Hugging Face's logo

Buckets:
MichaHenh
/
cil-sentiment-analysis

Files
xet
200 MB
10 files
Updated about 1 month ago
Name
Size
.gitattributes1.58 kB
xet
README.md2.56 kB
xet
adapter_config.json1.12 kB
xet
adapter_model.safetensors116 MB
xet
decoder_config.json205 Bytes
xet
metrics.json1.28 kB
xet
noise_weighting_summary.json313 Bytes
xet
oof_diagnostics.csv74.9 MB
xet
training_args.bin5.52 kB
xet
val_error_dataframe.csv9.37 MB
xet
README.md

cil-noise-weight-q5-xlmr-large-seed1

This model is a fine-tuned version of xlm-roberta-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7867
  • Accuracy: 0.6623
  • Map Mae: 0.3763
  • Bayes Mae: 0.3746
  • Expected Score Mae: 0.4254

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: 0.00015
  • train_batch_size: 64
  • eval_batch_size: 1024
  • seed: 1
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 1
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy Map Mae Bayes Mae Expected Score Mae
1.1054 0.1411 500 0.8717 0.6246 0.4273 0.4203 0.4875
0.8149 0.2822 1000 0.8269 0.6487 0.3873 0.3873 0.4555
0.7660 0.4233 1500 0.8029 0.6539 0.3825 0.3839 0.4438
0.7476 0.5643 2000 0.7872 0.6571 0.3860 0.3809 0.4351
0.7361 0.7054 2500 0.7922 0.6606 0.3796 0.3771 0.4285
0.7281 0.8465 3000 0.7943 0.6615 0.3771 0.3757 0.4254
0.7269 0.9876 3500 0.7867 0.6622 0.3763 0.3747 0.4254
0.7269 1.0 3544 0.7867 0.6623 0.3763 0.3746 0.4254

Framework versions

  • PEFT 0.19.1
  • Transformers 5.8.1
  • Pytorch 2.11.0+cu128
  • Datasets 4.8.5
  • Tokenizers 0.22.2
Total size
200 MB
Files
10
Last updated
May 29
Pre-warmed CDN
US EU US EU

Contributors