SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

fix typo in readme

Browse Source
This commit is contained in:
Thomas Wolf
2018-11-12 12:39:57 +01:00
committed by GitHub
parent fa1aa81f26
commit 5dfd19060a
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@@ -210,12 +210,12 @@ For example, fine-tuning BERT-large on SQuAD can be done on a server with 4 k-80
```bash ```bash
{"exact_match": 84.56953642384106, "f1": 91.04028647786927} {"exact_match": 84.56953642384106, "f1": 91.04028647786927}
``` ```
To get these results that we used a combination of: To get these results we used a combination of:
- multi-GPU training (automatically activated on a multi-GPU server), - multi-GPU training (automatically activated on a multi-GPU server),
- 2 steps of gradient accumulation and - 2 steps of gradient accumulation and
- perform the optimization step on CPU to store Adam's averages in RAM. - perform the optimization step on CPU to store Adam's averages in RAM.
Here are the full list of hyper-parameters we used for this run: Here are the full list of hyper-parameters for this run:
```bash ```bash
python ./run_squad.py \ python ./run_squad.py \
--vocab_file $BERT_LARGE_DIR/vocab.txt \ --vocab_file $BERT_LARGE_DIR/vocab.txt \