0c385c38e2
GPU hotpath optimization: zero GPU→CPU roundtrips during training. Adds DSR/n-step config, branching DQN GPU action selection, warp experience kernel, GPU statistics, and epsilon-greedy kernel. Merge conflicts resolved in 3 files: - dqn.rs: kept clippy fixes + branch structure - gpu_experience_collector.rs: removed duplicate methods, kept hotpath version - trainer.rs: fixed brace mismatch from conflict resolution Applied clippy fixes to new hotpath files (wildcard_enum_match_arm, missing_debug_implementations). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
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;