├── .gitignore ├── CMAP.png ├── README.md ├── attn_knockout ├── attn_knockout.ipynb ├── figures │ ├── last_pos.pdf │ └── query_box_pos.pdf └── results │ ├── from_last_to_correct_obj_pos.json │ ├── from_last_to_query_box_pos.json │ ├── from_last_to_random_pos.json │ ├── from_query_box_to_prev_query_box.json │ └── from_query_box_to_random_pos.json ├── circuit_flow.png ├── data ├── data_utils.py ├── dataset.jsonl └── objects.csv ├── environment.yml ├── experiment_1 ├── circuit_eval.ipynb ├── completeness.ipynb ├── minimality.py ├── path_patching.py ├── pp_utils.py └── results │ ├── circuits │ ├── float_circuit.json │ ├── goat_circuit.json │ └── llama_circuit.json │ ├── figures │ ├── causal_impact.pdf │ ├── circuit_flow.pdf │ ├── float_cir_completeness.pdf │ ├── float_llama_causal_impact.pdf │ ├── goat_cir_completeness.pdf │ └── llama_cir_completeness.pdf │ ├── minimality │ ├── float_circuit │ │ ├── pos_detector.json │ │ ├── pos_detector_15_significance.json │ │ ├── pos_transmitter.json │ │ ├── pos_transmitter_9_significance.json │ │ ├── struct_reader.json │ │ ├── struct_reader_12_significance.json │ │ ├── value_fetcher.json │ │ └── value_fetcher_24_significance.json │ ├── goat_circuit │ │ ├── pos_detector.json │ │ ├── pos_detector_15_significance.json │ │ ├── pos_transmitter.json │ │ ├── pos_transmitter_9_significance.json │ │ ├── struct_reader.json │ │ ├── struct_reader_12_significance.json │ │ ├── value_fetcher.json │ │ └── value_fetcher_24_significance.json │ └── llama_circuit │ │ ├── pos_detector.json │ │ ├── pos_detector_8_significance.json │ │ ├── pos_transmitter.json │ │ ├── pos_transmitter_3_significance.json │ │ ├── struct_reader.json │ │ ├── struct_reader_2_significance.json │ │ ├── value_fetcher.json │ │ └── value_fetcher_15_significance.json │ └── path_patching │ ├── float_circuit │ ├── pos_detector.pt │ ├── pos_transmitter.pt │ ├── struct_reader.pt │ └── value_fetcher.pt │ ├── goat_circuit │ ├── pos_detector.pt │ ├── pos_transmitter.pt │ ├── struct_reader.pt │ └── value_fetcher.pt │ └── llama_circuit │ ├── pos_detector.pt │ ├── pos_transmitter.pt │ ├── struct_reader.pt │ └── value_fetcher.pt ├── experiment_2 ├── DCM.py ├── activation_patching.py ├── functionality_utils.py └── results │ ├── DCM │ ├── float_circuit │ │ ├── pos_detector │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ ├── pos_transmitter │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ ├── struct_reader │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ └── value_fetcher │ │ │ ├── box_label_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ ├── goat_circuit │ │ ├── pos_detector │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ ├── pos_transmitter │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ ├── struct_reader │ │ │ ├── box_label_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ │ ├── 0.01 │ │ │ │ └── 0.01.txt │ │ └── value_fetcher │ │ │ ├── box_label_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ │ ├── object_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ │ └── positional │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ └── llama_circuit │ │ ├── pos_detector │ │ ├── box_label_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── object_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ └── positional │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── pos_transmitter │ │ ├── add_box_before_correct_segment │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── add_comma │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── add_seg_end │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── add_seg_start │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── add_tokens_btw_box_and_obj │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── altered_box_obj_association │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── altered_box_obj_order │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── box_label_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── incorrect_segment │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── no_comma │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── object_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── positional │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── random_text_end │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ └── random_text_start │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── struct_reader │ │ ├── box_label_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ ├── object_value │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ └── positional │ │ │ ├── 0.01 │ │ │ └── 0.01.txt │ │ └── value_fetcher │ │ ├── box_label_value │ │ ├── 0.01 │ │ └── 0.01.txt │ │ ├── object_value │ │ ├── 0.01 │ │ └── 0.01.txt │ │ └── positional │ │ ├── 0.01 │ │ └── 0.01.txt │ └── activation_patching │ ├── float_circuit │ ├── float_semantic_results.json │ ├── goat_semantic_results.json │ ├── llama_semantic_results.json │ └── vicuna_semantic_results.json │ ├── goat_circuit │ ├── float_semantic_results.json │ ├── goat_semantic_results.json │ ├── llama_semantic_results.json │ └── vicuna_semantic_results.json │ └── llama_circuit │ ├── float_semantic_results.json │ ├── goat_semantic_results.json │ ├── llama_add_semantic_results.json │ ├── llama_semantic_results.json │ └── vicuna_semantic_results.json └── experiment_3 ├── cmap.ipynb └── cmap_utils.py /.gitignore: -------------------------------------------------------------------------------- 1 | __pycache__/ 2 | *.sh -------------------------------------------------------------------------------- /CMAP.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/CMAP.png -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/README.md -------------------------------------------------------------------------------- /attn_knockout/attn_knockout.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/attn_knockout.ipynb -------------------------------------------------------------------------------- /attn_knockout/figures/last_pos.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/figures/last_pos.pdf -------------------------------------------------------------------------------- /attn_knockout/figures/query_box_pos.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/figures/query_box_pos.pdf -------------------------------------------------------------------------------- /attn_knockout/results/from_last_to_correct_obj_pos.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/results/from_last_to_correct_obj_pos.json -------------------------------------------------------------------------------- /attn_knockout/results/from_last_to_query_box_pos.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/results/from_last_to_query_box_pos.json -------------------------------------------------------------------------------- /attn_knockout/results/from_last_to_random_pos.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/results/from_last_to_random_pos.json -------------------------------------------------------------------------------- /attn_knockout/results/from_query_box_to_prev_query_box.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/results/from_query_box_to_prev_query_box.json -------------------------------------------------------------------------------- /attn_knockout/results/from_query_box_to_random_pos.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/attn_knockout/results/from_query_box_to_random_pos.json -------------------------------------------------------------------------------- /circuit_flow.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/circuit_flow.png -------------------------------------------------------------------------------- /data/data_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/data/data_utils.py -------------------------------------------------------------------------------- /data/dataset.jsonl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/data/dataset.jsonl -------------------------------------------------------------------------------- /data/objects.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/data/objects.csv -------------------------------------------------------------------------------- /environment.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/environment.yml -------------------------------------------------------------------------------- /experiment_1/circuit_eval.