├── .dockerignore ├── .github ├── actions │ └── setup-poetry-env │ │ └── action.yml └── workflows │ └── release.yml ├── .gitignore ├── .measurement_cache ├── token_pos_exp_1.pkl └── token_pos_tree_patch_2.pkl ├── .pre-commit-config.yaml ├── .prune_scores_cache ├── task-prune-scores-16-02-2024_22-22-43.pkl └── task-prune-scores-16-02-2024_23-27-49.pkl ├── .vscode ├── extensions.json └── settings.json ├── CITATION.cff ├── Dockerfile ├── README.md ├── Transformer Circuit Metrics are not Robust.pdf ├── auto_circuit ├── data.py ├── experiment_utils.py ├── metrics │ ├── area_under_curve.py │ ├── avoid_edges.py │ ├── completeness_metrics │ │ ├── same_under_knockouts.py │ │ └── train_same_under_knockouts.py │ ├── official_circuits │ │ ├── circuits │ │ │ ├── docstring_official.py │ │ │ ├── greaterthan_official.py │ │ │ ├── ioi_official.py │ │ │ ├── sports_players_official.py │ │ │ └── tracr │ │ │ │ ├── reverse_official.py │ │ │ │ └── xproportion_official.py │ │ ├── measure_roc.py │ │ └── roc_plot.py │ ├── prune_metrics │ │ ├── answer_diff.py │ │ ├── answer_diff_percent.py │ │ ├── answer_value.py │ │ ├── correct_answer_percent.py │ │ ├── kl_div.py │ │ ├── measure_prune_metrics.py │ │ ├── prune_metrics.py │ │ └── prune_metrics_plot.py │ └── prune_scores_similarity.py ├── model_utils │ ├── micro_model_utils.py │ ├── sparse_autoencoders │ │ ├── autoencoder_training.py │ │ ├── autoencoder_transformer.py │ │ └── sparse_autoencoder.py │ ├── task_projectors │ │ ├── projector_transformer.py │ │ ├── task_projector.py │ │ └── task_projector_training.py │ ├── tracr_model_utils.py │ └── transformer_lens_utils.py ├── prune.py ├── prune_algos │ ├── ACDC.py │ ├── activation_magnitude.py │ ├── circuit_probing.py │ ├── edge_attribution_patching.py │ ├── ground_truth.py │ ├── mask_gradient.py │ ├── parameter_integrated_gradients.py │ ├── prune_algos.py │ ├── random_edges.py │ └── subnetwork_probing.py ├── tasks.py ├── types.py ├── utils │ ├── ablation_activations.py │ ├── custom_tqdm.py │ ├── graph_utils.py │ ├── misc.py │ ├── patch_wrapper.py │ ├── patchable_model.py │ └── tensor_ops.py └── visualize.py ├── datasets ├── animal_diet_generator.py ├── animal_diet_long_prompts.json ├── animal_diet_short_prompts.json ├── capital_cities_generator.py ├── capital_cities_pythia-70m-deduped_prompts.json ├── docstring_generator.py ├── docstring_prompts.json ├── greaterthan_generator.py ├── greaterthan_gpt2-small_prompts.json ├── ioi │ ├── ioi_ABBA_template_0_prompts.json │ ├── ioi_ABBA_template_10_prompts.json │ ├── ioi_ABBA_template_11_prompts.json │ ├── ioi_ABBA_template_12_prompts.json │ ├── ioi_ABBA_template_13_prompts.json │ ├── ioi_ABBA_template_14_prompts.json │ ├── ioi_ABBA_template_1_prompts.json │ ├── ioi_ABBA_template_2_prompts.json │ ├── ioi_ABBA_template_3_prompts.json │ ├── ioi_ABBA_template_4_prompts.json │ ├── ioi_ABBA_template_5_prompts.json │ ├── ioi_ABBA_template_6_prompts.json │ ├── ioi_ABBA_template_7_prompts.json │ ├── ioi_ABBA_template_8_prompts.json │ ├── ioi_ABBA_template_9_prompts.json │ ├── ioi_BABA_template_0_prompts.json │ ├── ioi_BABA_template_10_prompts.json │ ├── ioi_BABA_template_11_prompts.json │ ├── ioi_BABA_template_12_prompts.json │ ├── ioi_BABA_template_13_prompts.json │ ├── ioi_BABA_template_14_prompts.json │ ├── ioi_BABA_template_1_prompts.json │ ├── ioi_BABA_template_2_prompts.json │ ├── ioi_BABA_template_3_prompts.json │ ├── ioi_BABA_template_4_prompts.json │ ├── ioi_BABA_template_5_prompts.json │ ├── ioi_BABA_template_6_prompts.json │ ├── ioi_BABA_template_7_prompts.json │ ├── ioi_BABA_template_8_prompts.json │ ├── ioi_BABA_template_9_prompts.json │ ├── ioi_generator.py │ ├── ioi_prompts.json │ ├── ioi_template_0_prompts.json │ ├── ioi_template_10_prompts.json │ ├── ioi_template_11_prompts.json │ ├── ioi_template_12_prompts.json │ ├── ioi_template_13_prompts.json │ ├── ioi_template_14_prompts.json │ ├── ioi_template_1_prompts.json │ ├── ioi_template_2_prompts.json │ ├── ioi_template_3_prompts.json │ ├── ioi_template_4_prompts.json │ ├── ioi_template_5_prompts.json │ ├── ioi_template_6_prompts.json │ ├── ioi_template_7_prompts.json │ ├── ioi_template_8_prompts.json │ ├── ioi_template_9_prompts.json │ └── ioi_vanilla_template_prompts.json ├── micro_model_input_generator.py ├── micro_model_inputs.json ├── micro_model_inputs_multiple_answers.json ├── mini_prompts.json ├── sports-players │ ├── american-football-players.txt │ ├── baseball-players.txt │ ├── basketball-players.txt │ ├── sports_players_generator.py │ ├── sports_players_pythia-2.8b-deduped_names.json │ ├── sports_players_pythia-2.8b-deduped_prompts.json │ └── sports_players_pythia-410m-deduped_prompts.json └── tracr │ ├── tracr_reverse_generator.py │ ├── tracr_reverse_len_5_prompts.json │ ├── tracr_xproportion_generator.py │ └── tracr_xproportion_len_5_prompts.json ├── docs ├── assets │ ├── Edge_Patching.png │ ├── Edge_Patching_Rounded.png │ ├── Factorized_Transformer.png │ ├── IOI_Attributions_Viz.png │ ├── IOI_Tokenwise_Viz.png │ ├── Node_Patching.png │ ├── Patching_Circuit.png │ ├── Patching_Complement.png │ ├── Residual_Transformer.png │ ├── Small_Circuit_Viz.png │ └── particles.json ├── gen_ref_pages.py ├── guides │ ├── 1)_Getting_Started.md │ ├── 2)_Circuit_Discovery.md │ ├── 3)_Other_Features.md │ ├── 4)_How_it_Works.md │ └── contributing.md ├── index.md ├── javascripts │ ├── SmallPRNG.js │ ├── dynamical.js │ ├── extra.js │ └── particles.min.js └── stylesheets │ └── extra.css ├── experiments ├── circuit_discovery.py ├── demos │ ├── circuit_discovery.py │ ├── other_features.py │ ├── patch_some_edges.py │ ├── token_specific_circuit_discovery.py │ └── zero_ablate_an_edge.py ├── docstring │ ├── docstring_faithfulness.py │ ├── docstring_logit_difference.py │ └── docstring_optimized.py ├── greaterthan │ └── greaterthan_faithfulness.py ├── ioi │ ├── edge_vs_tree_patch.py │ ├── ioi_edge_based_faithfulness.py │ ├── ioi_node_based_faithfulness.py │ ├── ioi_nodes_different_faithfulness_metrics.py │ ├── ioi_optimized_circuit.py │ └── learned_ioi_edges_all_toks.py ├── sports_players │ ├── sports_players_ablate_circuit.py │ ├── sports_players_full_circuit.py │ └── sports_players_probe.py └── tracr-experiments.py ├── mkdocs.yml ├── poetry.lock ├── poetry.toml ├── pyproject.toml └── tests ├── __init__.py ├── conftest.py ├── metrics ├── completeness_metrics │ └── test_same_under_knockouts.py └── prune_metrics │ ├── test_answer_diff.py │ ├── test_answer_diff_percent.py │ ├── test_answer_prob.py │ └── test_correct_answer_percent.py ├── model_utils ├── test_autoencoder_transformer.py ├── test_tracr_models.py └── test_unfactorized_edges.py ├── prune_algos ├── test_ACDC.py └── test_edge_attribution_patching.py ├── test_data.py ├── test_experiment_utils.py ├── test_prune.py ├── test_real_experiment.py └── utils ├── test_ablation_activations.py ├── test_graph_utils.py ├── test_instance_grads.py └── test_tensor_ops.py /.dockerignore: -------------------------------------------------------------------------------- 1 | # Ignore any file or directory that starts with . 