Edit model card

distilbert-base-uncased-finetuned-emotions

This model is a fine-tuned version of distilbert-base-uncased on the emotion dataset.

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: 2e-05
  • train_batch_size: 64
  • eval_batch_size: 64
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 30

Training results

Epochs and loss:

Epoch Training Loss Validation Loss
1 0.761500 No log
2 0.193600 No log
3 0.132400 No log
4 0.101800 No log
5 0.085500 No log
6 0.068500 No log
7 0.055500 No log
8 0.042100 No log
9 0.034100 No log
10 0.027300 No log
11 0.022700 No log
12 0.020400 No log
13 0.015000 No log
14 0.015700 No log
15 0.012700 No log
16 0.014500 No log
17 0.012700 No log
18 0.011900 No log
19 0.009400 No log
20 0.009600 No log
21 0.008600 No log
22 0.008000 No log
23 0.007700 No log
24 0.007800 No log
25 0.006500 No log
26 0.007100 No log
27 0.005600 No log
28 0.004800 No log
29 0.005700 No log
30 0.004000 No log

10 highest loss:

label predicted_label loss text
882 love sadness 12.565614 i feel badly about reneging on my commitment t...
1919 fear sadness 11.828767 i should admit when consuming alcohol myself i...
1801 love sadness 10.995152 i feel that he was being overshadowed by the s...
1950 surprise sadness 10.986624 i as representative of everything thats wrong ...
415 love sadness 10.301595 im kind of embarrassed about feeling that way ...
1124 anger sadness 10.188601 someone acting stupid in public
929 anger joy 10.099646 i feel food smarter already and slightly annoy...
1392 anger sadness 9.992455 i still dont know how i feel i hated getting w...
1870 joy love 9.959218 i guess i feel betrayed because i admired him ...
60 love joy 9.808908 i miss our talks our cuddling our kissing and ...

10 lowest loss:

label predicted_label loss text
21 sadness sadness 0.000019 i feel try to tell me im ungrateful tell me im...
369 sadness sadness 0.000020 i just need a few minutes to feel put upon and...
1120 sadness sadness 0.000020 i am feeling a little disheartened
1466 sadness sadness 0.000020 i feel so ungrateful to be wishing this pregna...
625 sadness sadness 0.000020 i feel unwelcome in this town as if my time he...
650 sadness sadness 0.000020 i am still feeling gloomy and down
473 sadness sadness 0.000020 i have this mixed up kinda feeling and i reall...
133 sadness sadness 0.000020 i and feel quite ungrateful for it but i m loo...
368 sadness sadness 0.000020 i have to admit that i m feeling quite gloomy ...
1295 sadness sadness 0.000020 i feel a little damaged

Inference Example

# Download the model from the Hub
classifier = pipeline("text-classification", model="JakeClark/distilbert-base-uncased-finetuned-emotions")
custom_text = "When we say evil, we’re not exaggerating. It’s here."
preds = classifier(custom_text, return_all_scores=True)
# Plot the prediction
labels = ['sadness', 'joy', 'love', 'anger', 'fear', 'surprise']
preds_df = pd.DataFrame(preds[0])
plt.bar(labels, 100 * preds_df["score"], color='C0')
plt.title(f'"{custom_text}"')
plt.ylabel("Class probability (%)")
plt.show()

Framework versions

  • Transformers 4.42.3
  • Pytorch 2.1.2
  • Datasets 2.20.0
  • Tokenizers 0.19.1
Downloads last month
59
Safetensors
Model size
67M params
Tensor type
F32
·
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.

Finetuned from

Dataset used to train JakeClark/distilbert-base-uncased-finetuned-emotions