hf: update Dockerfile & supervisord to point to q8_0.gguf model

docs: update modal docs to serve q8 instead of q4

finetune: update the params to run experiment on H200

streamlit: change the examples to be global

fix: filter only inland geometry for overture

chore: change logo to globe search

Randomize candidate dataset order

fix: reduce templates to country, region & county for mvp

Merge pull request #5 from developmentseed/host-on-modal

Hide misleading success banner

chore: Add logo, explain data & methodology

enh: Add tracking using plausible - tracks query & download button

enh: Add download button for geojson

enh: Run gazet on modal - llama server on GPU, API & UI on cpu

Merge pull request #3 from developmentseed/fix-mixed-queries

Merge pull request #4 from developmentseed/push-to-hf

fix: pull model from models/

fix: add prompts that work, remove llama cloud option

docs: Add deploy notes to hf

fix: register llama shared libs with ldconfig and LD_LIBRARY_PATH

fix: keep llama-server binary and backend .so files together

fix: correct llama-server binary path in Space Dockerfile

enh: Add HF spaces support with all-in-one Dockerfile

chore: Run eval on v3 dataset

chore: run v3 dataset creation

Lowercase subtype names, fix question type hints

Fix: No pairs are created for mixed queries

enh: Add templates to handle queries like subregion, region

prompt: Add notes on how to pick places from query better

fix: mixed templates are not generated because of stale parquet files

Tweak readme layout

Tweak readme layout

Add hf links to readme

Merge pull request #2 from developmentseed/fix-geom-handling-for-natural-earth

Fix natural earth geometries by converting wkb geometries to geojson

Add image to README for project visualization

Merge pull request #1 from developmentseed/slm-qwen3.5

enh: Add docker compose file to host llama server, fastapi app & streamlit together

chore: clean up the file references, remove unused scripts for gemma finetuning

Update training to work with conversationn data format

Add a couple of more sql templates for natural earth & chained queries

chore: clean sql generation, use conversation format, move prompts from user to system

Add model training script for qwen, update eval & gazet lm to work with qwen template

fix: use gemma-4-e2b for finetuning but gemma-4-e2b-it as tokenizer

Add finetune instructions for gemma

Add options to run gguf models using llama cpp

Revert ollama port to 11434

Run llama cpp server with gguf models & infer using cli

Add scripts to train & infer model on modal

ENH: Add dataset creation on scale using modal clusters