Update README.md

README.md

@@ -50,9 +50,9 @@ The following hyperparameters were used during pretraining:
 - total_train_batch_size: 192
 - max_seq_length: 4096
 - training_steps: 600000
-- warmup_steps: 10000
-- bf16: True
-- deepspeed: ds_config.json
+- warmup_steps: 6000
+- bf16: true
+- deepspeed: [ds_config.json](https://huggingface.co/nlp-waseda/bigbird-base-japanese/blob/main/ds_config.json)
 
 ## Performance on JGLUE
 
@@ -61,7 +61,7 @@ We tuned learning rate and training epochs for each model and task following [th
 
 For tasks other than MARC-ja, the maximum length is short, so the attention_type was set to "original_full" and fine-tuning was performed for tasks other than MARC-ja. For MARC-ja, both "block_sparse" and "original_full" were used.
 
-| Model                         | MARC-ja(original_full)/acc | JSTS/pearson | JSTS/spearman | JNLI/acc | JSQuAD/EM | JSQuAD/F1 | JComQA/acc |
+| Model                         | MARC-ja/acc | JSTS/pearson | JSTS/spearman | JNLI/acc | JSQuAD/EM | JSQuAD/F1 | JComQA/acc |
 |-------------------------------|--------------|---------------|----------|-----------|-----------|------------|------------|
 | Waseda RoBERTa base           |  0.965       | 0.913        | 0.876         | 0.905    | 0.853     | 0.916     | 0.853      |
 | Waseda RoBERTa large (seq512) | 0.969       | 0.925        | 0.890         | 0.928    | 0.910     | 0.955     | 0.900      |