markrodrigo/Llama-3.1-8B-Instruct-Spatial-SQL-1.0

Model Information

This model, Llama-3.1-8B-Instruct-Spatial-SQL-1.0, is an 8B, narrow use case, text to spatial SQL, lightly fine-tuned model. In general, its primary use case is the Natural Language command adaptation of particular geographic spatial functions as normally defined in pure SQL. Data input should be a combination of an English prefix in the form of a question, and a coordinate prompt injection, likely from an active mapping system application coordinate list. Output is PostGIS spatial SQL.

There are four primary geographic functions released in version 1.0.

Model developer: Mark Rodrigo

Github: https://github.com/mprodrigo/spatialsql

Docker: https://hub.docker.com/r/kartoza/postgis

Model Context Protocol (MCP): https://github.com/modelcontextprotocol/servers/tree/main/src/postgres

Model Architecture: The model is a QLoRA / Supervised Fine Tuning (SFT)

Model Input / Output Overview:

Input: Text plus coordinate prompt injection. </br> Output: PostGIS spatial SQL </br> NOTE: Inputs and outputs are in meters and or geographic decimal degrees WGS 84 coordinates.

Example Prompt / Prompt File

<|beginoftext|><|startheaderid|>system<|endheaderid|> <p></p> You are a helpful assistant. You are an expert at PostGIS and Postgresql and SQL and psql. <p></p> <|eotid|><|startheaderid|>user<|endheader_id|>

\### Instruction: Write a PostGIS SQL statement for the following. <p></p>

\### Input:

<p></p> {input}

<p></p>

\### Response:

<|eotid|><|startheaderid|>assistant<|endheader_id|>

Examples

AREA <p></p>

\### Input: What is the area for the polygon? : 'Polygon ((-3.7515154 40.3855551, -3.7514972 40.3856581, -3.7507005 40.3855767, -3.7507167 40.3854722, -3.7515154 40.3855551))' <p></p>

\### Input: ¿Cuál es el área para el polígono? : 'Polygon ((-3.7515154 40.3855551, -3.7514972 40.3856581, -3.7507005 40.3855767, -3.7507167 40.3854722, -3.7515154 40.3855551))' <p></p>

\### Input: 多邊形的面積是多少？ : 'Polygon ((-3.7515154 40.3855551, -3.7514972 40.3856581, -3.7507005 40.3855767, -3.7507167 40.3854722, -3.7515154 40.3855551))' <p></p>

\### Response: SELECT ST_Area(geog) As area FROM (select 'Polygon ((-3.7515154 40.3855551, -3.7514972 40.3856581, -3.7507005 40.3855767, -3.7507167 40.3854722, -3.7515154 40.3855551))' :: geography geog) subquery; <p></p>

CENTROID <p></p>

\### Input: What is the centroid for the polygon? : 'Polygon ((-3.6934636 40.4808785, -3.6933352 40.4811486, -3.6930125 40.4810598, -3.693141 40.4807897, -3.6934636 40.4808785))' <p></p>

\### Input: ¿Cuál es el centroide del polígono? : 'Polygon ((-3.6934636 40.4808785, -3.6933352 40.4811486, -3.6930125 40.4810598, -3.693141 40.4807897, -3.6934636 40.4808785))' <p></p>

Input: 多邊形的質心是什麼？ : 'Polygon ((-3.6934636 40.4808785, -3.6933352 40.4811486, -3.6930125 40.4810598, -3.693141 40.4807897, -3.6934636 40.4808785))'

<p></p>

\### Response: SELECT STAsText(STCentroid(geog)) As centroid FROM (select 'Polygon ((-3.6934636 40.4808785, -3.6933352 40.4811486, -3.6930125 40.4810598, -3.693141 40.4807897, -3.6934636 40.4808785))' :: geography geog) subquery;

<p></p> BUFFER <p></p>

\### Input: What is the thousand meter buffer for the following point? : 'Point(-8.7522658 41.3862664)' <p></p>

\### Input: ¿Cuál es el buffer de mil metros para lo siguiente punto? : 'Point(-8.7522658 41.3862664)' <p></p>

\### Input: 以下點的千米緩衝區是多少？ : 'Point(-8.7522658 41.3862664)' <p></p>

\### Response: SELECT STAsText(STBuffer(geog, 1000)) as buffer FROM (select 'Point(-8.7522658 41.3862664)' :: geography geog) subquery;

<p></p> LENGTH <p></p>

\### Input: How long is the line? : 'LINESTRING (-3.6976693 40.4263178, -3.6986082 40.4258729)' <p></p>

\### Input: ¿Cuánto dura la línea? : 'LINESTRING (-3.6976693 40.4263178, -3.6986082 40.4258729)' <p></p>

\### Input: 隊伍有多長？ : 'LINESTRING (-3.6976693 40.4263178, -3.6986082 40.4258729)' <p></p>

\### Response: SELECT ST_Length(geog) As length FROM (select 'LINESTRING (-3.6976693 40.4263178, -3.6986082 40.4258729)' :: geography geog) subquery; <p></p>

A Few Known Question Variation Examples

<p></p> AREA <p></p> What is the area for the geometry? <p></p> What is the area for this polygon? <p></p> CENTROID <p></p> What is the centroid for the geometry? <p></p> What is the center point of the polygon? <p></p> BUFFER <p></p> What is the 100 meter buffer for the following point? <p></p> Buffer the following point a thousand meters. <p></p> What is the 1000 meter buffer for the following point? <p></p> LENGTH <p></p> What is the length of the line? <p></p> How long is this line?

llama.cpp / Hyperparameter Recommendations For Inference

max context ~ 8,000 or lower <p></p> top k ~ 100 <p></p> temp ~ .4-.5 or lower

Agent Considerations

Agents are being considered as a separate project. Agents would mostly be related to pulling the coordinates from a mapping UI, and executing the SQL from responses against a PostGIS database.

Further Reference - link this

https://postgis.net/docs/manual-3.3/PostGISSpecialFunctionsIndex.html#PostGISGeographyFunctions

Evaluation data

More information needed

Training data

Custom synthetic

Training hyperparameters

The following hyperparameters were used during training:

• —learning_rate: 2e-04
• —trainbatchsize: 10
• —evalbatchsize: 3
• —distributed_type: multi-GPU
• —num_devices: 2
• —optimizer: Adam 8bit
• —lrschedulertype: linear
• —num_epochs: 3

Training results

Framework versions

• —Transformers 4.44.0
• —Pytorch 2.4.0
• —peft 0.12.0
• —Datasets 2.21.0
• —Tokenizers 0.19.1