2f8fa1ab19
7 root cause fixes for DQN hyperopt train/eval mismatch and reward corruption: - B1: Eval mode (noisy noise disabled, softmax action selection) - B3: Per-bar portfolio state sync in eval - C1: Extrinsic-only replay buffer (curiosity removed from rewards) - C2: Single exploration (noisy nets only, no epsilon/count bonus on Q-values) - C3: Neutral hold reward, search space 31D to 30D - C4: Sharpe-based early stopping (replaces val-loss plateau) 2720 tests, 0 failures, 0 clippy warnings.
ml
10-model ML ensemble for the Foxhunt HFT system, built on Candle v0.9.1.
Models
- DQN (Rainbow) — deep Q-network with prioritized replay, dueling heads, noisy nets
- PPO — proximal policy optimization with GAE, LSTM policies, clip-higher
- TFT — temporal fusion transformer for multi-horizon forecasting
- Mamba2 — state space model for sequence prediction
- Liquid Networks — biologically inspired networks for non-stationary data
- TLOB — transformer-based limit order book analysis
- KAN — Kolmogorov-Arnold networks
- xLSTM — extended LSTM architecture
- TGGN — temporal graph neural network
- Diffusion — diffusion-based generative model
Key Modules
ensemble— model ensemble coordination and confidence aggregationhyperopt— PSO-based hyperparameter optimization with per-model adapterstrainers— unified training loops (DQN, PPO, supervised)inference—InferenceAdaptertrait for predictioncheckpoint— model checkpointing and restorationevaluation— walk-forward evaluation pipeline
Usage
use ml::dqn::DQN;
use ml::ppo::PpoTrainer;