Update README.md

README.md

@@ -1,6 +1,6 @@
 # dv-wave
 
-This is a second attempt at a Dhivehi language model trained with
+This is my third version of a Dhivehi language model trained with
 Google Research's [ELECTRA](https://github.com/google-research/electra).
 
 Tokenization and pre-training CoLab: https://colab.research.google.com/drive/1ZJ3tU9MwyWj6UtQ-8G7QJKTn-hG1uQ9v?usp=sharing
@@ -10,17 +10,19 @@ Using SimpleTransformers to classify news https://colab.research.google.com/driv
 V1: similar performance to mBERT on news classification task after finetuning for 3 epochs (52%)
 
 V2: fixed tokenizers do_lower_case=False and strip_accents=False to preserve vowel signs of Dhivehi
-  dv-wave: 89% to mBERT: 52%
+
+> 8-topic news classification score: 88.6% compared to mBERT: 51.8%
+
+V3: trained longer on larger corpus (added OSCAR and Wikipedia)
+
+> news classification score: 89.2%
 
 ## Corpus
 
 Trained on @Sofwath's 307MB corpus of Dhivehi text: https://github.com/Sofwath/DhivehiDatasets
 
-This repo also contains the news classification task CSV
-
-[OSCAR](https://oscar-corpus.com/) was considered but has not been added to pretraining; as of
-this writing their web crawl has 126MB of Dhivehi text (79MB deduped).
+Plus [OSCAR](https://oscar-corpus.com/) and 2020-10-01 dump of dv.wikipedia.org
 
 ## Vocabulary
 
-Included as vocab.txt in the upload - vocab_size is 29874
+Included as vocab.txt in the upload