Sentiment Finetuned Phi-3

Developed by: Hypernap
License: apache-2.0
Finetuned from model : unsloth/phi-3.5-mini-instruct-bnb-4bit

This model is a fine-tuned version of the unsloth/phi-3.5-mini-instruct-bnb-4bit using a custom sentiment analysis dataset. It was trained with accelerated speed using Unsloth and Hugging Face's TRL library.

Model Details

This model is a fine-tuned version of the unsloth/phi-3.5-mini-instruct-bnb-4bit optimized for sentiment analysis tasks. The original Phi-3 model is a powerful language model, and this fine-tuned version further enhances its capabilities for tasks involving sentiment detection, classification and inference.

Intended Use:

This model is intended for use in tasks such as:

Sentiment Analysis: Classifying the sentiment of text as positive, negative, or neutral.
Customer Feedback Analysis: Analyzing reviews and feedback for sentiment.
Social Media Monitoring: Detecting the sentiment of posts and comments.
Text Classification: General text classification involving sentiment labels.
Opinion Mining: Understanding the sentiment within text data.

Training Details:

Fine-tuning Dataset: A custom sentiment dataset was used for fine-tuning. (Optional: If your dataset is public, you can include a link or a brief description here.)
Training Method: The model was fine-tuned using Unsloth and Hugging Face's TRL library, which provides optimized training for models.
Hardware: The model was trained using [Specify your hardware if you want].
Accelerated Training Using unsloth led to 2x faster training.

Model Evaluation:

(Optional) Provide links to evaluation metrics or example outputs if you have them available. You can include metrics like:
- Accuracy
- Precision, Recall and F1 scores
- Qualitative analysis of the outputs

Limitations:

The model's performance may vary on datasets significantly different from the training data.
It may struggle with sarcasm or nuanced expressions of sentiment.
The model is optimized for sentiment analysis tasks, it is not suitable as a generic language model.

Further Information:

If you have a repository where you keep your training code, datasets, or other relevant information, you can link it here.

Acknowledgements:

Unsloth for the optimized training library.
Hugging Face for the TRL library and model hosting.
[Optional] If you have used a specific dataset, give credit to the original creators.