foxhunt/LEGACY_256_CLEANUP_CHECKLIST.md

Legacy 256-Feature Test Cleanup Checklist

Start Date: __________ (Recommended: Week 2 after production deployment) Completion Target: __________ (Before ML retraining) Estimated Effort: 8-10 hours Priority: MEDIUM

---

Phase 1: Critical Feature Tests (7 hours)

Task 1: Update test_extract_256_dim_features.rs (2h)

• Create branch: git checkout -b cleanup/legacy-256-tests
• Open file: ml/tests/test_extract_256_dim_features.rs
• Line 40-48: Change assertions 256 → 225
• Line 9: Rename test function test_extract_256_dim_features → test_extract_225_dim_features
• Line 1-4: Update file header documentation
• Run test: cargo test --test test_extract_256_dim_features
• Verify: All tests pass
• Commit: git commit -am "Update test_extract_256_dim_features.rs to 225 features"

Task 2: Update dbn_256_feature_validation.rs (3h)

• Open file: ml/tests/dbn_256_feature_validation.rs
• Line 1-21: Update file header (256 → 225)
• Line 172: Change loop bounds 0..256 → 0..225
• Line 230, 242, 268, 294: Update assertions (256 → 225)
• Line 313, 331, 359: Update feature count assertions
• Line 411: Update performance targets
• Run test: cargo test --test dbn_256_feature_validation
• Verify: All tests pass with real DBN data
• Commit: git commit -am "Update dbn_256_feature_validation.rs to 225 features"

Task 3: Update test_dbn_sequence_256_features.rs (2h)

• Open file: ml/tests/test_dbn_sequence_256_features.rs
• Line 1-4: Update file header (256 → 225)
• Line 38: Change default d_model 256 → 225
• Line 69, 98: Update shape documentation
• Line 237-259: Add multi-d_model test [128, 225, 256, 512]
• Run test: cargo test --test test_dbn_sequence_256_features
• Verify: Backward compatibility with d_model=256 still works
• Commit: git commit -am "Update test_dbn_sequence_256_features.rs to default 225, add multi-d_model test"

---

Phase 2: Documentation Update (1 hour, parallel)

Task 4: Global Comment Update

• Find all files with 256 feature references:

  grep -r "256.*feature" ml/tests/ --include="*.rs" -l > /tmp/docs_to_update.txt
  

• For each file in /tmp/docs_to_update.txt:
  • Open file
  • Search: 256.*feature|256.*dim|256-dimensional
  • Replace: 225 features|225-dim|225-dimensional (in comments only)
  • Skip: Model config files (keep d_model=256)
  • Skip: Non-feature uses (buffer sizes, layer dimensions)
• Files to update (24 total):
  • ml/tests/feature_cache_tests.rs
  • ml/tests/microstructure_tests.rs
  • ml/tests/alternative_bars_integration_test.rs
  • ml/tests/calibration_dataset_test.rs
  • ml/tests/dbn_feature_config_test.rs
  • ml/tests/e2e_ensemble_integration.rs
  • ml/tests/model_registry_tests.rs
  • ml/tests/model_validation_comprehensive.rs
  • ml/tests/ppo_gae_test.rs
  • ml/tests/ppo_training_pipeline_test.rs
  • ml/tests/test_feature_cache_service.rs
  • ml/tests/test_tft_cuda_layernorm.rs
  • ml/tests/tft_attention_gradient_flow.rs
  • ml/tests/tft_attention_int8_quantization_test.rs
  • ml/tests/tft_e2e_training.rs
  • ml/tests/tft_inference_latency_benchmark.rs
  • ml/tests/tft_int8_inference_integration_test.rs
  • ml/tests/tft_int8_memory_benchmark_test.rs
  • ml/tests/tft_varmap_checkpoint_test.rs
  • ml/tests/training_chaos_tests.rs
  • ml/tests/wave_d_normalization_integration_test.rs
  • ml/tests/mamba_test.rs
  • ml/tests/mamba2_hardware_aware_test.rs
  • ml/tests/ensemble_4_model_trainable_integration.rs
• Commit: git commit -am "Update documentation: 256 → 225 features in comments"

---

Phase 3: Validation (2 hours)

Task 5: Run Updated Test Suite

• Run all updated tests:

  cargo test --test test_extract_256_dim_features -- --nocapture
  cargo test --test dbn_256_feature_validation -- --nocapture
  cargo test --test test_dbn_sequence_256_features -- --nocapture
  

• Verify output:
  • All tests pass
  • Feature vectors have 225 dimensions
  • Real data validation succeeds (ES.FUT, NQ.FUT, 6E.FUT, ZN.FUT)
  • Backward compatibility with d_model=256 works

Task 6: Verify Feature Extraction with Real DBN Data

• Run full ML test suite:

  cargo test -p ml --features "test-utils" | grep -i "feature"
  

• Check results:
  • Test pass rate ≥99% (target: 2,062/2,074 or better)
  • No 256-dimension assertion failures
  • All feature extraction tests use 225 dimensions
  • Model config tests still pass (d_model=256 valid)

Task 7: Update Documentation

• Update WAVE_D_VALIDATION_COMPLETE.md:

  echo "## Legacy Test Cleanup ($(date +%Y-%m-%d))" >> WAVE_D_VALIDATION_COMPLETE.md
  echo "- Updated 3 critical test files (1,209 lines) to 225 features" >> WAVE_D_VALIDATION_COMPLETE.md
  echo "- Updated 24 documentation files with comment corrections" >> WAVE_D_VALIDATION_COMPLETE.md
  echo "- Test suite: 100% pass rate, backward compatible with d_model=256" >> WAVE_D_VALIDATION_COMPLETE.md
  

• Update CLAUDE.md if needed (add completion note)
• Update LEGACY_256_TEST_CLEANUP.md status to COMPLETE

---

Final Validation Checklist

• All 3 critical test files pass with 225 features
• Feature extraction produces 225-dimensional vectors
• DBN validation tests succeed for all symbols
• DbnSequenceLoader supports d_model=225 (default)
• Backward compatibility: d_model=256 still works
• Test suite still at 99%+ pass rate
• No regression in model config tests
• Documentation updated
• Git branch merged:

  git checkout main
  git merge cleanup/legacy-256-tests
  git branch -d cleanup/legacy-256-tests
  

---

Completion Sign-off

Completed by: __________ Completion date: __________ Test pass rate: __________ Issues encountered: __________ Notes: __________

---

Reference

Full Documentation: /home/jgrusewski/Work/foxhunt/LEGACY_256_TEST_CLEANUP.md Wave D Validation: /home/jgrusewski/Work/foxhunt/WAVE_D_VALIDATION_COMPLETE.md Test Status: 99.4% pass rate (2,062/2,074) before cleanup