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/circuit_eval.ipynb -------------------------------------------------------------------------------- /experiment_1/completeness.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/completeness.ipynb -------------------------------------------------------------------------------- /experiment_1/minimality.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/minimality.py -------------------------------------------------------------------------------- /experiment_1/path_patching.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/path_patching.py -------------------------------------------------------------------------------- /experiment_1/pp_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/pp_utils.py -------------------------------------------------------------------------------- /experiment_1/results/circuits/float_circuit.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/circuits/float_circuit.json -------------------------------------------------------------------------------- /experiment_1/results/circuits/goat_circuit.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/circuits/goat_circuit.json -------------------------------------------------------------------------------- /experiment_1/results/circuits/llama_circuit.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/circuits/llama_circuit.json -------------------------------------------------------------------------------- /experiment_1/results/figures/causal_impact.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/causal_impact.pdf -------------------------------------------------------------------------------- /experiment_1/results/figures/circuit_flow.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/circuit_flow.pdf -------------------------------------------------------------------------------- /experiment_1/results/figures/float_cir_completeness.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/float_cir_completeness.pdf -------------------------------------------------------------------------------- /experiment_1/results/figures/float_llama_causal_impact.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/float_llama_causal_impact.pdf -------------------------------------------------------------------------------- /experiment_1/results/figures/goat_cir_completeness.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/goat_cir_completeness.pdf -------------------------------------------------------------------------------- /experiment_1/results/figures/llama_cir_completeness.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/figures/llama_cir_completeness.pdf -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/pos_detector.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/pos_detector.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/pos_detector_15_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/pos_detector_15_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/pos_transmitter.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/pos_transmitter.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/pos_transmitter_9_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/pos_transmitter_9_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/struct_reader.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/struct_reader.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/struct_reader_12_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/struct_reader_12_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/value_fetcher.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/value_fetcher.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/float_circuit/value_fetcher_24_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/float_circuit/value_fetcher_24_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/pos_detector.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/pos_detector.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/pos_detector_15_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/pos_detector_15_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/pos_transmitter.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/pos_transmitter.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/pos_transmitter_9_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/pos_transmitter_9_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/struct_reader.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/struct_reader.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/struct_reader_12_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/struct_reader_12_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/value_fetcher.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/value_fetcher.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/goat_circuit/value_fetcher_24_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/goat_circuit/value_fetcher_24_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/pos_detector.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/pos_detector.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/pos_detector_8_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/pos_detector_8_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/pos_transmitter.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/pos_transmitter.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/pos_transmitter_3_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/pos_transmitter_3_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/struct_reader.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/struct_reader.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/struct_reader_2_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/struct_reader_2_significance.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/value_fetcher.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/value_fetcher.json -------------------------------------------------------------------------------- /experiment_1/results/minimality/llama_circuit/value_fetcher_15_significance.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/minimality/llama_circuit/value_fetcher_15_significance.json -------------------------------------------------------------------------------- /experiment_1/results/path_patching/float_circuit/pos_detector.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/float_circuit/pos_detector.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/float_circuit/pos_transmitter.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/float_circuit/pos_transmitter.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/float_circuit/struct_reader.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/float_circuit/struct_reader.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/float_circuit/value_fetcher.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/float_circuit/value_fetcher.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/goat_circuit/pos_detector.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/goat_circuit/pos_detector.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/goat_circuit/pos_transmitter.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/goat_circuit/pos_transmitter.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/goat_circuit/struct_reader.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/goat_circuit/struct_reader.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/goat_circuit/value_fetcher.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/goat_circuit/value_fetcher.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/llama_circuit/pos_detector.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/llama_circuit/pos_detector.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/llama_circuit/pos_transmitter.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/llama_circuit/pos_transmitter.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/llama_circuit/struct_reader.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/llama_circuit/struct_reader.pt -------------------------------------------------------------------------------- /experiment_1/results/path_patching/llama_circuit/value_fetcher.pt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_1/results/path_patching/llama_circuit/value_fetcher.pt -------------------------------------------------------------------------------- /experiment_2/DCM.