Graham McIntire 07558d17eb Two-phase training: pretrain on algorithm scores, fine-tune on QSOs ... - 15 features: add surface_refractivity and latitude - Bigger network: 128→64→32 (3 hidden layers) - Phase 1: pretrain on 500K stratified algorithm scores (all seasons/locations) - Phase 2: fine-tune on 57K real QSO-HRRR matched data (percentile target) - Lower LR (0.0003) for fine-tuning to preserve pretrained knowledge - Model.train accepts :initial_state option for transfer learning		2026-04-01 09:27:27 -05:00
..
commercial	Add commercial link monitoring via SNMP polling	2026-03-30 13:02:59 -05:00
propagation	Two-phase training: pretrain on algorithm scores, fine-tune on QSOs	2026-04-01 09:27:27 -05:00
radio	Add /submit route for user QSO submission	2026-03-29 16:46:35 -05:00
terrain	Implement ITU-R P.526-16 terrain diffraction model	2026-03-31 16:04:23 -05:00
weather	Exclude slow GRIB2 fixture tests by default (81s -> 16s)	2026-03-31 10:00:13 -05:00
workers	QSO submission triggers enrichment directly, fix prod badarith crash	2026-03-31 12:29:39 -05:00
commercial_test.exs	Add commercial link monitoring via SNMP polling	2026-03-30 13:02:59 -05:00
propagation_test.exs	Use local solar time for time-of-day scoring, add PWAT factor and pressure refinements	2026-04-01 08:58:21 -05:00
radio_test.exs	Multi-point path enrichment for HRRR, IEMRE, and weather data	2026-03-30 16:09:01 -05:00
terrain_test.exs	Add SRTM terrain path profiles for QSOs	2026-03-29 16:22:29 -05:00
weather_test.exs	Multi-point path enrichment for HRRR, IEMRE, and weather data	2026-03-30 16:09:01 -05:00