distilbert-base-uncased_finetuned_on_emotions_data

This model is a fine-tuned version of distilbert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1561
• Accuracy: 0.933
• F1: 0.9328

Model description

his model is designed to analyze text and classify it into different emotional categories, such as joy, sadness, anger, etc. It has been trained on a dataset specifically labeled with emotions, allowing it to identify the emotional tone of the input text. The model works by processing the text and predicting which emotion best fits the given context

Intended uses & limitations

More information needed

limitations

• still this model is confused between fear and anger he model may confuse "fear" and "anger" because both emotions can be expressed in similar ways, especially in situations involving frustration, stress, or danger. Additionally, the language used to express these emotions might overlap, such as words like "nervous," "frustrated," or "threatened," which can be interpreted as either fear or anger depending on the context. This overlap in linguistic cues can make it challenging for the model to distinguish between the two emotions., joy & love
• similarely for love & Joy

Training and evaluation data

I've used emotion data available on huggingface Training data: emotion['train'] evaluation data: emotion['evaluation']

confusion matrix:

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: Use 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: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.7824	1.0	250	0.2717	0.9145	0.9149
0.2093	2.0	500	0.1788	0.93	0.9306
0.1379	3.0	750	0.1594	0.9345	0.9349
0.1106	4.0	1000	0.1561	0.933	0.9328

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0