Update README.md

README.md

@@ -1,12 +1,20 @@
-Script for training can be found here: https://github.com/vasudevgupta7/bigbird
+---
+language: en
+license: apache-2.0
+datasets: natural_questions
+
+---
+
+This checkpoint is obtained after training `BigBirdForQuestionAnswering` on [`natural_questions`](https://huggingface.co/datasets/natural_questions) dataset for ~ 2 weeks on 2 K80 GPUs. Script for training can be found here: https://github.com/vasudevgupta7/bigbird
+
+**Use this model just like any other model from 🤗Transformers**
 
 ```python
 from transformers import BigBirdForQuestionAnswering
 
-epoch_0 = "4daa96c5befa9b728c47a77280ef9484377d7791"
-epoch_1 = "b962e30f2367cbc5e35b2c0d64faa9bad469e2e2"
-revision = epoch_1
+model_id = "vasudevgupta/bigbird-roberta-natural-questions"
+model = BigBirdForQuestionAnswering.from_pretrained(model_id)
+tokenizer = BigBirdTokenizer.from_pretrained(model_id)
+```
 
-model = BigBirdForQuestionAnswering.from_pretrained(model_id, revision=revision)
-tokenizer = BigBirdTokenizer.from_pretrained(model_id, revision=revision)
-```
+In case you are interested in predicting category (null, long, short, yes, no) as well, use `BigBirdForNaturalQuestions` (instead of `BigBirdForQuestionAnswering`) from my training script.