davda54 commited on
Commit
2994875
1 Parent(s): f52bd43

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +48 -9
README.md CHANGED
@@ -15,25 +15,36 @@ pipeline_tag: text2text-generation
15
 
16
  # NorT5 x-small
17
 
 
 
 
 
18
 
19
  ## Other sizes:
20
- - [NorT5 xs (15M)](https://huggingface.co/ltg/nort5-xs)
21
- - [NorT5 small (40M)](https://huggingface.co/ltg/nort5-small)
22
- - [NorT5 base (123M)](https://huggingface.co/ltg/nort5-base)
23
- - [NorT5 large (323M)](https://huggingface.co/ltg/nort5-large)
 
 
 
 
 
 
 
24
 
25
 
26
  ## Example usage
27
 
28
- This model currently needs a custom wrapper from `modeling_nort5.py`. Then you can use it like this:
 
29
 
30
  ```python
31
  import torch
32
- from transformers import AutoTokenizer
33
- from modeling_norbert import NorT5ForConditionalGeneration
34
 
35
- tokenizer = AutoTokenizer.from_pretrained("path/to/folder")
36
- t5 = NorT5ForConditionalGeneration.from_pretrained("path/to/folder")
37
 
38
 
39
  # MASKED LANGUAGE MODELING
@@ -59,4 +70,32 @@ output_tensor = model.generate(input_tensor, max_new_tokens=50, num_beams=4, do_
59
  tokenizer.decode(output_tensor.squeeze())
60
 
61
  # should output: [BOS]ˈoppvarming, det vil si at det skjer en endring i temperaturen i et medium, f.eks. en ovn eller en radiator, slik at den blir varmere eller kaldere, eller at den blir varmere eller kaldere, eller at den blir
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
62
  ```
 
15
 
16
  # NorT5 x-small
17
 
18
+ <img src="https://huggingface.co/ltg/norbert3-base/resolve/main/norbert.png" width=12.5%>
19
+
20
+ The official release of a new generation of NorT5 language models described in paper [**NorBench — A Benchmark for Norwegian Language Models**](https://arxiv.org/abs/2305.03880). Plese read the paper to learn more details about the model.
21
+
22
 
23
  ## Other sizes:
24
+ - [NorT5 xs (32M)](https://huggingface.co/ltg/nort5-xs)
25
+ - [NorT5 small (88M)](https://huggingface.co/ltg/nort5-small)
26
+ - [NorT5 base (228M)](https://huggingface.co/ltg/nort5-base)
27
+ - [NorT5 large (808M)](https://huggingface.co/ltg/nort5-large)
28
+
29
+
30
+ ## Encoder-only NorBERT siblings:
31
+ - [NorBERT 3 xs (15M)](https://huggingface.co/ltg/norbert3-xs)
32
+ - [NorBERT 3 small (40M)](https://huggingface.co/ltg/norbert3-small)
33
+ - [NorBERT 3 base (123M)](https://huggingface.co/ltg/norbert3-base)
34
+ - [NorBERT 3 large (323M)](https://huggingface.co/ltg/norbert3-large)
35
 
36
 
37
  ## Example usage
38
 
39
+ This model currently needs a custom wrapper from `modeling_nort5.py`, you should therefore load the model with `trust_remote_code=True`.
40
+
41
 
42
  ```python
43
  import torch
44
+ from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 
45
 
46
+ tokenizer = AutoTokenizer.from_pretrained("ltg/nort5-xs")
47
+ t5 = AutoModelForSeq2SeqLM.from_pretrained("ltg/nort5-xs")
48
 
49
 
50
  # MASKED LANGUAGE MODELING
 
70
  tokenizer.decode(output_tensor.squeeze())
71
 
72
  # should output: [BOS]ˈoppvarming, det vil si at det skjer en endring i temperaturen i et medium, f.eks. en ovn eller en radiator, slik at den blir varmere eller kaldere, eller at den blir varmere eller kaldere, eller at den blir
73
+ ```
74
+
75
+
76
+ The following classes are currently implemented: `AutoModel`, `AutoModelForSeq2SeqLM`.
77
+
78
+ ## Cite us
79
+
80
+ ```bibtex
81
+ @inproceedings{samuel-etal-2023-norbench,
82
+ title = "{N}or{B}ench {--} A Benchmark for {N}orwegian Language Models",
83
+ author = "Samuel, David and
84
+ Kutuzov, Andrey and
85
+ Touileb, Samia and
86
+ Velldal, Erik and
87
+ {\O}vrelid, Lilja and
88
+ R{\o}nningstad, Egil and
89
+ Sigdel, Elina and
90
+ Palatkina, Anna",
91
+ booktitle = "Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)",
92
+ month = may,
93
+ year = "2023",
94
+ address = "T{\'o}rshavn, Faroe Islands",
95
+ publisher = "University of Tartu Library",
96
+ url = "https://aclanthology.org/2023.nodalida-1.61",
97
+ pages = "618--633",
98
+ abstract = "We present NorBench: a streamlined suite of NLP tasks and probes for evaluating Norwegian language models (LMs) on standardized data splits and evaluation metrics. We also introduce a range of new Norwegian language models (both encoder and encoder-decoder based). Finally, we compare and analyze their performance, along with other existing LMs, across the different benchmark tests of NorBench.",
99
+ }
100
+
101
  ```