- Reduce CI GPU test datasets 16x for walltime reduction - Reduce early-stop epochs 50→10, add --test-threads=1 - Serialize all GPU lib tests to prevent cuBLAS init race - Align state_dim to 16 for BF16 tensor core HMMA dispatch - BF16 precision tolerance in ml-dqn tests - Enable branching DQN + tracing subscriber in smoke tests - Prevent min_replay_size > buffer_size deadlock in early-stop tests - Prevent AutoReplaySizer from breaking gradient collapse warmup - Replace racy tokio::spawn checkpoint counter with AtomicUsize - Set warmup_steps=0 and max_training_steps_per_epoch=300 in early-stop tests - RealDataLoader respects TEST_DATA_DIR for CI PVC layout - Add collapse_warmup_capacity to gpu_smoketest DQNConfig - Drain CUDA context between test binaries - Detached HEAD checkout prevents local branch corruption - GPU pipeline tests: fix BF16 dtype and rank-1 squeeze assertions - OOD input handling tests use use_gpu: true Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
205 lines
6.2 KiB
Rust
205 lines
6.2 KiB
Rust
#![allow(
|
|
clippy::assertions_on_constants,
|
|
clippy::assertions_on_result_states,
|
|
clippy::clone_on_copy,
|
|
clippy::decimal_literal_representation,
|
|
clippy::doc_markdown,
|
|
clippy::empty_line_after_doc_comments,
|
|
clippy::field_reassign_with_default,
|
|
clippy::get_unwrap,
|
|
clippy::identity_op,
|
|
clippy::inconsistent_digit_grouping,
|
|
clippy::indexing_slicing,
|
|
clippy::integer_division,
|
|
clippy::len_zero,
|
|
clippy::let_underscore_must_use,
|
|
clippy::manual_div_ceil,
|
|
clippy::manual_let_else,
|
|
clippy::manual_range_contains,
|
|
clippy::modulo_arithmetic,
|
|
clippy::needless_range_loop,
|
|
clippy::non_ascii_literal,
|
|
clippy::redundant_clone,
|
|
clippy::shadow_reuse,
|
|
clippy::shadow_same,
|
|
clippy::shadow_unrelated,
|
|
clippy::single_match_else,
|
|
clippy::str_to_string,
|
|
clippy::string_slice,
|
|
clippy::tests_outside_test_module,
|
|
clippy::too_many_lines,
|
|
clippy::unnecessary_wraps,
|
|
clippy::unseparated_literal_suffix,
|
|
clippy::use_debug,
|
|
clippy::useless_vec,
|
|
clippy::wildcard_enum_match_arm,
|
|
clippy::else_if_without_else,
|
|
clippy::expect_used,
|
|
clippy::missing_const_for_fn,
|
|
clippy::similar_names,
|
|
clippy::type_complexity,
|
|
clippy::collapsible_else_if,
|
|
clippy::doc_lazy_continuation,
|
|
clippy::items_after_test_module,
|
|
clippy::map_clone,
|
|
clippy::multiple_unsafe_ops_per_block,
|
|
clippy::unwrap_or_default,
|
|
clippy::assign_op_pattern,
|
|
clippy::needless_borrow,
|
|
clippy::println_empty_string,
|
|
clippy::unnecessary_cast,
|
|
clippy::used_underscore_binding,
|
|
clippy::create_dir,
|
|
clippy::implicit_saturating_sub,
|
|
clippy::exit,
|
|
clippy::expect_fun_call,
|
|
clippy::too_many_arguments,
|
|
clippy::unnecessary_map_or,
|
|
clippy::unwrap_used,
|
|
dead_code,
|
|
unused_imports,
|
|
unused_variables,
|
|
clippy::cloned_ref_to_slice_refs,
|
|
clippy::neg_multiply,
|
|
clippy::while_let_loop,
|
|
clippy::bool_assert_comparison,
|
|
clippy::excessive_precision,
|
|
clippy::trivially_copy_pass_by_ref,
|
|
clippy::op_ref,
|
|
clippy::redundant_closure,
|
|
clippy::unnecessary_lazy_evaluations,
|
|
clippy::if_then_some_else_none,
|
|
clippy::unnecessary_to_owned,
|
|
clippy::single_component_path_imports,
|
|
)]
|
|
/// Wave 16M: Test log size reduction from 2.8MB → <1MB
|
|
///
|
|
/// Validates that 1-epoch training produces <100KB of INFO-level logs
|
|
/// (scaling to <1MB for 10 epochs).