2 | .* 3 | -------------------------------------------------------------------------------- /.github/actions/setup-poetry-env/action.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.github/actions/setup-poetry-env/action.yml -------------------------------------------------------------------------------- /.github/workflows/release.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.github/workflows/release.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.gitignore -------------------------------------------------------------------------------- /.measurement_cache/token_pos_exp_1.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.measurement_cache/token_pos_exp_1.pkl -------------------------------------------------------------------------------- /.measurement_cache/token_pos_tree_patch_2.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.measurement_cache/token_pos_tree_patch_2.pkl -------------------------------------------------------------------------------- /.pre-commit-config.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.pre-commit-config.yaml -------------------------------------------------------------------------------- /.prune_scores_cache/task-prune-scores-16-02-2024_22-22-43.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.prune_scores_cache/task-prune-scores-16-02-2024_22-22-43.pkl -------------------------------------------------------------------------------- /.prune_scores_cache/task-prune-scores-16-02-2024_23-27-49.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.prune_scores_cache/task-prune-scores-16-02-2024_23-27-49.pkl -------------------------------------------------------------------------------- /.vscode/extensions.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.vscode/extensions.json -------------------------------------------------------------------------------- /.vscode/settings.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/.vscode/settings.json -------------------------------------------------------------------------------- /CITATION.cff: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/CITATION.cff -------------------------------------------------------------------------------- /Dockerfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/Dockerfile -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/README.md -------------------------------------------------------------------------------- /Transformer Circuit Metrics are not Robust.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/Transformer Circuit Metrics are not Robust.pdf -------------------------------------------------------------------------------- /auto_circuit/data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/data.py -------------------------------------------------------------------------------- /auto_circuit/experiment_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/experiment_utils.py -------------------------------------------------------------------------------- /auto_circuit/metrics/area_under_curve.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/area_under_curve.py -------------------------------------------------------------------------------- /auto_circuit/metrics/avoid_edges.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/avoid_edges.py -------------------------------------------------------------------------------- /auto_circuit/metrics/completeness_metrics/same_under_knockouts.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/completeness_metrics/same_under_knockouts.py -------------------------------------------------------------------------------- /auto_circuit/metrics/completeness_metrics/train_same_under_knockouts.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/completeness_metrics/train_same_under_knockouts.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/docstring_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/docstring_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/greaterthan_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/greaterthan_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/ioi_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/ioi_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/sports_players_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/sports_players_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/tracr/reverse_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/tracr/reverse_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/circuits/tracr/xproportion_official.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/circuits/tracr/xproportion_official.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/measure_roc.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/measure_roc.py -------------------------------------------------------------------------------- /auto_circuit/metrics/official_circuits/roc_plot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/official_circuits/roc_plot.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/answer_diff.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/answer_diff.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/answer_diff_percent.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/answer_diff_percent.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/answer_value.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/answer_value.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/correct_answer_percent.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/correct_answer_percent.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/kl_div.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/kl_div.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/measure_prune_metrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/measure_prune_metrics.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/prune_metrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/prune_metrics.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_metrics/prune_metrics_plot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_metrics/prune_metrics_plot.py -------------------------------------------------------------------------------- /auto_circuit/metrics/prune_scores_similarity.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/metrics/prune_scores_similarity.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/micro_model_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/micro_model_utils.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/sparse_autoencoders/autoencoder_training.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/sparse_autoencoders/autoencoder_training.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/sparse_autoencoders/autoencoder_transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/sparse_autoencoders/autoencoder_transformer.