CoolFace
Modelpublic

davanstrien/clip-roberta-finetuned

sourceHugging Faceupdated 4y agoView on Hugging Face
0likes649downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

clip-roberta-finetuned

This model is a fine-tuned version of ./clip-roberta on the davanstrien/manuscriptnoisylabels_iiif dataset. It achieves the following results on the evaluation set:

  • —Loss: 2.5792

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 5e-05
  • —trainbatchsize: 128
  • —evalbatchsize: 256
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —num_epochs: 10.0
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation Loss
2.98410.075003.4112
2.720.1510003.3430
2.63190.2215003.2295
2.57810.2920003.1645
2.53390.3625003.1226
2.5030.4430003.0856
2.45810.5135003.0639
2.44940.5840003.0415
2.42750.6545003.0245
2.39090.7350002.9991
2.39020.855002.9931
2.37410.8760002.9612
2.35360.9565002.9509
2.33921.0270002.9289
2.30831.0975002.9214
2.30941.1680002.9153
2.28641.2485002.9034
2.28931.3190002.8963
2.26971.3895002.8847
2.27621.46100002.8665
2.26671.53105002.8536
2.25481.6110002.8472
2.2381.67115002.8491
2.24231.75120002.8257
2.24061.82125002.8287
2.22481.89130002.8193
2.2231.96135002.8101
2.19952.04140002.8027
2.18342.11145002.7880
2.17232.18150002.7783
2.16512.26155002.7739
2.15752.33160002.7825
2.15982.4165002.7660
2.16672.47170002.7578
2.15652.55175002.7580
2.15582.62180002.7561
2.16422.69185002.7512
2.13742.77190002.7361
2.14022.84195002.7385
2.13262.91200002.7235
2.12722.98205002.7183
2.09543.06210002.7156
2.08423.13215002.7065
2.08593.2220002.7089
2.08563.27225002.6962
2.07753.35230002.6931
2.08213.42235002.6933
2.07063.49240002.7011
2.06893.57245002.7009
2.08073.64250002.6825
2.06393.71255002.6744
2.07423.78260002.6777
2.07893.86265002.6689
2.05943.93270002.6566
2.0564.0275002.6676
2.02234.08280002.6711
2.01854.15285002.6568
2.0184.22290002.6567
2.00364.29295002.6545
2.02384.37300002.6559
2.00914.44305002.6450
2.00964.51310002.6389
2.00834.58315002.6401
2.00124.66320002.6399
2.01664.73325002.6289
1.99634.8330002.6348
1.99434.88335002.6240
2.00994.95340002.6190
1.98955.02345002.6308
1.95815.09350002.6385
1.95025.17355002.6237
1.94855.24360002.6248
1.96435.31365002.6279
1.95355.38370002.6185
1.95755.46375002.6146
1.94755.53380002.6093
1.94345.6385002.6090
1.9545.68390002.6027
1.95095.75395002.6107
1.94545.82400002.5980
1.94795.89405002.6016
1.95395.97410002.5971
1.91196.04415002.6228
1.89746.11420002.6169
1.90386.19425002.6027
1.90086.26430002.6027
1.91426.33435002.6011
1.87836.4440002.5960
1.88966.48445002.6111
1.89756.55450002.5889
1.90486.62455002.6007
1.90496.69460002.5972
1.89696.77465002.6053
1.91056.84470002.5893
1.89216.91475002.5883
1.89186.99480002.5792
1.86717.06485002.6041
1.85517.13490002.6070
1.85557.2495002.6148
1.85437.28500002.6077
1.84857.35505002.6131
1.84747.42510002.6039
1.84747.5515002.5973
1.84427.57520002.5946
1.83297.64525002.6069
1.85517.71530002.5923
1.84337.79535002.5922
1.8517.86540002.5993
1.83137.93545002.5960
1.82988.0550002.6058
1.81598.08555002.6286
1.8178.15560002.6348
1.80668.22565002.6411
1.79358.3570002.6338
1.8098.37575002.6290
1.8128.44580002.6258
1.798.51585002.6321
1.80468.59590002.6291
1.79758.66595002.6283
1.79688.73600002.6284
1.77798.81605002.6257
1.76648.88610002.6232
1.7928.95615002.6305
1.77259.02620002.6525
1.75639.1625002.6794
1.76069.17630002.6784
1.76669.24635002.6798
1.75519.31640002.6813
1.75789.39645002.6830
1.74839.46650002.6833
1.74319.53655002.6884
1.7439.61660002.6932
1.73959.68665002.6927
1.74739.75670002.6904
1.74139.82675002.6892
1.74379.9680002.6898
1.75469.97685002.6894

Framework versions

  • —Transformers 4.21.0.dev0
  • —Pytorch 1.12.0+cu102
  • —Datasets 2.3.2
  • —Tokenizers 0.12.1