Skip to content

Commit

Permalink
v0.2.1.post2
Browse files Browse the repository at this point in the history
  • Loading branch information
Tongjilibo committed Mar 17, 2024
1 parent 59a2603 commit 2f3fcd4
Show file tree
Hide file tree
Showing 3 changed files with 3 additions and 3 deletions.
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -68,7 +68,7 @@ pip install git+https://github.com/Tongjilibo/torch4keras.git
## 4. 版本历史
|更新日期| 版本 | 版本说明 |
|------| ----------------- |----------- |
|20240316|v0.2.1 |训练异常时保存权重,避免空数据集error,默认指标使用滑窗平滑,BaseModelDDP修改的更易用,mapping允许是函数更易用,Checkpoint增加save_on_train_end,增加SystemCallback,修改run_callback=False的bug, 适配build_MiniLLM_from_scratch|
|20240317|v0.2.1.post2 |训练异常时保存权重,避免空数据集error,默认指标使用滑窗平滑,BaseModelDDP修改的更易用,mapping允许是函数更易用,Checkpoint增加save_on_train_end,增加SystemCallback,修改run_callback=False的bug, 适配build_MiniLLM_from_scratch, 修复ddp中mix_precision和torch重名的bug|
|20240221|v0.2.0 | fit中修改.train()逻辑较少耗时|
|20240204|v0.1.9 | 增加Timeit, Timeit2, timeit等时间/速度监控|
|20240116|v0.1.8 | 重新整理snippets, 重写save_pretrained|
Expand Down
2 changes: 1 addition & 1 deletion docs/History.md
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
## 更新历史

- **20240315**: 训练异常时保存权重,避免空数据集error,默认指标使用滑窗平滑,BaseModelDDP修改的更易用,mapping允许是函数更易用,Checkpoint增加save_on_train_end,增加SystemCallback,修改run_callback=False的bug, 适配build_MiniLLM_from_scratch
- **20240317**: 训练异常时保存权重,避免空数据集error,默认指标使用滑窗平滑,BaseModelDDP修改的更易用,mapping允许是函数更易用,Checkpoint增加save_on_train_end,增加SystemCallback,修改run_callback=False的bug, 适配build_MiniLLM_from_scratch, 修复ddp中mix_precision和torch重名的bug
- **20240221**: fit中修改.train()逻辑减少耗时
- **20240204**: 增加Timeit, Timeit2, timeid等时间/速度监控
- **20240116**: 重新整理snippets, 重写save_pretrained
Expand Down
2 changes: 1 addition & 1 deletion setup.py
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@

setup(
name='torch4keras',
version='v0.2.1',
version='v0.2.1.post2',
description='Use torch like keras',
long_description=long_description,
long_description_content_type="text/markdown",
Expand Down

0 comments on commit 2f3fcd4

Please sign in to comment.