opelly/Stockfish
1
0
Fork 0
mirror of https://github.com/opelly27/Stockfish.git synced 2026-05-20 15:37:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

Skeleton for new evaluate learner

Browse Source
This commit is contained in:
Tomasz Sobczyk
2020-11-22 19:46:43 +01:00
committed by nodchip
parent 0d4b803b08
commit cc11375f6d
1 changed files with 49 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/nnue/evaluate_nnue_learner.cpp
+49 -20
View File
@@ -54,6 +54,12 @@ namespace Eval::NNUE {
const std::string& seed, const std::string& seed,
SynchronizedRegionLogger::Region& out) { SynchronizedRegionLogger::Region& out) {
#if defined (OPENBLAS_VERSION)
openblas_set_num_threads(1);
#elif defined (INTEL_MKL_VERSION)
mkl_set_num_threads(1);
#endif
out << "INFO (initialize_training): Initializing NN training for " out << "INFO (initialize_training): Initializing NN training for "
<< get_architecture_string() << std::endl; << get_architecture_string() << std::endl;
@@ -199,39 +205,62 @@ namespace Eval::NNUE {
bool collect_stats = verbose; bool collect_stats = verbose;
std::vector<double> abs_eval_diff_sum_local(thread_pool.size(), 0.0);
std::vector<double> abs_discrete_eval_sum_local(thread_pool.size(), 0.0);
std::vector<double> gradient_norm_local(thread_pool.size(), 0.0);
while (examples.size() >= batch_size) { while (examples.size() >= batch_size) {
std::vector<Example> batch(examples.end() - batch_size, examples.end()); std::vector<Example> batch(examples.end() - batch_size, examples.end());
examples.resize(examples.size() - batch_size); examples.resize(examples.size() - batch_size);
const auto network_output = trainer->propagate(thread_pool, batch); const auto network_output = trainer->step_start(thread_pool, batch);
std::vector<LearnFloatType> gradients(batch.size()); std::vector<LearnFloatType> gradients(batch.size());
for (std::size_t b = 0; b < batch.size(); ++b) {
const auto shallow = static_cast<Value>(round<std::int32_t>( thread_pool.for_each_index_chunk_with_workers(
batch[b].sign * network_output[b] * kPonanzaConstant)); std::size_t(0), batch.size(),
const auto discrete = batch[b].sign * batch[b].discrete_nn_eval; [&](Thread& th, std::size_t offset, std::size_t count) {
const auto& psv = batch[b].psv; const auto thread_id = th.thread_idx();
const double gradient =
batch[b].sign * calc_grad(shallow, (Value)psv.score, psv.game_result, psv.gamePly); trainer->propagate(th, offset, count);
gradients[b] = static_cast<LearnFloatType>(gradient * batch[b].weight);
for (std::size_t b = offset; b < offset + count; ++b) {
const auto shallow = static_cast<Value>(round<std::int32_t>(
batch[b].sign * network_output[b] * kPonanzaConstant));
const auto discrete = batch[b].sign * batch[b].discrete_nn_eval;
const auto& psv = batch[b].psv;
const double gradient =
batch[b].sign * calc_grad(shallow, (Value)psv.score, psv.game_result, psv.gamePly);
gradients[b] = static_cast<LearnFloatType>(gradient * batch[b].weight);
// The discrete eval will only be valid before first backpropagation, // The discrete eval will only be valid before first backpropagation,
// that is only for the first batch. // that is only for the first batch.
// Similarily we want only gradients from one batch. // Similarily we want only gradients from one batch.
if (collect_stats) if (collect_stats)
{ {
abs_eval_diff_sum += std::abs(discrete - shallow); abs_eval_diff_sum_local[thread_id] += std::abs(discrete - shallow);
abs_discrete_eval_sum += std::abs(discrete); abs_discrete_eval_sum_local[thread_id] += std::abs(discrete);
gradient_norm += std::abs(gradient); gradient_norm_local[thread_id] += std::abs(gradient);
}
}
trainer->backpropagate(th, gradients.data(), offset, count);
} }
} );
thread_pool.wait_for_workers_finished();
trainer->backpropagate(thread_pool, gradients.data(), learning_rate); trainer->step_end(thread_pool, learning_rate);
collect_stats = false; collect_stats = false;
} }
if (verbose)
{
abs_eval_diff_sum = std::accumulate(abs_eval_diff_sum_local.begin(), abs_eval_diff_sum_local.end(), 0.0);
abs_discrete_eval_sum = std::accumulate(abs_discrete_eval_sum_local.begin(), abs_discrete_eval_sum_local.end(), 0.0);
gradient_norm = std::accumulate(gradient_norm_local.begin(), gradient_norm_local.end(), 0.0);
}
if (verbose) { if (verbose) {
const double avg_abs_eval_diff = abs_eval_diff_sum / batch_size; const double avg_abs_eval_diff = abs_eval_diff_sum / batch_size;
const double avg_abs_discrete_eval = abs_discrete_eval_sum / batch_size; const double avg_abs_discrete_eval = abs_discrete_eval_sum / batch_size;