DarshanDeshpande commited on
Commit
406cae4
1 Parent(s): 5f4cddc

Added README.md

Browse files
Files changed (1) hide show
  1. README.md +59 -0
README.md ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - mr
4
+ tags:
5
+ - fill-mask
6
+ license: apache-2.0
7
+ datasets:
8
+ - Oscar Corpus, News, Stories
9
+ widget:
10
+ - text: "हा खरोखर चांगला [MASK] आहे."
11
+ ---
12
+
13
+ # Marathi DistilBERT
14
+
15
+ ## Model description
16
+
17
+ This model is an adaptation of DistilBert (Victor Sanh et al.) for Marathi language. This version of Marathi-DistilBert is trained from scratch on approximately 11.2 million sentences.
18
+
19
+ ```
20
+ DISCLAIMER
21
+
22
+ This model has not been thoroughly tested and may contain biased opinions or inappropriate language. User discretion is advised
23
+ ```
24
+
25
+ ## Training data
26
+ The training data has been extracted from a variety of sources, mainly including:
27
+ 1. Oscar Corpus
28
+ 2. Marathi Newspapers
29
+ 3. Marathi story books and articles
30
+ The data is cleaned by removing all languages other than Marathi, while preserving common punctuations
31
+
32
+ ## Training procedure
33
+ The model is trained from scratch using an Adam optimizer with a learning rate of 1e-4 and default β1 and β2 values of 0.9 and 0.999 respectively with a total batch size of 256 on a v3-8 TPU and mask probability of 15%.
34
+
35
+ ## Example
36
+ ```python
37
+ from transformers import pipeline
38
+ fill_mask = pipeline(
39
+ "fill-mask",
40
+ model="DarshanDeshpande/marathi-distilbert",
41
+ tokenizer="DarshanDeshpande/marathi-distilbert",
42
+ )
43
+ fill_mask("हा खरोखर चांगला [MASK] आहे.")
44
+ ```
45
+
46
+ ### BibTeX entry and citation info
47
+
48
+ ```bibtex
49
+ @misc{sanh2020distilbert,
50
+ title={DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter},
51
+ author={Victor Sanh and Lysandre Debut and Julien Chaumond and Thomas Wolf},
52
+ year={2020},
53
+ eprint={1910.01108},
54
+ archivePrefix={arXiv},
55
+ primaryClass={cs.CL}
56
+ }
57
+ ```
58
+
59
+ <h3>Author: Darshan Deshpande: <a href="https://github.com/DarshanDeshpande">GitHub</a>, <a href="https://www.linkedin.com/in/darshan-deshpande/">LinkedIn</a><h3>