| Model Type | | text-generation, natural language to SQL conversion |
|
| Use Cases |
| Areas: | | Research, Commercial applications |
|
| Primary Use Cases: | | Natural language to SQL generation tasks |
|
| Limitations: | | Requires tuning on specific schemas for optimal performance. |
|
|
| Additional Notes | | Fine-tuning on a given schema enhances performance. |
|
| Supported Languages | | en (Primary language for model tasks) |
|
| Training Details |
| Data Sources: | | 20,000 human-curated questions based on 10 different schemas |
|
| Methodology: | | Fine-tuned on a base StarCoder model |
|
| Hardware Used: | | Tested on an A100 40GB GPU with bfloat16 weights |
|
|
| Input Output |
| Input Format: | | Formatted prompt with task and database schema. |
|
| Accepted Modalities: | |
| Output Format: | |
| Performance Tips: | | Using specific schema tuning enhances model performance. |
|
|