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/sparse_autoencoders/sparse_autoencoder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/sparse_autoencoders/sparse_autoencoder.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/task_projectors/projector_transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/task_projectors/projector_transformer.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/task_projectors/task_projector.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/task_projectors/task_projector.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/task_projectors/task_projector_training.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/task_projectors/task_projector_training.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/tracr_model_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/tracr_model_utils.py -------------------------------------------------------------------------------- /auto_circuit/model_utils/transformer_lens_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/model_utils/transformer_lens_utils.py -------------------------------------------------------------------------------- /auto_circuit/prune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/ACDC.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/ACDC.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/activation_magnitude.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/activation_magnitude.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/circuit_probing.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/circuit_probing.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/edge_attribution_patching.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/edge_attribution_patching.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/ground_truth.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/ground_truth.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/mask_gradient.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/mask_gradient.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/parameter_integrated_gradients.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/parameter_integrated_gradients.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/prune_algos.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/prune_algos.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/random_edges.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/random_edges.py -------------------------------------------------------------------------------- /auto_circuit/prune_algos/subnetwork_probing.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/prune_algos/subnetwork_probing.py -------------------------------------------------------------------------------- /auto_circuit/tasks.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/tasks.py -------------------------------------------------------------------------------- /auto_circuit/types.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/types.py -------------------------------------------------------------------------------- /auto_circuit/utils/ablation_activations.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/ablation_activations.py -------------------------------------------------------------------------------- /auto_circuit/utils/custom_tqdm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/custom_tqdm.py -------------------------------------------------------------------------------- /auto_circuit/utils/graph_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/graph_utils.py -------------------------------------------------------------------------------- /auto_circuit/utils/misc.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/misc.py -------------------------------------------------------------------------------- /auto_circuit/utils/patch_wrapper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/patch_wrapper.py -------------------------------------------------------------------------------- /auto_circuit/utils/patchable_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/patchable_model.py -------------------------------------------------------------------------------- /auto_circuit/utils/tensor_ops.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/utils/tensor_ops.py -------------------------------------------------------------------------------- /auto_circuit/visualize.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/auto_circuit/visualize.py -------------------------------------------------------------------------------- /datasets/animal_diet_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/animal_diet_generator.py -------------------------------------------------------------------------------- /datasets/animal_diet_long_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/animal_diet_long_prompts.json -------------------------------------------------------------------------------- /datasets/animal_diet_short_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/animal_diet_short_prompts.json -------------------------------------------------------------------------------- /datasets/capital_cities_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/capital_cities_generator.py -------------------------------------------------------------------------------- /datasets/capital_cities_pythia-70m-deduped_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/capital_cities_pythia-70m-deduped_prompts.json -------------------------------------------------------------------------------- /datasets/docstring_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/docstring_generator.py -------------------------------------------------------------------------------- /datasets/docstring_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/docstring_prompts.json -------------------------------------------------------------------------------- /datasets/greaterthan_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/greaterthan_generator.py -------------------------------------------------------------------------------- /datasets/greaterthan_gpt2-small_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/greaterthan_gpt2-small_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_0_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_0_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_10_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_10_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_11_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_11_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_12_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_12_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_13_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_13_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_14_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_14_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_1_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_1_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_2_