metadata
license: apache-2.0
datasets:
- dair-ai/emotion
language:
- en
metrics:
- f1
- accuracy
base_model:
- distilbert-base-uncased
library_name: transformers
pipeline_tag: text-classification
tags:
- distilbert
- pytorch
- emotion
- trainer
widget:
- text: >-
Interview preparation, I hate talking about myself, one dull subject
matter!
- text: >-
I'm in such a happy mood today i feel almost delighted and i havent done
anything different today then i normally have it is wonderful
- text: >-
I had every intention of doing more gardening this morning while it was
still cool but i was just feeling so rotten
- text: >-
Wow! I'm really impressed that Ashley can speak 7 languages, whereas I
only speak one!
- text: >-
No one wants to win the wild card because you have to play the Cubs on the
road.
- text: >-
After Kylie had her heart broken by her ex-boyfriend, she felt so down and
blue. I tried to cheer her up, but she just wants to be sad for awhile.
- text: >-
Jamie was in a bar with his friends one night when he saw a beautiful
girl. He felt confident that night so he went to go talk to her.
distilbert-base-uncased-finetuned-emotion
This model is a fine-tuned variant of distilbert-base-uncased using the emotion dataset. The evaluation results demonstrate its performance:
- Loss: 0.1595
- Accuracy: 93.35%
- F1 Score: 93.35%
Hyperparameters
- learning_rate: 2e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- lr_scheduler_type: linear
- num_epochs: 2
Training results
Epoch | Training Loss | Validation Loss | Accuracy | F1 |
---|---|---|---|---|
1 | 0.1703 | 0.1709 | 0.9355 | 0.9361 |
2 | 0.1115 | 0.1595 | 0.9335 | 0.9335 |