450c23a6d0
- Remove ALL #[cfg(feature = "cuda")] guards (~400+ occurrences) - Remove ALL #[cfg_attr(not(feature = "cuda"), ignore)] test annotations (~250) - Make cuda default feature in 9 ML crates (ml, ml-core, ml-dqn, ml-ppo, etc.) - Convert nvrtc JIT compilation to precompiled nvcc (searchsorted, prefix_sum) - Move compile_ptx_for_device() to ml-core for shared access - Delete dead CPU code: multi_step.rs, self_supervised_pretraining.rs, training_guard_gpu_tests.rs, CPU PER buffer paths, CPU Q-diagnostics - Replace unwrap_or(Device::Cpu) with hard errors everywhere - Remove dead is_cuda() else branches in DQN/PPO/hyperopt trainers - Change config defaults from "cpu" to "cuda" (rainbow, tlob, pipeline) - Port IQL value network to GPU kernel (5 CUDA entry points) - Port HER goal relabeling to GPU kernel (warp-per-sample) - Wire DSR GPU-to-CPU sync in training loop - cfg!(feature = "cuda") → true in inference_validator Zero warnings, zero errors across entire workspace. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ml_training_service
Model training orchestration and lifecycle management for DQN, PPO, TFT, Mamba2, TLOB, and Liquid models with progress tracking and artifact storage.
Key Types
MlTrainingServiceImpl-- main gRPC serviceJobTracker-- training job state machineCheckpointManager-- model artifact persistence
Features
minimal(default) -- minimal ML feature set for financial modelsgpu-- SIMD GPU acceleration (requires CUDA)mock-data-- mock training data (testing, bypasses database)
Configuration
GRPC_PORT-- gRPC listen portDATABASE_URL-- PostgreSQL for job metadata and training history- Prometheus metrics on port 9094
Testing
SQLX_OFFLINE=true cargo test -p ml_training_service --lib