Commit 01e4305c authored by panhb's avatar panhb
Browse files

Update README.md

parent f7e394b6
Pipeline #1680 failed with stages
in 0 seconds
...@@ -98,11 +98,11 @@ conda create -n bert_tensorflow python=3.10 ...@@ -98,11 +98,11 @@ conda create -n bert_tensorflow python=3.10
pip install -r requirements.txt pip install -r requirements.txt
``` ```
安装过程可能顶掉DCU版本的tensorflow,可以到[开发者社区](https://cancon.hpccube.com:65024/4/main/tensorflow/dtk22.10)下载DCU版本对应包 安装过程可能顶掉DCU版本的tensorflow,可以到[开发者社区](https://developer.hpccube.com/tool/)下载DCU版本对应包
[tensorflow2.7](https://cancon.hpccube.com:65024/directlink/4/tensorflow/dtk22.10/tensorflow-2.7.0+git67f0ade9.dtk2210-cp37-cp37m-manylinux2014_x86_64.whl) [tensorflow2.13.1](https://cancon.hpccube.com:65024/directlink/4/tensorflow/DAS1.1/tensorflow-2.13.1+das1.1.git56b06c8.abi1.dtk2404-cp310-cp310-manylinux_2_31_x86_64.whl)
[DTK22.10.1](https://cancon.hpccube.com:65024/directlink/1/DTK-22.10.1/CentOS7.6/DTK-22.10.1-CentOS7.6-x86_64.tar.gz) [DTK22.10.1](https://cancon.hpccube.com:65024/directlink/1/latest/Ubuntu22.04/DTK-24.04.1-Ubuntu22.04-x86_64.tar.gz)
### python版本兼容 ### python版本兼容
...@@ -146,7 +146,7 @@ python create_finetuning_data.py \ ...@@ -146,7 +146,7 @@ python create_finetuning_data.py \
### 模型转化-MNLI ### 模型转化-MNLI
TF2.7.2与TF1.15.0模型存储、读取格式不同,官网给出的Bert一般是基于TF1.0的模型需要进行模型转化 TF2.13.1与TF1.15.0模型存储、读取格式不同,官网给出的Bert一般是基于TF1.0的模型需要进行模型转化
``` ```
python3 tf2_encoder_checkpoint_converter.py \ python3 tf2_encoder_checkpoint_converter.py \
...@@ -161,6 +161,8 @@ python3 tf2_encoder_checkpoint_converter.py \ ...@@ -161,6 +161,8 @@ python3 tf2_encoder_checkpoint_converter.py \
将转换完后的bert_model.ckpt-1.data-00000-of-00001 改为bert_model.ckpt.data-00000-of-00001 将转换完后的bert_model.ckpt-1.data-00000-of-00001 改为bert_model.ckpt.data-00000-of-00001
bert_model.ckpt-1.index改为 bert_model.ckpt.index bert_model.ckpt-1.index改为 bert_model.ckpt.index
如果报错 'no attriute experimental',则删除报错行中的experimental
``` ```
### 单卡运行-MNLI ### 单卡运行-MNLI
......
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