Konuktor92's picture
Add full training code (open source) for model v1
d7c810d verified
|
Raw History Blame
1.48 kB

TrendeX ML: training code (model v1)

Everything that produced TrendeX90/expected-greenness@v1 and TrendeX90/naryn-cube, in run order:

Step Script What it does
0 scripts/setup_box.sh, scripts/00_smoke.py environment; prove each NASA source opens
1 scripts/05_aoi.py Naryn oblast = 5 ADM2 districts + 20 km buffer, known-place boxes
2 scripts/10_download.py MODIS / NASADEM from LP DAAC, IMERG / GLDAS from GES DISC (earthaccess)
3 scripts/20_build_cube.py, scripts/25_qa_cube.py 1 km MODIS-grid cube + QA report
4 configs/model_v1.yaml protocol, committed in git before the first fit
5 scripts/30_train.py tuning, temporal + spatial tests, baselines, pre-registered gate, out-of-fold predictions
6 scripts/40_products.py residual trends (shared src/trendex_ml/stats/trend.py), placebo, season vs weather, Aqua-Terra
7 scripts/45_package.py, scripts/50_push_hf.py model repo with golden test; push to HF
8 scripts/55_scenarios.py, scripts/60_verify_hf.py precomputed what-if grid for the static site; verify HF from a clean machine

Run on one GPU box (RTX 4070, 40 cores) in about 3 hours including downloads; training itself took 7 minutes. requirements.lock is the exact environment. Credentials (Earthdata, HF) are read from `/.netrc` and the HF token cache; none are in this code. AI use: implementation was assisted by Claude (Anthropic).