From e10d389561b0fe969149fe856e025aabebc04f31 Mon Sep 17 00:00:00 2001 From: Keshan Date: Thu, 8 Oct 2020 02:10:52 +0530 Subject: [PATCH] [Model card] SinhalaBERTo model. (#7558) * [Model card] SinhalaBERTo model. This is the model card for keshan/SinhalaBERTo model. * Update model_cards/keshan/SinhalaBERTo/README.md Co-authored-by: Julien Chaumond --- model_cards/keshan/SinhalaBERTo/README.md | 37 +++++++++++++++++++++++ 1 file changed, 37 insertions(+) create mode 100644 model_cards/keshan/SinhalaBERTo/README.md diff --git a/model_cards/keshan/SinhalaBERTo/README.md b/model_cards/keshan/SinhalaBERTo/README.md new file mode 100644 index 00000000000..d1e71df59ab --- /dev/null +++ b/model_cards/keshan/SinhalaBERTo/README.md @@ -0,0 +1,37 @@ +--- +language: si +tags: +- SinhalaBERTo +- Sinhala +- roberta +datasets: +- oscar +--- +### Overview + +This is a slightly smaller model trained on [OSCAR](https://oscar-corpus.com/) Sinhala dedup dataset. As Sinhala is one of those low resource languages, there are only a handful of models been trained. So, this would be a great place to start training for more downstream tasks. + +## Model Specification + + +The model chosen for training is [Roberta](https://arxiv.org/abs/1907.11692) with the following specifications: + 1. vocab_size=52000 + 2. max_position_embeddings=514 + 3. num_attention_heads=12 + 4. num_hidden_layers=6 + 5. type_vocab_size=1 + +## How to Use +You can use this model directly with a pipeline for masked language modeling: + +```py +from transformers import AutoTokenizer, AutoModelWithLMHead, pipeline + +model = BertForMaskedLM.from_pretrained("keshan/SinhalaBERTo") +tokenizer = BertTokenizer.from_pretrained("keshan/SinhalaBERTo") + +fill_mask = pipeline('fill-mask', model=model, tokenizer=tokenizer) + +fill_mask("මම ගෙදර .") + +```