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_2_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_3_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_3_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_4_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_4_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_5_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_5_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_6_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_6_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_7_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_7_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_8_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_8_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_ABBA_template_9_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_ABBA_template_9_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_0_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_0_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_10_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_10_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_11_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_11_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_12_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_12_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_13_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_13_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_14_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_14_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_1_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_1_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_2_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_2_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_3_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_3_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_4_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_4_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_5_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_5_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_6_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_6_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_7_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_7_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_8_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_8_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_BABA_template_9_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_BABA_template_9_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_generator.py -------------------------------------------------------------------------------- /datasets/ioi/ioi_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_0_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_0_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_10_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_10_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_11_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_11_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_12_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_12_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_13_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_13_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_14_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_14_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_1_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_1_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_2_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_2_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_3_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_3_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_4_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_4_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_5_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_5_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_6_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_6_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_7_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_7_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_8_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_8_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_template_9_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_template_9_prompts.json -------------------------------------------------------------------------------- /datasets/ioi/ioi_vanilla_template_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/ioi/ioi_vanilla_template_prompts.json -------------------------------------------------------------------------------- /datasets/micro_model_input_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/micro_model_input_generator.py -------------------------------------------------------------------------------- /datasets/micro_model_inputs.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/micro_model_inputs.json -------------------------------------------------------------------------------- /datasets/micro_model_inputs_multiple_answers.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/micro_model_inputs_multiple_answers.json -------------------------------------------------------------------------------- /datasets/mini_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/mini_prompts.json -------------------------------------------------------------------------------- /datasets/sports-players/american-football-players.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/american-football-players.txt -------------------------------------------------------------------------------- /datasets/sports-players/baseball-players.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/baseball-players.txt -------------------------------------------------------------------------------- /datasets/sports-players/basketball-players.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/basketball-players.txt -------------------------------------------------------------------------------- /datasets/sports-players/sports_players_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/sports_players_generator.py -------------------------------------------------------------------------------- /datasets/sports-players/sports_players_pythia-2.8b-deduped_names.