| Model Type | | Instruction-tuned, Text generation, Assistant, Conversation |
|
| Use Cases |
| Areas: | | Research, Commercial applications |
|
| Applications: | | Text generation, Conversation assistance |
|
| Primary Use Cases: | | Creating conversational agents, Instruction-following tasks |
|
| Limitations: | | Hallucinations, Biases and toxicity, Repetition and verbosity |
|
|
| Additional Notes | | The model aims to generate accurate in-context responses; however, care should be taken regarding its well-documented limitations. |
|
| Supported Languages | | Portuguese (Native with all dialects) |
|
| Training Details |
| Data Sources: | | nicholasKluge/instruct-aira-dataset |
|
| Methodology: | | The model was trained with a dataset composed of prompts and completions generated synthetically by prompting already-tuned models (ChatGPT, Llama, Open-Assistant, etc). |
|
| Hardware Used: | |
| Model Architecture: | |
|
| Input Output |
| Input Format: | | String with special tokens |
|
| Accepted Modalities: | |
| Output Format: | |
| Performance Tips: | | Ensure repetition penalty, temperature, top_k, and top_p parameters are set to prevent repetitive or verbose outputs. |
|
|
| Release Notes |
| Version: | |
| Date: | |
| Notes: | | Initial release with instruction-tuned capabilities and text generation. |
|
|
|
