|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- BEE-spoke-data/fineweb-cryptid-5k |
|
|
--- |
|
|
|
|
|
New model tuning stratagy. Adding text to make this long enough. |
|
|
|
|
|
Model Description |
|
|
|
|
|
The Cryptid Detection Model is designed to generate and classify text related to cryptids, which are creatures from folklore and urban legends. The model is trained on a diverse dataset of cryptid-related content, including descriptions, stories, sightings, and various folklore sources. |
|
|
Intended Use |
|
|
|
|
|
Primary Use Case: Generating and classifying text about cryptids for entertainment, research, and educational purposes. |
|
|
Secondary Use Cases: Assisting in the creation of cryptid-related content for books, articles, and media. |
|
|
|
|
|
Input and Output |
|
|
|
|
|
Input: Text prompts or descriptions. |
|
|
Output: Generated text about cryptids or classifications of the input text as related to specific cryptids. |
|
|
|
|
|
Training Data |
|
|
|
|
|
The model was trained on a curated dataset of cryptid-related text, including but not limited to: |
|
|
|
|
|
Books and articles about cryptids. |
|
|
Online forums and discussion boards. |
|
|
Folklore databases. |
|
|
User-submitted stories and sightings. |
|
|
|
|
|
Data Preprocessing |
|
|
|
|
|
Text cleaning: Removal of special characters, HTML tags, and excessive whitespace. |
|
|
Tokenization: Breaking down text into tokens for training. |
|
|
|
|
|
Model Performance |
|
|
|
|
|
Metrics: [Accuracy, F1 Score, Precision, Recall, etc.] |
|
|
Evaluation: The model was evaluated on a validation set consisting of [describe the validation set]. |
|
|
|
|
|
Limitations and Biases |
|
|
|
|
|
Biases: The model may reflect biases present in the training data, such as regional biases in folklore or common myths. |
|
|
Limitations: The model may not accurately generate or classify less common or very specific cryptids. |
|
|
|
|
|
Ethical Considerations |
|
|
|
|
|
The model is intended for entertainment and educational purposes. It should not be used as a factual source for scientific research or investigation. |
|
|
Users should be aware of the potential for generating content that might be misinterpreted as factual. |
|
|
|
|
|
Future Work |
|
|
|
|
|
Expanding the training dataset to include more diverse sources. |
|
|
Improving classification accuracy for less common cryptids. |
|
|
Adding functionality for multilingual support. |
