chinese sentences from Flores test set, Flores devtest set, and NTREX dataset.
|
137 |
+
- aina-translator-ca-zh: A specialized bilingual model for Catalan-Chinese translations.
|
138 |
+
- Google Translate: A widely-used general-purpose machine translation system.
|
139 |
+
- OpenAI GPT-4: A large-scale language model capable of performing a wide range of tasks in conversational settings, including high-quality translation.
|
140 |
+
|
141 |
+
To evaluate the quality of translations without relying on human annotations, we employ two reference-free evaluation models:
|
142 |
+
|
143 |
+
- [Unbabel/wmt23-cometkiwi-da-xxl](https://huggingface.co/Unbabel/wmt23-cometkiwi-da-xxl)
|
144 |
+
- [Unbabel/XCOMET-XXL](https://huggingface.co/Unbabel/XCOMET-XXL)
|
145 |
+
|
146 |
+
These models provide direct assessment scores for each translation. The scores from both models are averaged to determine the relative quality of each translation. Based on this evaluation, the highest-scoring ("chosen") and lowest-scoring ("rejected") translations are identified for each source sentence, forming contrastive pairs. The CPO dataset comprises a total of 4,006 such pairs of "chosen" and "rejected" translations.
|
147 |
+
|
148 |
+
|
149 |
+
#### Training
|
150 |
+
|
151 |
+
The training was executed on NVIDIA GPUs utilizing the Hugging Face Transformers framework.
|
152 |
+
The model was trained for 245,000 updates.
|
153 |
+
|
154 |
+
Following fine-tuning on the M2M100 1.2B model, Contrastive Preference Optimization (CPO) was performed using our CPO dataset and the Hugging Face CPO Trainer. This phase involved 1,500 updates.
|
155 |
+
|
156 |
+
## Evaluation
|
157 |
+
|
158 |
+
### Variable and metrics
|
159 |
+
|
160 |
+
Below are the evaluation results on the Projecte Aina's Catalan-Chinese test set (unpublished), compared to Google Translate for the CA-ZH direction. The evaluation was conducted using [`tower-eval`](https://github.com/deep-spin/tower-eval) following the standard setting (beam search with beam size 5, limiting the translation length to 200 tokens). We report the following metrics:
|
161 |
+
|
162 |
+
- BLEU: Sacrebleu implementation, version:2.4.0
|
163 |
+
- ChrF: Sacrebleu implementation.
|
164 |
+
- Comet: Model checkpoint: "Unbabel/wmt22-comet-da".
|
165 |
+
- Comet-kiwi: Model checkpoint: "Unbabel/wmt22-cometkiwi-da".
|
166 |
+
|
167 |
+
|
168 |
+
### Evaluation results
|
169 |
+
|
170 |
+
Below are the evaluation results on the machine translation from Catalan to Chinese compared to [Google Translate](https://translate.google.com/):
|
171 |
+
|
172 |
+
|
173 |
+
#### Projecte Aina's Catalan-Chinese evaluation dataset
|
174 |
+
|
175 |
+
| | Bleu ↑ | ChrF ↑ | Comet ↑ | Comet-kiwi ↑ |
|
176 |
+
|:-----------------------|-------:|------:|-------:|--------:|
|
177 |
+
| aina-translator-ca-zh | 43.88 | 40.19 | **0.87** | **0.81** |
|
178 |
+
| Google Translate | **44.64** | **41.15** | **0.87** | 0.80 |
|
179 |
+
|
180 |
+
|
181 |
+
|
182 |
+
## Additional information
|
183 |
+
|
184 |
+
### Author
|
185 |
+
The Language Technologies Unit from Barcelona Supercomputing Center.
|
186 |
+
|
187 |
+
### Contact
|
188 |
+
For further information, please send an email to <[email protected]>.
|
189 |
+
|
190 |
+
### Copyright
|
191 |
+
Copyright(c) 2023 by Language Technologies Unit, Barcelona Supercomputing Center.
|
192 |
+
|
193 |
+
### License
|
194 |
+
[Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
195 |
+
|
196 |
+
### Funding
|
197 |
+
This work has been promoted and financed by the Generalitat de Catalunya through the [Aina project](https://projecteaina.cat/).
|
198 |
+
|
199 |
+
### Disclaimer
|
200 |
+
|
201 |
+
<details>
|
202 |
+
<summary>Click to expand</summary>
|
203 |
+
|
204 |
+
The model published in this repository is intended for a generalist purpose and is available to third parties under a permissive Apache License, Version 2.0.
|
205 |
+
|
206 |
+
Be aware that the model may have biases and/or any other undesirable distortions.
|
207 |
+
|
208 |
+
When third parties deploy or provide systems and/or services to other parties using this model (or any system based on it)
|
209 |
+
or become users of the model, they should note that it is their responsibility to mitigate the risks arising from its use and,
|
210 |
+
in any event, to comply with applicable regulations, including regulations regarding the use of Artificial Intelligence.
|
211 |
+
|
212 |
+
In no event shall the owner and creator of the model (Barcelona Supercomputing Center)
|
213 |
+
be liable for any results arising from the use made by third parties.
|
214 |
+
|
215 |
+
</details>
|