added model card
Browse files
README.md
ADDED
@@ -0,0 +1,113 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
language:
|
4 |
+
- de
|
5 |
+
pipeline_tag: text-generation
|
6 |
+
tags:
|
7 |
+
- german
|
8 |
+
- deutsch
|
9 |
+
- simplification
|
10 |
+
- vereinfachung
|
11 |
+
---
|
12 |
+
# Model Card for Model ID
|
13 |
+
|
14 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
15 |
+
|
16 |
+
We fine-tuned [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) with a set of ca. 800 newspaper articles which have been simplified by the Austrian Press Agency.
|
17 |
+
Our aim was to have a model which can simplify German-language text.
|
18 |
+
|
19 |
+
## Model Details
|
20 |
+
|
21 |
+
### Model Description
|
22 |
+
|
23 |
+
<!-- Provide a longer summary of what this model is. -->
|
24 |
+
|
25 |
+
|
26 |
+
|
27 |
+
- **Developed by:** Members of the [Public Interest AI research group](https://publicinterest.ai/), [HIIG Berlin](https://www.hiig.de/)
|
28 |
+
- **Model type:** simplification model, text generation
|
29 |
+
- **Language(s) (NLP):** German
|
30 |
+
- **License:** Apache 2.0
|
31 |
+
- **Finetuned from model:** meta-llama/Meta-Llama-3-8B-Instruct
|
32 |
+
|
33 |
+
### Model Sources
|
34 |
+
|
35 |
+
<!-- Provide the basic links for the model. -->
|
36 |
+
|
37 |
+
- **Repository:** https://github.com/fhewett/simba
|
38 |
+
<!-- - **Paper [optional]:** [More Information Needed] -->
|
39 |
+
- **Project website:** https://publicinterest.ai/tool/simba
|
40 |
+
|
41 |
+
## Uses
|
42 |
+
|
43 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
44 |
+
|
45 |
+
### Direct Use
|
46 |
+
|
47 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
48 |
+
|
49 |
+
This model works best for simplifying German-language newspaper articles (news items, not commentaries or editorials). It may work for other types of texts.
|
50 |
+
|
51 |
+
### Downstream Use
|
52 |
+
|
53 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
54 |
+
We have fine-tuned using only newspaper articles. We have not yet performed extensive out-of-domain testing, but believe that the model's capabilities could be improved by fine-tuning on more diverse data. Contact us if you have a dataset which you think could work (parallel texts, German standard & German simplified).
|
55 |
+
|
56 |
+
<!-- ### Out-of-Scope Use -->
|
57 |
+
|
58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
59 |
+
|
60 |
+
## Bias, Risks, and Limitations
|
61 |
+
|
62 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
63 |
+
|
64 |
+
As with most text generation models, the model sometimes produces information that is incorrect.
|
65 |
+
|
66 |
+
### Recommendations
|
67 |
+
|
68 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
69 |
+
|
70 |
+
Please check manually that your output text corresponds to the input text, as factual inconsistencies may have arisen.
|
71 |
+
|
72 |
+
## How to Get Started with the Model
|
73 |
+
|
74 |
+
Use the code below to get started with the model.
|
75 |
+
|
76 |
+
[More Information Needed]
|
77 |
+
|
78 |
+
## Training Details
|
79 |
+
|
80 |
+
### Training Data
|
81 |
+
|
82 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
83 |
+
|
84 |
+
A sample of the data used to train our model can be found [here](https://github.com/fhewett/apa-rst/tree/main/original_texts).
|
85 |
+
|
86 |
+
#### Training Hyperparameters
|
87 |
+
|
88 |
+
- **Training regime:** Our training script can be found [here](https://github.com/fhewett/simba/blob/main/models/train_simba.py). <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
89 |
+
|
90 |
+
<!-- #### Speeds, Sizes, Times [optional] -->
|
91 |
+
|
92 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
93 |
+
|
94 |
+
## Evaluation
|
95 |
+
|
96 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
97 |
+
|
98 |
+
#### Summary
|
99 |
+
|
100 |
+
|
101 |
+
<!-- ## Citation [optional]
|
102 |
+
|
103 |
+
**BibTeX:**
|
104 |
+
|
105 |
+
[More Information Needed]
|
106 |
+
|
107 |
+
**APA:**
|
108 |
+
|
109 |
+
[More Information Needed]-->
|
110 |
+
|
111 |
+
## Model Card Contact
|
112 |
+
|
113 |
+
simba -at- hiig.de
|