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/DCM.py -------------------------------------------------------------------------------- /experiment_2/activation_patching.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/activation_patching.py -------------------------------------------------------------------------------- /experiment_2/functionality_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/functionality_utils.py -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_detector/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_detector/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/pos_transmitter/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/pos_transmitter/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/struct_reader/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/struct_reader/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/float_circuit/value_fetcher/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/float_circuit/value_fetcher/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_detector/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_detector/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/pos_transmitter/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/pos_transmitter/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/struct_reader/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/struct_reader/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/goat_circuit/value_fetcher/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/goat_circuit/value_fetcher/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_detector/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_detector/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_box_before_correct_segment/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_box_before_correct_segment/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_box_before_correct_segment/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_box_before_correct_segment/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_comma/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_comma/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_comma/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_comma/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_end/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_end/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_end/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_end/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_start/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_start/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_start/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_seg_start/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_tokens_btw_box_and_obj/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_tokens_btw_box_and_obj/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/add_tokens_btw_box_and_obj/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/add_tokens_btw_box_and_obj/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_association/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_association/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_association/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_association/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_order/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_order/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_order/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/altered_box_obj_order/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/incorrect_segment/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/incorrect_segment/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/incorrect_segment/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/incorrect_segment/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/no_comma/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/no_comma/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/no_comma/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/no_comma/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_end/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_end/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_end/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_end/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_start/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_start/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_start/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/pos_transmitter/random_text_start/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/struct_reader/positional/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/struct_reader/positional/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/box_label_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/value_fetcher/box_label_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/box_label_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/value_fetcher/box_label_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/object_value/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/value_fetcher/object_value/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/object_value/0.01.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/value_fetcher/object_value/0.01.txt -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/positional/0.01: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/DCM/llama_circuit/value_fetcher/positional/0.01 -------------------------------------------------------------------------------- /experiment_2/results/DCM/llama_circuit/value_fetcher/positional/0.01.txt: -------------------------------------------------------------------------------- 1 | Heads: [] 2 | Test Accuracy: 0.134 3 | -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/float_circuit/float_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/float_circuit/float_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/float_circuit/goat_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/float_circuit/goat_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/float_circuit/llama_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/float_circuit/llama_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/float_circuit/vicuna_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/float_circuit/vicuna_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/goat_circuit/float_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/goat_circuit/float_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/goat_circuit/goat_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/goat_circuit/goat_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/goat_circuit/llama_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/goat_circuit/llama_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/goat_circuit/vicuna_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/goat_circuit/vicuna_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/llama_circuit/float_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/llama_circuit/float_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/llama_circuit/goat_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/llama_circuit/goat_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/llama_circuit/llama_add_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/llama_circuit/llama_add_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/llama_circuit/llama_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/llama_circuit/llama_semantic_results.json -------------------------------------------------------------------------------- /experiment_2/results/activation_patching/llama_circuit/vicuna_semantic_results.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_2/results/activation_patching/llama_circuit/vicuna_semantic_results.json -------------------------------------------------------------------------------- /experiment_3/cmap.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_3/cmap.ipynb -------------------------------------------------------------------------------- /experiment_3/cmap_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Nix07/finetuning/HEAD/experiment_3/cmap_utils.py --------------------------------------------------------------------------------