Davlan commited on
Commit
0867bb0
1 Parent(s): 0ecb52d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +36 -0
README.md CHANGED
@@ -1,3 +1,39 @@
1
  ---
2
  license: afl-3.0
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: afl-3.0
3
  ---
4
+
5
+
6
+ # afro-xlmr-small
7
+
8
+ AfroXLMR-mini was created by [first reducing the vocabulary token size](https://aclanthology.org/2020.sustainlp-1.16/) of [XLM-R-miniLM](https://huggingface.co/nreimers/mMiniLMv2-L12-H384-distilled-from-XLMR-Large) from 250K to 70k, followed by MLM adaptation on 17 African languages (Afrikaans, Amharic, Hausa, Igbo, Malagasy, Chichewa, Oromo, Naija, Kinyarwanda, Kirundi, Shona, Somali, Sesotho, Swahili, isiXhosa, Yoruba, and isiZulu) covering the major African language families and 3 high resource languages (Arabic, French, and English).
9
+
10
+ ## Eval results on MasakhaNER (F-score)
11
+ language| XLM-R-miniLM| XLM-R-base |XLM-R-large| afro-xlmr-base | afro-xlmr-small | afro-xlmr-mini
12
+ -|-|-|-|-|-|-
13
+ amh |69.5|70.6|76.2|76.1|70.1|69.7
14
+ hau |74.5|89.5|90.5|91.2|91.4|87.7
15
+ ibo |81.9|84.8|84.1|87.4|86.6|83.5
16
+ kin |68.6|73.3|73.8|78.0|77.5|74.1
17
+ lug |64.7|79.7|81.6|82.9|83.2|77.4
18
+ luo |11.7|74.9|73.6|75.1|75.4|17.5
19
+ pcm |83.2|87.3|89.0|89.6|89.0|85.5
20
+ swa |86.3|87.4|89.4|88.6|88.7|86.0
21
+ wol |51.7|63.9|67.9|67.4|65.9|59.0
22
+ yor |72.0|78.3|78.9|82.1|81.3|75.1
23
+
24
+ ### BibTeX entry and citation info
25
+ ```
26
+ @misc{afro_maft,
27
+ doi = {10.48550/ARXIV.2204.06487},
28
+ url = {https://arxiv.org/abs/2204.06487},
29
+ author = {Alabi, Jesujoba O. and Adelani, David Ifeoluwa and Mosbach, Marius and Klakow, Dietrich},
30
+ keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
31
+ title = {Multilingual Language Model Adaptive Fine-Tuning: A Study on African Languages},
32
+ publisher = {arXiv},
33
+ year = {2022},
34
+ copyright = {Creative Commons Attribution 4.0 International}
35
+ }
36
+
37
+ ```
38
+
39
+