VityaVitalich commited on
Commit
cb7e01c
1 Parent(s): 18fcf61

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +30 -0
README.md CHANGED
@@ -24,3 +24,33 @@ configs:
24
  - split: test
25
  path: data/test-*
26
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
  - split: test
25
  path: data/test-*
26
  ---
27
+
28
+ # Dataset card for WordNet-TaxoLLaMA
29
+
30
+ TaxoLLaMA is a model capable of solving Lexical Semantics task with SoTA metrics.
31
+ The model was fine-tuned on instructive dataset WordNet-TaxoLLaMA. It consists of hypernym-hyponym pairs sampled from WordNet 3.0. As well, it contains definitions, that were used during training to help model disambiguate senses.
32
+
33
+ ## Input Format
34
+
35
+ The TaxoLLaMA model was trained to use the following format :
36
+ ```
37
+ <s>[INST] <<SYS>> You are a helpfull assistant. List all the possible words divided with a coma. Your answer should not include anything except the words divided by a coma<</SYS>>
38
+ hyponym: tiger (large feline of forests in most of Asia having a tawny coat with black stripes)| hypernyms: [/INST]
39
+ ```
40
+
41
+ We recommend you to follow this format, however you are free to change it to suite your task!
42
+
43
+ ## Citation
44
+
45
+ If you find TaxoLLaMA or WordNet-TaxoLLaMA is useful in your work, please cite it with:
46
+
47
+ ```
48
+ @misc{moskvoretskii2024taxollama,
49
+ title={TaxoLLaMA: WordNet-based Model for Solving Multiple Lexical Sematic Tasks},
50
+ author={Viktor Moskvoretskii and Ekaterina Neminova and Alina Lobanova and Alexander Panchenko and Irina Nikishina},
51
+ year={2024},
52
+ eprint={2403.09207},
53
+ archivePrefix={arXiv},
54
+ primaryClass={cs.CL}
55
+ }
56
+ ```