|
--- |
|
library_name: transformers |
|
tags: [text-to-sql] |
|
--- |
|
|
|
# TinyLlama-7B (Fine-tuned for Text-to-SQL) |
|
|
|
TinyLlama-7B is a fine-tuned version of the Llama-2 7B model, specifically trained to handle Text-to-SQL tasks. It is designed to efficiently translate natural language queries into structured SQL queries, making it ideal for use in applications requiring database interactions from natural language instructions. With a smaller model size compared to larger variants, TinyLlama-7B offers a balance between performance and efficiency for environments with limited computational resources. |
|
|
|
## Model Details |
|
|
|
- **Model Name**: TinyLlama-7B (Fine-tuned for Text-to-SQL) |
|
- **Base Model**: Llama-2 7B |
|
- **Model Type**: Fine-tuned Transformer-based language model |
|
- **Parameter Size**: 7 billion parameters |
|
- **Fine-tuning Tasks**: Text-to-SQL, Natural Language to Structured Query Translation |
|
- **License**: Custom commercial license (please refer to original Llama-2 model license) |
|
|
|
## Intended Use |
|
|
|
### Use Cases: |
|
- **Text-to-SQL**: Translating natural language questions into executable SQL queries for database retrieval tasks. |
|
- **Database Management**: Assisting with creating, modifying, and querying databases using natural language. |
|
- **General NLP**: Handling basic language tasks like question answering, summarization, and text classification when combined with SQL-related tasks. |
|
|
|
### Out-of-scope Uses: |
|
- **Harmful Content Generation**: Generating biased or harmful content, or violating local laws. |
|
- **Languages Other Than English**: Primary focus is on English for database queries and SQL generation. |
|
- **Non-SQL Tasks**: Not intended for use in tasks outside of text-to-SQL or related language generation. |
|
|
|
## Model Performance |
|
|
|
TinyLlama-7B has been fine-tuned specifically for the task of converting natural language queries into SQL queries. The fine-tuning data included a diverse set of SQL query generation tasks, making the model capable of handling complex queries, joins, aggregations, and filtering operations. |
|
|
|
**Evaluation on Text-to-SQL Tasks**: |
|
- **High accuracy in translating natural language into SQL**: TinyLlama-7B performs well in generating accurate SQL queries from user inputs, even for complex requests involving multiple tables and conditions. |
|
- **Efficient and Fast**: The smaller size ensures lower latency and faster inference time compared to larger models, making it more suitable for environments with limited resources. |
|
|
|
## Training Data |
|
|
|
TinyLlama-7B was fine-tuned using a variety of publicly available datasets focused on SQL generation, including text-to-SQL datasets and query-generation tasks. The model was trained on data that includes diverse table schemas, SQL queries, and natural language questions to improve its performance on Text-to-SQL tasks. |
|
|
|
|
|
|
