Open
Conversation
…empty core eg.: class 2 never loses to class 1
…ire input is unSAT
- Remove unused sqlite3.Date import in main.py - Fix unreachable code in explainer.py (return before return 1) - Clean up ~350 lines of commented debug code across multiple files: - main.py: remove experimental config comments (297 → 16 lines) - experiment.py: remove commented input() statements - encoder.py: remove debug print statements - model.py: remove stale TODOs and debug code - dataset.py: remove commented Caltech101Dataset code - Reorganize README with proper structure: - Add table of contents - Add LGN Explanation Framework section with CLI reference - Document supported datasets, arguments, and usage examples - Consolidate installation notes - Remove internal notes and TODOs 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Replace repetitive switch-case patterns with DATASET_REGISTRY dict - Add LEGACY_DATASETS dict for image datasets without proper classes - Consolidate get_dataset(), input_dim_of_dataset(), num_classes_of_dataset() - Add docstrings explaining function behavior and requirements - Export registry constants from __init__.py for extensibility - Keep Caltech101Dataset placeholder for backwards compatibility 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Move mid-file imports to top in lgn/model/model.py - Remove unused imports: - logging from lgn/dataset/dataset.py - time from lgn/explanation/explainer.py - Dict, ModelArgs from experiment/experiment.py - dataclass from lgn/model/model.py - Extract magic numbers to named constants: - NUM_BINARY_OPS = 16 in lgn/model/model.py - TORCH_NUM_THREADS = 1 in experiment/experiment.py - Add explanatory comment for torch.set_num_threads() 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add comprehensive type annotations to improve code clarity and IDE support: - lgn/dataset/dataset.py: - Add TypedDict for LegacyDatasetInfo - Type DATASET_REGISTRY and LEGACY_DATASETS - Add return types to all functions - experiment/experiment.py: - Add types to Experiment class methods - Type debug(), run(), setup_preset_args(), etc. - difflogic/difflogic.py: - Add types to LogicLayer and GroupSum classes - Type forward methods with Union[Tensor, PackBitsTensor] - Add Optional for nullable parameters - lgn/model/model.py: - Type get_model() and compile_model() functions - lgn/encoding/encoding.py: - Add types to Encoding class constructor and methods - Type all getter methods with proper return types 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add SPLIT_DATASETS registry for datasets with explicit train/test splits - Add RANDOM_SPLIT_DATASETS set for datasets using random splitting - Add FACTORY_DATASETS for datasets needing factory functions (MNIST) - Extract common logic into helper functions: - _create_data_loaders() for DataLoader creation - _load_split_dataset() for datasets with splits - _load_random_split_dataset() for random split datasets - Add comprehensive type hints and docstrings - Add DEFAULT_TRAIN_RATIO constant (0.8) - Improve error messages with available dataset list - Reduce code duplication from 88 lines to cleaner 127 lines 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace 16 individual parameters with a typed configuration dataclass: - Type-safe Literal types for valid option values - Factory methods (small, full, custom) for common configurations - Backward compatible: legacy parameters still work - Clean conversion to experiment args via to_exp_args() 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Extract single-responsibility classes from Context: - MemoryProfiler: tracemalloc-based memory tracking - DeduplicationTracker: encoding deduplication statistics - SolvingStats: SAT solver clause/variable statistics - OutputFormatter: CSV and table output formatting Context maintains backward compatibility by delegating to these new classes while preserving the original API. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add comprehensive type annotations to: - experiment/helpers/util.py: seed_all, get_enc_type, loaders - experiment/helpers/sat_context.py: SatContext class - experiment/helpers/results.py: Results class with all methods - lgn/encoding/encoders/encoder.py: Encoder class - lgn/encoding/encoders/util.py: get_eq_constraints, _get_layers Includes docstrings for all public functions and classes. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Note that this repository is a modified version of the original difflogic repository, extended as part of the Master's thesis "Efficient Formal Explainability of Logic Gate Networks" at Monash University by Argens Ng, supervised by Alexey Ignatiev. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add Quick Start section with virtualenv setup instructions - Fix experiment commands (experiments/main.py -> main.py) - Remove references to non-existent main_baseline.py and --get_formula - Fix test instructions (cd experiments -> python -m pytest) - Add SSL certificate fix section for macOS users - Update main.py to support CLI arguments via Experiment.run_with_cmd() - Add missing datasets (compas, lending) to argparse choices - Remove difflogic from requirements.txt (local dev uses source) - Fix INSTALLATION_SUPPORT.md path reference 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove test_util.py (meaningless assertion) - Remove test_sat.py (only skipped/commented tests) - Remove test_session.py (empty file) - Convert test_bdd.py to modern pytest style: - Replace unittest.TestCase with pytest fixtures - Fix broken setup_class method - Remove unused imports - Add fixture docstrings 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
refactor: remove dead code and improve README documentation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.