diff --git a/TXTfpbsupervisedBERT/Metainfo.txt b/TXTfpbsupervisedBERT/Metainfo.txt
new file mode 100644
index 0000000..baaef55
--- /dev/null
+++ b/TXTfpbsupervisedBERT/Metainfo.txt
@@ -0,0 +1,21 @@
+Name of QuantLet : TXTfpbsupervisedBERT
+
+Published in : TXT
+
+Description : ‘Sentiment classification using BERT (Devlin et al, 2018). 
+The results are evaluated using Malo et al. (2014)’s « Sentences_66Agree.txt ».  We use Support Vector Machine (SVM) as a baseline. 
+
+This training set is available at 
+https://www.researchgate.net/publication/251231364_FinancialPhraseBank-v10
+
+The tutorial used for training BERT can be found here: https://medium.com/@abhikjha/fastai-integration-with-bert-a0a66b1cecbe 
+The adaptation of the code to our case for training BERT can be found here: https://colab.research.google.com/drive/11zeT5R3mGAiaQ9Vu5J7ujoFGOMbky6qV 
+
+
+Keywords : text mining, sentiment, classification, transfer learning, BERT, support vector machine
+
+See also : TXTfpbsupervised for SVM baseline
+
+Author : Manuel Tonneau
+
+Submitted : Thu, Sep 5 2019 by Manuel Tonneau