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/sports_players_pythia-2.8b-deduped_names.json -------------------------------------------------------------------------------- /datasets/sports-players/sports_players_pythia-2.8b-deduped_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/sports_players_pythia-2.8b-deduped_prompts.json -------------------------------------------------------------------------------- /datasets/sports-players/sports_players_pythia-410m-deduped_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/sports-players/sports_players_pythia-410m-deduped_prompts.json -------------------------------------------------------------------------------- /datasets/tracr/tracr_reverse_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/tracr/tracr_reverse_generator.py -------------------------------------------------------------------------------- /datasets/tracr/tracr_reverse_len_5_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/tracr/tracr_reverse_len_5_prompts.json -------------------------------------------------------------------------------- /datasets/tracr/tracr_xproportion_generator.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/tracr/tracr_xproportion_generator.py -------------------------------------------------------------------------------- /datasets/tracr/tracr_xproportion_len_5_prompts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/datasets/tracr/tracr_xproportion_len_5_prompts.json -------------------------------------------------------------------------------- /docs/assets/Edge_Patching.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Edge_Patching.png -------------------------------------------------------------------------------- /docs/assets/Edge_Patching_Rounded.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Edge_Patching_Rounded.png -------------------------------------------------------------------------------- /docs/assets/Factorized_Transformer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Factorized_Transformer.png -------------------------------------------------------------------------------- /docs/assets/IOI_Attributions_Viz.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/IOI_Attributions_Viz.png -------------------------------------------------------------------------------- /docs/assets/IOI_Tokenwise_Viz.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/IOI_Tokenwise_Viz.png -------------------------------------------------------------------------------- /docs/assets/Node_Patching.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Node_Patching.png -------------------------------------------------------------------------------- /docs/assets/Patching_Circuit.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Patching_Circuit.png -------------------------------------------------------------------------------- /docs/assets/Patching_Complement.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Patching_Complement.png -------------------------------------------------------------------------------- /docs/assets/Residual_Transformer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Residual_Transformer.png -------------------------------------------------------------------------------- /docs/assets/Small_Circuit_Viz.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/Small_Circuit_Viz.png -------------------------------------------------------------------------------- /docs/assets/particles.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/assets/particles.json -------------------------------------------------------------------------------- /docs/gen_ref_pages.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/gen_ref_pages.py -------------------------------------------------------------------------------- /docs/guides/1)_Getting_Started.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/guides/1)_Getting_Started.md -------------------------------------------------------------------------------- /docs/guides/2)_Circuit_Discovery.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/guides/2)_Circuit_Discovery.md -------------------------------------------------------------------------------- /docs/guides/3)_Other_Features.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/guides/3)_Other_Features.md -------------------------------------------------------------------------------- /docs/guides/4)_How_it_Works.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/guides/4)_How_it_Works.md -------------------------------------------------------------------------------- /docs/guides/contributing.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/guides/contributing.md -------------------------------------------------------------------------------- /docs/index.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/index.md -------------------------------------------------------------------------------- /docs/javascripts/SmallPRNG.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/javascripts/SmallPRNG.js -------------------------------------------------------------------------------- /docs/javascripts/dynamical.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/javascripts/dynamical.js -------------------------------------------------------------------------------- /docs/javascripts/extra.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/javascripts/extra.js -------------------------------------------------------------------------------- /docs/javascripts/particles.min.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/javascripts/particles.min.js -------------------------------------------------------------------------------- /docs/stylesheets/extra.css: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/docs/stylesheets/extra.css -------------------------------------------------------------------------------- /experiments/circuit_discovery.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/circuit_discovery.py -------------------------------------------------------------------------------- /experiments/demos/circuit_discovery.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/demos/circuit_discovery.py -------------------------------------------------------------------------------- /experiments/demos/other_features.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/demos/other_features.py -------------------------------------------------------------------------------- /experiments/demos/patch_some_edges.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/demos/patch_some_edges.py -------------------------------------------------------------------------------- /experiments/demos/token_specific_circuit_discovery.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/demos/token_specific_circuit_discovery.py -------------------------------------------------------------------------------- /experiments/demos/zero_ablate_an_edge.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/demos/zero_ablate_an_edge.py -------------------------------------------------------------------------------- /experiments/docstring/docstring_faithfulness.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/docstring/docstring_faithfulness.py -------------------------------------------------------------------------------- /experiments/docstring/docstring_logit_difference.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/docstring/docstring_logit_difference.py -------------------------------------------------------------------------------- /experiments/docstring/docstring_optimized.