go-emotions-plus-other-datasets-fine-tuned-distilroberta

This model is a fine-tuned version of distilbert/distilroberta-base on the these datasets:

It achieves the following results on the evaluation set:

  • Loss: 0.0719
  • Micro Precision: 0.7358
  • Micro Recall: 0.5840
  • Micro F1: 0.6512
  • Macro Precision: 0.5957
  • Macro Recall: 0.4191
  • Macro F1: 0.4670
  • Weighted Precision: 0.7120
  • Weighted Recall: 0.5840
  • Weighted F1: 0.6286
  • Hamming Loss: 0.0266

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 3.0

Training results

Training Loss Epoch Step Validation Loss Micro Precision Micro Recall Micro F1 Macro Precision Macro Recall Macro F1 Weighted Precision Weighted Recall Weighted F1 Hamming Loss
No log 1.0 10377 0.0788 0.7494 0.4946 0.5959 0.5505 0.3191 0.3567 0.7217 0.4946 0.5559 0.0285
No log 2.0 20754 0.0723 0.7354 0.5782 0.6474 0.6452 0.3792 0.4259 0.7312 0.5782 0.6154 0.0268
No log 3.0 31131 0.0719 0.7358 0.5840 0.6512 0.5957 0.4191 0.4670 0.7120 0.5840 0.6286 0.0266

Test results

Threshold = 0.5

Emotion Precision Recall F1-Score Support
admiration 0.65 0.69 0.67 504
amusement 0.72 0.89 0.80 264
anger 0.78 0.70 0.74 1585
annoyance 0.51 0.11 0.18 320
approval 0.58 0.31 0.41 351
caring 0.50 0.27 0.35 135
confusion 0.51 0.33 0.40 153
curiosity 0.52 0.49 0.50 284
desire 0.46 0.25 0.33 83
disappointment 0.58 0.09 0.16 151
disapproval 0.51 0.28 0.36 267
disgust 0.72 0.64 0.68 1222
embarrassment 0.78 0.19 0.30 37
excitement 0.54 0.36 0.43 103
fear 0.78 0.75 0.77 787
gratitude 0.92 0.89 0.90 352
grief 0.00 0.00 0.00 6
joy 0.87 0.77 0.82 2298
love 0.72 0.60 0.65 1305
nervousness 0.00 0.00 0.00 23
optimism 0.71 0.56 0.63 1329
pride 0.00 0.00 0.00 16
realization 0.46 0.11 0.18 145
relief 0.59 0.08 0.14 160
remorse 0.56 0.77 0.65 56
sadness 0.78 0.67 0.72 2212
surprise 0.59 0.28 0.38 572
neutral 0.69 0.56 0.62 2668
Micro Avg 0.74 0.60 0.66 17388
Macro Avg 0.57 0.42 0.46 17388
Weighted Avg 0.72 0.60 0.65 17388
Samples Avg 0.64 0.61 0.61 17388

Framework versions

  • Transformers 4.47.0
  • Pytorch 2.3.1+cu121
  • Datasets 2.20.0
  • Tokenizers 0.21.0
Downloads last month
5
Safetensors
Model size
82.1M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for sangkm/go-emotions-plus-other-datasets-fine-tuned-distilroberta

Finetuned
(548)
this model

Dataset used to train sangkm/go-emotions-plus-other-datasets-fine-tuned-distilroberta