Skip to content

Commit b19244d

Browse files
authored
Update README.md (#4205)
1 parent 67a045c commit b19244d

File tree

1 file changed

+6
-0
lines changed

1 file changed

+6
-0
lines changed

model_zoo/ernie-1.0/README.md

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -367,6 +367,12 @@ wget https://bj.bcebos.com/paddlenlp/models/transformers/data_tools/wudao_200g_s
367367
wget https://bj.bcebos.com/paddlenlp/models/transformers/data_tools/wudao_200g_sample_ernie-3.0-base-zh_idx.npz
368368
cd -
369369
```
370+
同时我们也提供了 `ernie-1.0-base-zh` 的悟道一个小规模样本的数据:
371+
```
372+
https://paddlenlp.bj.bcebos.com/models/transformers/data_tools/wudao_200g_sample_ernie-1.0-base-zh_ids.npy
373+
https://paddlenlp.bj.bcebos.com/models/transformers/data_tools/wudao_200g_sample_ernie-1.0-base-zh_idx.npz
374+
```
375+
370376
可以指定`tokenizer_name_or_path=ernie-3.0-bash-zh`,`input_dir=./data` 用下面的脚本训练。
371377

372378
这里启动的是单机8卡任务,整体全局的batch_size 512 (64*8)。如果指定ips参数,进行多机运行,如 `python3 -u -m paddle.distributed.launch --gpus "0,1,2,3,4,5,6,7" --ips 192.168.1.101,192.168.1.101 `

0 commit comments

Comments
 (0)