Unverified Commit d8dbf3b7 authored by Stas Bekman's avatar Stas Bekman Committed by GitHub
Browse files

[s2s] clean up + doc (#6184)


Co-authored-by: default avatarSam Shleifer <sshleifer@gmail.com>
parent a39dfe4f
...@@ -66,6 +66,15 @@ Summarization Tips: ...@@ -66,6 +66,15 @@ Summarization Tips:
Datasets: Seq2SeqDataset will be used for all models besides MBart, for which MBartDataset will be used.** Datasets: Seq2SeqDataset will be used for all models besides MBart, for which MBartDataset will be used.**
A new dataset is needed to support multilingual tasks. A new dataset is needed to support multilingual tasks.
### Command Line Options
To see all the possible command line options, run:
```bash
./finetune.sh --help # this calls python finetune.py --help
```
### Finetuning Training Params ### Finetuning Training Params
To override the pretrained model's training params, you can pass them to `./finetune.sh`: To override the pretrained model's training params, you can pass them to `./finetune.sh`:
......
...@@ -2,6 +2,7 @@ ...@@ -2,6 +2,7 @@
export PYTHONPATH="../":"${PYTHONPATH}" export PYTHONPATH="../":"${PYTHONPATH}"
# the proper usage is documented in the README, you need to specify data_dir, output_dir and model_name_or_path # the proper usage is documented in the README, you need to specify data_dir, output_dir and model_name_or_path
# run ./finetune.sh --help to see all the possible options
python finetune.py \ python finetune.py \
--learning_rate=3e-5 \ --learning_rate=3e-5 \
--fp16 \ --fp16 \
...@@ -10,8 +11,4 @@ python finetune.py \ ...@@ -10,8 +11,4 @@ python finetune.py \
--do_predict \ --do_predict \
--n_val 1000 \ --n_val 1000 \
--val_check_interval 0.1 \ --val_check_interval 0.1 \
--encoder_layerdrop 0.1 \
--decoder_layerdrop 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
$@ $@
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment