| Model Card: AdemGPT | |
| 1. General Information | |
| Model Name: AdemGPT | |
| Description: AdemGPT is a pre-trained generative language model that seeks to generate coherent and relevant text based on a wide spectrum of linguistic tasks. | |
| 2. Authors and Affiliations | |
| Authors: [Trat80] | |
| Affiliations: [N/A] | |
| 3. Model Functionality | |
| Supported Tasks: Text generation, Answering questions, Text to text, etc. | |
| Supported Languages: Mainly Spanish. | |
| Examples of Use: Generation of summaries, creative writing, informal conversation, among others. | |
| 4. Dataset and Training | |
| Dataset Origin: Created from multiple sources of text in Spanish (books, online articles, conversations, etc.). | |
| Dataset Size: Contains millions of examples of text in Spanish. | |
| Training Procedures: The GPT-3 architecture was used and trained for several weeks in a high-performance environment. | |
| 5. Model Performance | |
| Evaluation Metrics: Text coherence, precision in questions and answers, language fluency, etc. | |
| Results: Achieved high scores on text generation tests and language processing tasks. | |
| 6. Ethical Considerations | |
| Bias Considerations: Efforts have been made to mitigate bias, but there may be some inherent biases in the training data. | |
| Privacy and Security: The model does not store user information and caution should be taken when using it with sensitive data. | |
| 7. Limitations of the Model | |
| Known Limitations: Cannot provide information in other languages and may have difficulty with very specialized or technical concepts. | |
| 8. License and Conditions of Use | |
| License: [cc-by-nc-sa4.0] | |
| Conditions of Use: The model is available for non-commercial and educational use. It is recommended to review the license terms. |