leftyer/mindformers
1
0
Fork 0
mirror of https://gitee.com/mindspore/mindformers.git synced 2025-12-06 11:29:59 +08:00
Code Issues Packages Projects Releases Wiki Activity

bugfix: api doc bugfix

Browse Source
This commit is contained in:
husichao
2025-09-12 16:47:39 +08:00
parent 78e2e6cc2b
commit c6dd16bf12
4 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/api/api_python/core/mindformers.core.CrossEntropyLoss.rst
View File

@@ -1,7 +1,7 @@
mindformers.core.CrossEntropyLoss
=================================
.. py:class:: mindformers.core.CrossEntropyLoss(parallel_config=default_dpmp_config, check_for_nan_in_loss_and_grad=False, monitor_device_local_loss=False, calculate_per_token_loss=False, seq_split_num=1, **kwargs)
.. py:class:: mindformers.core.CrossEntropyLoss(parallel_config=default_dpmp_config, calculate_per_token_loss=False, seq_split_num=1, **kwargs)
计算预测值和目标值之间的交叉熵损失。

2
docs/api/api_python/core/mindformers.core.MFLossMonitor.rst
View File

@@ -1,7 +1,7 @@
mindformers.core.MFLossMonitor
==============================
.. py:class:: mindformers.core.MFLossMonitor(learning_rate=None, per_print_times=1, micro_batch_num=1, micro_batch_interleave_num=1, origin_epochs=None, dataset_size=None, initial_epoch=0, initial_step=0, global_batch_size=0, gradient_accumulation_steps=1, check_for_nan_in_loss_and_grad=False, calculate_per_token_loss=False)
.. py:class:: mindformers.core.MFLossMonitor(learning_rate=None, per_print_times=1, micro_batch_num=1, micro_batch_interleave_num=1, origin_epochs=None, dataset_size=None, initial_epoch=0, initial_step=0, global_batch_size=0, gradient_accumulation_steps=1, check_for_nan_in_loss_and_grad=False, calculate_per_token_loss=False, print_separate_loss=False)
监控训练过程中loss等相关参数的回调函数。

2
docs/api/api_python/wrapper/mindformers.wrapper.MFPipelineWithLossScaleCell.rst
View File

@@ -1,7 +1,7 @@
mindformers.wrapper.MFPipelineWithLossScaleCell
===============================================
.. py:class:: mindformers.wrapper.MFPipelineWithLossScaleCell(network, optimizer, use_clip_grad=True, max_grad_norm=1.0, scale_sense=1.0, micro_batch_num=1, local_norm=False, calculate_per_token_loss=False, global_norm_spike_threshold=1.0, use_skip_data_by_global_norm=False, **kwargs)
.. py:class:: mindformers.wrapper.MFPipelineWithLossScaleCell(network, optimizer, use_clip_grad=True, max_grad_norm=1.0, scale_sense=1.0, micro_batch_num=1, local_norm=False, calculate_per_token_loss=False, global_norm_spike_threshold=1.0, use_skip_data_by_global_norm=False, print_separate_loss=False, **kwargs)
为MindFormers的单步训练单元扩充流水线并行的损失缩放功能。

2
docs/api/api_python/wrapper/mindformers.wrapper.MFTrainOneStepCell.rst
View File

@@ -1,7 +1,7 @@
mindformers.wrapper.MFTrainOneStepCell
======================================
.. py:class:: mindformers.wrapper.MFTrainOneStepCell(network, optimizer, use_clip_grad=False, max_grad_norm=1.0, scale_sense=1.0, local_norm=False, calculate_per_token_loss=False, global_norm_spike_threshold=1.0, use_skip_data_by_global_norm=False, **kwargs)
.. py:class:: mindformers.wrapper.MFTrainOneStepCell(network, optimizer, use_clip_grad=False, max_grad_norm=1.0, scale_sense=1.0, local_norm=False, calculate_per_token_loss=False, global_norm_spike_threshold=1.0, use_skip_data_by_global_norm=False, print_separate_loss=False, **kwargs)
MindFormers的单步训练包装接口。
使用损失缩放、梯度裁剪、梯度累积、指数移动平均等策略进行网络训练。