70ba9341fa
- cublasGemmEx BF16×BF16→BF16 method (CUBLAS_COMPUTE_32F, tensor core path) - BF16 bias+relu and bias-only CUDA kernels (add_bias_relu_bf16_kernel) - 15 BF16 activation buffers allocated in CublasForward (online + target) - f32_to_bf16_kernel loaded for states conversion - bf16_weight_ptrs() helper for BF16 flat buffer offset computation - forward_online_bf16() method — complete BF16 forward pass (not yet wired) - cudarc f16 feature enabled, nvrtc removed 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;