Links to article

README.md

@@ -11,6 +11,8 @@ tags:
 
 This is a [sentence-transformers](https://www.SBERT.net) model: It maps Swedish sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. This model is a bilingual Swedish-English model trained according to instructions in the paper [Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation](https://arxiv.org/pdf/2004.09813.pdf) and the [documentation](https://www.sbert.net/examples/training/multilingual/README.html) accompanying its companion python package. We have used the strongest available pretrained English Bi-Encoder ([paraphrase-mpnet-base-v2](https://www.sbert.net/docs/pretrained_models.html#sentence-embedding-models)) as a teacher model, and the pretrained Swedish [KB-BERT](https://huggingface.co/KB/bert-base-swedish-cased) as the student model. 
 
+A more detailed description of the model can be found in an article we published on the [KBLab blog](https://kb-labb.github.io/posts/2021-08-23-a-swedish-sentence-transformer/). 
+
 <!--- Describe your model here -->
 
 ## Usage (Sentence-Transformers)
@@ -76,7 +78,7 @@ print(sentence_embeddings)
 
 <!--- Describe how your model was evaluated -->
 
-The model was evaluated on [SweParaphrase v1.0](https://spraakbanken.gu.se/en/resources/sweparaphrase) by calculating Pearson and Spearman correlation between predicted model similarity scores and the human similarity score labels. The model achieved a Pearson correlation coefficient of **0.918** and a Spearman's rank correlation coefficient of **0.911**.
+The model was primarily evaluated on [SweParaphrase v1.0](https://spraakbanken.gu.se/en/resources/sweparaphrase). This test set is part of [SuperLim](https://spraakbanken.gu.se/en/resources/superlim) -- a Swedish evaluation suite for natural langage understanding tasks.  We calculated Pearson and Spearman correlation between predicted model similarity scores and the human similarity score labels. The model achieved a Pearson correlation coefficient of **0.918** and a Spearman's rank correlation coefficient of **0.911**.
 
 The following code snippet can be used to reproduce the above results:
 
@@ -119,12 +121,15 @@ sentence_pair_scores = cosine_scores.diag()
 df["model_score"] = sentence_pair_scores.cpu().tolist()
 print(df[["score", "model_score"]].corr(method="spearman"))
 print(df[["score", "model_score"]].corr(method="pearson"))
-```
 
+Examples how to evaluate the model on other test sets of the SuperLim suites can be found on the following links: [evaluate_faq.py](https://github.com/kb-labb/swedish-sbert/blob/main/evaluate_faq.py) (Swedish FAQ), [evaluate_swesat.py](https://github.com/kb-labb/swedish-sbert/blob/main/evaluate_swesat.py) (SweSAT synonyms), [evaluate_supersim.py](https://github.com/kb-labb/swedish-sbert/blob/main/evaluate_supersim.py) (SuperSim).
+```
 
 ## Training
 
-Around 14.6 million sentences from English-Swedish parallel corpuses were used to train the model. Data was sourced from the [Open Parallel Corpus](https://opus.nlpl.eu/) (OPUS) and downloaded via the python package [opustools](https://pypi.org/project/opustools/). Datasets used were: JW300, EUbooks, Europarl, EUbookshop, EMEA, TED2020, Tatoeba and OpenSubtitles. 
+An article with more details on data and the model can be found on the [KBLab blog](https://kb-labb.github.io/posts/2021-08-23-a-swedish-sentence-transformer/). 
+
+Around 14.6 million sentences from English-Swedish parallel corpuses were used to train the model. Data was sourced from the [Open Parallel Corpus](https://opus.nlpl.eu/) (OPUS) and downloaded via the python package [opustools](https://pypi.org/project/opustools/). Datasets used were: JW300, Europarl, EUbookshop, EMEA, TED2020, Tatoeba and OpenSubtitles. 
 
 The model was trained with the parameters:
 
@@ -174,6 +179,8 @@ SentenceTransformer(
 <!--- Describe where people can find more information -->
 This model was trained by KBLab, a data lab at the National Library of Sweden. 
 
+You can cite the article on our blog: https://kb-labb.github.io/posts/2021-08-23-a-swedish-sentence-transformer/ .
+
 ## Acknowledgements
 
-We gratefully acknowledge the HPC RIVR consortium (www.hpc-rivr.si) and EuroHPC JU (eurohpc-ju.europa.eu) for funding this research by providing computing resources of the HPC system Vega at the Institute of Information Science (www.izum.si).
+We gratefully acknowledge the HPC RIVR consortium (www.hpc-rivr.si) and EuroHPC JU ([eurohpc-ju.europa.eu](eurohpc-ju.europa.eu)) for funding this research by providing computing resources of the HPC system Vega at the Institute of Information Science (www.izum.si).