Skip to content

v1.0.2 beta: Multiple input and output formats, saving confidence scores, and model performance evaluation
Compare
Choose a tag to compare
@qiuqiao qiuqiao released this 03 Mar 14:55
· 28 commits to main since this release
v1.0.2
51a869d

The v1.0.2 version is compatible with models from v1.0.

New Features

  • Saving Confidence Scores: During inference, specify the --save_confidence parameter to save the confidence score for each sample.
  • Model Evaluation: A new evaluate.py file has been added for evaluating model performance. See readme.md for usage instructions.

Improved Features

  • Increased Support for Output Formats During Inference: Specify the --out_formats parameter during inference to freely choose the output format. Refer to the readme for usage instructions.
  • Customize the Input File Extension for Inference: Specify the --in_format parameter during inference to choose the file extension you want to use.

Bug Fixes

  • Fixed many bugs.

v1.0.2 beta：多种推理时的输入输出格式，保存置信度，模型性能评估

v1.0.2版本兼容v1.0版本的模型。

新功能

  • 保存置信度： 推理时，指定--save_confidence参数即可保存每一条样本的置信度。
  • 模型性能评估： 新增evaluate.py文件，用于评估模型性能，使用方法参考readme。

改进功能

  • 增加推理时的输出格式支持： 推理时指定--out_formats参数即可自由选择输出格式，使用方法参见readme。
  • 自定义推理时输入的文件后缀名： 推理时指定--in_format参数，即可指定要使用的文件后缀名。

bug修复

  • 修复了许多bug
Assets 2
Loading