AnimateDread afee4cc4c1 fix: lower WEIGHT_DECAY to 0.001 to prevent argmax degeneration to Neutral ...

Reduce weight decay from 0.01 to 0.001 across all backends (Network.cl, Network.mqh, WarriorCPU.cpp) to fix a training collapse issue. The original 0.01 AdamW default caused discriminative weights to decay below the calibration-capped class-prior offsets, resulting in a monotonically shrinking per-bar logit spread and eventual constant Neutral predictions (argmax degenerated once evidence tilt dropped under the prior tilt). The new value 0.001 lifts the evidence ceiling 10× while still bounding long-run weight growth, restoring effective discrimination. Note: this change must remain in sync across all four backends.

2026-07-19 17:05:58 -04:00

200 KiB

Raw Permalink History

View raw