Here is a list of all files with brief descriptions:

data
prepare_dataset.py
docs
engine
src
book
polyglot_book.cpp	Missing description
polyglot_book.hpp	Polyglot opening book parser and probe utility
polyglot_keys.cpp	Missing description
polyglot_keys.hpp	Precomputed random keys for Polyglot Zobrist hashing
eval
BatchEvalContext.h	Missing description
EvalContextMoECache.cpp	Missing description
EvalContextMoECache.h	Thread-local evaluator for the Factorized Mixture of Experts (MoE) network
Evaluator.cpp	Missing description
Evaluator.h	Runtime evaluator manager
IEvaluator.h	Abstract interface for board evaluation algorithms
MoECacheModel.hpp	Factorized Mixture of Experts (MoE) neural network architecture and inference components
MoERouting.h	Mixture of Experts (MoE) routing logic
PersistentThreadPool.hpp
SimpleEvalContext.cpp	Missing description
SimpleEvalContext.h	Hand-crafted evaluation function for testing and fallback
WDLConverter.hpp	Converts neural network WDL outputs to centipawns
search
ISearch.h	Search interface and shared data structures
MCTS.cpp	Missing description
MCTS.h	Monte Carlo Tree Search implementation
MovePicker.cpp	Missing description
MovePicker.h	Move selection and ordering for the search algorithm
Negamax.cpp	Missing description
Negamax.h	Alpha-Beta search implementation with enhancements
SearchHeuristics.h	Search data structures
TablebaseFallback.cpp	Missing description
TablebaseFallback.h	Integration of Syzygy tablebases at the root search node
TimeManager.cpp
TimeManager.h	Sophisticated time allocation and management for the chess engine
TT.cpp	Missing description
TT.h	Transposition Table implementation using a 3+1 Cluster Architecture
tablebases
Tablebase.cpp	Missing description
Tablebase.h	Syzygy tablebase probing via Fathom
tools
benchmark_eval.cpp
compare_evals.cpp
eval_raw.cpp
uci
UCI.cpp	Missing description
UCI.h	Universal Chess Interface protocol implementation
UciOptions.cpp	Missing description
UciOptions.h	Management of UCI engine options
feature_api.cpp	WebAssembly API for feature extraction
main.cpp	Chess Engine Entry Point
Types.h	Common type definitions and constants for the chess engine
preprocessing
include
EvalNormalizer.hpp	Utilities for normalizing chess evaluations into win probabilities
ExpertRouter.hpp	Computes expert weights for the Residual MoE architecture based on position features
FactorizedFeatureExtractor.hpp	Feature extraction logic for generating packed factorized features
FeatureExtractor.hpp	Standard sparse/dense feature extraction logic for chess positions
WDLNormalizer.hpp	Conversion from raw evaluation strings to WDL (Win/Draw/Loss) probabilities
Writers.hpp	Utility classes for writing packed dataset features and labels to disk
src
FactorizedFeatureExtractor.cpp	Implementation of factorized feature extraction
FeatureExtractor.cpp	Implementation of standard feature extraction
main.cpp	Main entry point for the preprocessing pipeline
Writers.cpp	Implementation of dataset writing utilities
training
dataset.py
export.py
loss.py
model.py
train.py
utils.py