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/docstring/docstring_optimized.py -------------------------------------------------------------------------------- /experiments/greaterthan/greaterthan_faithfulness.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /experiments/ioi/edge_vs_tree_patch.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/edge_vs_tree_patch.py -------------------------------------------------------------------------------- /experiments/ioi/ioi_edge_based_faithfulness.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/ioi_edge_based_faithfulness.py -------------------------------------------------------------------------------- /experiments/ioi/ioi_node_based_faithfulness.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/ioi_node_based_faithfulness.py -------------------------------------------------------------------------------- /experiments/ioi/ioi_nodes_different_faithfulness_metrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/ioi_nodes_different_faithfulness_metrics.py -------------------------------------------------------------------------------- /experiments/ioi/ioi_optimized_circuit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/ioi_optimized_circuit.py -------------------------------------------------------------------------------- /experiments/ioi/learned_ioi_edges_all_toks.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/ioi/learned_ioi_edges_all_toks.py -------------------------------------------------------------------------------- /experiments/sports_players/sports_players_ablate_circuit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/sports_players/sports_players_ablate_circuit.py -------------------------------------------------------------------------------- /experiments/sports_players/sports_players_full_circuit.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/sports_players/sports_players_full_circuit.py -------------------------------------------------------------------------------- /experiments/sports_players/sports_players_probe.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/sports_players/sports_players_probe.py -------------------------------------------------------------------------------- /experiments/tracr-experiments.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/experiments/tracr-experiments.py -------------------------------------------------------------------------------- /mkdocs.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/mkdocs.yml -------------------------------------------------------------------------------- /poetry.lock: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/poetry.lock -------------------------------------------------------------------------------- /poetry.toml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/poetry.toml -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/pyproject.toml -------------------------------------------------------------------------------- /tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tests/conftest.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/conftest.py -------------------------------------------------------------------------------- /tests/metrics/completeness_metrics/test_same_under_knockouts.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/metrics/completeness_metrics/test_same_under_knockouts.py -------------------------------------------------------------------------------- /tests/metrics/prune_metrics/test_answer_diff.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/metrics/prune_metrics/test_answer_diff.py -------------------------------------------------------------------------------- /tests/metrics/prune_metrics/test_answer_diff_percent.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/metrics/prune_metrics/test_answer_diff_percent.py -------------------------------------------------------------------------------- /tests/metrics/prune_metrics/test_answer_prob.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/metrics/prune_metrics/test_answer_prob.py -------------------------------------------------------------------------------- /tests/metrics/prune_metrics/test_correct_answer_percent.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/metrics/prune_metrics/test_correct_answer_percent.py -------------------------------------------------------------------------------- /tests/model_utils/test_autoencoder_transformer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/model_utils/test_autoencoder_transformer.py -------------------------------------------------------------------------------- /tests/model_utils/test_tracr_models.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/model_utils/test_tracr_models.py -------------------------------------------------------------------------------- /tests/model_utils/test_unfactorized_edges.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/model_utils/test_unfactorized_edges.py -------------------------------------------------------------------------------- /tests/prune_algos/test_ACDC.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/prune_algos/test_ACDC.py -------------------------------------------------------------------------------- /tests/prune_algos/test_edge_attribution_patching.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/prune_algos/test_edge_attribution_patching.py -------------------------------------------------------------------------------- /tests/test_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/test_data.py -------------------------------------------------------------------------------- /tests/test_experiment_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/test_experiment_utils.py -------------------------------------------------------------------------------- /tests/test_prune.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/test_prune.py -------------------------------------------------------------------------------- /tests/test_real_experiment.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/test_real_experiment.py -------------------------------------------------------------------------------- /tests/utils/test_ablation_activations.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/utils/test_ablation_activations.py -------------------------------------------------------------------------------- /tests/utils/test_graph_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/utils/test_graph_utils.py -------------------------------------------------------------------------------- /tests/utils/test_instance_grads.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/utils/test_instance_grads.py -------------------------------------------------------------------------------- /tests/utils/test_tensor_ops.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/UFO-101/auto-circuit/HEAD/tests/utils/test_tensor_ops.py --------------------------------------------------------------------------------