HuggingFace镜像/Layout-finetuned-fr-model-50instances20-100epochs-5e-05lr
模型介绍文件和版本分析
下载使用量0

Layout-finetuned-fr-model-50instances20-100epochs-5e-05lr

该模型是 microsoft/layoutxlm-base 在一个未知数据集上的微调版本。 它在评估集上取得了以下结果：

  • 损失：0.0000

模型描述

需要更多信息

预期用途与局限性

需要更多信息

训练和评估数据

需要更多信息

训练过程

训练超参数

训练过程中使用了以下超参数：

  • 学习率：5e-05
  • 训练批次大小：4
  • 评估批次大小：8
  • 随机种子：42
  • 优化器：使用 OptimizerNames.ADAMW_TORCH，betas=(0.9, 0.999)，epsilon=1e-08，无额外优化器参数
  • 学习率调度器类型：reduce_lr_on_plateau
  • 学习率调度器预热比例：0.06
  • 训练轮数：100

训练结果

训练损失轮次步数验证损失
3.37070.7692100.8298
0.331.5385200.0024
0.00222.3077300.0003
0.08143.0769400.0002
0.00043.8462500.0001
0.00034.6154600.0001
0.00025.3846700.0001
0.00026.1538800.0001
0.00026.9231900.0001
0.00027.69231000.0001
0.00028.46151100.0001
0.00029.23081200.0001
0.000110.01300.0001
0.000110.76921400.0001
0.000111.53851500.0001
0.000112.30771600.0001
0.000113.07691700.0001
0.000113.84621800.0001
0.000114.61541900.0000
0.000115.38462000.0000
0.000116.15382100.0000
0.000116.92312200.0000
0.000117.69232300.0000
0.000118.46152400.0000
0.000119.23082500.0000
0.000120.02600.0000
0.000120.76922700.0000
0.000121.53852800.0000
0.000122.30772900.0000
0.000123.07693000.0000
0.000123.84623100.0000
0.000124.61543200.0000
0.000125.38463300.0000
0.000126.15383400.0000
0.000126.92313500.0000
0.000127.69233600.0000
0.000128.46153700.0000
0.000129.23083800.0000
0.000130.03900.0000
0.000130.76924000.0000
0.000131.53854100.0000
0.000132.30774200.0000
0.000133.07694300.0000
0.000133.84624400.0000
0.000134.61544500.0000
0.000135.38464600.0000
0.000136.15384700.0000
0.036.92314800.0000
0.037.69234900.0000
0.038.46155000.0000
0.039.23085100.0000
0.040.05200.0000
0.040.76925300.0000
0.041.53855400.0000
0.042.30775500.0000
0.043.07695600.0000
0.043.84625700.0000
0.044.61545800.0000
0.045.38465900.0000
0.046.15386000.0000
0.046.92316100.0000
0.047.69236200.0000
0.048.46156300.0000
0.049.23086400.0000
0.050.06500.0000
0.050.76926600.0000
0.051.53856700.0000
0.052.30776800.0000
0.053.07696900.0000
0.053.84627000.0000
0.054.61547100.0000
0.055.38467200.0000
0.056.15387300.0000
0.056.92317400.0000
0.057.69237500.0000
0.058.46157600.0000
0.059.23087700.0000
0.060.07800.0000
0.060.76927900.0000
0.061.53858000.0000
0.062.30778100.0000
0.063.07698200.0000
0.063.84628300.0000
0.064.61548400.0000
0.065.38468500.0000
0.066.15388600.0000
0.066.92318700.0000
0.067.69238800.0000
0.068.46158900.0000
0.069.23089000.0000
0.070.09100.0000
0.070.76929200.0000
0.071.53859300.0000
0.072.30779400.0000
0.073.07699500.0000
0.073.84629600.0000
0.074.61549700.0000
0.075.38469800.0000
0.076.15389900.0000
0.076.923110000.0000
0.077.692310100.0000
0.078.461510200.0000
0.079.230810300.0000
0.080.010400.0000
0.080.769210500.0000
0.081.538510600.0000
0.082.307710700.0000
0.083.076910800.0000
0.083.846210900.0000
0.084.615411000.0000
0.085.384611100.0000
0.086.153811200.0000
0.086.923111300.0000
0.087.692311400.0000
0.088.461511500.0000
0.089.230811600.0000
0.090.011700.0000
0.090.769211800.0000
0.091.538511900.0000
0.092.307712000.0000
0.093.076912100.0000
0.093.846212200.0000
0.094.615412300.0000
0.095.384612400.0000
0.096.153812500.0000
0.096.923112600.0000
0.097.692312700.0000
0.098.461512800.0000
0.099.230812900.0000
0.0100.013000.0000

框架版本

  • Transformers 4.48.0
  • Pytorch 2.4.1.post100
  • Datasets 3.2.0
  • Tokenizers 0.21.0