|
|
///
|
|
/// **Test Strategy**:
|
|
/// 1. Run 1-epoch training with RUST_LOG=info
|
|
/// 2. Capture log output to file
|
|
/// 3. Verify file size <100KB (10x scaling → <1MB for 10 epochs)
|
|
/// 4. Verify essential metrics still visible (epoch summary, trading stats)
|
|
|
|
use std::fs::File;
|
|
use std::io::Write;
|
|
use std::process::Command;
|
|
use tracing::info;
|
|
|
|
#[test]
|
|
fn test_log_size_under_1mb() {
|
|
// Create temp log file
|
|
let log_file = "/tmp/test_log_size.log";
|
|
|
|
// Run 1-epoch training with INFO-level logging
|
|
let output = Command::new("cargo")
|
|
.args(&[
|
|
"run",
|
|
"-p",
|
|
"ml",
|
|
"--example",
|
|
"train_baseline_rl",
|
|
"--release",
|
|
"--features",
|
|
"cuda",
|
|
"--",
|
|
"--model",
|
|
"dqn",
|
|
"--epochs",
|
|
"1",
|
|
])
|
|
.env("RUST_LOG", "info")
|
|
.output()
|
|
.expect("Failed to run training");
|
|
|
|
// Write output to file
|
|
let mut file = File::create(log_file).expect("Failed to create log file");
|
|
file.write_all(&output.stdout).expect("Failed to write stdout");
|
|
file.write_all(&output.stderr).expect("Failed to write stderr");
|
|
|
|
// Check log size
|
|
let metadata = std::fs::metadata(log_file).expect("Failed to read log file");
|
|
let size_bytes = metadata.len();
|
|
let size_kb = size_bytes as f64 / 1_024.0;
|
|
let size_mb = size_bytes as f64 / 1_048_576.0;
|
|
|
|
info!(size_kb, size_mb, "Log size");
|
|
|
|
// 1 epoch should produce <100KB (10 epochs → <1MB)
|
|
assert!(
|
|
size_kb < 100.0,
|
|
"1-epoch log should be <100KB, got {:.2}KB ({:.3}MB). 10-epoch projection: {:.2}MB",
|
|
size_kb,
|
|
size_mb,
|
|
size_mb * 10.0
|
|
);
|
|
|
|
// Verify essential metrics are still present (INFO level)
|
|
let log_content = std::fs::read_to_string(log_file).expect("Failed to read log file");
|
|
|
|
// Essential metrics that MUST be visible at INFO level
|
|
assert!(
|
|
log_content.contains("Epoch 1/1"),
|
|
"Missing epoch summary in INFO logs"
|
|
);
|
|
assert!(
|
|
log_content.contains("train_loss=") || log_content.contains("Training Stats"),
|
|
"Missing training loss in INFO logs"
|
|
);
|
|
assert!(
|
|
log_content.contains("Trading Stats") || log_content.contains("P&L"),
|
|
"Missing trading stats in INFO logs"
|
|
);
|
|
assert!(
|
|
log_content.contains("Action diversity") || log_content.contains("diversity="),
|
|
"Missing action diversity in INFO logs"
|
|
);
|
|
|
|
info!(size_kb, projected_10epoch_mb = size_mb * 10.0, "Test passed: log size within target");
|
|
}
|
|
|
|
#[test]
|
|
fn test_debug_logs_available() {
|
|
// Create temp log file
|
|
let log_file = "/tmp/test_log_debug.log";
|
|
|
|
// Run 1-epoch training with DEBUG-level logging
|
|
let output = Command::new("cargo")
|
|
.args(&[
|
|
"run",
|
|
"-p",
|
|
"ml",
|
|
"--example",
|
|
"train_baseline_rl",
|
|
"--release",
|
|
"--features",
|
|
"cuda",
|
|
"--",
|
|
"--model",
|
|
"dqn",
|
|
"--epochs",
|
|
"1",
|
|
])
|
|
.env("RUST_LOG", "debug")
|
|
.output()
|
|
.expect("Failed to run training");
|
|
|
|
// Write output to file
|
|
let mut file = File::create(log_file).expect("Failed to create log file");
|
|
file.write_all(&output.stdout).expect("Failed to write stdout");
|
|
file.write_all(&output.stderr).expect("Failed to write stderr");
|
|
|
|
// Verify DEBUG logs contain step-level details
|
|
let log_content = std::fs::read_to_string(log_file).expect("Failed to read log file");
|
|
|
|
// DEBUG-level details should be present
|
|
assert!(
|
|
log_content.contains("Q-values:") || log_content.contains("Diagnostics:"),
|
|
"Missing step-level diagnostics in DEBUG logs"
|
|
);
|
|
|
|
info!("DEBUG logs contain step-level details");
|
|
}
|