---
license: apache-2.0
language:
- zh
metrics:
- accuracy
base_model:
- google-bert/bert-base-chinese
pipeline_tag: text-classification
library_name: transformers
---


```

评估结果:
--------------------------------------------------
  eval_loss.......................... 0.9872
  eval_accuracy...................... 0.7845
  eval_precision..................... 0.7861
  eval_recall........................ 0.7845
  eval_f1............................ 0.7851
  eval_1star_f1...................... 0.8589
  eval_2star_f1...................... 0.7269
  eval_3star_f1...................... 0.6344
  eval_4star_f1...................... 0.8488
  eval_5star_f1...................... 0.7884
  eval_runtime....................... 32.6454
  eval_samples_per_second............ 275.5980
  eval_steps_per_second.............. 4.3190
  epoch.............................. 6.0000

分类报告（每类详细指标）:
======================================================================
              precision    recall  f1-score   support

          1星     0.8533    0.8646    0.8589      1056
          2星     0.7451    0.7096    0.7269      1615
          3星     0.6194    0.6501    0.6344      1532
          4星     0.8570    0.8407    0.8488      3822
          5星     0.7655    0.8128    0.7884       972

    accuracy                         0.7845      8997
   macro avg     0.7681    0.7755    0.7715      8997
weighted avg     0.7861    0.7845    0.7851      8997


每类性能分析:
----------------------------------------------------------------------
类别       样本数        精确率        召回率        F1分数      
----------------------------------------------------------------------
1星     1056       0.8533     0.8646     0.8589    
2星     1615       0.7451     0.7096     0.7269    
3星     1532       0.6194     0.6501     0.6344    
4星     3822       0.8570     0.8407     0.8488    
5星     972        0.7655     0.8128     0.7884    

```