e19ce1dd9b
ppo.rs update_gpu(): - Advantage normalization: ReductionKernels::stats() (5 scalars) + affine() - Value loss: sub() → sqr() → mean_all() (1 scalar readback) - Policy loss: sub → clamp → exp → mul → GPU min → neg → mean_all - Zero full-buffer downloads (was: 4 arrays × N elements) gpu_portfolio.rs: - New simulate_batch_gpu() returns CudaSlice (DtoD clone from kernel) - Zero CPU download for portfolio features/rewards/done cuda_pipeline/mod.rs: - build_state_tensor/build_batch_states: DtoD assembly - dtod_copy_into/dtod_copy_into_at_offset helpers - Zero memcpy_dtoh in state construction Co-Authored-By: Claude Opus 4.6 (1M context) <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;